Added an extention to the relation

- can compile simple stuff

exercises/03/vslc/src/tree.c

@@ -154,8 +154,8 @@ prune_children(node_t **simplified, node_t *root)
     switch (root->type)
     {
     case GLOBAL:
-    case ARGUMENT_LIST:
-    case PARAMETER_LIST:
+    //case ARGUMENT_LIST:
+    //case PARAMETER_LIST:
     case STATEMENT:
     case PRINT_ITEM:
     case PRINT_STATEMENT:
@@ -287,7 +287,6 @@ flatten(node_t **simplified, node_t *root)
     *simplified = result;
 }
 
-
 static void
 simplify_tree ( node_t **simplified, node_t *root )
 {
@@ -302,4 +301,4 @@ simplify_tree ( node_t **simplified, node_t *root )
     flatten(&root, root);
 
     *simplified = root;
-}
+}

exercises/04/vslc/include/ir.h

@@ -10,21 +10,42 @@ typedef struct n {
     struct n **children;
 } node_t;
 
-// Export the initializer function, it is needed by the parser
-void node_init (
-    node_t *nd, node_index_t type, void *data, uint64_t n_children, ...
+/**Export the initializer function, it is needed by the parser
+ * @param *nd node to initialize
+ * @param type type of node (see nodetype.h)
+ * @param *data associated data. Declared void to allow any type
+ * @param n_children number of children
+ * @param ... variable argument list of child nodes (node_t *)
+ * 
+ * @return Pointer to the initialized node
+ * */
+node_t* node_init (
+    node_t*         nd, 
+    node_index_t    type, 
+    void*           data, 
+    uint64_t        n_children,
+    ...
 );
 
 typedef enum {
-    SYM_GLOBAL_VAR, SYM_FUNCTION, SYM_PARAMETER, SYM_LOCAL_VAR
+    SYM_GLOBAL_VAR, 
+    SYM_FUNCTION, 
+    SYM_PARAMETER, 
+    SYM_LOCAL_VAR
 } symtype_t;
 
 typedef struct s {
-    char *name;
-    symtype_t type;
-    node_t *node;
-    size_t seq;
-    size_t nparms;
-    tlhash_t *locals;
+    char*       name;
+    symtype_t   type;
+    node_t*     node;
+    size_t      seq;
+    size_t      nparms;
+    tlhash_t*   locals;
 } symbol_t;
 #endif
+
+#define GLOBAL_BUCKET_SIZE          32
+#define LOCAL_BUCKET_SIZE           16
+
+#define DEFAULT_STRING_LIST_SIZE    8
+#define DEFAULT_NO_SCOPES           1

exercises/04/vslc/include/vslc.h

@@ -25,6 +25,7 @@ int yyerror ( const char *error );
 /* These are defined in the parser generated by bison */
 extern int yylineno;
 extern int yylex ( void );
+extern int yylex_destroy( void );
 extern char yytext[];
 
 /* Global state */

exercises/04/vslc/src/ir.c

@@ -1,76 +1,602 @@
 #include <vslc.h>
 
+#define ERRPRT(format, args...) {fprintf(stderr, "[ERROR] "); fprintf(stderr ,format, ##args);}
+
+
 // Externally visible, for the generator
 extern tlhash_t *global_names;
 extern char **string_list;
 extern size_t n_string_list, stringc;
 
-// Implementation choices, only relevant internally
-static void find_globals ( void );
-/** @param function Function's symbol table entry
- *  @param root Function's root node */
+// Functions from the skeleton
+
+static uint64_t find_globals ( void );
 static void bind_names ( symbol_t *function, node_t *root );
 
+// Helper functions, see description in the definition
+
+static void print_global_tree(symbol_t* global);
+static void print_string_list(void);
+static void destroy_global(symbol_t* global);
+static void push_scope(void);
+static void pop_scope(void);
+static void insert_symbol(tlhash_t *hash_table, symbol_t* symbol);
+static void insert_local_to_scope(symbol_t *local);
+static void insert_local_to_func(symbol_t *function, symbol_t *root);
+static void insert_local_var(symbol_t *function, node_t *root);
+static void collect_string(node_t *root);
+static symbol_t* lookup_var(symbol_t *function, char* var);
+
+// Local "global" variables
+static const char *symbol_names[4] = {
+    "GLOBAL_VAR", 
+    "FUNCTION", 
+    "PARAMETER", 
+    "LOCAL_VAR"
+};
+static uint64_t no_scopes, cur_scope_depth;
+static tlhash_t **scopes;
+
+
+/**
+ * Gather information and create a symbol table.
+ * 
+ * Used in vslc.c
+ */
 void
 create_symbol_table ( void )
 {
-    /* TODO: traverse the syntax tree and create the symbol table */
 
-    // ! Example code solely to demonstrate usage of tlhash. Make sure to remove
-    // ! or comment this out when implementing your solution.
+    // Initialize string array
+    n_string_list   = DEFAULT_STRING_LIST_SIZE;
+    string_list     = malloc(n_string_list * sizeof(char*));
+    stringc         = 0;
 
-    // Initialize table
-    tlhash_t *my_table = (tlhash_t*)malloc(sizeof(tlhash_t));
-    tlhash_init(my_table, 64);
+    // Initialize scope array
+    no_scopes       = DEFAULT_NO_SCOPES;
+    scopes          = malloc(no_scopes * sizeof(tlhash_t));
+    cur_scope_depth = 0;
 
-    char *my_key0 = "key";    // Keep in mind that these are stack allocated for simplicity, yours should not
-    char *my_val0 = "valuable"; // Also, I'm using strings as values, you will be using symbol_t pointers
-    char *my_key1 = "another_key";
-    char *my_val1 = "more valuable";
-    
-    // Insert some values to the table. Remember that the length of a string
-    // interpreted as an array is the string length plus one '\0' character
-    tlhash_insert(my_table, my_key0, strlen(my_key0)+1, my_val0);
-    tlhash_insert(my_table, my_key1, strlen(my_key1)+1, my_val1);
+    // Traverse the root node for globals
+    uint64_t no_globals = find_globals();
 
-    // Iterate keys and lookup their values
-    size_t size = tlhash_size(my_table);
-    char **keys = (char **)malloc(size);
-    tlhash_keys(my_table, keys);
-    char *val;
-    for (int i = 0; i < size; i++) {
-        tlhash_lookup(my_table, keys[i], strlen(keys[i])+1, &val);
-        printf("my_table[%s] = \"%s\"\n", keys[i], val);
+    // Prepare a temp list of globals and fetch all globals
+    symbol_t **global_list = malloc(no_globals * sizeof(symbol_t));
+    tlhash_values(global_names, (void **)global_list );
+
+    /* Iterate over the temporary list, bind names in each function */
+    for (uint64_t g = 0; g < no_globals; g++ )
+    {
+        if (global_list[g]->type == SYM_FUNCTION)
+            bind_names(global_list[g], global_list[g]->node);
     }
-
-    // Free allocated memory when done with the symbol table
-    tlhash_finalize(my_table);
-    free(my_table);
-    free(keys);
+    // Free the temp list
+    free(global_list);
 }
 
 
+/**
+ * Prints the symbol table and the string array
+ * 
+ * Used in vslc.c
+ */
 void
 print_symbol_table ( void )
 {
-    /* TODO: output its contents */
+    /* Get the number of symbols, size up a temporary list and fill it */
+    uint64_t no_globals = tlhash_size(global_names);
+    symbol_t **global_list = malloc(no_globals * sizeof(symbol_t));
+    tlhash_values(global_names, (void **)global_list );
+
+    /* Iterate over the temporary list, printing entries */
+    for (uint64_t g = 0; g < no_globals; g++ )
+        // Print the tree structure for each global
+        print_global_tree(global_list[g]);
+    
+    free(global_list);
+
+    // Print strings
+    print_string_list();
 }
 
+
+/**
+ * Prins the tree of a global
+ * 
+ * @param global pointer to the global to be printed
+ */
+static void 
+print_global_tree(symbol_t* global)
+{
+    // Check if null ptr
+    if (!global)
+        return;
+    
+    // Print global root
+    printf("─%s: %-16s [nparams=%2ld, seq=%2ld, node=%p]\n", 
+        symbol_names[global->type], 
+        global->name, 
+        global->nparms,
+        global->seq,
+        global->node
+    );
+    
+    // If the global does not have params or locals, return
+    if (!global->nparms && !global->locals)
+        {putchar('\n');return;}
+    
+    // Need to fetch the whole size, since nparams 
+    // only count the params, not all locals
+    uint64_t no_locals = tlhash_size(global->locals);
+    symbol_t **locals_list = malloc(no_locals * sizeof(symbol_t));
+    tlhash_values(global->locals, (void **)locals_list );
+    // Go through all locals
+    for (int l = 0; l < no_locals; l++)
+    {   // Do some simple sorting, so seq num is in order
+        for (int ll = 0; ll < no_locals; ll++)
+        {
+            if (locals_list[ll]->seq == l)
+            {
+                printf(" %s─[%s]: %-22s\t[seq=%2ld, node=%p]\n",
+                    (l < (no_locals - 1)) ? "├" : "└", 
+                    symbol_names[locals_list[ll]->type],
+                    locals_list[ll]->name,
+                    locals_list[ll]->seq,
+                    locals_list[ll]->node
+                );
+                break;
+            }
+        }
+    }
+    putchar('\n');
+    free(locals_list);
+}
+
+
+/**
+ * Prints the array of strings
+ * 
+ */
+static void
+print_string_list(void)
+{   // Print out all the collected strings
+    printf("─STRINGS [%ld]\n", stringc);
+    for (uint64_t i = 0; i < stringc; i++)
+        printf(" %s─[%ld]: %s\n", 
+            (i < (stringc - 1)) ? "├" : "└", 
+            i, 
+            string_list[i]
+        );
+}
+
+
+/**
+ * Destroys all the dynamicly allocated memory and all the hash tables.
+ * Frees up the array of strings as well.
+ * 
+ * Used in vslc.c
+ */
 void
 destroy_symbol_table ( void )
 {
-    /* TODO: release memory allocated to the symbol table */
+    // FREE STRINGS
+    // Free all strings that are kept in the array
+    for (uint64_t c = 0; c < stringc; c++)
+        free(string_list[c]);
+    // Free the actual list
+    free(string_list);
+
+    // FREE SCOPES
+    // At the end of program, all scopes have to be popped
+    // Therefore only free the list
+    free(scopes);
+
+    // FREE GLOBAL NAMES
+    if (!global_names)
+        return;
+
+    // Fetch list of globals
+    uint64_t no_globals = tlhash_size(global_names);
+    symbol_t **global_list = malloc(no_globals * sizeof(symbol_t));
+    tlhash_values(global_names, (void **)global_list );
+
+    // Destroy all global elements
+    for (uint64_t g = 0; g < no_globals; g++)
+        destroy_global(global_list[g]);
+    
+    // Destory the global hash table
+    tlhash_finalize(global_names);
+    // Free the global hash table
+    free(global_names);
+
+    // Free the temp list
+    free(global_list);
 }
 
-void
+
+/**
+ * Destroys the supplied global symbol by
+ * finalizing each of the local tables
+ * 
+ * @param global pointer to the global symbol to be destroyed
+ */
+static void 
+destroy_global(symbol_t* global)
+{
+    if (!global)
+        return;
+    if (!global->locals)
+    {
+        free(global);
+        return;
+    }
+    
+    uint64_t no_locals = tlhash_size(global->locals);
+    symbol_t **locals_list = malloc(no_locals * sizeof(symbol_t));
+    tlhash_values(global->locals, (void **)locals_list );
+
+    for (int l = 0; l < no_locals; l++)
+        free(locals_list[l]);
+    
+    tlhash_finalize(global->locals);
+    free(global->locals);
+    free(global);
+
+    free(locals_list);
+}
+
+
+/**
+ * Goes trough the root node and finds all global variables and functions
+ * 
+ * @return Returns the number of globals found (functions + variables)
+ */
+static uint64_t
 find_globals ( void )
 {
-    /* TODO: Populate symbol table with global variables and functions */
+    tlhash_init(global_names = malloc(sizeof(tlhash_t)), GLOBAL_BUCKET_SIZE);
+
+    uint64_t no_functions = 0, no_global_vars = 0;
+    node_t *global_list = root;
+
+    // Check if not nullptr
+    if (!global_list)
+        return 0;
+
+    symbol_t* global_symbol;
+    for (uint64_t global_i = 0; global_i < global_list->n_children; global_i++)
+    {
+        node_t *current_global = global_list->children[global_i];
+        switch (current_global->type)
+        {
+        case VARIABLE_LIST:
+
+            // Go through the variable list and get all the global variables
+            for (uint64_t var_i = 0; var_i < current_global->n_children; var_i++)
+            {
+                global_symbol = malloc(sizeof(symbol_t));
+                *global_symbol = (symbol_t){
+                    .type   = SYM_GLOBAL_VAR,
+                    .name   = current_global->children[var_i]->data,
+                    .node   = current_global->children[var_i],
+                    .seq    = 0,
+                    .nparms = 0,
+                    .locals = NULL
+                };
+                insert_symbol(global_names, global_symbol);
+                no_global_vars++;
+            }
+            break;
+        case FUNCTION:
+            node_t *function = current_global;
+
+            // Function node allways have the same structure,
+            // [0] are the identifier
+            // [1] are the variable list, within a paramerer_list
+            // [2] are the actual block
+            if (!function->children[0])
+                break;
+
+            // Create the function symbol
+            global_symbol = malloc(sizeof(symbol_t));
+            *global_symbol = (symbol_t){
+                .type   = SYM_FUNCTION,
+                .name   = current_global->children[0]->data,
+                .node   = current_global->children[2],
+                .seq    = no_functions++,
+                .nparms = 0,
+                .locals = malloc(sizeof(tlhash_t))
+            };
+
+            // Initialize the local variable table
+            tlhash_init(global_symbol->locals, LOCAL_BUCKET_SIZE);
+
+            // Insert the pointer to the newly created symbol
+            insert_symbol(global_names, global_symbol);
+
+            // If there are no parameters in function, break.
+            if (!current_global->children[1]->n_children)
+                break;
+
+
+            // Find all params and insert into hash table in global_symbol
+            symbol_t *param_sym;
+            node_t *param_list = current_global->children[1]->children[0];
+            global_symbol->nparms = param_list->n_children;
+
+            for (uint64_t param_i = 0; param_i < param_list->n_children; param_i++)
+            {
+                param_sym = malloc(sizeof(symbol_t));
+                *param_sym = (symbol_t){
+                    .type   = SYM_PARAMETER,
+                    .name   = param_list->children[param_i]->data,
+                    .node   = param_list->children[param_i],
+                    .seq    = param_i,
+                    .nparms = 0,
+                    .locals = NULL
+                };
+
+                insert_symbol(global_symbol->locals, param_sym);
+            }
+            break;
+        }
+    }
+    return no_functions + no_global_vars;
 }
 
 
+/**
+ * Inserts a symbol into a hash table, key is defined in the name field in the symbol supplied.
+ * 
+ * @param hash_table    pointer to the hash table the symbol is inserted into
+ * @param symbol        pointer to the symbol to be inserted
+ */
 void 
-bind_names ( symbol_t *function, node_t *root )
+insert_symbol(tlhash_t *hash_table, symbol_t* symbol)
 {
-    /* TODO: Bind names and string literals in local symbol table */
+    tlhash_insert(hash_table, symbol->name, strlen(symbol->name), symbol);
+}
+
+
+/**
+ * @brief Traverse a node root, and find all variables and strings
+ * 
+ * @param function  pointer to the current function
+ * @param root      pointer to the root node
+ */
+static void 
+bind_names ( symbol_t *function, node_t *root )
+{   // NULL check
+    if (!function)
+        return;
+    if (!root)
+        return;
+
+    // Can't declare variables inside switch unless 
+    // it is in a new scope
+    node_t *declarations;
+    // We want do top to bottom traverse, so do not 
+    // call recusivly unless we need to go deeper
+    switch (root->type)
+    {
+    // If new BLOCK start, push the scope and recurse from here.
+    case BLOCK:
+        push_scope();
+        for (uint64_t i = 0; i < root->n_children; i++)
+            bind_names(function, root->children[i]);
+        pop_scope();
+        break;
+
+    // If DECLARATION_LIST, find all the identifiers 
+    // and insert local into scope and function 
+    case DECLARATION_LIST:
+        if (!root->children[0])
+            break;
+
+        declarations = root->children[0];
+        for (uint64_t i = 0; i < declarations->n_children; i++)
+            // Insert each of the local variables in the declaration
+            insert_local_var(function, declarations->children[i]);
+
+        break;
+    
+    // If IDENTIFIER_DATA, look up the identifier in all the scopes. 
+    // If not found (NULL), crash the compiler with a somewhat helpful message.
+    case IDENTIFIER_DATA:
+        if (!root->data)
+            break;
+
+        if (!(root->entry = lookup_var(function, root->data)))
+        {
+            ERRPRT("Could not find %s in scope!\n", (char*)root->data)
+            exit(EXIT_FAILURE);
+        }
+        break;
+
+    // If STRING_DATA, collect the string and point the 
+    // data in the corresponding node to the array index
+    case STRING_DATA:
+        collect_string(root);
+        break;
+    
+    // If none of the above, go deeper if possible.
+    default:
+        for (uint64_t i = 0; i < root->n_children; i++)
+            bind_names(function, root->children[i]);
+        break;
+    }
+}
+
+/**
+ * Pushes a new hash table to the scope stack.
+ * 
+ * Increases the size of the stack if too small.
+ * 
+ */
+static void
+push_scope(void)
+{
+    // Allocate memory for the hash table and initialize
+    scopes[cur_scope_depth] = malloc(sizeof(tlhash_t));
+    tlhash_init(scopes[cur_scope_depth++], LOCAL_BUCKET_SIZE);
+
+    // Grow the amount of scopes if not enough
+    if (cur_scope_depth >= no_scopes)
+    {
+        no_scopes *= 2;
+        tlhash_t **new_scopes = realloc(scopes, no_scopes * sizeof(tlhash_t));
+        if (!new_scopes)
+        {
+            ERRPRT("Could not realloc scopes!\n");
+            exit(EXIT_FAILURE);
+        }
+        scopes = new_scopes;
+    }
+}
+
+
+/**
+ * Pops the dynamicy allocated hash table for the current scope depth
+ * 
+ */
+static void
+pop_scope(void)
+{
+    tlhash_finalize(scopes[--cur_scope_depth]);
+    free(scopes[cur_scope_depth]);
+}
+
+
+/**
+ * Allocates and inserts a local symbol into 
+ * the scope stack and into the function
+ * 
+ * @param function  pointer to the current function
+ * @param root      pointer to the root node for the symbol
+ */
+static void
+insert_local_var(symbol_t *function, node_t *root)
+{   // Null ptr check
+    if (!root->data)
+        return;
+
+    // Get the sequence num, is the size
+    size_t sequence = tlhash_size(function->locals);
+
+    symbol_t *variable = malloc(sizeof(symbol_t));
+    *variable = (symbol_t){
+        .type   = SYM_LOCAL_VAR,
+        .name   = root->data,
+        .node   = root,            
+        .seq    = sequence,     //! Use sequence as name in var list of function, strictly growing
+        .nparms = 0,
+        .locals = NULL
+    };
+    insert_local_to_scope(variable);
+    insert_local_to_func(function, variable);
+}
+
+
+/**
+ * Inserts a symbol to the top most scope in stack
+ * 
+ * @param local pointer to the local to be inserted
+ */
+static void 
+insert_local_to_scope(symbol_t *local)
+{
+    insert_symbol(scopes[cur_scope_depth - 1], local);
+}
+
+
+/**
+ * Insert local symbol to the functions table of local variables
+ * uses the seq num as key as this is strictly growing
+ * 
+ * @param function  pointer to the function to insert the symbol
+ * @param local     pointer to the symbol to be inserted in the table 
+ */
+void 
+insert_local_to_func(symbol_t *function, symbol_t *local)
+{
+    tlhash_insert(
+        function->locals,   //! Insert local to the function var table
+        &local->seq,         //! The key is a number, unique, strictly growing
+        sizeof(local->seq),  //! Size of key
+        local                //! The local symbol
+    );
+}
+
+
+/**
+ * Collects strings to the string array and 
+ * points the data in the associated node 
+ * to the array position
+ * 
+ * @param root  pointer to the root node of the string
+ */
+static void
+collect_string(node_t *root)
+{   // Null ptr check
+    if (!root->data)
+        return;
+    
+    // Get the string and allocate room for array index of string
+    string_list[stringc] = root->data;
+    root->data = malloc(sizeof(size_t));
+    // Set the data ptr
+    *((size_t*)root->data) = stringc++;
+
+    // Grow string array if nessecary
+    if (stringc >= n_string_list)
+    {
+        n_string_list *= 2;
+        char **new_string_list = realloc(string_list, n_string_list * sizeof(char*));
+        if (!new_string_list)
+        {
+            ERRPRT("Could not realloc string list!\n");
+            exit(EXIT_FAILURE);
+        }
+        string_list = new_string_list;
+    }
+}
+
+
+/**
+ * Looks up a variable identifier in all the scopes. 
+ * Start with the scopes, then the parameters and 
+ * the the globals
+ * 
+ * @param function  pointer to the function
+ * @param var       identifier to the variable
+ * @return Returns the pointer to the "closest" matched identifier. NULL if not found.
+ */
+static symbol_t*
+lookup_var(symbol_t *function, char* var)
+{
+    // Symbol to store the stymbol to be found
+    symbol_t* symbol = NULL;
+    // Result stores the result of the hash lookups
+    int result;
+
+    // Try the local scopes first
+    for (int64_t d = cur_scope_depth - 1; d >= 0; d--)
+    {
+        result = tlhash_lookup(scopes[d], var, strlen(var), (void**)&symbol);
+        if (result == TLHASH_SUCCESS)
+            return symbol;
+    }
+
+    // Then move to parameters
+    result = tlhash_lookup(function->locals, var, strlen(var), (void**)&symbol);
+    if (result == TLHASH_SUCCESS)
+        return symbol;
+    
+    // Last try global parameters
+    result = tlhash_lookup(global_names, var, strlen(var), (void**)&symbol);
+    if (result == TLHASH_SUCCESS)
+        return symbol;
+
+    // If nothing is found, return NULL
+    return NULL;
 }

exercises/04/vslc/src/parser.y

@@ -1,176 +1,303 @@
 %{
 #include <vslc.h>
 
-#define N0C(n,t,d) do { \
-    node_init ( n = malloc(sizeof(node_t)), t, d, 0 ); \
-} while ( false )
-#define N1C(n,t,d,a) do { \
-    node_init ( n = malloc(sizeof(node_t)), t, d, 1, a ); \
-} while ( false )
-#define N2C(n,t,d,a,b) do { \
-    node_init ( n = malloc(sizeof(node_t)), t, d, 2, a, b ); \
-} while ( false )
-#define N3C(n,t,d,a,b,c) do { \
-    node_init ( n = malloc(sizeof(node_t)), t, d, 3, a, b, c ); \
-} while ( false )
-
+#define NODE(type, data, n_children, children...) node_init(malloc(sizeof(node_t)), type, data, n_children,  ##children)
 %}
 
-%left '|'
-%left '^'
-%left '&'
+%define api.value.type {node_t}
+%token FUNC PRINT RETURN CONTINUE IF THEN ELSE WHILE DO OPENBLOCK CLOSEBLOCK
+%token VAR NUMBER IDENTIFIER STRING
+
+%left '|' '&' '^'
 %left '+' '-'
 %left '*' '/'
 %nonassoc UMINUS
 %right '~'
-%expect 1
+//%expect 1
 
-%token FUNC PRINT RETURN CONTINUE IF THEN ELSE WHILE DO OPENBLOCK CLOSEBLOCK
-%token VAR NUMBER IDENTIFIER STRING
+%nonassoc IF THEN
+%nonassoc ELSE
+
+/* Tried fixing vscode complaining about the type for the non-terminals, didn't work
+%union {
+  node_t* node;
+}
+
+%type <node> global_list global 
+%type <node> statement_list print_list expression_list variable_list argument_list parameter_list declaration_list
+%type <node> function statement block 
+%type <node> assignment_statement return_statement print_statement null_statement if_statement while_statement
+%type <node> relation expression declaration print_item identifier number string
+*/ 
 
 %%
-program :
-      global_list { N1C ( root, PROGRAM, NULL, $1 ); }
-    ;
-global_list :
-      global { N1C ( $$, GLOBAL_LIST, NULL, $1 ); }
-    | global_list global { N2C ( $$, GLOBAL_LIST, NULL, $1, $2 ); }
-    ;
+
+program:
+    global_list {
+      root = NODE(PROGRAM, NULL, 1, $1);
+    }
+  ;
+
+global_list:
+    global {
+      $$ = NODE(GLOBAL_LIST, NULL, 1, $1);
+    }
+  | global_list global {
+      $$ = NODE(GLOBAL_LIST, NULL, 2, $1, $2);
+    }
+  ;
+
 global:
-      function { N1C ( $$, GLOBAL, NULL, $1 ); }
-    | declaration { N1C ( $$, GLOBAL, NULL, $1 ); }
-    ;
-statement_list :
-      statement { N1C ( $$, STATEMENT_LIST, NULL, $1 ); }
-    | statement_list statement { N2C ( $$, STATEMENT_LIST, NULL, $1, $2 ); }
-    ;
-print_list :
-      print_item { N1C ( $$, PRINT_LIST, NULL, $1 ); }
-    | print_list ',' print_item { N2C ( $$, PRINT_LIST, NULL, $1, $3 ); }
-    ;
-expression_list :
-      expression { N1C ( $$, EXPRESSION_LIST, NULL, $1 ); }
-    | expression_list ',' expression { N2C($$, EXPRESSION_LIST, NULL, $1, $3); }
-    ;
-variable_list :
-      identifier { N1C ( $$, VARIABLE_LIST, NULL, $1 ); }
-    | variable_list ',' identifier { N2C ( $$, VARIABLE_LIST, NULL, $1, $3 ); }
-    ;
-argument_list :
-      expression_list { N1C ( $$, ARGUMENT_LIST, NULL, $1 ); }
-    | /* epsilon */ { $$ = NULL; }
-    ;
-parameter_list :
-      variable_list { N1C ( $$, PARAMETER_LIST, NULL, $1 ); }
-    | /* epsilon */ { $$ = NULL; }
-    ;
-declaration_list :
-      declaration { N1C ( $$, DECLARATION_LIST, NULL, $1 ); }
-    | declaration_list declaration { N2C ($$, DECLARATION_LIST, NULL, $1, $2); }
-    ;
-function :
-      FUNC identifier '(' parameter_list ')' statement
-        { N3C ( $$, FUNCTION, NULL, $2, $4, $6 ); }
-    ;
-statement :
-      assignment_statement { N1C ( $$, STATEMENT, NULL, $1 ); }
-    | return_statement { N1C ( $$, STATEMENT, NULL, $1 ); }
-    | print_statement { N1C ( $$, STATEMENT, NULL, $1 ); }
-    | if_statement { N1C ( $$, STATEMENT, NULL, $1 ); }
-    | while_statement { N1C ( $$, STATEMENT, NULL, $1 ); }
-    | null_statement { N1C ( $$, STATEMENT, NULL, $1 ); }
-    | block { N1C ( $$, STATEMENT, NULL, $1 ); }
-    ;
-block :
-      OPENBLOCK declaration_list statement_list CLOSEBLOCK
-        { N2C ($$, BLOCK, NULL, $2, $3); }
-    | OPENBLOCK statement_list CLOSEBLOCK { N1C ($$, BLOCK, NULL, $2 ); }
-    ;
-assignment_statement :
-      identifier ':' '=' expression
-        { N2C ( $$, ASSIGNMENT_STATEMENT, NULL, $1, $4 ); }
-    | identifier '+' '=' expression
-        { N2C ( $$, ADD_STATEMENT, NULL, $1, $4 ); }
-    | identifier '-' '=' expression
-        { N2C ( $$, SUBTRACT_STATEMENT, NULL, $1, $4 ); }
-    | identifier '*' '=' expression
-        { N2C ( $$, MULTIPLY_STATEMENT, NULL, $1, $4 ); }
-    | identifier '/' '=' expression
-        { N2C ( $$, DIVIDE_STATEMENT, NULL, $1, $4 ); }
-    ;
-return_statement :
-      RETURN expression
-        { N1C ( $$, RETURN_STATEMENT, NULL, $2 ); }
-    ;
-print_statement :
-      PRINT print_list
-        { N1C ( $$, PRINT_STATEMENT, NULL, $2 ); }
-    ;
-null_statement :
-      CONTINUE
-        { N0C ( $$, NULL_STATEMENT, NULL ); }
-    ;
-if_statement :
-      IF relation THEN statement
-        { N2C ( $$, IF_STATEMENT, NULL, $2, $4 ); }
-    | IF relation THEN statement ELSE statement
-        { N3C ( $$, IF_STATEMENT, NULL, $2, $4, $6 ); }
-    ;
-while_statement :
-      WHILE relation DO statement
-        { N2C ( $$, WHILE_STATEMENT, NULL, $2, $4 ); }
-    ;
+    function {
+      $$ = NODE(GLOBAL, NULL, 1, $1);
+    }
+  | declaration {
+      $$ = NODE(GLOBAL, NULL, 1, $1);
+    }
+  ;
+
+statement_list:
+    statement {
+      $$ = NODE(STATEMENT_LIST, NULL, 1, $1);
+    }
+  | statement_list statement {
+      $$ = NODE(STATEMENT_LIST, NULL, 2, $1, $2);
+    }
+  ;
+
+print_list: 
+    print_item {
+      $$ = NODE(PRINT_LIST, NULL, 1, $1);
+    }
+  | print_list ',' print_item {
+      $$ = NODE(PRINT_LIST, NULL, 2, $1, $3);
+    }
+  ;
+
+expression_list:
+    expression {
+      $$ = NODE(EXPRESSION_LIST, NULL, 1, $1);
+    }
+  | expression_list ',' expression {
+      $$ = NODE(EXPRESSION_LIST, NULL, 2, $1, $3);
+    }
+  ;
+
+variable_list:
+    identifier {
+      $$ = NODE(VARIABLE_LIST, NULL, 1, $1);
+    }
+  | variable_list ',' identifier {
+      $$ = NODE(VARIABLE_LIST, NULL, 2, $1, $3);
+    }
+  ;
+
+argument_list:
+    expression_list {
+      $$ = NODE(ARGUMENT_LIST, NULL, 1, $1);
+    }
+  | /* epsilon */ {
+      $$ = NODE(ARGUMENT_LIST, NULL, 0);
+    }
+  ;
+
+parameter_list:
+    variable_list {
+      $$ = NODE(PARAMETER_LIST, NULL, 1, $1);
+    }
+  | /* epsilon */ {
+      $$ = NODE(PARAMETER_LIST, NULL, 0);
+    }
+  ;
+
+declaration_list:
+    declaration {
+      $$ = NODE(DECLARATION_LIST, NULL, 1, $1);
+    }
+  | declaration_list declaration {
+      $$ = NODE(DECLARATION_LIST, NULL, 2, $1, $2);
+    }
+  ;
+
+function:
+    FUNC identifier '(' parameter_list ')' statement  {
+      $$ = NODE(FUNCTION, NULL, 3, $2, $4, $6);
+    }
+  ;
+
+statement:
+    assignment_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | return_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | print_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | if_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | while_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | null_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | block {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  ;
+
+block:
+    OPENBLOCK declaration_list statement_list CLOSEBLOCK {
+      $$ = NODE(BLOCK, NULL, 2, $2, $3);
+    }
+  | OPENBLOCK statement_list CLOSEBLOCK {
+      $$ = NODE(BLOCK, NULL, 1, $2);
+    }
+  ;
+
+assignment_statement:
+    identifier ':' '=' expression {
+      $$ = NODE(ASSIGNMENT_STATEMENT, NULL, 2, $1, $4);
+    }
+  | identifier '+' '=' expression {
+      $$ = NODE(ADD_STATEMENT, NULL, 2, $1, $4);
+    }
+  | identifier '-' '=' expression {
+      $$ = NODE(SUBTRACT_STATEMENT, NULL, 2, $1, $4);
+    }
+  | identifier '*' '=' expression {
+      $$ = NODE(MULTIPLY_STATEMENT, NULL, 2, $1, $4);
+    }
+  | identifier '/' '=' expression {
+      $$ = NODE(DIVIDE_STATEMENT, NULL, 2, $1, $4);
+    }
+  ;
+
+return_statement:
+    RETURN expression {
+      $$ = NODE(RETURN_STATEMENT, NULL, 1, $2);
+    }
+  ;
+
+print_statement:
+    PRINT print_list {
+      $$ = NODE(PRINT_STATEMENT, NULL, 1, $2);
+    }
+  ;
+
+null_statement:
+    CONTINUE {
+      $$ = NODE(NULL_STATEMENT, NULL, 0);
+    }
+  ;
+
+if_statement:
+    IF relation THEN statement {
+      $$ = NODE(IF_STATEMENT, NULL, 2, $2, $4);
+    }
+  | IF relation THEN statement ELSE statement {
+      $$ = NODE(IF_STATEMENT, NULL, 3, $2, $4, $6);
+    }
+  ;
+
+while_statement:
+    WHILE relation DO statement {
+      $$ = NODE(WHILE_STATEMENT, NULL, 2, $2, $4);
+    }
+  ;
+
 relation:
-      expression '=' expression
-        { N2C ( $$, RELATION, strdup("="), $1, $3 ); }
-    | expression '<' expression
-        { N2C ( $$, RELATION, strdup("<"), $1, $3 ); }
-    | expression '>' expression
-        { N2C ( $$, RELATION, strdup(">"), $1, $3 ); }
-    ;
-expression :
-      expression '|' expression
-        { N2C ( $$, EXPRESSION, strdup("|"), $1, $3 ); }
-    | expression '^' expression
-        { N2C ( $$, EXPRESSION, strdup("^"), $1, $3 ); }
-    | expression '&' expression
-        { N2C ( $$, EXPRESSION, strdup("&"), $1, $3 ); }
-    | expression '+' expression
-        { N2C ( $$, EXPRESSION, strdup("+"), $1, $3 ); }
-    | expression '-' expression
-        { N2C ( $$, EXPRESSION, strdup("-"), $1, $3 ); }
-    | expression '*' expression
-        { N2C ( $$, EXPRESSION, strdup("*"), $1, $3 ); }
-    | expression '/' expression
-        { N2C ( $$, EXPRESSION, strdup("/"), $1, $3 ); }
-    | '-' expression %prec UMINUS
-        { N1C ( $$, EXPRESSION, strdup("-"), $2 ); }
-    | '~' expression %prec UMINUS
-        { N1C ( $$, EXPRESSION, strdup("~"), $2 ); }
-    | '(' expression ')' { $$ = $2; }
-    | number { N1C ( $$, EXPRESSION, NULL, $1 ); }
-    | identifier
-        { N1C ( $$, EXPRESSION, NULL, $1 ); }
-    | identifier '(' argument_list ')'
-        { N2C ( $$, EXPRESSION, NULL, $1, $3 ); }
-    ;
-declaration :
-      VAR variable_list { N1C ( $$, DECLARATION, NULL, $2 ); }
-    ;
-print_item :
-      expression
-        { N1C ( $$, PRINT_ITEM, NULL, $1 ); }
-    | string
-        { N1C ( $$, PRINT_ITEM, NULL, $1 ); }
-    ;
-identifier: IDENTIFIER { N0C($$, IDENTIFIER_DATA, strdup(yytext) ); }
-number: NUMBER
-      {
-        int64_t *value = malloc ( sizeof(int64_t) );
-        *value = strtol ( yytext, NULL, 10 );
-        N0C($$, NUMBER_DATA, value );
-      }
-string: STRING { N0C($$, STRING_DATA, strdup(yytext) ); }
+    expression '=' expression {
+      $$ = NODE(RELATION, strdup("="), 2, $1, $3);
+    }
+  | expression '<' expression {
+      $$ = NODE(RELATION, strdup("<"), 2, $1, $3);
+    }
+  | expression '>' expression {
+      $$ = NODE(RELATION, strdup(">"), 2, $1, $3);
+    }
+  ;
+
+expression:
+    expression '|' expression {
+      $$ = NODE(EXPRESSION, strdup("|"), 2, $1, $3);
+    }
+  | expression '^' expression {
+      $$ = NODE(EXPRESSION, strdup("^"), 2, $1, $3);
+    }
+  | expression '&' expression {
+      $$ = NODE(EXPRESSION, strdup("&"), 2, $1, $3);
+    }
+  | expression '+' expression {
+      $$ = NODE(EXPRESSION, strdup("+"), 2, $1, $3);
+    }
+  | expression '-' expression {
+      $$ = NODE(EXPRESSION, strdup("-"), 2, $1, $3);
+    }
+  | expression '*' expression {
+      $$ = NODE(EXPRESSION, strdup("*"), 2, $1, $3);
+    }
+  | expression '/' expression {
+      $$ = NODE(EXPRESSION, strdup("/"), 2, $1, $3);
+    }
+  | '-' expression %prec UMINUS {
+      $$ = NODE(EXPRESSION, strdup("-"), 1, $2);
+    }
+  | '~' expression {
+      $$ = NODE(EXPRESSION, strdup("~"), 1, $2);
+    }
+  | '(' expression ')' {
+      $$ = NODE(EXPRESSION, /*NULL*/ strdup("group"), 1, $2);
+    }
+  | number {
+      $$ = NODE(EXPRESSION, /*NULL*/ strdup("number"), 1, $1);
+    }
+  | identifier {
+      $$ = NODE(EXPRESSION, /*NULL*/ strdup("identifier"), 1, $1);
+    }
+  | identifier '(' argument_list ')' {
+      $$ = NODE(EXPRESSION, /*NULL*/ strdup("function_call"), 2, $1, $3);
+    }
+  ;
+
+declaration: 
+    VAR variable_list {
+      $$ = NODE(DECLARATION, NULL, 1, $2);
+    }
+  ;
+
+print_item:
+    expression {
+      $$ = NODE(PRINT_ITEM, NULL, 1, $1);
+    }
+  | string {
+      $$ = NODE(PRINT_ITEM, NULL, 1, $1);
+    }
+  ;
+
+identifier:
+    IDENTIFIER {
+      $$ = NODE(IDENTIFIER_DATA, strdup(yytext), 0); // Zero children
+    }
+  ;
+
+number:
+    NUMBER {
+      uint64_t* p_number = malloc(sizeof(uint64_t));
+      *p_number = strtol(yytext, NULL, 10);
+      $$ = NODE(NUMBER_DATA, p_number, 0); // Zero children
+    }
+  ;
+
+string:
+    STRING {
+      $$ = NODE(STRING_DATA, strdup(yytext), 0); // Zero children
+    }
+  ;
+
 %%
 
 int

exercises/04/vslc/src/tree.c

@@ -12,6 +12,10 @@ tree_print(node_t* root, stem head);
 
 static void destroy_subtree ( node_t *discard );
 
+static void prune_children(node_t **simplified, node_t *root);
+static void resolve_constant_expressions(node_t **simplified, node_t *root);
+static void flatten(node_t **simplified, node_t *root);
+
 
 /* External interface */
 void
@@ -40,7 +44,8 @@ print_syntax_tree ( void )
 }
 
 
-void
+// Changed so it returns the pointer to the new node, can be used as before, but makes the parser file cleaner
+node_t*
 node_init (node_t *nd, node_index_t type, void *data, uint64_t n_children, ...)
 {
     va_list child_list;
@@ -55,6 +60,8 @@ node_init (node_t *nd, node_index_t type, void *data, uint64_t n_children, ...)
     for ( uint64_t i=0; i<n_children; i++ )
         nd->children[i] = va_arg ( child_list, node_t * );
     va_end ( child_list );
+
+    return nd;
 }
 
 
@@ -82,9 +89,10 @@ tree_print(node_t* root, stem head)
         return;
     }
     printf("─%s", node_string[root->type]);
-    if ( root->type == IDENTIFIER_DATA ||
-         root->type == STRING_DATA ||
-         root->type == EXPRESSION ) 
+    if ( root->type == IDENTIFIER_DATA  ||
+         root->type == STRING_DATA      ||
+         root->type == EXPRESSION       ||
+         root->type == RELATION)  
         printf("(%s)", (char *) root->data);
     else if (root->type == NUMBER_DATA)
         printf("(%ld)", *((int64_t *)root->data));
@@ -152,81 +160,237 @@ destroy_subtree ( node_t *discard )
 
 
 static void
-simplify_tree ( node_t **simplified, node_t *root )
+flatten(node_t **simplified, node_t *root)
 {
-    if ( root == NULL )
+    /* This will flatten left-expanded lists */
+    if (!root)
         return;
 
-    /* Simplify subtrees before examining this node */
-    for ( uint64_t i=0; i<root->n_children; i++ )
-        simplify_tree ( &root->children[i], root->children[i] );
+    /* Do this recursivly */
+    for (int i = 0; i < root->n_children; i++)
+        flatten(&root->children[i], root->children[i]);
 
-    node_t *discard, *result = root;
-    switch ( root->type )
+    node_t **new_children, *result = root;
+    switch (root->type)
     {
-        /* Structures of purely syntactic function */
-        case PARAMETER_LIST: case ARGUMENT_LIST:
-        case STATEMENT: case PRINT_ITEM: case GLOBAL:
-            result = root->children[0];
-            node_finalize ( root );
+    case GLOBAL_LIST:
+    case STATEMENT_LIST:
+    case PRINT_LIST:
+    case EXPRESSION_LIST:
+    case VARIABLE_LIST:
+    case DECLARATION_LIST:
+        // Check if node have more than two children
+        if (root->n_children < 2)
             break;
-        case PRINT_STATEMENT:
-            result = root->children[0];
+        
+        result = root->children[0];
+        result->n_children++;
+
+        // Realloc the array of children to the new size 
+        if (!(new_children = realloc(result->children, result->n_children * sizeof(node_t*))))
+            break;
+        // if successs, insert the new array
+        result->children = new_children;
+
+        // Insert child at the end
+        result->children[result->n_children - 1] = root->children[1];
+
+        node_finalize(root);
+        break;
+    }
+
+    *simplified = result;
+}
+
+
+static void
+prune_children(node_t **simplified, node_t *root) 
+{
+    if (!root)
+        return;
+
+    /* Do this recursivly */
+    for (int i = 0; i < root->n_children; i++)
+        prune_children(&root->children[i], root->children[i]);
+
+    node_t *result = root;
+    switch (root->type)
+    {
+    case PROGRAM:
+    case GLOBAL:
+    //case ARGUMENT_LIST:     // For this to work, need to change order of operations
+    //case PARAMETER_LIST:    // For this to work, need to change order of operations
+    //case VARIABLE_LIST:
+    //case EXPRESSION_LIST:
+    case DECLARATION:
+    case STATEMENT:
+    case PRINT_ITEM:
+    case PRINT_STATEMENT:
+        result = root->children[0];
+
+        // The print_statement only contains a print_list, still need a print_statement.
+        if (root->type == PRINT_STATEMENT)
             result->type = PRINT_STATEMENT;
+        
+        node_finalize(root);
+        break;
+    }
+
+    *simplified = result;
+}
+
+
+static void
+resolve_constant_expressions(node_t **simplified, node_t *root)
+{
+    if (!root)
+        return;
+
+    /* Do this recursivly */
+    for (int i = 0; i < root->n_children; i++)
+        resolve_constant_expressions(&root->children[i], root->children[i]);
+    
+    if (root->type != EXPRESSION)
+        return;
+    
+    node_t *result = root;
+    
+    switch (root->n_children)
+    {
+    case 1:
+        result = root->children[0];
+        
+        if (root->data && 
+            result->type == NUMBER_DATA && 
+            result->data)
+        {
+            switch (*((char*)root->data))
+            {
+            case '-':
+                *((int64_t*)result->data) *= -1;
+                break;
+            case '~':
+                *((int64_t*)result->data) = ~*((int64_t*)result->data);
+                break;
+            }
+        }
+
+        node_finalize(root);
+        break;
+    
+    case 2:
+        // Both children needs to be numbers to resolve constants
+        if (root->children[0]->type == NUMBER_DATA && 
+            root->children[1]->type == NUMBER_DATA)
+        {
+            // Check if children does not contain null pointers
+            if (!root->children[0]->data)
+                break;
+            if (!root->children[1]->data)
+                break;
+
+            // Check if data field is not null pointer
+            if (!root->data)
+                break;
+            
+            result = root->children[0];
+            int64_t
+                *lhs = result->data,
+                *rhs = root->children[1]->data;
+
+            switch (*(char*)root->data)
+            {
+            /* Assignments */
+            case '|': *lhs |= *rhs; break;
+            case '^': *lhs ^= *rhs; break;
+            case '&': *lhs &= *rhs; break;
+            case '+': *lhs += *rhs; break;
+            case '-': *lhs -= *rhs; break;
+            case '*': *lhs *= *rhs; break;
+            case '/': *lhs /= *rhs; break;
+            }
+
+            node_finalize(root->children[1]);
             node_finalize(root);
-        /* Flatten lists:
-         * Take left child, append right child, substitute left for root.
-         */
-        case STATEMENT_LIST: case DECLARATION_LIST: case GLOBAL_LIST:
-        case PRINT_LIST: case EXPRESSION_LIST: case VARIABLE_LIST:
-            if ( root->n_children >= 2 )
-            {
-                result = root->children[0];
-                result->n_children += 1;
-                result->children = realloc (
-                    result->children, result->n_children * sizeof(node_t *)
-                );
-                result->children[result->n_children-1] = root->children[1];
-                node_finalize ( root );
-            }
-            break;
-        case EXPRESSION:
-            switch ( root->n_children )
-            {
-                case 1:
-                    if ( root->children[0]->type == NUMBER_DATA )
-                    {
-                        result = root->children[0];
-                        if ( root->data != NULL )
-                            *((int64_t *)result->data) *= -1;
-                        node_finalize (root);
-                    }
-                    else if ( root->data == NULL )
-                    {
-                        result = root->children[0];
-                        node_finalize (root);
-                    }
-                    break;
-                case 2:
-                    if ( root->children[0]->type == NUMBER_DATA &&
-                         root->children[1]->type == NUMBER_DATA
-                    ) {
-                        result = root->children[0];
-                        int64_t
-                            *x = result->data,
-                            *y = root->children[1]->data;
-                        switch ( *((char *)root->data) )
-                        {
-                            case '+': *x += *y; break;
-                            case '-': *x -= *y; break;
-                            case '*': *x *= *y; break;
-                            case '/': *x /= *y; break;
-                        }
-                        node_finalize ( root->children[1] );
-                        node_finalize ( root );
-                    }
-                    break;
-            }
+        }            
+        
+        
+        break;
     }
     *simplified = result;
 }
+
+
+static void
+resolve_constant_relations( node_t** simplified, node_t* root)
+{
+    if (!root)
+        return;
+
+    /* Do this recursivly */
+    for (int i = 0; i < root->n_children; i++)
+        resolve_constant_relations(&root->children[i], root->children[i]);
+    
+    if (root->type != RELATION)//|| root->type != RELATION)
+        return;
+
+    node_t *result = root;
+
+    if (root->n_children != 2)
+        return;
+
+    // Both children must be constant numbers
+    if (root->children[0]->type != NUMBER_DATA ||
+        root->children[1]->type != NUMBER_DATA)
+        return;
+
+    // Check if children does not contain null pointers
+    if (!root->children[0]->data)
+        return;
+    if (!root->children[1]->data)
+        return;
+
+    // Check if data field is not null pointer
+    if (!root->data)
+        return;
+
+    result = root->children[0];
+    int64_t
+        *lhs = result->data,
+        *rhs = root->children[1]->data;
+
+    switch (*(char*)root->data)
+    {
+    /* Relations */
+    case '=': *lhs = (*lhs == *rhs); break;
+    case '<': *lhs = (*lhs <  *rhs); break;
+    case '>': *lhs = (*lhs >  *rhs); break;
+    }
+
+    node_finalize(root->children[1]);
+    node_finalize(root);
+
+    *simplified = result;
+}
+
+
+static void
+simplify_tree ( node_t **simplified, node_t *root )
+{
+    if (!root) 
+        return;
+
+    /*  
+        Each of  the functions do their operations recursivly.
+        This opens up for a lot more flexibility, like removing 
+        variable list after it is flatten
+    */
+    flatten(&root, root);
+    prune_children(&root, root);
+    resolve_constant_expressions(&root, root);
+
+    // The following is experimental, will resolve the constant relations
+    resolve_constant_relations(&root, root);
+
+    *simplified = root;
+}

exercises/04/vslc/src/vslc.c

@@ -6,19 +6,19 @@
 
 /* Global state */
 
-node_t *root;               // Syntax tree
-tlhash_t *global_names;     // Symbol table
-char **string_list;         // List of strings in the source
-size_t n_string_list = 8;   // Initial string list capacity (grow on demand)
-size_t stringc = 0;         // Initial string count
+node_t *root;                   // Syntax tree
+tlhash_t *global_names;         // Symbol table
+char **string_list;             // List of strings in the source
+size_t n_string_list    = 8;    // Initial string list capacity (grow on demand)
+size_t stringc          = 0;    // Initial string count
 
 /* Command line option parsing for the main function */
 static void options ( int argc, char **argv );
 bool
-    print_full_tree = false,
-    print_simplified_tree = false,
+    print_full_tree             = false,
+    print_simplified_tree       = false,
     print_symbol_table_contents = false,
-    new_print_style = true;
+    new_print_style             = true;
 
 
 /* Entry point */
@@ -27,7 +27,8 @@ main ( int argc, char **argv )
 {
     options ( argc, argv );
 
-    yyparse();  // Generated from grammar/bison, constructs syntax tree
+    yyparse();                  // Generated from grammar/bison, constructs syntax tree
+    yylex_destroy();            // Free heap used by flex
 
     if ( print_full_tree )
         print_syntax_tree ();
@@ -35,7 +36,7 @@ main ( int argc, char **argv )
     if ( print_simplified_tree )
         print_syntax_tree ();
 
-    create_symbol_table ();   // In ir.c
+    create_symbol_table ();     // In ir.c
     if ( print_symbol_table_contents )
         print_symbol_table();
 

exercises/04/vslc/vsl_programs/Makefile

@@ -9,8 +9,8 @@ ps2: $(PS2_EXAMPLES)
 ps3: $(PS3_EXAMPLES)
 ps4: $(PS4_EXAMPLES)
 
-%.ast: %.vsl
-	$(VSLC) -t < $^ > $@
+%.ast: %.vsl clean
+	$(VSLC) -T -s < $^ > $@
 
 clean:
 	-rm -r */*.ast

exercises/04/vslc/vsl_programs/ps3-simplify/constants.vsl

@@ -1,6 +1,6 @@
 
 func main() begin
-    var a
+    var a, b
     a := 1 + 2 + 4 + 5 + 6 + 7 + 8 + 9
     b := (10 + 10 * 4) * (2 + 2 * (1 + 1)) / 10 + 2 * 5 + 6 / 3
 

exercises/04/vslc/vsl_programs/ps3-simplify/lists.vsl

@@ -1,5 +1,5 @@
 
-func my_fun(a, b, c, d, e, f, g, h) begin
+func my_func(a, b, c, d, e, f, g, h) begin
     var i, j, k, l, m
 
     i := a + b + d
@@ -7,10 +7,22 @@ func my_fun(a, b, c, d, e, f, g, h) begin
     if i = f then begin
         print "hmmm"
     end
+
+    if 1 = 1 then begin
+        print "true"
+    end
+
+    if 1 = 2 then begin
+        print "false"
+    end 
+
+    if 1 < 2 then begin
+        print "true"
+    end
 end
 
 func main() begin
     var n, o, p, q, r, s, t, u, v, w
     n := 5
-    n += my_func(1, 2, 3, 5, 8, 13, 21, 34)
+    n += my_func(1, 2, 3, 5, 8, 13, 21, w)
 end

exercises/05/.gitignore

@@ -0,0 +1,17 @@
+# Buildfiles
+*.o
+vslc/src/vslc
+parser.c
+scanner.c
+y.tab.h
+
+# Submission
+*.tar.xz
+
+# VSL treefiles
+*.ast
+*.sast
+*.sym
+*.bin
+*.s
+*.S

exercises/05/.vscode/launch.json

@@ -0,0 +1,49 @@
+{
+    "version": "0.2.0",
+    "configurations": [
+      {
+        "name": "(gdb) Launch task",
+        "type": "cppdbg",
+        "request": "launch",
+        "program": "${workspaceRoot}/tasks/10_green_bottles",
+        "args": [],
+        "stopAtEntry": false,
+        "cwd": "${fileDirname}",
+        "environment": [],
+        "externalConsole": false,
+        "MIMode": "gdb",
+        "setupCommands": [
+            {
+                "description": "Enable pretty-printing for gdb",
+                "text": "-enable-pretty-printing",
+                "ignoreFailures": true
+            },
+            {
+                "description":  "Set Disassembly Flavor to Intel",
+                "text": "-gdb-set disassembly-flavor intel",
+                "ignoreFailures": true
+            }
+        ]
+      },
+      {
+        // for Linux
+        "name": "(gdb) Launch",
+        "type": "cppdbg",
+        "request": "launch",
+        "program": "${workspaceRoot}/vslc/src/vslc",
+        "args": ["<", "${workspaceFolder}/vslc/vsl_programs/ps5-codegen1/lists.vsl"],
+        "stopAtEntry": false,
+        "cwd": "${workspaceRoot}",
+        "environment": [],
+        "externalConsole": true,
+        "MIMode": "gdb",
+        "setupCommands": [
+          {
+            "description": "Enable pretty-printing for gdb",
+            "text": "-enable-pretty-printing",
+            "ignoreFailures": true
+          }
+        ]
+      }
+    ]
+  }

exercises/05/vslc/Makefile

@@ -0,0 +1,12 @@
+LEX=flex
+YACC=bison
+YFLAGS+=--defines=src/y.tab.h -o y.tab.c
+CFLAGS+=-std=c99 -g -Isrc -Iinclude -D_POSIX_C_SOURCE=200809L -DYYSTYPE="node_t *"
+
+src/vslc: src/vslc.c src/parser.o src/scanner.o src/nodetypes.o src/tree.o src/ir.o src/generator.o src/tlhash.c
+src/y.tab.h: src/parser.c
+src/scanner.c: src/y.tab.h src/scanner.l
+clean:
+	-rm -f src/parser.c src/scanner.c src/*.tab.* src/*.o
+purge: clean
+	-rm -f src/vslc

exercises/05/vslc/include/ir.h

@@ -0,0 +1,51 @@
+#ifndef IR_H
+#define IR_H
+
+/* This is the tree node structure */
+typedef struct n {
+    node_index_t type;
+    void *data;
+    struct s *entry;
+    uint64_t n_children;
+    struct n **children;
+} node_t;
+
+/**Export the initializer function, it is needed by the parser
+ * @param *nd node to initialize
+ * @param type type of node (see nodetype.h)
+ * @param *data associated data. Declared void to allow any type
+ * @param n_children number of children
+ * @param ... variable argument list of child nodes (node_t *)
+ * 
+ * @return Pointer to the initialized node
+ * */
+node_t* node_init (
+    node_t*         nd, 
+    node_index_t    type, 
+    void*           data, 
+    uint64_t        n_children,
+    ...
+);
+
+typedef enum {
+    SYM_GLOBAL_VAR, 
+    SYM_FUNCTION, 
+    SYM_PARAMETER, 
+    SYM_LOCAL_VAR
+} symtype_t;
+
+typedef struct s {
+    char*       name;
+    symtype_t   type;
+    node_t*     node;
+    size_t      seq;
+    size_t      nparms;
+    tlhash_t*   locals;
+} symbol_t;
+#endif
+
+#define GLOBAL_BUCKET_SIZE          32
+#define LOCAL_BUCKET_SIZE           16
+
+#define DEFAULT_STRING_LIST_SIZE    8
+#define DEFAULT_NO_SCOPES           1

exercises/05/vslc/include/nodetypes.h

@@ -0,0 +1,37 @@
+#ifndef NODETYPES_H
+#define NODETYPES_H
+typedef enum {
+    PROGRAM,
+    GLOBAL_LIST,
+    GLOBAL,
+    STATEMENT_LIST,
+    PRINT_LIST,
+    EXPRESSION_LIST,
+    VARIABLE_LIST,
+    ARGUMENT_LIST,
+    PARAMETER_LIST,
+    DECLARATION_LIST,
+    FUNCTION,
+    STATEMENT,
+    BLOCK,
+    ASSIGNMENT_STATEMENT,
+    ADD_STATEMENT,
+    SUBTRACT_STATEMENT,
+    MULTIPLY_STATEMENT,
+    DIVIDE_STATEMENT,
+    RETURN_STATEMENT,
+    PRINT_STATEMENT,
+    NULL_STATEMENT,
+    IF_STATEMENT,
+    WHILE_STATEMENT,
+    EXPRESSION,
+    RELATION,
+    DECLARATION,
+    PRINT_ITEM,
+    IDENTIFIER_DATA,
+    NUMBER_DATA,
+    STRING_DATA
+} node_index_t;
+
+extern char *node_string[26];
+#endif

exercises/05/vslc/include/tlhash.h

@@ -0,0 +1,28 @@
+#ifndef TLHASH_H
+#define TLHASH_H
+#include <stddef.h>
+typedef struct el {
+    void *key, *value;
+    size_t key_length;
+    struct el *next;
+} tlhash_element_t;
+
+typedef struct {
+    size_t n_buckets, size;
+    tlhash_element_t **buckets;
+} tlhash_t;
+
+int tlhash_init ( tlhash_t *tab, size_t n_buckets );
+int tlhash_finalize ( tlhash_t *tab );
+int tlhash_insert ( tlhash_t *tab, void *key, size_t keylen, void *val );
+int tlhash_lookup ( tlhash_t *tab, void *key, size_t keylen, void **val );
+int tlhash_remove ( tlhash_t *tab, void *key, size_t key_length );
+size_t tlhash_size ( tlhash_t *tab );
+void tlhash_keys ( tlhash_t *tab, void **keys );
+void tlhash_values ( tlhash_t *tab, void **values );
+
+#define TLHASH_SUCCESS 0    /* Success */
+#define TLHASH_ENOMEM 1     /* No memory available */
+#define TLHASH_ENOENT 2     /* No such table entry */
+#define TLHASH_EEXIST 3     /* Table entry already exists */
+#endif

exercises/05/vslc/include/vslc.h

@@ -0,0 +1,51 @@
+#ifndef VSLC_H
+#define VSLC_H
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+#include <stdbool.h>
+#include <stdarg.h>
+
+// Prototypes for the hash table functions
+#include "tlhash.h"
+
+// Numbers and names for the types of syntax tree nodes
+#include "nodetypes.h"
+
+// Definition of the tree node type
+#include "ir.h"
+
+// Token definitions and other things from bison, needs def. of node type
+#include "y.tab.h"
+
+/* This is generated from the bison grammar, calls on the flex specification */
+int yyerror ( const char *error );
+
+/* These are defined in the parser generated by bison */
+extern int yylineno;
+extern int yylex ( void );
+extern int yylex_destroy( void );
+extern char yytext[];
+
+/* Global state */
+extern node_t *root;
+
+// Moving global defs to global header
+
+extern tlhash_t *global_names;  // Defined in ir.c, used by generator.c
+extern char **string_list;      // Defined in ir.c, used by generator.c
+extern size_t stringc;          // Defined in ir.c, used by generator.c
+
+/* Global routines, called from main in vslc.c */
+void simplify_syntax_tree ( void );
+void print_syntax_tree ( void );
+void destroy_syntax_tree ( void );
+
+void create_symbol_table ( void );
+void print_symbol_table ( void );
+void destroy_symbol_table ( void );
+
+void generate_program ( void );
+
+#endif

exercises/05/vslc/src/generator.c

@@ -0,0 +1,612 @@
+#include <vslc.h>
+
+#define ASM(opcode, args...)    puts("\t"#opcode"\t"#args)
+#define LABEL(label)            printf("_%s:\n", (char*)label)
+
+#define NO_REG_RECORD           6
+#define NO_CALLE_SAVED_REG      10
+
+/**Generate table of strings in a rodata section. */
+void generate_stringtable ( void );
+/**Declare global variables in a bss section */
+void generate_global_variables ( void );
+/**Generate function entry code
+ * @param function symbol table entry of function */
+void generate_function ( symbol_t *function );
+/**Generate code for a node in the AST, to be called recursively from
+ * generate_function
+ * @param node root node of current code block */
+static void generate_node ( node_t *node );
+/**Initializes program (already implemented) */
+void generate_main ( symbol_t *first );
+
+#define MIN(a,b) (((a)<(b)) ? (a):(b))
+
+static const char *record[NO_REG_RECORD] = {
+    "%rdi", "%rsi", "%rdx", "%rcx", "%r8", "%r9"
+};
+
+static const char *calle_saved_reg[NO_CALLE_SAVED_REG] = {
+    "%rax", "%rcx", "%rdx", "%rdi", "%rsi", "%rsp", "%r8", "%r9", "%r10", "%r11"
+};
+
+// Helper funcs for generating different nodes
+
+/**
+ * Generates assembly for printing
+ * 
+ * @param node print statement node with children to print
+ */
+static void generate_print(node_t* node);
+
+/**
+ * Generate identfier for a variable in memory
+ * 
+ * @param node identifier we want the addres of
+ */
+static void generate_var_ident(node_t *node);
+
+/**
+ * Main function to calculate and solve the expressions.
+ * Based on a stack machine. Result is stored on stack.
+ * 
+ * @param node root node for expression
+ */
+static void solve_expressions(node_t *node);
+
+/**
+ * Generates a funciton call
+ * 
+ * @param node root node for function
+ */
+static void generate_function_call(node_t *node);
+
+/**
+ * Generates the return part of a function
+ *  
+ * @param node node containing the return statement
+ */
+static void generate_function_return(node_t *node);
+
+/**
+ * Used for calculating and evaluating the add/sub/mul/div statements.
+ * Turns the statement into an expression, adds the result of rhs to lhs 
+ * and stores the value back to the indentifier
+ * 
+ * @param node      node to the statement
+ * @param operator  one of the following +, -, *, /
+ */
+static void solve_statements(node_t *node, char *operator);
+
+/**
+ * Generate assembly to fetch a variable on stack
+ * 
+ * @param node node to the variable to be fetched
+ * @param dest where to put the value
+ */
+static void fetch_variable(node_t *node, const char* dest);
+
+/**
+ * Same as fetch_variable, but stores it back to memory.
+ * 
+ * @param node variable to be stored
+ * @param src  from where should the data come from
+ */
+static void writeback_variable(node_t *node, char* src);
+
+// Helper func for fetching all symbols in a table
+static uint64_t fetch_symbols(tlhash_t* symbol_table, symbol_t*** symbol_list);
+
+void
+generate_program ( void )
+{
+    generate_stringtable();
+    generate_global_variables();
+
+    symbol_t **global_list;
+    uint64_t no_globals = fetch_symbols(global_names, &global_list);
+
+    bool main_generated = false;
+    uint64_t seq0_index = -1;
+    for (uint64_t g = 0; g < no_globals; g++)
+    {
+        if (global_list[g]->type != SYM_FUNCTION)
+            continue;
+        
+        // If the name of the function is main
+        if (!strcmp(global_list[g]->name, "main"))
+        {
+            generate_main(global_list[g]);
+            main_generated = true;
+        }
+
+        if (!global_list[g]->seq)
+            seq0_index = g;
+    }
+
+    // If no main was found, use the first function instead. 
+    // That means the function with seq = 0
+    if (!main_generated)
+        generate_main(global_list[seq0_index]);
+
+    for (uint64_t g = 0; g < no_globals; g++)
+    {
+        if (global_list[g]->type == SYM_FUNCTION)
+            generate_function(global_list[g]);
+    }
+
+    free(global_list);
+}
+
+void
+generate_stringtable ( void )
+{
+    /* These can be used to emit numbers, strings and a run-time
+     * error msg. from main
+     */
+    puts("# DATA SECTION");
+    puts(".data");
+    puts(".intout:\t.asciz \"\%ld \"");
+    puts(".strout:\t.asciz \"\%s \"");
+    puts(".errout:\t.asciz \"Wrong number of arguments\"");
+
+    for (uint64_t s = 0; s < stringc; s++) 
+    {
+        printf(".STR%03ld:\t.asciz %s\n", s, string_list[s]);
+    }
+    putchar('\n');
+}
+
+void
+generate_global_variables ( void )
+{
+    symbol_t **global_list;
+    uint64_t no_globals = fetch_symbols(global_names, &global_list);
+
+    puts("# GLOBAL VARIABLES");
+    puts(".bss");
+    puts(".align 8");
+
+    for (uint64_t g = 0; g < no_globals; g++) {
+        if (global_list[g]->type == SYM_GLOBAL_VAR)
+            printf(".%s:\n", global_list[g]->name);
+    }
+    putchar('\n');
+    free(global_list);
+
+}
+
+void
+generate_function ( symbol_t *function )
+{
+    // TODO: Generate code for declaring and entering function, then generate its body
+
+    printf("# func %s(nparams: %ld)\n", function->name, function->nparms);
+    puts(".text");
+    printf(".global _%s\n", function->name);
+    LABEL(function->name);
+    ASM(pushq, %rbp);
+    ASM(movq, %rsp, %rbp);
+
+    // Push params to stack
+    for (int arg = 0; arg < MIN(NO_REG_RECORD,function->nparms); arg++)
+        printf("\tpushq\t%s\n", record[arg] );
+
+    // How many local variables are inside function
+    uint64_t no_locals = function->locals->size - function->nparms;
+
+    // IF the stack alignment is not 16 bytes, 
+    // add one now as all local var also is 0
+    if ((MIN(6,function->nparms) + no_locals) % 2)
+        ASM(pushq, $0);
+
+    // Make room for the local vars
+    while(no_locals--)
+        ASM(pushq, $0);
+
+    // Now the stack ptr should be 16 byte aligned.
+
+    generate_node(function->node);
+
+    putchar('\n');
+}
+
+
+void
+generate_node ( node_t *node)
+{
+    // TODO: Generate code corresponding to node
+
+    // All statements have the same structure.
+    // [0] is the lhs, needs to be identifier, parser ensures this
+    // [1] is thr rhs
+    switch (node->type)
+    {
+    case ASSIGNMENT_STATEMENT:
+        solve_expressions(node->children[1]);
+        ASM(popq, %rax);
+        writeback_variable(node->children[0], "%rax");
+        break;
+
+    case ADD_STATEMENT:
+        // The following way is the naive way of doing an assignment
+        /*
+        fetch_variable(node->children[0], "%rax");
+        ASM(pushq, %rax);
+        solve_expressions(node->children[1]);
+        ASM(popq, %r10);
+        ASM(popq, %rax);
+        ASM(addq, %r10, %rax);
+        writeback_variable(node->children[0], "%rax");
+        */
+        /*  The thing is that add/sub/mul/div assignments 
+            have the same structure as expressions.
+            We can therefore just say that the assignment is 
+            an expression, but remembering to do the writeback afterwards.
+        */
+        puts("# Add statement");
+        solve_statements(node, "+");
+        break;
+
+    case SUBTRACT_STATEMENT:
+        puts("# Subtract statement");
+        solve_statements(node, "-");
+        break;
+
+    case MULTIPLY_STATEMENT:
+        puts("# Multiply statement");
+        solve_statements(node, "*");
+        break;
+
+    case DIVIDE_STATEMENT:
+        puts("# Divide statement");
+        solve_statements(node, "/");
+        break;
+
+    case PRINT_STATEMENT:
+        puts("# Print statement");
+        generate_print(node);
+        break;
+
+    case RETURN_STATEMENT:
+        puts("# Return statement");
+        generate_function_return(node);
+        break;
+
+
+
+    case IF_STATEMENT:
+    case WHILE_STATEMENT:
+        /* DO NOTHING YET */
+        break;
+    case NULL_STATEMENT:
+        /* USED IN WHILE/IF */
+        break;
+
+    
+    case DECLARATION_LIST:
+        /* List of blocks we dont need to traverse */
+        break;
+
+    default:
+        for (int c = 0; c < node->n_children; c++)
+            generate_node(node->children[c]);
+        break;
+
+    }
+}
+
+
+void
+generate_print(node_t* node)
+{
+    // Push rdi and rsi to stack incase there are data in them
+    //ASM(pushq, %rdi);
+    //ASM(pushq, %rsi);
+    for (uint64_t p = 0; p < node->n_children; p++)
+    {
+        node_t *curr_print = node->children[p];
+
+        switch (curr_print->type)
+        {
+        case EXPRESSION:
+            solve_expressions(curr_print);
+            ASM(popq, %rax);
+            ASM(movq, $.intout, %rdi);
+            ASM(movq, %rax, %rsi);
+            break;
+
+        case STRING_DATA:
+            ASM(movq, $.strout, %rdi);
+            printf("\tmovq\t$.STR%03ld, %%rsi\n", *(uint64_t*)curr_print->data);
+            break;
+
+        case IDENTIFIER_DATA:
+            ASM(movq, $.intout, %rdi);
+            fetch_variable(curr_print, "%rsi");
+            break;
+        default:
+            break;
+        }
+        ASM(call, printf);
+    }
+    // Adds a newline
+    ASM(movq, $'\n', %rdi);
+    ASM(call, putchar);
+    
+    //ASM(popq, %rsi);
+    //ASM(popq, %rdi);
+}
+
+
+// This will put the value of var in node in dest
+void
+fetch_variable(node_t *node, const char* dest)
+{
+    printf("\tmovq\t");
+    generate_var_ident(node);
+    printf(", %s\t\t# Fetched: %s\n", dest, node->entry->name);
+}
+
+// This will put the value in dest to the var in node
+void
+writeback_variable(node_t *node, char* src)
+{
+    printf("\tmovq\t%s,", src);
+    generate_var_ident(node);
+    printf("\t\t# Writeback: %s\n", node->entry->name);
+}
+
+
+void
+generate_var_ident(node_t *node)
+{
+    symbol_t *ident_sym = node->entry;
+    switch (ident_sym->type)
+    {
+    case SYM_GLOBAL_VAR:
+        printf("$.%s", ident_sym->name);
+        break;
+
+    case SYM_PARAMETER:
+        // If it is a paramter is one of the first 6, seacrch below bp
+        if (ident_sym->seq < 6)
+            printf("%ld(%%rbp)", -8 * (ident_sym->seq + 1));
+        else
+            // This requires that the parameters on 
+            // stack is in reversed order... easier to implement
+            printf("%ld(%%rbp)", 8 * (ident_sym->seq - 6 + 1 ));
+        break;
+
+    case SYM_LOCAL_VAR:
+        printf("%ld(%%rbp)", -8 * (ident_sym->seq + 1));
+        break;
+    }
+}
+
+
+// This should allways push the result to stack
+void
+solve_expressions(node_t *node)
+{
+    if (node->data)
+    {   // Check if the expression is a function call
+        bool is_function_call = !strcmp(node->data, "function_call");
+        if (is_function_call)
+        {
+            generate_function_call(node);
+            return;
+        }
+    }
+
+
+    switch (node->n_children)
+    {
+    case 0:
+        switch (node->type)
+        {
+        case IDENTIFIER_DATA:
+            fetch_variable(node, "%rax");
+            ASM(pushq, %rax);
+            break;
+        case NUMBER_DATA:
+            printf("\tmovq\t$%ld,%%rax\n",*(int64_t*)node->data);
+            ASM(pushq, %rax);
+            break;
+        }
+        break;
+    case 1:
+        solve_expressions(node->children[0]);
+        ASM(popq, %rax);
+
+        switch (*(char*)node->data)
+        {
+        case '-':
+            ASM(negq, %rax);
+            break;
+        case '~':
+            ASM(notq, %rax);
+            break;
+        }
+        ASM(pushq, %rax);
+
+        break;
+    case 2:
+
+        // First fetch lhs of expr and then rhs
+        // Push results on stack
+        for (int i = 0; i < 2; i++)
+            solve_expressions(node->children[i]);
+
+
+        // Put rhs in %r10
+        ASM(popq, %r10);
+        // put lhs in %rax
+        ASM(popq, %rax);
+
+        switch (*(char*)node->data)
+        {
+            /* Assignments */
+            case '|': ASM(orq,  %r10, %rax);    break;  // Bitwise or of %rax and %r10
+            case '^': ASM(xorq, %r10, %rax);    break;  // Bitwise xor of %rax and %r10
+            case '&': ASM(andq, %r10, %rax);    break;  // Bitwise and of %rax and %r10
+            case '+': ASM(addq, %r10, %rax);    break;  // Add %rax and %r10
+            case '-': ASM(subq, %r10, %rax);    break;  // Subtract %r10 from %rax
+            case '*': ASM(imulq, %r10);         break;  // Mulitply %rax with %r10
+            case '/': 
+                ASM(cqto);          // Convert rax to octaword, %rdx:%rax
+                ASM(idivq, %r10);   // Divide %rdx:%rax by %r10
+                break;
+        }
+
+        // Push result to stack.
+        ASM(pushq, %rax);
+        break;
+    }
+}
+
+void
+generate_function_call(node_t *node)
+{
+    printf("# Function call\n");
+
+    node_t *arg_list = node->children[1];
+    if (arg_list->n_children)
+        arg_list = arg_list->children[0];
+
+    for (int arg = 0; arg < MIN(NO_REG_RECORD, arg_list->n_children); arg++)
+    {
+        if (arg_list->children[arg]->type == NUMBER_DATA)
+            printf("\tmovq\t$%ld, %s\n", 
+                *(int64_t*)arg_list->children[arg]->data, 
+                record[arg]
+            );
+        else
+            fetch_variable(arg_list->children[arg], record[arg]);
+    }
+
+    if (arg_list->n_children > NO_REG_RECORD)
+    {
+        for (int arg = arg_list->n_children - 1; arg >= NO_REG_RECORD; arg--)
+        {
+            if (arg_list->children[arg]->type == NUMBER_DATA)
+                printf("\tpushq\t$%ld\n", 
+                    *(int64_t*)arg_list->children[arg]->data
+                );
+            else
+            {
+                printf("\tpushq\t");
+                generate_var_ident(arg_list->children[arg]);
+                putchar('\n');
+            }
+        }
+        if (arg_list->n_children % 2)
+            ASM(pushq, $0);
+    }
+
+    printf("\tcall\t_%s\n", (char*)node->children[0]->data);
+    ASM(pushq, %rax);
+    printf("# End of function call\n");
+
+    /*
+    for (int reg = 0; reg < NO_CALLE_SAVED_REG; reg++)
+        printf("\tpushq\t%s \t\t# Pushing %s to stack\n", 
+            calle_saved_reg[reg], 
+            calle_saved_reg[reg]
+        );
+
+    for (int reg = NO_CALLE_SAVED_REG; reg > 0; reg--)
+    printf("\tpopq\t%s \t\t# Poping %s from stack\n", 
+        calle_saved_reg[reg], 
+        calle_saved_reg[reg]
+    );*/
+
+}
+
+void
+generate_function_return(node_t *node)
+{
+    solve_expressions(node->children[0]);
+    ASM(popq, %rax);
+    ASM(leave);
+    ASM(ret);
+}
+
+void
+solve_statements(node_t *node, char *operator)
+{
+    node->type = EXPRESSION;
+    node->data = strdup(operator);
+
+    solve_expressions(node);
+
+    ASM(popq, %rax);
+    writeback_variable(node->children[0], "%rax");
+}
+
+/**Generates the main function with argument parsing and calling of our
+ * main function (first, if no function is named main)
+ * @param first Symbol table entry of our main function */
+void
+generate_main ( symbol_t *first )
+{
+    puts("###### Entry point for GAS #####");
+    puts ( ".globl main" );
+    puts ( ".section .text" );
+    puts ( "main:" );
+    puts ( "\tpushq   %rbp" );
+    puts ( "\tmovq    %rsp, %rbp" );
+
+    printf ( "\tsubq\t$1,%%rdi\n" );
+    printf ( "\tcmpq\t$%zu,%%rdi\n", first->nparms );
+    printf ( "\tjne \tABORT\n" );
+    printf ( "\tcmpq\t$0,%%rdi\n" );
+    printf ( "\tjz  \tSKIP_ARGS\n" );
+
+    printf ( "\tmovq\t%%rdi,%%rcx\n" );
+    printf ( "\taddq\t$%zu, %%rsi\n", 8*first->nparms );
+    printf ( "PARSE_ARGV:\n" );
+    printf ( "\tpushq\t%%rcx\n" );
+    printf ( "\tpushq\t%%rsi\n" );
+
+    printf ( "\tmovq\t(%%rsi),%%rdi\n" );
+    printf ( "\tmovq\t$0,%%rsi\n" );
+    printf ( "\tmovq\t$10,%%rdx\n" );
+    printf ( "\tcall\tstrtol\n" );
+
+/*  Now a new argument is an integer in rax */
+
+    printf ( "\tpopq\t%%rsi\n" );
+    printf ( "\tpopq\t%%rcx\n" );
+    printf ( "\tpushq\t%%rax\n" );
+    printf ( "\tsubq\t$8, %%rsi\n" );
+    printf ( "\tloop\tPARSE_ARGV\n" );
+
+    /* Now the arguments are in order on stack */
+    for (int arg = 0; arg < MIN(6,first->nparms); arg++)
+        printf ( "\tpopq\t%s\n", record[arg] );
+
+    printf ( "SKIP_ARGS:\n" );
+    printf ( "\tcall\t_%s\n", first->name );
+    printf ( "\tjmp \tEND\n" );
+    printf ( "ABORT:\n" );
+    printf ( "\tmovq\t$.errout, %%rdi\n" );
+    printf ( "\tcall\tputs\n" );
+
+    printf ( "END:\n" );
+    puts ( "\tmovq \t%rax, %rdi" );
+    puts ( "\tcall \texit" );
+    puts("###### FUNCTIONS FROM VSL BELOW #####");
+    putchar('\n');
+
+}
+
+static uint64_t 
+fetch_symbols(tlhash_t* symbol_table, symbol_t*** symbol_list)
+{
+    uint64_t no_symbols = tlhash_size(symbol_table);
+    *symbol_list = malloc(no_symbols * sizeof(symbol_t));
+    tlhash_values(symbol_table, (void **)*symbol_list );
+
+    return no_symbols;
+}

exercises/05/vslc/src/ir.c

@@ -0,0 +1,602 @@
+#include <vslc.h>
+
+#define ERRPRT(format, args...) {fprintf(stderr, "[ERROR] "); fprintf(stderr ,format, ##args);}
+
+
+// Externally visible, for the generator
+extern tlhash_t *global_names;
+extern char **string_list;
+extern size_t n_string_list, stringc;
+
+// Functions from the skeleton
+
+static uint64_t find_globals ( void );
+static void bind_names ( symbol_t *function, node_t *root );
+
+// Helper functions, see description in the definition
+
+static void print_global_tree(symbol_t* global);
+static void print_string_list(void);
+static void destroy_global(symbol_t* global);
+static void push_scope(void);
+static void pop_scope(void);
+static void insert_symbol(tlhash_t *hash_table, symbol_t* symbol);
+static void insert_local_to_scope(symbol_t *local);
+static void insert_local_to_func(symbol_t *function, symbol_t *root);
+static void insert_local_var(symbol_t *function, node_t *root);
+static void collect_string(node_t *root);
+static symbol_t* lookup_var(symbol_t *function, char* var);
+
+// Local "global" variables
+static const char *symbol_names[4] = {
+    "GLOBAL_VAR", 
+    "FUNCTION", 
+    "PARAMETER", 
+    "LOCAL_VAR"
+};
+static uint64_t no_scopes, cur_scope_depth;
+static tlhash_t **scopes;
+
+
+/**
+ * Gather information and create a symbol table.
+ * 
+ * Used in vslc.c
+ */
+void
+create_symbol_table ( void )
+{
+
+    // Initialize string array
+    n_string_list   = DEFAULT_STRING_LIST_SIZE;
+    string_list     = malloc(n_string_list * sizeof(char*));
+    stringc         = 0;
+
+    // Initialize scope array
+    no_scopes       = DEFAULT_NO_SCOPES;
+    scopes          = malloc(no_scopes * sizeof(tlhash_t));
+    cur_scope_depth = 0;
+
+    // Traverse the root node for globals
+    uint64_t no_globals = find_globals();
+
+    // Prepare a temp list of globals and fetch all globals
+    symbol_t **global_list = malloc(no_globals * sizeof(symbol_t));
+    tlhash_values(global_names, (void **)global_list );
+
+    /* Iterate over the temporary list, bind names in each function */
+    for (uint64_t g = 0; g < no_globals; g++ )
+    {
+        if (global_list[g]->type == SYM_FUNCTION)
+            bind_names(global_list[g], global_list[g]->node);
+    }
+    // Free the temp list
+    free(global_list);
+}
+
+
+/**
+ * Prints the symbol table and the string array
+ * 
+ * Used in vslc.c
+ */
+void
+print_symbol_table ( void )
+{
+    /* Get the number of symbols, size up a temporary list and fill it */
+    uint64_t no_globals = tlhash_size(global_names);
+    symbol_t **global_list = malloc(no_globals * sizeof(symbol_t));
+    tlhash_values(global_names, (void **)global_list );
+
+    /* Iterate over the temporary list, printing entries */
+    for (uint64_t g = 0; g < no_globals; g++ )
+        // Print the tree structure for each global
+        print_global_tree(global_list[g]);
+    
+    free(global_list);
+
+    // Print strings
+    print_string_list();
+}
+
+
+/**
+ * Prins the tree of a global
+ * 
+ * @param global pointer to the global to be printed
+ */
+static void 
+print_global_tree(symbol_t* global)
+{
+    // Check if null ptr
+    if (!global)
+        return;
+    
+    // Print global root
+    printf("─%s: %-16s [nparams=%2ld, seq=%2ld, node=%p]\n", 
+        symbol_names[global->type], 
+        global->name, 
+        global->nparms,
+        global->seq,
+        global->node
+    );
+    
+    // If the global does not have params or locals, return
+    if (!global->nparms && !global->locals)
+        {putchar('\n');return;}
+    
+    // Need to fetch the whole size, since nparams 
+    // only count the params, not all locals
+    uint64_t no_locals = tlhash_size(global->locals);
+    symbol_t **locals_list = malloc(no_locals * sizeof(symbol_t));
+    tlhash_values(global->locals, (void **)locals_list );
+    // Go through all locals
+    for (int l = 0; l < no_locals; l++)
+    {   // Do some simple sorting, so seq num is in order
+        for (int ll = 0; ll < no_locals; ll++)
+        {
+            if (locals_list[ll]->seq == l)
+            {
+                printf(" %s─[%s]: %-22s\t[seq=%2ld, node=%p]\n",
+                    (l < (no_locals - 1)) ? "├" : "└", 
+                    symbol_names[locals_list[ll]->type],
+                    locals_list[ll]->name,
+                    locals_list[ll]->seq,
+                    locals_list[ll]->node
+                );
+                break;
+            }
+        }
+    }
+    putchar('\n');
+    free(locals_list);
+}
+
+
+/**
+ * Prints the array of strings
+ * 
+ */
+static void
+print_string_list(void)
+{   // Print out all the collected strings
+    printf("─STRINGS [%ld]\n", stringc);
+    for (uint64_t i = 0; i < stringc; i++)
+        printf(" %s─[%ld]: %s\n", 
+            (i < (stringc - 1)) ? "├" : "└", 
+            i, 
+            string_list[i]
+        );
+}
+
+
+/**
+ * Destroys all the dynamicly allocated memory and all the hash tables.
+ * Frees up the array of strings as well.
+ * 
+ * Used in vslc.c
+ */
+void
+destroy_symbol_table ( void )
+{
+    // FREE STRINGS
+    // Free all strings that are kept in the array
+    for (uint64_t c = 0; c < stringc; c++)
+        free(string_list[c]);
+    // Free the actual list
+    free(string_list);
+
+    // FREE SCOPES
+    // At the end of program, all scopes have to be popped
+    // Therefore only free the list
+    free(scopes);
+
+    // FREE GLOBAL NAMES
+    if (!global_names)
+        return;
+
+    // Fetch list of globals
+    uint64_t no_globals = tlhash_size(global_names);
+    symbol_t **global_list = malloc(no_globals * sizeof(symbol_t));
+    tlhash_values(global_names, (void **)global_list );
+
+    // Destroy all global elements
+    for (uint64_t g = 0; g < no_globals; g++)
+        destroy_global(global_list[g]);
+    
+    // Destory the global hash table
+    tlhash_finalize(global_names);
+    // Free the global hash table
+    free(global_names);
+
+    // Free the temp list
+    free(global_list);
+}
+
+
+/**
+ * Destroys the supplied global symbol by
+ * finalizing each of the local tables
+ * 
+ * @param global pointer to the global symbol to be destroyed
+ */
+static void 
+destroy_global(symbol_t* global)
+{
+    if (!global)
+        return;
+    if (!global->locals)
+    {
+        free(global);
+        return;
+    }
+    
+    uint64_t no_locals = tlhash_size(global->locals);
+    symbol_t **locals_list = malloc(no_locals * sizeof(symbol_t));
+    tlhash_values(global->locals, (void **)locals_list );
+
+    for (int l = 0; l < no_locals; l++)
+        free(locals_list[l]);
+    
+    tlhash_finalize(global->locals);
+    free(global->locals);
+    free(global);
+
+    free(locals_list);
+}
+
+
+/**
+ * Goes trough the root node and finds all global variables and functions
+ * 
+ * @return Returns the number of globals found (functions + variables)
+ */
+static uint64_t
+find_globals ( void )
+{
+    tlhash_init(global_names = malloc(sizeof(tlhash_t)), GLOBAL_BUCKET_SIZE);
+
+    uint64_t no_functions = 0, no_global_vars = 0;
+    node_t *global_list = root;
+
+    // Check if not nullptr
+    if (!global_list)
+        return 0;
+
+    symbol_t* global_symbol;
+    for (uint64_t global_i = 0; global_i < global_list->n_children; global_i++)
+    {
+        node_t *current_global = global_list->children[global_i];
+        switch (current_global->type)
+        {
+        case VARIABLE_LIST:
+
+            // Go through the variable list and get all the global variables
+            for (uint64_t var_i = 0; var_i < current_global->n_children; var_i++)
+            {
+                global_symbol = malloc(sizeof(symbol_t));
+                *global_symbol = (symbol_t){
+                    .type   = SYM_GLOBAL_VAR,
+                    .name   = current_global->children[var_i]->data,
+                    .node   = current_global->children[var_i],
+                    .seq    = 0,
+                    .nparms = 0,
+                    .locals = NULL
+                };
+                insert_symbol(global_names, global_symbol);
+                no_global_vars++;
+            }
+            break;
+        case FUNCTION:
+            node_t *function = current_global;
+
+            // Function node allways have the same structure,
+            // [0] are the identifier
+            // [1] are the variable list, within a paramerer_list
+            // [2] are the actual block
+            if (!function->children[0])
+                break;
+
+            // Create the function symbol
+            global_symbol = malloc(sizeof(symbol_t));
+            *global_symbol = (symbol_t){
+                .type   = SYM_FUNCTION,
+                .name   = current_global->children[0]->data,
+                .node   = current_global->children[2],
+                .seq    = no_functions++,
+                .nparms = 0,
+                .locals = malloc(sizeof(tlhash_t))
+            };
+
+            // Initialize the local variable table
+            tlhash_init(global_symbol->locals, LOCAL_BUCKET_SIZE);
+
+            // Insert the pointer to the newly created symbol
+            insert_symbol(global_names, global_symbol);
+
+            // If there are no parameters in function, break.
+            if (!current_global->children[1]->n_children)
+                break;
+
+
+            // Find all params and insert into hash table in global_symbol
+            symbol_t *param_sym;
+            node_t *param_list = current_global->children[1]->children[0];
+            global_symbol->nparms = param_list->n_children;
+
+            for (uint64_t param_i = 0; param_i < param_list->n_children; param_i++)
+            {
+                param_sym = malloc(sizeof(symbol_t));
+                *param_sym = (symbol_t){
+                    .type   = SYM_PARAMETER,
+                    .name   = param_list->children[param_i]->data,
+                    .node   = param_list->children[param_i],
+                    .seq    = param_i,
+                    .nparms = 0,
+                    .locals = NULL
+                };
+
+                insert_symbol(global_symbol->locals, param_sym);
+            }
+            break;
+        }
+    }
+    return no_functions + no_global_vars;
+}
+
+
+/**
+ * Inserts a symbol into a hash table, key is defined in the name field in the symbol supplied.
+ * 
+ * @param hash_table    pointer to the hash table the symbol is inserted into
+ * @param symbol        pointer to the symbol to be inserted
+ */
+void 
+insert_symbol(tlhash_t *hash_table, symbol_t* symbol)
+{
+    tlhash_insert(hash_table, symbol->name, strlen(symbol->name), symbol);
+}
+
+
+/**
+ * @brief Traverse a node root, and find all variables and strings
+ * 
+ * @param function  pointer to the current function
+ * @param root      pointer to the root node
+ */
+static void 
+bind_names ( symbol_t *function, node_t *root )
+{   // NULL check
+    if (!function)
+        return;
+    if (!root)
+        return;
+
+    // Can't declare variables inside switch unless 
+    // it is in a new scope
+    node_t *declarations;
+    // We want do top to bottom traverse, so do not 
+    // call recusivly unless we need to go deeper
+    switch (root->type)
+    {
+    // If new BLOCK start, push the scope and recurse from here.
+    case BLOCK:
+        push_scope();
+        for (uint64_t i = 0; i < root->n_children; i++)
+            bind_names(function, root->children[i]);
+        pop_scope();
+        break;
+
+    // If DECLARATION_LIST, find all the identifiers 
+    // and insert local into scope and function 
+    case DECLARATION_LIST:
+        if (!root->children[0])
+            break;
+
+        declarations = root->children[0];
+        for (uint64_t i = 0; i < declarations->n_children; i++)
+            // Insert each of the local variables in the declaration
+            insert_local_var(function, declarations->children[i]);
+
+        break;
+    
+    // If IDENTIFIER_DATA, look up the identifier in all the scopes. 
+    // If not found (NULL), crash the compiler with a somewhat helpful message.
+    case IDENTIFIER_DATA:
+        if (!root->data)
+            break;
+
+        if (!(root->entry = lookup_var(function, root->data)))
+        {
+            ERRPRT("Could not find %s in scope!\n", (char*)root->data)
+            exit(EXIT_FAILURE);
+        }
+        break;
+
+    // If STRING_DATA, collect the string and point the 
+    // data in the corresponding node to the array index
+    case STRING_DATA:
+        collect_string(root);
+        break;
+    
+    // If none of the above, go deeper if possible.
+    default:
+        for (uint64_t i = 0; i < root->n_children; i++)
+            bind_names(function, root->children[i]);
+        break;
+    }
+}
+
+/**
+ * Pushes a new hash table to the scope stack.
+ * 
+ * Increases the size of the stack if too small.
+ * 
+ */
+static void
+push_scope(void)
+{
+    // Allocate memory for the hash table and initialize
+    scopes[cur_scope_depth] = malloc(sizeof(tlhash_t));
+    tlhash_init(scopes[cur_scope_depth++], LOCAL_BUCKET_SIZE);
+
+    // Grow the amount of scopes if not enough
+    if (cur_scope_depth >= no_scopes)
+    {
+        no_scopes *= 2;
+        tlhash_t **new_scopes = realloc(scopes, no_scopes * sizeof(tlhash_t));
+        if (!new_scopes)
+        {
+            ERRPRT("Could not realloc scopes!\n");
+            exit(EXIT_FAILURE);
+        }
+        scopes = new_scopes;
+    }
+}
+
+
+/**
+ * Pops the dynamicy allocated hash table for the current scope depth
+ * 
+ */
+static void
+pop_scope(void)
+{
+    tlhash_finalize(scopes[--cur_scope_depth]);
+    free(scopes[cur_scope_depth]);
+}
+
+
+/**
+ * Allocates and inserts a local symbol into 
+ * the scope stack and into the function
+ * 
+ * @param function  pointer to the current function
+ * @param root      pointer to the root node for the symbol
+ */
+static void
+insert_local_var(symbol_t *function, node_t *root)
+{   // Null ptr check
+    if (!root->data)
+        return;
+
+    // Get the sequence num, is the size
+    size_t sequence = tlhash_size(function->locals);
+
+    symbol_t *variable = malloc(sizeof(symbol_t));
+    *variable = (symbol_t){
+        .type   = SYM_LOCAL_VAR,
+        .name   = root->data,
+        .node   = root,            
+        .seq    = sequence,     //! Use sequence as name in var list of function, strictly growing
+        .nparms = 0,
+        .locals = NULL
+    };
+    insert_local_to_scope(variable);
+    insert_local_to_func(function, variable);
+}
+
+
+/**
+ * Inserts a symbol to the top most scope in stack
+ * 
+ * @param local pointer to the local to be inserted
+ */
+static void 
+insert_local_to_scope(symbol_t *local)
+{
+    insert_symbol(scopes[cur_scope_depth - 1], local);
+}
+
+
+/**
+ * Insert local symbol to the functions table of local variables
+ * uses the seq num as key as this is strictly growing
+ * 
+ * @param function  pointer to the function to insert the symbol
+ * @param local     pointer to the symbol to be inserted in the table 
+ */
+void 
+insert_local_to_func(symbol_t *function, symbol_t *local)
+{
+    tlhash_insert(
+        function->locals,   //! Insert local to the function var table
+        &local->seq,         //! The key is a number, unique, strictly growing
+        sizeof(local->seq),  //! Size of key
+        local                //! The local symbol
+    );
+}
+
+
+/**
+ * Collects strings to the string array and 
+ * points the data in the associated node 
+ * to the array position
+ * 
+ * @param root  pointer to the root node of the string
+ */
+static void
+collect_string(node_t *root)
+{   // Null ptr check
+    if (!root->data)
+        return;
+    
+    // Get the string and allocate room for array index of string
+    string_list[stringc] = root->data;
+    root->data = malloc(sizeof(size_t));
+    // Set the data ptr
+    *((size_t*)root->data) = stringc++;
+
+    // Grow string array if nessecary
+    if (stringc >= n_string_list)
+    {
+        n_string_list *= 2;
+        char **new_string_list = realloc(string_list, n_string_list * sizeof(char*));
+        if (!new_string_list)
+        {
+            ERRPRT("Could not realloc string list!\n");
+            exit(EXIT_FAILURE);
+        }
+        string_list = new_string_list;
+    }
+}
+
+
+/**
+ * Looks up a variable identifier in all the scopes. 
+ * Start with the scopes, then the parameters and 
+ * the the globals
+ * 
+ * @param function  pointer to the function
+ * @param var       identifier to the variable
+ * @return Returns the pointer to the "closest" matched identifier. NULL if not found.
+ */
+static symbol_t*
+lookup_var(symbol_t *function, char* var)
+{
+    // Symbol to store the stymbol to be found
+    symbol_t* symbol = NULL;
+    // Result stores the result of the hash lookups
+    int result;
+
+    // Try the local scopes first
+    for (int64_t d = cur_scope_depth - 1; d >= 0; d--)
+    {
+        result = tlhash_lookup(scopes[d], var, strlen(var), (void**)&symbol);
+        if (result == TLHASH_SUCCESS)
+            return symbol;
+    }
+
+    // Then move to parameters
+    result = tlhash_lookup(function->locals, var, strlen(var), (void**)&symbol);
+    if (result == TLHASH_SUCCESS)
+        return symbol;
+    
+    // Last try global parameters
+    result = tlhash_lookup(global_names, var, strlen(var), (void**)&symbol);
+    if (result == TLHASH_SUCCESS)
+        return symbol;
+
+    // If nothing is found, return NULL
+    return NULL;
+}

exercises/05/vslc/src/nodetypes.c

@@ -0,0 +1,34 @@
+#define STRING(x) #x
+char *node_string[30] = {
+    STRING(PROGRAM),
+    STRING(GLOBAL_LIST),
+    STRING(GLOBAL),
+    STRING(STATEMENT_LIST),
+    STRING(PRINT_LIST),
+    STRING(EXPRESSION_LIST),
+    STRING(VARIABLE_LIST),
+    STRING(ARGUMENT_LIST),
+    STRING(PARAMETER_LIST),
+    STRING(DECLARATION_LIST),
+    STRING(FUNCTION),
+    STRING(STATEMENT),
+    STRING(BLOCK),
+    STRING(ASSIGNMENT_STATEMENT),
+    STRING(ADD_STATEMENT),
+    STRING(SUBTRACT_STATEMENT),
+    STRING(MULTIPLY_STATEMENT),
+    STRING(DIVIDE_STATEMENT),
+    STRING(RETURN_STATEMENT),
+    STRING(PRINT_STATEMENT),
+    STRING(NULL_STATEMENT),
+    STRING(IF_STATEMENT),
+    STRING(WHILE_STATEMENT),
+    STRING(EXPRESSION),
+    STRING(RELATION),
+    STRING(DECLARATION),
+    STRING(PRINT_ITEM),
+    STRING(IDENTIFIER_DATA),
+    STRING(NUMBER_DATA),
+    STRING(STRING_DATA)
+};
+#undef STRING

exercises/05/vslc/src/parser.y

@@ -0,0 +1,308 @@
+%{
+#include <vslc.h>
+
+#define NODE(type, data, n_children, children...) node_init(malloc(sizeof(node_t)), type, data, n_children,  ##children)
+%}
+
+%define api.value.type {node_t}
+%token FUNC PRINT RETURN CONTINUE IF THEN ELSE WHILE DO OPENBLOCK CLOSEBLOCK
+%token VAR NUMBER IDENTIFIER STRING
+
+%left '|' '&' '^'
+%left '+' '-'
+%left '*' '/'
+%nonassoc UMINUS
+%right '~'
+//%expect 1
+
+%nonassoc IF THEN
+%nonassoc ELSE
+
+/* Tried fixing vscode complaining about the type for the non-terminals, didn't work
+%union {
+  node_t* node;
+}
+
+%type <node> global_list global 
+%type <node> statement_list print_list expression_list variable_list argument_list parameter_list declaration_list
+%type <node> function statement block 
+%type <node> assignment_statement return_statement print_statement null_statement if_statement while_statement
+%type <node> relation expression declaration print_item identifier number string
+*/ 
+
+%%
+
+program:
+    global_list {
+      root = NODE(PROGRAM, NULL, 1, $1);
+    }
+  ;
+
+global_list:
+    global {
+      $$ = NODE(GLOBAL_LIST, NULL, 1, $1);
+    }
+  | global_list global {
+      $$ = NODE(GLOBAL_LIST, NULL, 2, $1, $2);
+    }
+  ;
+
+global:
+    function {
+      $$ = NODE(GLOBAL, NULL, 1, $1);
+    }
+  | declaration {
+      $$ = NODE(GLOBAL, NULL, 1, $1);
+    }
+  ;
+
+statement_list:
+    statement {
+      $$ = NODE(STATEMENT_LIST, NULL, 1, $1);
+    }
+  | statement_list statement {
+      $$ = NODE(STATEMENT_LIST, NULL, 2, $1, $2);
+    }
+  ;
+
+print_list: 
+    print_item {
+      $$ = NODE(PRINT_LIST, NULL, 1, $1);
+    }
+  | print_list ',' print_item {
+      $$ = NODE(PRINT_LIST, NULL, 2, $1, $3);
+    }
+  ;
+
+expression_list:
+    expression {
+      $$ = NODE(EXPRESSION_LIST, NULL, 1, $1);
+    }
+  | expression_list ',' expression {
+      $$ = NODE(EXPRESSION_LIST, NULL, 2, $1, $3);
+    }
+  ;
+
+variable_list:
+    identifier {
+      $$ = NODE(VARIABLE_LIST, NULL, 1, $1);
+    }
+  | variable_list ',' identifier {
+      $$ = NODE(VARIABLE_LIST, NULL, 2, $1, $3);
+    }
+  ;
+
+argument_list:
+    expression_list {
+      $$ = NODE(ARGUMENT_LIST, NULL, 1, $1);
+    }
+  | /* epsilon */ {
+      $$ = NODE(ARGUMENT_LIST, NULL, 0);
+    }
+  ;
+
+parameter_list:
+    variable_list {
+      $$ = NODE(PARAMETER_LIST, NULL, 1, $1);
+    }
+  | /* epsilon */ {
+      $$ = NODE(PARAMETER_LIST, NULL, 0);
+    }
+  ;
+
+declaration_list:
+    declaration {
+      $$ = NODE(DECLARATION_LIST, NULL, 1, $1);
+    }
+  | declaration_list declaration {
+      $$ = NODE(DECLARATION_LIST, NULL, 2, $1, $2);
+    }
+  ;
+
+function:
+    FUNC identifier '(' parameter_list ')' statement  {
+      $$ = NODE(FUNCTION, NULL, 3, $2, $4, $6);
+    }
+  ;
+
+statement:
+    assignment_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | return_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | print_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | if_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | while_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | null_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | block {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  ;
+
+block:
+    OPENBLOCK declaration_list statement_list CLOSEBLOCK {
+      $$ = NODE(BLOCK, NULL, 2, $2, $3);
+    }
+  | OPENBLOCK statement_list CLOSEBLOCK {
+      $$ = NODE(BLOCK, NULL, 1, $2);
+    }
+  ;
+
+assignment_statement:
+    identifier ':' '=' expression {
+      $$ = NODE(ASSIGNMENT_STATEMENT, NULL, 2, $1, $4);
+    }
+  | identifier '+' '=' expression {
+      $$ = NODE(ADD_STATEMENT, NULL, 2, $1, $4);
+    }
+  | identifier '-' '=' expression {
+      $$ = NODE(SUBTRACT_STATEMENT, NULL, 2, $1, $4);
+    }
+  | identifier '*' '=' expression {
+      $$ = NODE(MULTIPLY_STATEMENT, NULL, 2, $1, $4);
+    }
+  | identifier '/' '=' expression {
+      $$ = NODE(DIVIDE_STATEMENT, NULL, 2, $1, $4);
+    }
+  ;
+
+return_statement:
+    RETURN expression {
+      $$ = NODE(RETURN_STATEMENT, NULL, 1, $2);
+    }
+  ;
+
+print_statement:
+    PRINT print_list {
+      $$ = NODE(PRINT_STATEMENT, NULL, 1, $2);
+    }
+  ;
+
+null_statement:
+    CONTINUE {
+      $$ = NODE(NULL_STATEMENT, NULL, 0);
+    }
+  ;
+
+if_statement:
+    IF relation THEN statement {
+      $$ = NODE(IF_STATEMENT, NULL, 2, $2, $4);
+    }
+  | IF relation THEN statement ELSE statement {
+      $$ = NODE(IF_STATEMENT, NULL, 3, $2, $4, $6);
+    }
+  ;
+
+while_statement:
+    WHILE relation DO statement {
+      $$ = NODE(WHILE_STATEMENT, NULL, 2, $2, $4);
+    }
+  ;
+
+relation:
+    expression '=' expression {
+      $$ = NODE(RELATION, strdup("="), 2, $1, $3);
+    }
+  | expression '<' expression {
+      $$ = NODE(RELATION, strdup("<"), 2, $1, $3);
+    }
+  | expression '>' expression {
+      $$ = NODE(RELATION, strdup(">"), 2, $1, $3);
+    }
+  ;
+
+expression:
+    expression '|' expression {
+      $$ = NODE(EXPRESSION, strdup("|"), 2, $1, $3);
+    }
+  | expression '^' expression {
+      $$ = NODE(EXPRESSION, strdup("^"), 2, $1, $3);
+    }
+  | expression '&' expression {
+      $$ = NODE(EXPRESSION, strdup("&"), 2, $1, $3);
+    }
+  | expression '+' expression {
+      $$ = NODE(EXPRESSION, strdup("+"), 2, $1, $3);
+    }
+  | expression '-' expression {
+      $$ = NODE(EXPRESSION, strdup("-"), 2, $1, $3);
+    }
+  | expression '*' expression {
+      $$ = NODE(EXPRESSION, strdup("*"), 2, $1, $3);
+    }
+  | expression '/' expression {
+      $$ = NODE(EXPRESSION, strdup("/"), 2, $1, $3);
+    }
+  | '-' expression %prec UMINUS {
+      $$ = NODE(EXPRESSION, strdup("-"), 1, $2);
+    }
+  | '~' expression {
+      $$ = NODE(EXPRESSION, strdup("~"), 1, $2);
+    }
+  | '(' expression ')' {
+      $$ = NODE(EXPRESSION, NULL /*strdup("group")*/, 1, $2);
+    }
+  | number {
+      $$ = NODE(EXPRESSION, NULL /*strdup("number")*/, 1, $1);
+    }
+  | identifier {
+      $$ = NODE(EXPRESSION, NULL /*strdup("identifier")*/, 1, $1);
+    }
+  | identifier '(' argument_list ')' {
+      $$ = NODE(EXPRESSION, /*NULL*/ strdup("function_call"), 2, $1, $3);
+    }
+  ;
+
+declaration: 
+    VAR variable_list {
+      $$ = NODE(DECLARATION, NULL, 1, $2);
+    }
+  ;
+
+print_item:
+    expression {
+      $$ = NODE(PRINT_ITEM, NULL, 1, $1);
+    }
+  | string {
+      $$ = NODE(PRINT_ITEM, NULL, 1, $1);
+    }
+  ;
+
+identifier:
+    IDENTIFIER {
+      $$ = NODE(IDENTIFIER_DATA, strdup(yytext), 0); // Zero children
+    }
+  ;
+
+number:
+    NUMBER {
+      uint64_t* p_number = malloc(sizeof(uint64_t));
+      *p_number = strtol(yytext, NULL, 10);
+      $$ = NODE(NUMBER_DATA, p_number, 0); // Zero children
+    }
+  ;
+
+string:
+    STRING {
+      $$ = NODE(STRING_DATA, strdup(yytext), 0); // Zero children
+    }
+  ;
+
+%%
+
+int
+yyerror ( const char *error )
+{
+    fprintf ( stderr, "%s on line %d\n", error, yylineno );
+    exit ( EXIT_FAILURE );
+}

exercises/05/vslc/src/scanner.l

@@ -0,0 +1,30 @@
+%{
+#include <vslc.h>
+%}
+%option noyywrap
+%option array
+%option yylineno
+
+WHITESPACE [\ \t\v\r\n]
+COMMENT \/\/[^\n]+
+QUOTED \"([^\"\n]|\\\")*\"
+%%
+{WHITESPACE}+           { /* Eliminate whitespace */ }
+{COMMENT}               { /* Eliminate comments */ }
+func                    { return FUNC; }
+print                   { return PRINT; }
+return                  { return RETURN; }
+continue                { return CONTINUE; }
+if                      { return IF; }
+then                    { return THEN; }
+else                    { return ELSE; }
+while                   { return WHILE; }
+do                      { return DO; }
+begin                   { return OPENBLOCK; }
+end                     { return CLOSEBLOCK; }
+var                     { return VAR; }
+[0-9]+                  { return NUMBER; }
+[A-Za-z_][0-9A-Za-z_]*  { return IDENTIFIER; }
+{QUOTED}                { return STRING; }
+.                       { return yytext[0]; }
+%%

exercises/05/vslc/src/tlhash.c

@@ -0,0 +1,281 @@
+#include <string.h>
+#include <stdlib.h>
+#include <stdint.h>
+
+#include <tlhash.h>
+
+/*********************************************************************
+ * Declarations of the utility functions for obtaining hashes, found *
+ * at the bottom of this file.                                       *
+ *********************************************************************/
+
+/* Little-endian, for x86-s */
+#define CRC32_IEEE802_3 (0xedb88320)
+static const uint32_t crc32_ieee802_3[256];
+static const uint32_t *crc_table = (uint32_t *)crc32_ieee802_3;
+
+static uint32_t crc32 ( void *input, size_t length );
+
+
+/********************************
+ * External interface functions *
+ ********************************/
+
+
+/* Initializer 
+ * Returns
+ *  ENOMEM - if allocation of table entries fails.
+ */
+int
+tlhash_init ( tlhash_t *tab, size_t n_buckets )
+{
+    size_t i;
+    tab->n_buckets = n_buckets;
+    tab->size = 0;
+    tab->buckets = (tlhash_element_t **) calloc (
+        n_buckets, sizeof(tlhash_element_t *)
+    );
+    if ( tab->buckets == NULL )
+        return TLHASH_ENOMEM;
+    for ( i=0; i<n_buckets; i++ )
+        tab->buckets[i] = NULL;
+    return TLHASH_SUCCESS;
+}
+
+
+/* Finalizer
+ * Returns
+ *  ENOENT - if there is no table to free.
+ */
+int
+tlhash_finalize ( tlhash_t *tab )
+{
+    size_t i;
+    if ( tab == NULL )
+        return TLHASH_ENOENT;
+    for ( i=0; i<tab->n_buckets; i++ )
+    {
+        tlhash_element_t *element = tab->buckets[i], *next;
+        while ( element != NULL )
+        {
+            next = element->next;
+            free ( element->key );
+            free ( element );
+            tab->size -= 1;
+            element = next;
+        }
+    }
+
+    free ( tab->buckets );
+    return TLHASH_SUCCESS;
+}
+
+
+/* Insert - find hash value, modulate over buckets, append to linked list
+ * Returns
+ *  EEXIST - if an element is already indexed by this key
+ *  ENOMEM - if allocation of element or key copy fails
+ */
+int
+tlhash_insert (
+    tlhash_t *tab, void *key, size_t key_length, void *value
+)
+{
+    void *test_entry;
+    int test = tlhash_lookup ( tab, key, key_length, &test_entry );
+    if ( test != TLHASH_ENOENT )
+        return TLHASH_EEXIST;
+    uint32_t hash = crc32 ( key, key_length );
+    size_t bucket = hash % tab->n_buckets;
+    tlhash_element_t *element = malloc ( sizeof(tlhash_element_t) );
+    if ( element == NULL )
+        return TLHASH_ENOMEM;
+    void *key_copy = malloc ( key_length );
+    if ( key_copy == NULL )
+    {
+        free ( element );
+        return TLHASH_ENOMEM;
+    }
+    memcpy ( key_copy, key, key_length );
+    element->key         = key_copy;
+    element->key_length  = key_length;
+    element->value       = value;
+    element->next        = tab->buckets[bucket];
+    tab->buckets[bucket] = element;
+    tab->size += 1;
+    return TLHASH_SUCCESS;
+}
+
+
+/* Lookup - find hash value, modulate over buckets, search linked list
+ * Returns
+ *  ENOENT - if no element is indexed by this key
+ */
+int
+tlhash_lookup (
+    tlhash_t *tab, void *key, size_t key_length, void **value
+)
+{
+    uint32_t hash = crc32 ( key, key_length );
+    size_t bucket = hash % tab->n_buckets;
+    tlhash_element_t *el = tab->buckets[bucket];
+
+    *value = NULL;
+    while ( el != NULL )
+    {
+        if ( el->key_length == key_length && ! memcmp(el->key,key,key_length) )
+        {
+            *value = el->value;
+            break;
+        }
+        el = el->next;
+    }
+    if ( el != NULL )
+        return TLHASH_SUCCESS;
+    else
+        return TLHASH_ENOENT;
+}
+
+
+/* Removal - find hash value, modulate over buckets, delete entry
+ * Returns
+ *  ENOENT - no such element to remove was found.
+ */
+int
+tlhash_remove ( tlhash_t *tab, void *key, size_t key_length )
+{
+    uint32_t hash = crc32 ( key, key_length );
+    size_t bucket = hash % tab->n_buckets;
+    tlhash_element_t *el = tab->buckets[bucket], *prev = NULL;
+
+    while ( el != NULL )
+    {
+        if ( el->key_length == key_length && ! memcmp(el->key,key,key_length) )
+        {
+            /* We have a match. */
+            if ( prev != NULL ) /* Remove from list if it's not the head */
+                prev->next = (void *)el->next;
+            else                /* Substitute it if it IS the head */
+                tab->buckets[bucket] = el->next;
+            /* Free the container and key copy allocated by this lib */
+            free ( el->key );
+            free ( el );
+            break;
+        }
+        prev = el;
+        el = el->next;
+    }
+    if ( el == NULL )
+        return TLHASH_ENOENT;
+    else
+    {
+        tab->size -= 1;
+        return TLHASH_SUCCESS;
+    }
+}
+
+
+size_t
+tlhash_size ( tlhash_t *tab )
+{
+    return tab->size;
+}
+
+
+void
+tlhash_keys ( tlhash_t *tab, void **keys )
+{
+    size_t b, i = 0;
+    for ( b=0; b<tab->n_buckets; b++ )
+    {
+        tlhash_element_t *el = tab->buckets[b];
+        while ( el != NULL )
+        {
+            keys[i] = el->key;
+            i += 1;
+            el = el->next;
+        }
+    }
+}
+
+
+void
+tlhash_values ( tlhash_t *tab, void **values )
+{
+    size_t b, i = 0;
+    for ( b=0; b<tab->n_buckets; b++ )
+    {
+        tlhash_element_t *el = tab->buckets[b];
+        while ( el != NULL )
+        {
+            values[i] = el->value;
+            i += 1;
+            el = el->next;
+        }
+    }
+}
+
+
+/***************************************
+ * Hashing function and IEEE data blob *
+ ***************************************/
+
+
+static uint32_t
+crc32 ( void *input, size_t length )
+{
+    const uint8_t *data = (uint8_t *)input;
+    size_t i = 0;
+    uint32_t hash = 0xFFFFFFFF;
+    for ( i = 0; i<length; i++ )
+        hash = (hash>>8) ^ crc_table [ data[i] ^ (uint8_t)hash ];
+    return (hash^0xFFFFFFFF);
+}
+
+
+static const uint32_t
+crc32_ieee802_3[256] = {
+    0x00000000, 0x77073096, 0xee0e612c, 0x990951ba, 0x076dc419, 0x706af48f,
+    0xe963a535, 0x9e6495a3, 0x0edb8832, 0x79dcb8a4, 0xe0d5e91e, 0x97d2d988,
+    0x09b64c2b, 0x7eb17cbd, 0xe7b82d07, 0x90bf1d91, 0x1db71064, 0x6ab020f2,
+    0xf3b97148, 0x84be41de, 0x1adad47d, 0x6ddde4eb, 0xf4d4b551, 0x83d385c7,
+    0x136c9856, 0x646ba8c0, 0xfd62f97a, 0x8a65c9ec, 0x14015c4f, 0x63066cd9,
+    0xfa0f3d63, 0x8d080df5, 0x3b6e20c8, 0x4c69105e, 0xd56041e4, 0xa2677172,
+    0x3c03e4d1, 0x4b04d447, 0xd20d85fd, 0xa50ab56b, 0x35b5a8fa, 0x42b2986c,
+    0xdbbbc9d6, 0xacbcf940, 0x32d86ce3, 0x45df5c75, 0xdcd60dcf, 0xabd13d59,
+    0x26d930ac, 0x51de003a, 0xc8d75180, 0xbfd06116, 0x21b4f4b5, 0x56b3c423,
+    0xcfba9599, 0xb8bda50f, 0x2802b89e, 0x5f058808, 0xc60cd9b2, 0xb10be924,
+    0x2f6f7c87, 0x58684c11, 0xc1611dab, 0xb6662d3d, 0x76dc4190, 0x01db7106,
+    0x98d220bc, 0xefd5102a, 0x71b18589, 0x06b6b51f, 0x9fbfe4a5, 0xe8b8d433,
+    0x7807c9a2, 0x0f00f934, 0x9609a88e, 0xe10e9818, 0x7f6a0dbb, 0x086d3d2d,
+    0x91646c97, 0xe6635c01, 0x6b6b51f4, 0x1c6c6162, 0x856530d8, 0xf262004e,
+    0x6c0695ed, 0x1b01a57b, 0x8208f4c1, 0xf50fc457, 0x65b0d9c6, 0x12b7e950,
+    0x8bbeb8ea, 0xfcb9887c, 0x62dd1ddf, 0x15da2d49, 0x8cd37cf3, 0xfbd44c65,
+    0x4db26158, 0x3ab551ce, 0xa3bc0074, 0xd4bb30e2, 0x4adfa541, 0x3dd895d7,
+    0xa4d1c46d, 0xd3d6f4fb, 0x4369e96a, 0x346ed9fc, 0xad678846, 0xda60b8d0,
+    0x44042d73, 0x33031de5, 0xaa0a4c5f, 0xdd0d7cc9, 0x5005713c, 0x270241aa,
+    0xbe0b1010, 0xc90c2086, 0x5768b525, 0x206f85b3, 0xb966d409, 0xce61e49f,
+    0x5edef90e, 0x29d9c998, 0xb0d09822, 0xc7d7a8b4, 0x59b33d17, 0x2eb40d81,
+    0xb7bd5c3b, 0xc0ba6cad, 0xedb88320, 0x9abfb3b6, 0x03b6e20c, 0x74b1d29a,
+    0xead54739, 0x9dd277af, 0x04db2615, 0x73dc1683, 0xe3630b12, 0x94643b84,
+    0x0d6d6a3e, 0x7a6a5aa8, 0xe40ecf0b, 0x9309ff9d, 0x0a00ae27, 0x7d079eb1,
+    0xf00f9344, 0x8708a3d2, 0x1e01f268, 0x6906c2fe, 0xf762575d, 0x806567cb,
+    0x196c3671, 0x6e6b06e7, 0xfed41b76, 0x89d32be0, 0x10da7a5a, 0x67dd4acc,
+    0xf9b9df6f, 0x8ebeeff9, 0x17b7be43, 0x60b08ed5, 0xd6d6a3e8, 0xa1d1937e,
+    0x38d8c2c4, 0x4fdff252, 0xd1bb67f1, 0xa6bc5767, 0x3fb506dd, 0x48b2364b,
+    0xd80d2bda, 0xaf0a1b4c, 0x36034af6, 0x41047a60, 0xdf60efc3, 0xa867df55,
+    0x316e8eef, 0x4669be79, 0xcb61b38c, 0xbc66831a, 0x256fd2a0, 0x5268e236,
+    0xcc0c7795, 0xbb0b4703, 0x220216b9, 0x5505262f, 0xc5ba3bbe, 0xb2bd0b28,
+    0x2bb45a92, 0x5cb36a04, 0xc2d7ffa7, 0xb5d0cf31, 0x2cd99e8b, 0x5bdeae1d,
+    0x9b64c2b0, 0xec63f226, 0x756aa39c, 0x026d930a, 0x9c0906a9, 0xeb0e363f,
+    0x72076785, 0x05005713, 0x95bf4a82, 0xe2b87a14, 0x7bb12bae, 0x0cb61b38,
+    0x92d28e9b, 0xe5d5be0d, 0x7cdcefb7, 0x0bdbdf21, 0x86d3d2d4, 0xf1d4e242,
+    0x68ddb3f8, 0x1fda836e, 0x81be16cd, 0xf6b9265b, 0x6fb077e1, 0x18b74777,
+    0x88085ae6, 0xff0f6a70, 0x66063bca, 0x11010b5c, 0x8f659eff, 0xf862ae69,
+    0x616bffd3, 0x166ccf45, 0xa00ae278, 0xd70dd2ee, 0x4e048354, 0x3903b3c2,
+    0xa7672661, 0xd06016f7, 0x4969474d, 0x3e6e77db, 0xaed16a4a, 0xd9d65adc,
+    0x40df0b66, 0x37d83bf0, 0xa9bcae53, 0xdebb9ec5, 0x47b2cf7f, 0x30b5ffe9,
+    0xbdbdf21c, 0xcabac28a, 0x53b39330, 0x24b4a3a6, 0xbad03605, 0xcdd70693,
+    0x54de5729, 0x23d967bf, 0xb3667a2e, 0xc4614ab8, 0x5d681b02, 0x2a6f2b94,
+    0xb40bbe37, 0xc30c8ea1, 0x5a05df1b, 0x2d02ef8d
+};

exercises/05/vslc/src/tree.c

@@ -0,0 +1,400 @@
+#include <vslc.h>
+
+static void node_print ( node_t *root, int nesting );
+static void simplify_tree ( node_t **simplified, node_t *root );
+static void node_finalize ( node_t *discard );
+
+typedef struct stem_t *stem;
+struct stem_t { const char *str; stem next; };
+static void
+tree_print(node_t* root, stem head);
+
+
+static void destroy_subtree ( node_t *discard );
+
+static void prune_children(node_t **simplified, node_t *root);
+static void resolve_constant_expressions(node_t **simplified, node_t *root);
+static void flatten(node_t **simplified, node_t *root);
+
+
+/* External interface */
+void
+destroy_syntax_tree ( void )
+{
+    destroy_subtree ( root );
+}
+
+
+void
+simplify_syntax_tree ( void )
+{
+    simplify_tree ( &root, root );
+}
+
+
+extern bool new_print_style;
+void
+print_syntax_tree ( void )
+{
+    if (new_print_style)
+        tree_print ( root, 0 );
+    // Old tree printing
+    else
+        node_print ( root, 0 );
+}
+
+
+// Changed so it returns the pointer to the new node, can be used as before, but makes the parser file cleaner
+node_t*
+node_init (node_t *nd, node_index_t type, void *data, uint64_t n_children, ...)
+{
+    va_list child_list;
+    *nd = (node_t) {
+        .type = type,
+        .data = data,
+        .entry = NULL,
+        .n_children = n_children,
+        .children = (node_t **) malloc ( n_children * sizeof(node_t *) )
+    };
+    va_start ( child_list, n_children );
+    for ( uint64_t i=0; i<n_children; i++ )
+        nd->children[i] = va_arg ( child_list, node_t * );
+    va_end ( child_list );
+
+    return nd;
+}
+
+
+static void
+tree_print(node_t* root, stem head) 
+{
+    static const char *sdown = " │", *slast = " └", *snone = "  ";
+    struct stem_t col = {0, 0}, *tail;
+
+    // Print stems of branches coming further down
+    for (tail = head; tail; tail = tail->next) {
+        if (!tail->next) {
+            if (!strcmp(sdown, tail->str))
+                printf(" ├");
+            else 
+                printf("%s", tail->str);
+            break;
+        }
+        printf("%s", tail->str);
+    }
+    
+    if (root == NULL) {
+        // Secure against null pointers sent as root
+        printf("─(nil)\n");
+        return;
+    }
+    printf("─%s", node_string[root->type]);
+    if ( root->type == IDENTIFIER_DATA  ||
+         root->type == STRING_DATA      ||
+         root->type == EXPRESSION       ||
+         root->type == RELATION)  
+        printf("(%s)", (char *) root->data);
+    else if (root->type == NUMBER_DATA)
+        printf("(%ld)", *((int64_t *)root->data));
+    putchar('\n');
+ 
+    if (!root->n_children) return;
+ 
+    if (tail && tail->str == slast)
+        tail->str = snone;
+ 
+    if (!tail)  tail = head = &col;
+    else        tail->next = &col;
+
+    for ( int64_t i=0; i < root->n_children; i++ ) {
+        col.str = root->n_children - i - 1 ? sdown : slast;
+        tree_print(root->children[i], head);
+    }
+    tail->next = 0;
+}
+
+/* Internal choices */
+static void
+node_print ( node_t *root, int nesting )
+{
+    if ( root != NULL )
+    {
+        printf ( "%*c%s", nesting, ' ', node_string[root->type] );
+        if ( root->type == IDENTIFIER_DATA ||
+             root->type == STRING_DATA ||
+             root->type == EXPRESSION ) 
+            printf ( "(%s)", (char *) root->data );
+        else if ( root->type == NUMBER_DATA )
+            printf ( "(%ld)", *((int64_t *)root->data) );
+        putchar ( '\n' );
+        for ( int64_t i=0; i<root->n_children; i++ )
+            node_print ( root->children[i], nesting+1 );
+    }
+    else
+        printf ( "%*c%p\n", nesting, ' ', root );
+}
+
+
+static void
+node_finalize ( node_t *discard )
+{
+    if ( discard != NULL )
+    {
+        free ( discard->data );
+        free ( discard->children );
+        free ( discard );
+    }
+}
+
+
+static void
+destroy_subtree ( node_t *discard )
+{
+    if ( discard != NULL )
+    {
+        for ( uint64_t i=0; i<discard->n_children; i++ )
+            destroy_subtree ( discard->children[i] );
+        node_finalize ( discard );
+    }
+}
+
+
+static void
+flatten(node_t **simplified, node_t *root)
+{
+    /* This will flatten left-expanded lists */
+    if (!root)
+        return;
+
+    /* Do this recursivly */
+    for (int i = 0; i < root->n_children; i++)
+        flatten(&root->children[i], root->children[i]);
+
+    node_t **new_children, *result = root;
+    switch (root->type)
+    {
+    case GLOBAL_LIST:
+    case STATEMENT_LIST:
+    case PRINT_LIST:
+    case EXPRESSION_LIST:
+    case VARIABLE_LIST:
+    case DECLARATION_LIST:
+        // Check if node have more than two children
+        if (root->n_children < 2)
+            break;
+        
+        result = root->children[0];
+        result->n_children++;
+
+        // Realloc the array of children to the new size 
+        if (!(new_children = realloc(result->children, result->n_children * sizeof(node_t*))))
+            break;
+        // if successs, insert the new array
+        result->children = new_children;
+
+        // Insert child at the end
+        result->children[result->n_children - 1] = root->children[1];
+
+        node_finalize(root);
+        break;
+    }
+
+    *simplified = result;
+}
+
+
+static void
+prune_children(node_t **simplified, node_t *root) 
+{
+    if (!root)
+        return;
+
+    /* Do this recursivly */
+    for (int i = 0; i < root->n_children; i++)
+        prune_children(&root->children[i], root->children[i]);
+
+    node_t *result = root;
+    switch (root->type)
+    {
+    case PROGRAM:
+    case GLOBAL:
+    //case ARGUMENT_LIST:     // For this to work, need to change order of operations
+    //case PARAMETER_LIST:    // For this to work, need to change order of operations
+    //case VARIABLE_LIST:
+    //case EXPRESSION_LIST:
+    case DECLARATION:
+    case STATEMENT:
+    case PRINT_ITEM:
+    case PRINT_STATEMENT:
+        result = root->children[0];
+
+        // The print_statement only contains a print_list, still need a print_statement.
+        if (root->type == PRINT_STATEMENT)
+            result->type = PRINT_STATEMENT;
+        
+        node_finalize(root);
+        break;
+    }
+
+    *simplified = result;
+}
+
+
+static void
+resolve_constant_expressions(node_t **simplified, node_t *root)
+{
+    if (!root)
+        return;
+
+    /* Do this recursivly */
+    for (int i = 0; i < root->n_children; i++)
+        resolve_constant_expressions(&root->children[i], root->children[i]);
+    
+    if (root->type != EXPRESSION)
+        return;
+    
+    node_t *result = root;
+    
+    switch (root->n_children)
+    {
+    case 1:
+        result = root->children[0];
+        
+        if (root->data && 
+            result->type == NUMBER_DATA && 
+            result->data)
+        {
+            switch (*((char*)root->data))
+            {
+            case '-':
+                *((int64_t*)result->data) *= -1;
+                break;
+            case '~':
+                *((int64_t*)result->data) = ~*((int64_t*)result->data);
+                break;
+            }
+            node_finalize(root);
+        }
+        else if (!root->data)
+            node_finalize(root);
+        else
+            result = root;
+
+        break;
+    
+    case 2:
+        // Both children needs to be numbers to resolve constants
+        if (root->children[0]->type == NUMBER_DATA && 
+            root->children[1]->type == NUMBER_DATA)
+        {
+            // Check if children does not contain null pointers
+            if (!root->children[0]->data)
+                break;
+            if (!root->children[1]->data)
+                break;
+
+            // Check if data field is not null pointer
+            if (!root->data)
+                break;
+            
+            result = root->children[0];
+            int64_t
+                *lhs = result->data,
+                *rhs = root->children[1]->data;
+
+            switch (*(char*)root->data)
+            {
+            /* Assignments */
+            case '|': *lhs |= *rhs; break;
+            case '^': *lhs ^= *rhs; break;
+            case '&': *lhs &= *rhs; break;
+            case '+': *lhs += *rhs; break;
+            case '-': *lhs -= *rhs; break;
+            case '*': *lhs *= *rhs; break;
+            case '/': *lhs /= *rhs; break;
+            }
+
+            node_finalize(root->children[1]);
+            node_finalize(root);
+        }            
+        
+        
+        break;
+    }
+    *simplified = result;
+}
+
+
+static void
+resolve_constant_relations( node_t** simplified, node_t* root)
+{
+    if (!root)
+        return;
+
+    /* Do this recursivly */
+    for (int i = 0; i < root->n_children; i++)
+        resolve_constant_relations(&root->children[i], root->children[i]);
+    
+    if (root->type != RELATION)//|| root->type != RELATION)
+        return;
+
+    node_t *result = root;
+
+    if (root->n_children != 2)
+        return;
+
+    // Both children must be constant numbers
+    if (root->children[0]->type != NUMBER_DATA ||
+        root->children[1]->type != NUMBER_DATA)
+        return;
+
+    // Check if children does not contain null pointers
+    if (!root->children[0]->data)
+        return;
+    if (!root->children[1]->data)
+        return;
+
+    // Check if data field is not null pointer
+    if (!root->data)
+        return;
+
+    result = root->children[0];
+    int64_t
+        *lhs = result->data,
+        *rhs = root->children[1]->data;
+
+    switch (*(char*)root->data)
+    {
+    /* Relations */
+    case '=': *lhs = (*lhs == *rhs); break;
+    case '<': *lhs = (*lhs <  *rhs); break;
+    case '>': *lhs = (*lhs >  *rhs); break;
+    }
+
+    node_finalize(root->children[1]);
+    node_finalize(root);
+
+    *simplified = result;
+}
+
+
+static void
+simplify_tree ( node_t **simplified, node_t *root )
+{
+    if (!root) 
+        return;
+
+    /*  
+        Each of  the functions do their operations recursivly.
+        This opens up for a lot more flexibility, like removing 
+        variable list after it is flatten
+    */
+    flatten(&root, root);
+    prune_children(&root, root);
+    resolve_constant_expressions(&root, root);
+
+    // The following is experimental, will resolve the constant relations
+    resolve_constant_relations(&root, root);
+
+    *simplified = root;
+}

exercises/05/vslc/src/vslc.c

@@ -0,0 +1,83 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <getopt.h>
+#include <vslc.h>
+
+
+/* Global state */
+
+node_t *root;                   // Syntax tree
+tlhash_t *global_names;         // Symbol table
+char **string_list;             // List of strings in the source
+size_t n_string_list    = 8;    // Initial string list capacity (grow on demand)
+size_t stringc          = 0;    // Initial string count
+
+/* Command line option parsing for the main function */
+static void options ( int argc, char **argv );
+bool
+    print_full_tree             = false,
+    print_simplified_tree       = false,
+    print_symbol_table_contents = false,
+    print_generated_program     = true,
+    new_print_style             = true;
+
+
+/* Entry point */
+int
+main ( int argc, char **argv )
+{
+    options ( argc, argv );
+
+    yyparse();                  // Generated from grammar/bison, constructs syntax tree
+    yylex_destroy();            // Free heap used by flex
+
+    if ( print_full_tree )
+        print_syntax_tree ();
+    simplify_syntax_tree ();    // In tree.c
+    if ( print_simplified_tree )
+        print_syntax_tree ();
+
+    create_symbol_table ();     // In ir.c
+    if ( print_symbol_table_contents )
+        print_symbol_table();
+
+    if ( print_generated_program )
+        generate_program ();    // In generator.c
+
+
+    destroy_syntax_tree ();     // In tree.c
+    destroy_symbol_table ();    // In ir.c
+}
+
+
+static const char *usage =
+"Command line options\n"
+"\t-h\tOutput this text and halt\n"
+"\t-t\tOutput the full syntax tree\n"
+"\t-T\tOutput the simplified syntax tree\n"
+"\t-s\tOutput the symbol table contents\n"
+"\t-q\tQuiet: suppress output from the code generator\n"
+"\t-u\tDo not use print style more like the tree command\n";
+
+
+static void
+options ( int argc, char **argv )
+{
+    int o;
+    while ( (o=getopt(argc,argv,"htTsqu")) != -1 )
+    {
+        switch ( o )
+        {
+            case 'h':
+                printf ( "%s:\n%s", argv[0], usage );
+                exit ( EXIT_FAILURE );
+                break;
+            case 't':   print_full_tree = true;             break;
+            case 'T':   print_simplified_tree = true;       break;
+            case 's':   print_symbol_table_contents = true; break;
+            case 'q':   print_generated_program = false;    break;
+            case 'u':   new_print_style = false;            break;
+        }
+    }
+}
+

exercises/05/vslc/vsl_programs/Makefile

@@ -0,0 +1,42 @@
+VSLC := ../src/vslc
+AS := gcc
+
+# This updated makefile calls the relevant print for each assignment
+# The added target compile will attempt to compile all vsl source files,
+# while the compile-ps5 target attempts to compile a selection of sources
+# that should work after ps5 is done.
+
+# Call `vslc -h` to see the available flags, and call `vslc [flags] < file.vsl`
+# to compile a single file.
+
+PS2_EXAMPLES := $(patsubst ps2-parser/%.vsl, ps2-parser/%.ast, $(wildcard ps2-parser/*.vsl))
+PS3_EXAMPLES := $(patsubst ps3-simplify/%.vsl, ps3-simplify/%.sast, $(wildcard ps3-simplify/*.vsl))
+PS4_EXAMPLES := $(patsubst ps4-symtab/%.vsl, ps4-symtab/%.sym, $(wildcard ps4-symtab/*.vsl))
+PS5_EXAMPLES := $(patsubst ps5-codegen1/%.vsl, ps5-codegen1/%.S, $(wildcard ps5-codegen1/*.vsl))
+
+PS5_OBJECTS := $(PS5_EXAMPLES:.S=.bin)
+OBJECTS := $(PS5_OBJECTS) $(PS4_EXAMPLES:.sym=.bin) $(PS3_EXAMPLES:.sast=.bin) $(PS2_EXAMPLES:.ast=.bin)
+
+all: clean $(PS2_EXAMPLES) $(PS3_EXAMPLES) $(PS4_EXAMPLES) $(PS5_EXAMPLES)
+ps2: $(PS2_EXAMPLES)
+ps3: $(PS3_EXAMPLES)
+ps4: $(PS4_EXAMPLES)
+ps5: $(PS5_EXAMPLES)
+
+%.ast: %.vsl
+	$(VSLC) -t -q < $^ > $@ 2> $@
+%.sast: %.vsl
+	$(VSLC) -T -q < $^ > $@ 2> $@
+%.sym: %.vsl
+	$(VSLC) -s -q < $^ > $@ 2> $@
+%.S: %.vsl
+	$(VSLC) < $^ > $@
+# This target is only tested on x86-linux
+%.bin: %.S
+	$(AS) -no-pie -o $@ $^
+
+ps5-compile: $(PS5_OBJECTS)
+compile: $(OBJECTS)
+
+clean:
+	-rm -r */*.ast */*.sast */*.sym */*.bin */*.s */*.S

exercises/05/vslc/vsl_programs/ps2-parser/assignments.vsl

@@ -0,0 +1,15 @@
+// checking that comments are ignored
+
+// This program checks the assignment operators
+
+func main()
+begin
+    var a
+    a := 3
+    a += 1
+    a /= 2
+    a *= 32
+    a -= 2
+    print a
+    return 0
+end

exercises/05/vslc/vsl_programs/ps2-parser/function_call.vsl

@@ -0,0 +1,10 @@
+func add(a, b) begin
+    a := -a
+    return a + b
+end
+
+func main() 
+begin
+    print add(40, 2)
+    return 0
+end

exercises/05/vslc/vsl_programs/ps2-parser/helloworld.vsl

@@ -0,0 +1,4 @@
+func helloworld() begin
+    print "Hello, World!"
+    return 0
+end

exercises/05/vslc/vsl_programs/ps2-parser/if_else.vsl

@@ -0,0 +1,26 @@
+func main()
+begin
+    var a, b, c, d
+    c := 1
+    a := 3
+    b := a + c // 4
+    d := a * 100 + 50
+    print "a", a
+    print "b", b
+    print "c", c
+    print "d", d
+    if a = 14 then
+        print 1, "N", d / 5 + a, "RPR", a, "TERS "
+    else
+        print "COMP", c, "L", "ERS "
+    
+    print b, "R", a, " "
+
+    if a < b then
+        if d > 42 then
+            print b, "W", d, "ME"
+        else
+            print "L", b, "M", c
+    // A dangling else, what could go wrong?
+    return 0
+end

exercises/05/vslc/vsl_programs/ps2-parser/variables.vsl

@@ -0,0 +1,21 @@
+
+var global_var
+
+func my_func(param)
+begin
+    var local_var, local_var2
+    local_var := 1
+    return local_var
+end
+
+var glob1, glob2
+
+func main()
+begin
+    var main_local_var
+    begin
+        var main_local_nested_var
+        main_local_nested_var := main_local_var + my_func(1)
+    end
+    return 0
+end

exercises/05/vslc/vsl_programs/ps2-parser/while.vsl

@@ -0,0 +1,11 @@
+// check parsing of do-while loop
+
+func main()
+begin   
+    var i
+    i := 2
+    while i < 42 do
+        i := i * i
+    print i
+    return 0
+end

exercises/05/vslc/vsl_programs/ps3-simplify/constants.vsl

@@ -0,0 +1,10 @@
+
+func main() begin
+    var a, b
+    a := 1 + 2 + 4 + 5 + 6 + 7 + 8 + 9
+    b := (10 + 10 * 4) * (2 + 2 * (1 + 1)) / 10 + 2 * 5 + 6 / 3
+
+    if a = b then
+        print "The answer is", b
+    return 0
+end

exercises/05/vslc/vsl_programs/ps3-simplify/function_call.vsl

@@ -0,0 +1,9 @@
+func add(a, b, c) begin
+    return a + b
+end
+
+func main() 
+begin
+    print add(40, 2, 4)
+    return 0
+end

exercises/05/vslc/vsl_programs/ps3-simplify/globals.vsl

@@ -0,0 +1,12 @@
+var global_var0, global_var1
+
+func my_func(param0, param1) begin
+    var a
+    return 0
+end
+
+func main() begin
+    var a
+    print "a string"
+    return 0
+end

exercises/05/vslc/vsl_programs/ps3-simplify/lists.vsl

@@ -0,0 +1,22 @@
+
+func my_func(a, b, c, d, e, f, g, h) begin
+    var i, j, k, l, m
+
+
+
+    i := a + ~(i + 4)
+
+    i := a + (b + d) * (3 + b)
+
+    if i = f then begin
+        print "hmmm"
+    end
+    return i
+end
+
+func main() begin
+    var n, o, p, q, r, s, t, u, v, w
+    n := 5
+    n += my_func(1, 2, 3, 5, 8, 13, 21, 34)
+    return 0
+end

exercises/05/vslc/vsl_programs/ps3-simplify/shadow.vsl

@@ -0,0 +1,28 @@
+// Checking symbol shadowing
+
+var a
+
+func shadow() begin
+    var a, b
+    a := 1
+    begin
+        var a
+        a := 2
+        b := 40
+        begin
+            var a
+            a := b + 2
+            print a, b
+        end
+        print a
+        begin
+            var b
+            b := 38
+            a := b + 3
+            print a, b
+        end
+        print a
+    end
+    print b
+    return 0
+end

exercises/05/vslc/vsl_programs/ps4-symtab/globals.vsl

@@ -0,0 +1,12 @@
+var global_var0, global_var1
+
+func my_func(param0, param1) begin
+    var a
+    return 0
+end
+
+func main() begin
+    var a
+    print "a string"
+    return 0
+end

exercises/05/vslc/vsl_programs/ps4-symtab/lists.vsl

@@ -0,0 +1,20 @@
+
+func my_func(a, b, c, d, e, f, g, h) begin
+    var i, j, k, l, m
+
+    i := --i
+
+    i := a + b + d
+
+    if i = f then begin
+        print "hmmm"
+    end
+    return i
+end
+
+func main() begin
+    var n, o, p, q, r, s, t, u, v, w
+    n := 5
+    n += my_func(1, 2, 3, 5, 8, 13, 21, 34)
+    return 0
+end

exercises/05/vslc/vsl_programs/ps4-symtab/shadow.vsl

@@ -0,0 +1,28 @@
+// Checking symbol shadowing
+
+var a
+
+func shadow() begin
+    var a, b
+    a := 1
+    begin
+        var a
+        a := 2
+        b := 40
+        begin
+            var a
+            a := b + 2
+            print a, b
+        end
+        print a
+        begin
+            var b
+            b := 38
+            a := b + 3
+            print a, b
+        end
+        print a
+    end
+    print b
+    return 0
+end

exercises/05/vslc/vsl_programs/ps5-codegen1/function_call.vsl

@@ -0,0 +1,9 @@
+func add(a, b, c) begin
+    return a + b
+end
+
+func main() 
+begin
+    print add(40, 2, 4)
+    return 0
+end

exercises/05/vslc/vsl_programs/ps5-codegen1/globals.vsl

@@ -0,0 +1,12 @@
+var global_var0, global_var1
+
+func my_func(param0, param1) begin
+    var a
+    return 0
+end
+
+func main() begin
+    var a
+    print "a string"
+    return 0
+end

exercises/05/vslc/vsl_programs/ps5-codegen1/helloworld.vsl

@@ -0,0 +1,4 @@
+func helloworld() begin
+    print "Hello, World!"
+    return 0
+end

exercises/05/vslc/vsl_programs/ps5-codegen1/lists.vsl

@@ -0,0 +1,22 @@
+
+func my_func(a, b, c, d, e, f, g, h) begin
+    var i, j, k, l, m
+
+
+    i := a + ~(i + 4)
+
+    i := a + (b + d) * (3 + b)
+
+    if i = f then begin
+        print "hmmm"
+    end
+    return i
+end
+
+func main() begin
+    var n, o, p, q, r, s, t, u, v, w
+    n := 5
+    n += my_func(1, 2, 3, 5, 8, 13, 21, 34)
+    print n
+    return 0
+end

exercises/05/vslc/vsl_programs/ps5-codegen1/ps5.vsl

@@ -0,0 +1,30 @@
+// This program tests activation records, function call and return
+func funcall ()
+begin
+    var x,y,z
+    x := 5
+    y := 10
+    print "Calling my_function with parameters", x, y
+    z := my_function ( x, y )
+    print "The returned result is", z
+    z := my_other_function ()
+    print "The other returned result is", z
+    return 0
+end
+
+func my_function ( s, t )
+begin
+    var u
+    u := s*s + t*t
+    print "Parameter s is", s
+    print "Parameter t is", t
+    print "The sum of their squares is", u
+    return u
+end
+
+func my_other_function ()
+begin
+    var x
+    x := 42
+    return x
+end

exercises/05/vslc/vsl_programs/ps5-codegen1/shadow.vsl

@@ -0,0 +1,28 @@
+// Checking symbol shadowing
+
+var a
+
+func shadow() begin
+    var a, b
+    a := 1
+    begin
+        var a
+        a := 2
+        b := 40
+        begin
+            var a
+            a := b + 2
+            print a, b
+        end
+        print a
+        begin
+            var b
+            b := 38
+            a := b + 3
+            print a, b
+        end
+        print a
+    end
+    print b
+    return 0
+end

exercises/06/.gitignore

@@ -0,0 +1,17 @@
+# Buildfiles
+*.o
+vslc/src/vslc
+parser.c
+scanner.c
+y.tab.h
+
+# Submission
+*.tar.xz
+
+# VSL treefiles
+*.ast
+*.sast
+*.sym
+*.bin
+*.s
+*.S

exercises/06/.vscode/launch.json

@@ -0,0 +1,28 @@
+{
+    // Use IntelliSense to learn about possible attributes.
+    // Hover to view descriptions of existing attributes.
+    // For more information, visit: https://go.microsoft.com/fwlink/?linkid=830387
+    "version": "0.2.0",
+    "configurations": [
+        {
+            // for Linux
+            "name": "(gdb) Launch euclid codegen 2",
+            "type": "cppdbg",
+            "request": "launch",
+            "program": "${workspaceRoot}/vslc/src/vslc",
+            "args": ["<", "${workspaceFolder}/vslc/vsl_programs/ps6-codegen2/euclid.vsl"],
+            "stopAtEntry": false,
+            "cwd": "${workspaceRoot}",
+            "environment": [],
+            "externalConsole": true,
+            "MIMode": "gdb",
+            "setupCommands": [
+              {
+                "description": "Enable pretty-printing for gdb",
+                "text": "-enable-pretty-printing",
+                "ignoreFailures": true
+              }
+            ]
+          }
+    ]
+}

exercises/06/vslc/Makefile

@@ -0,0 +1,12 @@
+LEX=flex
+YACC=bison
+YFLAGS+=--defines=src/y.tab.h -o y.tab.c
+CFLAGS+=-std=c99 -g -Isrc -Iinclude -D_POSIX_C_SOURCE=200809L -DYYSTYPE="node_t *"
+
+src/vslc: src/vslc.c src/parser.o src/scanner.o src/nodetypes.o src/tree.o src/ir.o src/generator.o src/tlhash.c
+src/y.tab.h: src/parser.c
+src/scanner.c: src/y.tab.h src/scanner.l
+clean:
+	-rm -f src/parser.c src/scanner.c src/*.tab.* src/*.o
+purge: clean
+	-rm -f src/vslc

exercises/06/vslc/include/ir.h

@@ -0,0 +1,51 @@
+#ifndef IR_H
+#define IR_H
+
+/* This is the tree node structure */
+typedef struct n {
+    node_index_t type;
+    void *data;
+    struct s *entry;
+    uint64_t n_children;
+    struct n **children;
+} node_t;
+
+/**Export the initializer function, it is needed by the parser
+ * @param *nd node to initialize
+ * @param type type of node (see nodetype.h)
+ * @param *data associated data. Declared void to allow any type
+ * @param n_children number of children
+ * @param ... variable argument list of child nodes (node_t *)
+ * 
+ * @return Pointer to the initialized node
+ * */
+node_t* node_init (
+    node_t*         nd, 
+    node_index_t    type, 
+    void*           data, 
+    uint64_t        n_children,
+    ...
+);
+
+typedef enum {
+    SYM_GLOBAL_VAR, 
+    SYM_FUNCTION, 
+    SYM_PARAMETER, 
+    SYM_LOCAL_VAR
+} symtype_t;
+
+typedef struct s {
+    char*       name;
+    symtype_t   type;
+    node_t*     node;
+    size_t      seq;
+    size_t      nparms;
+    tlhash_t*   locals;
+} symbol_t;
+#endif
+
+#define GLOBAL_BUCKET_SIZE          32
+#define LOCAL_BUCKET_SIZE           16
+
+#define DEFAULT_STRING_LIST_SIZE    8
+#define DEFAULT_NO_SCOPES           1

exercises/06/vslc/include/nodetypes.h

@@ -0,0 +1,37 @@
+#ifndef NODETYPES_H
+#define NODETYPES_H
+typedef enum {
+    PROGRAM,
+    GLOBAL_LIST,
+    GLOBAL,
+    STATEMENT_LIST,
+    PRINT_LIST,
+    EXPRESSION_LIST,
+    VARIABLE_LIST,
+    ARGUMENT_LIST,
+    PARAMETER_LIST,
+    DECLARATION_LIST,
+    FUNCTION,
+    STATEMENT,
+    BLOCK,
+    ASSIGNMENT_STATEMENT,
+    ADD_STATEMENT,
+    SUBTRACT_STATEMENT,
+    MULTIPLY_STATEMENT,
+    DIVIDE_STATEMENT,
+    RETURN_STATEMENT,
+    PRINT_STATEMENT,
+    NULL_STATEMENT,
+    IF_STATEMENT,
+    WHILE_STATEMENT,
+    EXPRESSION,
+    RELATION,
+    DECLARATION,
+    PRINT_ITEM,
+    IDENTIFIER_DATA,
+    NUMBER_DATA,
+    STRING_DATA
+} node_index_t;
+
+extern char *node_string[26];
+#endif

exercises/06/vslc/include/tlhash.h

@@ -0,0 +1,28 @@
+#ifndef TLHASH_H
+#define TLHASH_H
+#include <stddef.h>
+typedef struct el {
+    void *key, *value;
+    size_t key_length;
+    struct el *next;
+} tlhash_element_t;
+
+typedef struct {
+    size_t n_buckets, size;
+    tlhash_element_t **buckets;
+} tlhash_t;
+
+int tlhash_init ( tlhash_t *tab, size_t n_buckets );
+int tlhash_finalize ( tlhash_t *tab );
+int tlhash_insert ( tlhash_t *tab, void *key, size_t keylen, void *val );
+int tlhash_lookup ( tlhash_t *tab, void *key, size_t keylen, void **val );
+int tlhash_remove ( tlhash_t *tab, void *key, size_t key_length );
+size_t tlhash_size ( tlhash_t *tab );
+void tlhash_keys ( tlhash_t *tab, void **keys );
+void tlhash_values ( tlhash_t *tab, void **values );
+
+#define TLHASH_SUCCESS 0    /* Success */
+#define TLHASH_ENOMEM 1     /* No memory available */
+#define TLHASH_ENOENT 2     /* No such table entry */
+#define TLHASH_EEXIST 3     /* Table entry already exists */
+#endif

exercises/06/vslc/include/vslc.h

@@ -0,0 +1,51 @@
+#ifndef VSLC_H
+#define VSLC_H
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+#include <stdbool.h>
+#include <stdarg.h>
+
+// Prototypes for the hash table functions
+#include "tlhash.h"
+
+// Numbers and names for the types of syntax tree nodes
+#include "nodetypes.h"
+
+// Definition of the tree node type
+#include "ir.h"
+
+// Token definitions and other things from bison, needs def. of node type
+#include "y.tab.h"
+
+/* This is generated from the bison grammar, calls on the flex specification */
+int yyerror ( const char *error );
+
+/* These are defined in the parser generated by bison */
+extern int yylineno;
+extern int yylex ( void );
+extern int yylex_destroy( void );
+extern char yytext[];
+
+/* Global state */
+extern node_t *root;
+
+// Moving global defs to global header
+
+extern tlhash_t *global_names;  // Defined in ir.c, used by generator.c
+extern char **string_list;      // Defined in ir.c, used by generator.c
+extern size_t stringc;          // Defined in ir.c, used by generator.c
+
+/* Global routines, called from main in vslc.c */
+void simplify_syntax_tree ( void );
+void print_syntax_tree ( void );
+void destroy_syntax_tree ( void );
+
+void create_symbol_table ( void );
+void print_symbol_table ( void );
+void destroy_symbol_table ( void );
+
+void generate_program ( void );
+
+#endif

exercises/06/vslc/src/generator.c

@@ -0,0 +1,829 @@
+#include <vslc.h>
+
+#define ERRPRT(format, args...)     fprintf(stderr, "[ERROR] "); fprintf(stderr ,format, ##args)
+
+#define ASM(opcode, args...)        puts("\t"#opcode"\t"#args)
+#define LABEL(label)                printf("_%s:\n", (char*)label)
+#define COMMENT(format, args...)    printf("# "format"\n", ##args)
+
+// The PUSH and POP macros also increments/decrements the stack_depth to keep track of the stack
+#define PUSH(param)                 printf("\tpushq\t%s\t\t\t\t# PUSH: %ld\n", #param, ++stack_depth)
+#define POP(param)                  printf("\tpopq\t%s\t\t\t\t# POP:  %ld\n", #param, --stack_depth)
+
+#define NO_REG_RECORD           6
+
+// Keep track of sequence of stack depth, ifs and whiles
+static uint64_t 
+    stack_depth,
+    if_seq,
+    while_seq,
+    closest_while;
+
+/**Generate table of strings in a rodata section. */
+void generate_stringtable ( void );
+/**Declare global variables in a bss section */
+void generate_global_variables ( void );
+/**Generate function entry code
+ * @param function symbol table entry of function */
+void generate_function ( symbol_t *function );
+/**Generate code for a node in the AST, to be called recursively from
+ * generate_function
+ * @param node root node of current code block */
+static void generate_node ( node_t *node );
+/**Initializes program (already implemented) */
+void generate_main ( symbol_t *first );
+
+#define MIN(a,b) (((a)<(b)) ? (a):(b))
+
+static const char *record[NO_REG_RECORD] = {
+    "%rdi", "%rsi", "%rdx", "%rcx", "%r8", "%r9"
+};
+
+// Helper funcs for generating different nodes
+
+/**
+ * Generates assembly for printing
+ * 
+ * @param node print statement node with children to print
+ */
+static void generate_print(node_t* node);
+
+/**
+ * Generate identfier for a variable in memory
+ * 
+ * @param node identifier we want the addres of
+ */
+static void generate_var_ident(node_t *node);
+
+/**
+ * Main function to calculate and solve the expressions.
+ * Based on a stack machine. Result is stored on stack.
+ * 
+ * @param node root node for expression
+ */
+static void solve_expressions(node_t *node);
+
+/**
+ * Generates a funciton call
+ * 
+ * @param node root node for function
+ */
+static void generate_function_call(node_t *node);
+
+/**
+ * Generates the return part of a function
+ *  
+ * @param node node containing the return statement
+ */
+static void generate_function_return(node_t *node);
+
+/**
+ * Used for calculating and evaluating the add/sub/mul/div statements.
+ * Turns the statement into an expression, adds the result of rhs to lhs 
+ * and stores the value back to the indentifier
+ * 
+ * @param node      node to the statement
+ * @param operator  one of the following +, -, *, /
+ */
+static void solve_statements(node_t *node, char *operator);
+
+/**
+ * Solves a relation and leaves the result (1 or 0, True or False) in %rax
+ * If the relation node is just a number, just put the number in %rax.
+ * This is used with a cmp instruction to check if the relation is true.
+ * 
+ * This makes it possible to compute the constant relations.
+ * 
+ * @param node pointer to the node the suspected relation is in
+ */
+static void solve_relations(node_t *node);
+
+/**
+ * Solves a node that is an if statement
+ * Uses the solve realtions function to solve the relations
+ * 
+ * @param node pointer to the if node
+ */
+static void generate_if_statement(node_t *node);
+
+/**
+ * Solves a node that is an while statement
+ * Uses the solve realtions function to solve the relations
+ * 
+ * @param node pointer to the while node
+ */
+static void generate_while_statement(node_t *node);
+
+/**
+ * Inserts a jump to the inner most while.
+ * This also works when multiple whiles are nested. 
+ * 
+ */
+static void solve_continue_statement();
+
+/**
+ * Generate assembly to fetch a variable on stack
+ * 
+ * @param node node to the variable to be fetched
+ * @param dest where to put the value
+ */
+static void fetch_variable(node_t *node, const char* dest);
+
+/**
+ * Same as fetch_variable, but stores it back to memory.
+ * 
+ * @param node variable to be stored
+ * @param src  from where should the data come from
+ */
+static void writeback_variable(node_t *node, char* src);
+
+// Helper func for fetching all symbols in a table
+static uint64_t fetch_symbols(tlhash_t* symbol_table, symbol_t*** symbol_list);
+
+void
+generate_program ( void )
+{
+    // Generate the string table at the top
+    generate_stringtable();
+
+    // Fetch all the global elements (functions and global vars)
+    symbol_t **global_list;
+    uint64_t no_globals = fetch_symbols(global_names, &global_list);
+
+    // Find the number of actual global variables
+    uint64_t no_global_vars = 0;
+    for (uint64_t g = 0; g < no_globals; g++)
+        if (global_list[g]->type == SYM_GLOBAL_VAR) no_global_vars++;
+
+    // Generate globbal variables if there are any
+    if (no_global_vars)
+        generate_global_variables();
+
+    // Find the function called main and keep track of if it found an generated
+    bool main_generated = false;
+    uint64_t seq0_index = -1;
+    for (uint64_t g = 0; g < no_globals; g++)
+    {
+        if (global_list[g]->type != SYM_FUNCTION)
+            continue;
+        
+        // If the name of the function is main
+        if (!strcmp(global_list[g]->name, "main"))
+        {
+            generate_main(global_list[g]);
+            main_generated = true;
+        }
+
+        if (!global_list[g]->seq)
+            seq0_index = g;
+    }
+
+    // If no main was found, use the first function instead. 
+    // That means the function with seq = 0
+    if (!main_generated)
+        generate_main(global_list[seq0_index]);
+
+    // Then generate all the functions from vsl
+    for (uint64_t g = 0; g < no_globals; g++)
+    {
+        if (global_list[g]->type == SYM_FUNCTION)
+            generate_function(global_list[g]);
+    }
+
+    free(global_list);
+}
+
+void
+generate_stringtable ( void )
+{
+    /* These can be used to emit numbers, strings and a run-time
+     * error msg. from main
+     */
+    puts("# DATA SECTION");
+    puts(".section .data");
+    puts(".intout:\t.asciz \"\%ld \"");
+    puts(".strout:\t.asciz \"\%s \"");
+    puts(".errout:\t.asciz \"Wrong number of arguments\"");
+
+    for (uint64_t s = 0; s < stringc; s++) 
+    {
+        printf(".STR%03ld:\t.asciz %s\n", s, string_list[s]);
+    }
+    putchar('\n');
+}
+
+void
+generate_global_variables ( void )
+{
+    symbol_t **global_list;
+    uint64_t no_globals = fetch_symbols(global_names, &global_list);
+
+    puts("# GLOBAL VARIABLES");
+    puts(".bss");
+    puts(".align 8");
+
+    for (uint64_t g = 0; g < no_globals; g++)
+        if (global_list[g]->type == SYM_GLOBAL_VAR)
+            printf(".%s:\n", global_list[g]->name);
+    
+    putchar('\n');
+    free(global_list);
+
+}
+
+void
+generate_function ( symbol_t *function )
+{
+    // Keep track of the stack size in each of the functions
+    stack_depth = 0;
+
+    printf("# func %s(nparams: %ld)\n", function->name, function->nparms);
+    puts(".section .text");
+    printf(".global _%s\n", function->name);
+    LABEL(function->name);
+    PUSH(%rbp);
+
+    ASM(movq, %rsp, %rbp);
+
+    // Push params to stack
+    for (int arg = 0; arg < MIN(NO_REG_RECORD,function->nparms); arg++)
+        printf("\tpushq\t%s\t\t\t\t# PUSH: %ld\n", 
+            record[arg],
+            ++stack_depth
+        );
+
+    // How many local variables are inside function
+    uint64_t no_locals = function->locals->size - function->nparms;
+
+    // Make room for the local vars
+    while(no_locals--)
+        PUSH($0);
+
+    // IF the stack alignment is not 16 bytes, 
+    // add one now as all local var also is 0
+    if (stack_depth % 2)
+        PUSH($0);
+
+
+    // Now the stack ptr should be 16 byte aligned.
+
+    // Then generate the body of the function
+    generate_node(function->node);
+
+    putchar('\n');
+}
+
+
+void
+generate_node ( node_t *node)
+{
+    // All statements have the same structure.
+    // [0] is the lhs, needs to be identifier, parser ensures this
+    // [1] is thr rhs
+    switch (node->type)
+    {
+    case ASSIGNMENT_STATEMENT:
+        // First solve the rhs
+        solve_expressions(node->children[1]);
+        // Then store in lhs
+        writeback_variable(node->children[0], "%rax");
+        break;
+
+    case ADD_STATEMENT:
+        // The following way is the naive way of doing an assignment
+        /*
+        fetch_variable(node->children[0], "%rax");
+        ASM(pushq, %rax);
+        solve_expressions(node->children[1]);
+        ASM(popq, %r10);
+        ASM(popq, %rax);
+        ASM(addq, %r10, %rax);
+        writeback_variable(node->children[0], "%rax");
+        */
+        /*  The thing is that add/sub/mul/div assignments 
+            have the same structure as expressions.
+            We can therefore just say that the assignment is 
+            an expression, but remembering to do the writeback afterwards.
+        */
+        puts("# Add statement");
+        solve_statements(node, "+");
+        break;
+
+    case SUBTRACT_STATEMENT:
+        puts("# Subtract statement");
+        solve_statements(node, "-");
+        break;
+
+    case MULTIPLY_STATEMENT:
+        puts("# Multiply statement");
+        solve_statements(node, "*");
+        break;
+
+    case DIVIDE_STATEMENT:
+        puts("# Divide statement");
+        solve_statements(node, "/");
+        break;
+
+    case PRINT_STATEMENT:
+        puts("# Print statement");
+        generate_print(node);
+        break;
+
+    case RETURN_STATEMENT:
+        puts("# Return statement");
+        generate_function_return(node);
+        break;
+
+    case IF_STATEMENT:
+        generate_if_statement(node);
+        break;
+
+    case WHILE_STATEMENT:
+        generate_while_statement(node);
+        break;
+
+    case NULL_STATEMENT:
+        solve_continue_statement();
+        break;
+    
+    case DECLARATION_LIST:
+        /* List of blocks we dont need to traverse */
+        break;
+
+    default:
+        // Otherwise, generate the traverse
+        for (int c = 0; c < node->n_children; c++)
+            generate_node(node->children[c]);
+        break;
+
+    }
+}
+
+
+// Generate the print node
+void
+generate_print(node_t* node)
+{
+
+    for (uint64_t p = 0; p < node->n_children; p++)
+    {
+        node_t *curr_print = node->children[p];
+
+        switch (curr_print->type)
+        {
+        case EXPRESSION:
+            solve_expressions(curr_print);
+            ASM(movq, $.intout, %rdi);
+            ASM(movq, %rax, %rsi);
+            break;
+
+        case STRING_DATA:
+            ASM(movq, $.strout, %rdi);
+            printf("\tmovq\t$.STR%03ld, %%rsi\n", *(uint64_t*)curr_print->data);
+            break;
+
+        case IDENTIFIER_DATA:
+            ASM(movq, $.intout, %rdi);
+            fetch_variable(curr_print, "%rsi");
+            break;
+        default:
+            break;
+        }
+        
+        COMMENT("printf call");
+        ASM(call, printf);
+
+    }
+    // Adds a newline
+    ASM(movq, $'\n', %rdi);
+    ASM(call, putchar);
+}
+
+
+// This will put the value of var in node in dest
+void
+fetch_variable(node_t *node, const char* dest)
+{
+    printf("\tmovq\t");
+    generate_var_ident(node);
+    printf(", %s\t\t# Fetched: %s and put in %s\n", 
+        dest, 
+        node->entry->name, 
+        dest
+    );
+}
+
+// This will put the value in dest to the var in node
+void
+writeback_variable(node_t *node, char* src)
+{
+    printf("\tmovq\t%s,", src);
+    generate_var_ident(node);
+    printf("\t\t# Writeback '%s' from %s\n", 
+        node->entry->name, 
+        src
+    );
+}
+
+// Generate variable identifier, 
+// if local var -> find offset from sb
+// if parameter -> find offset from sb
+// if global    -> insert global tag
+void
+generate_var_ident(node_t *node)
+{
+    symbol_t *ident_sym = node->entry;
+    switch (ident_sym->type)
+    {
+    case SYM_GLOBAL_VAR:
+        printf("$.%s", ident_sym->name);
+        break;
+
+    case SYM_PARAMETER:
+        // If it is a paramter is one of the first 6, seacrch below bp
+        if (ident_sym->seq < 6)
+            printf("%ld(%%rbp)", -8 * (ident_sym->seq + 1));
+        else
+            // This requires that the parameters on 
+            // stack is in reversed order. easier to implement...
+            printf("%ld(%%rbp)", 8 * (ident_sym->seq - 6 + 1 ));
+        break;
+
+    case SYM_LOCAL_VAR:
+        printf("%ld(%%rbp)", -8 * (ident_sym->seq + 1));
+        break;
+    }
+}
+
+
+// This should allways push the result to stack
+// no no no noooo, it should leave it in rax! As it does :))))
+void
+solve_expressions(node_t *node)
+{
+    if (node->data)
+    {   // Check if the expression is a function call
+        bool is_function_call = !strcmp(node->data, "function_call");
+        if (is_function_call)
+        {
+            generate_function_call(node);
+            return;
+        }
+    }
+
+
+    switch (node->n_children)
+    {
+    case 0:
+        switch (node->type)
+        {
+        case IDENTIFIER_DATA:
+            fetch_variable(node, "%rax");
+            break;
+        case NUMBER_DATA:
+            printf("\tmovq\t$%ld,%%rax\n",*(int64_t*)node->data);
+            break;
+        }
+        break;
+
+    case 1:
+        solve_expressions(node->children[0]);
+
+        switch (*(char*)node->data)
+        {
+        case '-':
+            ASM(negq, %rax);
+            break;
+        case '~':
+            ASM(notq, %rax);
+            break;
+        }
+
+        break;
+    case 2:
+
+        // First fetch lhs of expr and then rhs
+        // Push results on stack
+        for (int i = 0; i < 2; i++)
+        {
+            solve_expressions(node->children[i]);
+            PUSH(%rax);
+        }
+
+
+        // Put rhs in %r10
+        POP(%r10);
+        // put lhs in %rax
+        POP(%rax);
+
+        // All operators below leaves result in %rax
+        switch (*(char*)node->data)
+        {
+            /* Assignments */
+            case '|': ASM(orq,  %r10, %rax);    break;  // Bitwise or of %rax and %r10
+            case '^': ASM(xorq, %r10, %rax);    break;  // Bitwise xor of %rax and %r10
+            case '&': ASM(andq, %r10, %rax);    break;  // Bitwise and of %rax and %r10
+            case '+': ASM(addq, %r10, %rax);    break;  // Add %rax and %r10
+            case '-': ASM(subq, %r10, %rax);    break;  // Subtract %r10 from %rax
+            case '*': ASM(imulq, %r10);         break;  // Mulitply %rax with %r10
+            case '/': 
+                ASM(cqto);          // Convert rax to octaword, %rdx:%rax
+                ASM(idivq, %r10);   // Divide %rdx:%rax by %r10
+                break;
+        }
+        break;
+    }
+}
+
+void
+generate_function_call(node_t *node)
+{
+    printf("# Function call\n");
+
+    bool isStack16ByteAligned = !(stack_depth % 2);
+
+    // If the stack is 16 byte alligned here, offset 
+    // by 1 because call pushes return addr to stack
+    // Therefore after call, the stack is 16 byte aligned
+    if (isStack16ByteAligned)
+        PUSH($0);
+
+    // Arg list is allways the second children, lies within a parameter list
+    // Type of this is therefore PARAMETER_LIST
+    node_t *arg_list = node->children[1];
+    if (arg_list->n_children)
+        arg_list = arg_list->children[0]; // This is the acutal parameter list
+
+    for (int arg = 0; arg < MIN(NO_REG_RECORD, arg_list->n_children); arg++)
+    {
+        switch (arg_list->children[arg]->type)
+        {
+        case NUMBER_DATA:
+            printf("\tmovq\t$%ld, %s\n", 
+                *(int64_t*)arg_list->children[arg]->data, 
+                record[arg]
+            );
+            break;
+        
+        case EXPRESSION:
+            solve_expressions(arg_list->children[arg]);
+            printf("\tmovq\t%%rax, %s\n",record[arg]);
+            break;
+        
+        default:
+            fetch_variable(arg_list->children[arg], record[arg]);
+            break;
+        }
+            
+    }
+
+    if (arg_list->n_children > NO_REG_RECORD)
+    {
+        // If there is an odd number of args to push to stack, add 1
+        if (arg_list->n_children % 2)
+            PUSH($0);
+
+        for (int arg = arg_list->n_children - 1; arg >= NO_REG_RECORD; arg--)
+        {
+            switch (arg_list->children[arg]->type)
+            {
+            case NUMBER_DATA:
+                printf("\tpushq\t$%ld\t\t\t\t# PUSH: %ld\n", 
+                    *(int64_t*)arg_list->children[arg]->data,
+                    ++stack_depth
+                );
+                break;
+            
+            // Arg can be an expression
+            case EXPRESSION:
+                solve_expressions(arg_list->children[arg]);
+                PUSH(%rax);
+                break;
+            
+            default:
+                printf("\tpushq\t");
+                generate_var_ident(arg_list->children[arg]);
+                printf("\t\t\t\t# PUSH: %ld", ++stack_depth);
+                putchar('\n');
+                break;
+            }
+        }
+
+    }
+
+    printf("\tcall\t_%s\n", (char*)node->children[0]->data);
+
+    // Aaaand pop the stack to return back to stack alignment
+    if (isStack16ByteAligned)
+        POP(%rcx);
+    
+    printf("# End of function call\n");
+
+}
+
+void
+generate_function_return(node_t *node)
+{
+    solve_expressions(node->children[0]);
+    ASM(leave);
+    ASM(ret);
+}
+
+void
+solve_statements(node_t *node, char *operator)
+{
+    node->type = EXPRESSION;
+    node->data = strdup(operator);
+
+    solve_expressions(node);
+
+    writeback_variable(node->children[0], "%rax");
+}
+
+// Takes in a relation/number node and sets %rax to true if the statement is true
+void
+solve_relations(node_t *relation_root)
+{
+    switch (relation_root->type)
+    {
+    case NUMBER_DATA:
+        // Numberdata is boring, just leave value in %rax, only 1 is interperted as true
+        solve_expressions(relation_root);
+        break;
+    
+    case RELATION:
+        if (relation_root->n_children != 2)
+        {
+            ERRPRT("Relation requires two expressions, one lhs and one rhs\n");
+            exit(EXIT_FAILURE);
+        }
+
+        for (int i = 0; i < relation_root->n_children; i++)
+        {
+            solve_expressions(relation_root->children[i]);
+            PUSH(%rax);
+        }
+
+        POP(%r10);
+        POP(%rax);
+
+        ASM(cmp, %r10, %rax);
+        ASM(movq, $0, %rax);
+
+        switch (*(char*)relation_root->data)
+        {
+        case '=':   ASM(sete, %al); break;  // Set %al (0th byte of %rax) to 1 if lhs == rhs
+        case '>':   ASM(setg, %al); break;  // Set %al (0th byte of %rax) to 1 if lhs > rhs
+        case '<':   ASM(setl, %al); break;  // Set %al (0th byte of %rax) to 1 if lhs < rhs
+        }
+        break;
+    }
+}
+
+void
+generate_if_statement(node_t *node)
+{
+    uint64_t current_if_seq = if_seq++;
+
+    COMMENT("Begin IF %ld", current_if_seq);
+
+    // The realtion is allways in the first part of the IF
+    solve_relations(node->children[0]);
+    
+    // Compare to 0 (False)
+    ASM(cmp, $0, %rax);
+    // If False, jump to either ELSE or ENDIF based on no children in if
+    printf("\tje  \t%s%03ld\n", (node->n_children > 2) ? "ELSE" : "ENDIF", current_if_seq);
+    
+    // Then generate body
+    generate_node(node->children[1]);
+
+    // If else block aswell, add that
+    if (node->n_children > 2) 
+    {
+        printf("\tjmp \tENDIF%03ld\n", current_if_seq);
+        printf("ELSE%03ld:\n", current_if_seq);
+
+        // Generate ELSE body
+        generate_node(node->children[2]);
+    }
+
+    COMMENT("End IF %ld", current_if_seq);
+    printf("ENDIF%03ld:\n", current_if_seq);
+}
+
+void
+generate_while_statement(node_t *node)
+{
+    // Keep local var of which WHILE this is
+    uint64_t current_while_seq = while_seq++;
+    
+    // Also keep the previous closest while 
+    // in case a continue happens after 
+    // this (and this is inside an another while loop)
+    uint64_t prev_closest_while = closest_while;
+    closest_while = current_while_seq;
+
+    COMMENT("Begin WHILE %ld", current_while_seq);   
+    printf("WHILE%03ld:\n", current_while_seq);
+
+    // Relation is allways the first entry in a while
+    solve_relations(node->children[0]);
+
+    // Compare to 0 (False)
+    ASM(cmp, $0, %rax);
+    // If false, then exit while
+    printf("\tje  \tENDWHILE%03ld\n", current_while_seq);
+
+    // Generate body
+    generate_node(node->children[1]);
+
+    // Restore the previous while
+    closest_while = prev_closest_while;
+
+    printf("\tjmp \tWHILE%03ld\n", current_while_seq);
+    printf("ENDWHILE%03ld:\n", current_while_seq);
+}
+
+void
+solve_continue_statement()
+{
+    COMMENT("Continue to WHILE%03ld", closest_while);
+    printf("\tjmp \tWHILE%03ld\n", closest_while);
+}
+
+static uint64_t 
+fetch_symbols(tlhash_t* symbol_table, symbol_t*** symbol_list)
+{
+    uint64_t no_symbols = tlhash_size(symbol_table);
+    *symbol_list = malloc(no_symbols * sizeof(symbol_t));
+    tlhash_values(symbol_table, (void **)*symbol_list );
+
+    return no_symbols;
+}
+
+/**Generates the main function with argument parsing and calling of our
+ * main function (first, if no function is named main)
+ * @param first Symbol table entry of our main function */
+void
+generate_main ( symbol_t *first )
+{
+    puts("###### Entry point for GAS #####");
+    puts ( ".globl main" );
+    puts ( ".section .text" );
+    puts ( "main:" );
+    puts ( "\tpushq   %rbp" );  // Added this for stack alignment
+    puts ( "\tpushq   %rbp" );
+    puts ( "\tmovq    %rsp, %rbp" );
+
+    printf ( "\tsubq\t$1,%%rdi\n" );
+    printf ( "\tcmpq\t$%zu,%%rdi\n", first->nparms );
+    printf ( "\tjne \tABORT\n" );
+    printf ( "\tcmpq\t$0,%%rdi\n" );
+    printf ( "\tjz  \tSKIP_ARGS\n" );
+
+    printf ( "\tmovq\t%%rdi,%%rcx\n" );
+    printf ( "\taddq\t$%zu, %%rsi\n", 8*first->nparms );
+
+    // Modification to mail call, remove this if no params supplied
+    if (first->nparms)
+    {
+        printf ( "PARSE_ARGV:\n" );
+        printf ( "\tpushq\t%%rcx\n" );
+        printf ( "\tpushq\t%%rsi\n" );
+
+        printf ( "\tmovq\t(%%rsi),%%rdi\n" );
+        printf ( "\tmovq\t$0,%%rsi\n" );
+        printf ( "\tmovq\t$10,%%rdx\n" );
+        printf ( "\tcall\tstrtol\n" );
+
+    /*  Now a new argument is an integer in rax */
+
+        printf ( "\tpopq\t%%rsi\n" );
+        printf ( "\tpopq\t%%rcx\n" );
+        printf ( "\tpushq\t%%rax\n" );
+        printf ( "\tsubq\t$8, %%rsi\n" );
+        printf ( "\tloop\tPARSE_ARGV\n" );
+
+        /* Now the arguments are in order on stack */
+        for (int arg = 0; arg < MIN(6,first->nparms); arg++)
+            printf ( "\tpopq\t%s\n", record[arg] );
+
+    }
+
+    printf ( "SKIP_ARGS:\n" );
+    printf ( "\tcall\t_%s\n", first->name );
+    printf ( "\tjmp \tEND\n" );
+    printf ( "ABORT:\n" );
+    printf ( "\tmovq\t$.errout, %%rdi\n" );
+    printf ( "\tcall\tputs\n" );
+
+    printf ( "END:\n" );
+    puts ( "\tmovq \t%rax, %rdi" );
+    puts ( "\tcall \texit" );
+    puts("###### FUNCTIONS FROM VSL BELOW #####");
+    putchar('\n');
+
+}
+
+

exercises/06/vslc/src/generator.c.bak

@@ -0,0 +1,641 @@
+#include <vslc.h>
+
+#define ASM(opcode, args...)        puts("\t"#opcode"\t"#args)
+#define LABEL(label)                printf("_%s:\n", (char*)label)
+#define COMMENT(format, args...)    printf("# "format"\n", ##args)
+
+#define PUSH(param)                 printf("\tpushq\t%s\t\t\t\t# Scope depth: %ld\n", #param, ++scope_depth)
+#define POP(param)                  printf("\tpopq\t%s\t\t\t\t# Scope depth: %ld\n", #param, --scope_depth)
+
+#define NO_REG_RECORD           6
+#define NO_CALLE_SAVED_REG      10
+
+uint64_t scope_depth = 0;
+
+/**Generate table of strings in a rodata section. */
+void generate_stringtable ( void );
+/**Declare global variables in a bss section */
+void generate_global_variables ( void );
+/**Generate function entry code
+ * @param function symbol table entry of function */
+void generate_function ( symbol_t *function );
+/**Generate code for a node in the AST, to be called recursively from
+ * generate_function
+ * @param node root node of current code block */
+static void generate_node ( node_t *node );
+/**Initializes program (already implemented) */
+void generate_main ( symbol_t *first );
+
+#define MIN(a,b) (((a)<(b)) ? (a):(b))
+
+static const char *record[NO_REG_RECORD] = {
+    "%rdi", "%rsi", "%rdx", "%rcx", "%r8", "%r9"
+};
+
+static const char *calle_saved_reg[NO_CALLE_SAVED_REG] = {
+    "%rax", "%rcx", "%rdx", "%rdi", "%rsi", "%rsp", "%r8", "%r9", "%r10", "%r11"
+};
+
+// Helper funcs for generating different nodes
+
+/**
+ * Generates assembly for printing
+ * 
+ * @param node print statement node with children to print
+ */
+static void generate_print(node_t* node);
+
+/**
+ * Generate identfier for a variable in memory
+ * 
+ * @param node identifier we want the addres of
+ */
+static void generate_var_ident(node_t *node);
+
+/**
+ * Main function to calculate and solve the expressions.
+ * Based on a stack machine. Result is stored on stack.
+ * 
+ * @param node root node for expression
+ */
+static void solve_expressions(node_t *node);
+
+/**
+ * Generates a funciton call
+ * 
+ * @param node root node for function
+ */
+static void generate_function_call(node_t *node);
+
+/**
+ * Generates the return part of a function
+ *  
+ * @param node node containing the return statement
+ */
+static void generate_function_return(node_t *node);
+
+/**
+ * Used for calculating and evaluating the add/sub/mul/div statements.
+ * Turns the statement into an expression, adds the result of rhs to lhs 
+ * and stores the value back to the indentifier
+ * 
+ * @param node      node to the statement
+ * @param operator  one of the following +, -, *, /
+ */
+static void solve_statements(node_t *node, char *operator);
+
+/**
+ * Generate assembly to fetch a variable on stack
+ * 
+ * @param node node to the variable to be fetched
+ * @param dest where to put the value
+ */
+static void fetch_variable(node_t *node, const char* dest);
+
+/**
+ * Same as fetch_variable, but stores it back to memory.
+ * 
+ * @param node variable to be stored
+ * @param src  from where should the data come from
+ */
+static void writeback_variable(node_t *node, char* src);
+
+// Helper func for fetching all symbols in a table
+static uint64_t fetch_symbols(tlhash_t* symbol_table, symbol_t*** symbol_list);
+
+void
+generate_program ( void )
+{
+    generate_stringtable();
+    generate_global_variables();
+
+    symbol_t **global_list;
+    uint64_t no_globals = fetch_symbols(global_names, &global_list);
+
+    bool main_generated = false;
+    uint64_t seq0_index = -1;
+    for (uint64_t g = 0; g < no_globals; g++)
+    {
+        if (global_list[g]->type != SYM_FUNCTION)
+            continue;
+        
+        // If the name of the function is main
+        if (!strcmp(global_list[g]->name, "main"))
+        {
+            generate_main(global_list[g]);
+            main_generated = true;
+        }
+
+        if (!global_list[g]->seq)
+            seq0_index = g;
+    }
+
+    // If no main was found, use the first function instead. 
+    // That means the function with seq = 0
+    if (!main_generated)
+        generate_main(global_list[seq0_index]);
+
+    for (uint64_t g = 0; g < no_globals; g++)
+    {
+        if (global_list[g]->type == SYM_FUNCTION)
+            generate_function(global_list[g]);
+    }
+
+    free(global_list);
+}
+
+void
+generate_stringtable ( void )
+{
+    /* These can be used to emit numbers, strings and a run-time
+     * error msg. from main
+     */
+    puts("# DATA SECTION");
+    puts(".section .data");
+    puts(".intout:\t.asciz \"\%ld \"");
+    puts(".strout:\t.asciz \"\%s \"");
+    puts(".errout:\t.asciz \"Wrong number of arguments\"");
+
+    for (uint64_t s = 0; s < stringc; s++) 
+    {
+        printf(".STR%03ld:\t.asciz %s\n", s, string_list[s]);
+    }
+    putchar('\n');
+}
+
+void
+generate_global_variables ( void )
+{
+    symbol_t **global_list;
+    uint64_t no_globals = fetch_symbols(global_names, &global_list);
+
+    puts("# GLOBAL VARIABLES");
+    puts(".bss");
+    puts(".align 8");
+
+    for (uint64_t g = 0; g < no_globals; g++) {
+        if (global_list[g]->type == SYM_GLOBAL_VAR)
+            printf(".%s:\n", global_list[g]->name);
+    }
+    putchar('\n');
+    free(global_list);
+
+}
+
+void
+generate_function ( symbol_t *function )
+{
+    // TODO: Generate code for declaring and entering function, then generate its body
+
+    printf("# func %s(nparams: %ld)\n", function->name, function->nparms);
+    puts(".section .text");
+    printf(".global _%s\n", function->name);
+    LABEL(function->name);
+    //ASM(pushq, %rbp);
+    PUSH(%rbp);
+    ASM(movq, %rsp, %rbp);
+
+    // Push params to stack
+    for (int arg = 0; arg < MIN(NO_REG_RECORD,function->nparms); arg++)
+        printf("\tpushq\t%s\t\t\t\t# Scope depth: %ld\n", record[arg], ++scope_depth);
+
+    // How many local variables are inside function
+    uint64_t no_locals = function->locals->size - function->nparms;
+
+    // IF the stack alignment is not 16 bytes, 
+    // add one now as all local var also is 0
+    if ((MIN(6,function->nparms) + no_locals) % 2)
+        PUSH($0);
+        //ASM(pushq, $0);
+
+    // Make room for the local vars
+    while(no_locals--)
+        PUSH($0);
+        //ASM(pushq, $0);
+
+    // Now the stack ptr should be 16 byte aligned.
+
+    generate_node(function->node);
+
+    putchar('\n');
+}
+
+
+void
+generate_node ( node_t *node)
+{
+    // TODO: Generate code corresponding to node
+
+    // All statements have the same structure.
+    // [0] is the lhs, needs to be identifier, parser ensures this
+    // [1] is thr rhs
+    switch (node->type)
+    {
+    case ASSIGNMENT_STATEMENT:
+        solve_expressions(node->children[1]);
+        //ASM(popq, %rax);
+        POP(%rax);
+        writeback_variable(node->children[0], "%rax");
+        break;
+
+    case ADD_STATEMENT:
+        // The following way is the naive way of doing an assignment
+        /*
+        fetch_variable(node->children[0], "%rax");
+        ASM(pushq, %rax);
+        solve_expressions(node->children[1]);
+        ASM(popq, %r10);
+        ASM(popq, %rax);
+        ASM(addq, %r10, %rax);
+        writeback_variable(node->children[0], "%rax");
+        */
+        /*  The thing is that add/sub/mul/div assignments 
+            have the same structure as expressions.
+            We can therefore just say that the assignment is 
+            an expression, but remembering to do the writeback afterwards.
+        */
+        puts("# Add statement");
+        solve_statements(node, "+");
+        break;
+
+    case SUBTRACT_STATEMENT:
+        puts("# Subtract statement");
+        solve_statements(node, "-");
+        break;
+
+    case MULTIPLY_STATEMENT:
+        puts("# Multiply statement");
+        solve_statements(node, "*");
+        break;
+
+    case DIVIDE_STATEMENT:
+        puts("# Divide statement");
+        solve_statements(node, "/");
+        break;
+
+    case PRINT_STATEMENT:
+        puts("# Print statement");
+        generate_print(node);
+        break;
+
+    case RETURN_STATEMENT:
+        puts("# Return statement");
+        generate_function_return(node);
+        break;
+
+
+
+    case IF_STATEMENT:
+    case WHILE_STATEMENT:
+        /* DO NOTHING YET */
+        break;
+    case NULL_STATEMENT:
+        /* USED IN WHILE/IF */
+        break;
+
+    
+    case DECLARATION_LIST:
+        /* List of blocks we dont need to traverse */
+        break;
+
+    default:
+        for (int c = 0; c < node->n_children; c++)
+            generate_node(node->children[c]);
+        break;
+
+    }
+}
+
+
+void
+generate_print(node_t* node)
+{
+    // Push rdi and rsi to stack incase there are data in them
+    //ASM(pushq, %rdi);
+    //ASM(pushq, %rsi);
+    for (uint64_t p = 0; p < node->n_children; p++)
+    {
+        node_t *curr_print = node->children[p];
+
+        switch (curr_print->type)
+        {
+        case EXPRESSION:
+            solve_expressions(curr_print);
+            //ASM(popq, %rax);
+            POP(%rax);
+            ASM(movq, $.intout, %rdi);
+            ASM(movq, %rax, %rsi);
+            break;
+
+        case STRING_DATA:
+            ASM(movq, $.strout, %rdi);
+            printf("\tmovq\t$.STR%03ld, %%rsi\n", *(uint64_t*)curr_print->data);
+            break;
+
+        case IDENTIFIER_DATA:
+            ASM(movq, $.intout, %rdi);
+            fetch_variable(curr_print, "%rsi");
+            break;
+        default:
+            break;
+        }
+        PUSH(%rax);
+        ASM(movq, $0, %rax);
+        COMMENT("Actual print. Stack size: %ld", scope_depth);
+        ASM(call, printf);
+        POP(%rax);
+
+    }
+    // Adds a newline
+    PUSH(%rax);
+    ASM(movq, $0, %rax);
+    ASM(movq, $'\n', %rdi);
+    ASM(call, putchar);
+    POP(%rax);
+}
+
+
+// This will put the value of var in node in dest
+void
+fetch_variable(node_t *node, const char* dest)
+{
+    printf("\tmovq\t");
+    generate_var_ident(node);
+    printf(", %s\t\t# Fetched: %s\n", dest, node->entry->name);
+}
+
+// This will put the value in dest to the var in node
+void
+writeback_variable(node_t *node, char* src)
+{
+    printf("\tmovq\t%s,", src);
+    generate_var_ident(node);
+    printf("\t\t# Writeback: %s\n", node->entry->name);
+}
+
+
+void
+generate_var_ident(node_t *node)
+{
+    symbol_t *ident_sym = node->entry;
+    switch (ident_sym->type)
+    {
+    case SYM_GLOBAL_VAR:
+        printf("$.%s", ident_sym->name);
+        break;
+
+    case SYM_PARAMETER:
+        // If it is a paramter is one of the first 6, seacrch below bp
+        if (ident_sym->seq < 6)
+            printf("%ld(%%rbp)", -8 * (ident_sym->seq + 1));
+        else
+            // This requires that the parameters on 
+            // stack is in reversed order... easier to implement
+            printf("%ld(%%rbp)", 8 * (ident_sym->seq - 6 + 1 ));
+        break;
+
+    case SYM_LOCAL_VAR:
+        printf("%ld(%%rbp)", -8 * (ident_sym->seq + 1));
+        break;
+    }
+}
+
+
+// This should allways push the result to stack
+void
+solve_expressions(node_t *node)
+{
+    if (node->data)
+    {   // Check if the expression is a function call
+        bool is_function_call = !strcmp(node->data, "function_call");
+        if (is_function_call)
+        {
+            generate_function_call(node);
+            return;
+        }
+    }
+
+
+    switch (node->n_children)
+    {
+    case 0:
+        switch (node->type)
+        {
+        case IDENTIFIER_DATA:
+            fetch_variable(node, "%rax");
+            PUSH(%rax);
+            //ASM(pushq, %rax);
+            break;
+        case NUMBER_DATA:
+            printf("\tmovq\t$%ld,%%rax\n",*(int64_t*)node->data);
+            PUSH(%rax);
+            //ASM(pushq, %rax);
+            break;
+        }
+        break;
+    case 1:
+        solve_expressions(node->children[0]);
+        //ASM(popq, %rax);
+        POP(%rax);
+
+        switch (*(char*)node->data)
+        {
+        case '-':
+            ASM(negq, %rax);
+            break;
+        case '~':
+            ASM(notq, %rax);
+            break;
+        }
+        //ASM(pushq, %rax);
+        PUSH(%rax);
+
+        break;
+    case 2:
+
+        // First fetch lhs of expr and then rhs
+        // Push results on stack
+        for (int i = 0; i < 2; i++)
+            solve_expressions(node->children[i]);
+
+
+        // Put rhs in %r10
+        //ASM(popq, %r10);
+        POP(%r10);
+        // put lhs in %rax
+        //ASM(popq, %rax);
+        POP(%rax);
+
+        switch (*(char*)node->data)
+        {
+            /* Assignments */
+            case '|': ASM(orq,  %r10, %rax);    break;  // Bitwise or of %rax and %r10
+            case '^': ASM(xorq, %r10, %rax);    break;  // Bitwise xor of %rax and %r10
+            case '&': ASM(andq, %r10, %rax);    break;  // Bitwise and of %rax and %r10
+            case '+': ASM(addq, %r10, %rax);    break;  // Add %rax and %r10
+            case '-': ASM(subq, %r10, %rax);    break;  // Subtract %r10 from %rax
+            case '*': ASM(imulq, %r10);         break;  // Mulitply %rax with %r10
+            case '/': 
+                ASM(cqto);          // Convert rax to octaword, %rdx:%rax
+                ASM(idivq, %r10);   // Divide %rdx:%rax by %r10
+                break;
+        }
+
+        // Push result to stack.
+        //ASM(pushq, %rax);
+        PUSH(%rax);
+        break;
+    }
+}
+
+void
+generate_function_call(node_t *node)
+{
+    printf("# Function call\n");
+
+    node_t *arg_list = node->children[1];
+    if (arg_list->n_children)
+        arg_list = arg_list->children[0];
+
+    for (int arg = 0; arg < MIN(NO_REG_RECORD, arg_list->n_children); arg++)
+    {
+        if (arg_list->children[arg]->type == NUMBER_DATA)
+            printf("\tmovq\t$%ld, %s\n", 
+                *(int64_t*)arg_list->children[arg]->data, 
+                record[arg]
+            );
+        else
+            fetch_variable(arg_list->children[arg], record[arg]);
+    }
+
+    if (arg_list->n_children > NO_REG_RECORD)
+    {
+        for (int arg = arg_list->n_children - 1; arg >= NO_REG_RECORD; arg--)
+        {
+            if (arg_list->children[arg]->type == NUMBER_DATA)
+                printf("\tpushq\t$%ld\t\t\t\t# Scope depth: %ld\n", 
+                    *(int64_t*)arg_list->children[arg]->data,
+                    ++scope_depth
+                );
+            else
+            {
+                printf("\tpushq\t");
+                generate_var_ident(arg_list->children[arg]);
+                printf("\t\t\t\t# Scope depth: %ld", ++scope_depth);
+                putchar('\n');
+            }
+        }
+        if (arg_list->n_children % 2)
+            PUSH($0);
+            //ASM(pushq, $0);
+    }
+
+    printf("\tcall\t_%s\n", (char*)node->children[0]->data);
+    PUSH(%rax);
+    //ASM(pushq, %rax);
+    printf("# End of function call\n");
+
+    /*
+    for (int reg = 0; reg < NO_CALLE_SAVED_REG; reg++)
+        printf("\tpushq\t%s \t\t# Pushing %s to stack\n", 
+            calle_saved_reg[reg], 
+            calle_saved_reg[reg]
+        );
+
+    for (int reg = NO_CALLE_SAVED_REG; reg > 0; reg--)
+    printf("\tpopq\t%s \t\t# Poping %s from stack\n", 
+        calle_saved_reg[reg], 
+        calle_saved_reg[reg]
+    );*/
+
+}
+
+void
+generate_function_return(node_t *node)
+{
+    solve_expressions(node->children[0]);
+    //ASM(popq, %rax);
+    POP(%rax);
+    ASM(leave);
+    ASM(ret);
+}
+
+void
+solve_statements(node_t *node, char *operator)
+{
+    node->type = EXPRESSION;
+    node->data = strdup(operator);
+
+    solve_expressions(node);
+
+    //ASM(popq, %rax);
+    POP(%rax);
+    writeback_variable(node->children[0], "%rax");
+}
+
+/**Generates the main function with argument parsing and calling of our
+ * main function (first, if no function is named main)
+ * @param first Symbol table entry of our main function */
+void
+generate_main ( symbol_t *first )
+{
+    puts("###### Entry point for GAS #####");
+    puts ( ".globl main" );
+    puts ( ".section .text" );
+    puts ( "main:" );
+    puts ( "\tpushq   %rbp" );
+    puts ( "\tmovq    %rsp, %rbp" );
+
+    printf ( "\tsubq\t$1,%%rdi\n" );
+    printf ( "\tcmpq\t$%zu,%%rdi\n", first->nparms );
+    printf ( "\tjne \tABORT\n" );
+    printf ( "\tcmpq\t$0,%%rdi\n" );
+    printf ( "\tjz  \tSKIP_ARGS\n" );
+
+    printf ( "\tmovq\t%%rdi,%%rcx\n" );
+    printf ( "\taddq\t$%zu, %%rsi\n", 8*first->nparms );
+    printf ( "PARSE_ARGV:\n" );
+    printf ( "\tpushq\t%%rcx\n" );
+    printf ( "\tpushq\t%%rsi\n" );
+
+    printf ( "\tmovq\t(%%rsi),%%rdi\n" );
+    printf ( "\tmovq\t$0,%%rsi\n" );
+    printf ( "\tmovq\t$10,%%rdx\n" );
+    printf ( "\tcall\tstrtol\n" );
+
+/*  Now a new argument is an integer in rax */
+
+    printf ( "\tpopq\t%%rsi\n" );
+    printf ( "\tpopq\t%%rcx\n" );
+    printf ( "\tpushq\t%%rax\n" );
+    printf ( "\tsubq\t$8, %%rsi\n" );
+    printf ( "\tloop\tPARSE_ARGV\n" );
+
+    /* Now the arguments are in order on stack */
+    for (int arg = 0; arg < MIN(6,first->nparms); arg++)
+        printf ( "\tpopq\t%s\n", record[arg] );
+
+    printf ( "SKIP_ARGS:\n" );
+    printf ( "\tcall\t_%s\n", first->name );
+    printf ( "\tjmp \tEND\n" );
+    printf ( "ABORT:\n" );
+    printf ( "\tmovq\t$.errout, %%rdi\n" );
+    printf ( "\tcall\tputs\n" );
+
+    printf ( "END:\n" );
+    puts ( "\tmovq \t%rax, %rdi" );
+    puts ( "\tcall \texit" );
+    puts("###### FUNCTIONS FROM VSL BELOW #####");
+    putchar('\n');
+
+}
+
+static uint64_t 
+fetch_symbols(tlhash_t* symbol_table, symbol_t*** symbol_list)
+{
+    uint64_t no_symbols = tlhash_size(symbol_table);
+    *symbol_list = malloc(no_symbols * sizeof(symbol_t));
+    tlhash_values(symbol_table, (void **)*symbol_list );
+
+    return no_symbols;
+}

exercises/06/vslc/src/ir.c

@@ -0,0 +1,602 @@
+#include <vslc.h>
+
+#define ERRPRT(format, args...) {fprintf(stderr, "[ERROR] "); fprintf(stderr ,format, ##args);}
+
+
+// Externally visible, for the generator
+extern tlhash_t *global_names;
+extern char **string_list;
+extern size_t n_string_list, stringc;
+
+// Functions from the skeleton
+
+static uint64_t find_globals ( void );
+static void bind_names ( symbol_t *function, node_t *root );
+
+// Helper functions, see description in the definition
+
+static void print_global_tree(symbol_t* global);
+static void print_string_list(void);
+static void destroy_global(symbol_t* global);
+static void push_scope(void);
+static void pop_scope(void);
+static void insert_symbol(tlhash_t *hash_table, symbol_t* symbol);
+static void insert_local_to_scope(symbol_t *local);
+static void insert_local_to_func(symbol_t *function, symbol_t *root);
+static void insert_local_var(symbol_t *function, node_t *root);
+static void collect_string(node_t *root);
+static symbol_t* lookup_var(symbol_t *function, char* var);
+
+// Local "global" variables
+static const char *symbol_names[4] = {
+    "GLOBAL_VAR", 
+    "FUNCTION", 
+    "PARAMETER", 
+    "LOCAL_VAR"
+};
+static uint64_t no_scopes, cur_scope_depth;
+static tlhash_t **scopes;
+
+
+/**
+ * Gather information and create a symbol table.
+ * 
+ * Used in vslc.c
+ */
+void
+create_symbol_table ( void )
+{
+
+    // Initialize string array
+    n_string_list   = DEFAULT_STRING_LIST_SIZE;
+    string_list     = malloc(n_string_list * sizeof(char*));
+    stringc         = 0;
+
+    // Initialize scope array
+    no_scopes       = DEFAULT_NO_SCOPES;
+    scopes          = malloc(no_scopes * sizeof(tlhash_t));
+    cur_scope_depth = 0;
+
+    // Traverse the root node for globals
+    uint64_t no_globals = find_globals();
+
+    // Prepare a temp list of globals and fetch all globals
+    symbol_t **global_list = malloc(no_globals * sizeof(symbol_t));
+    tlhash_values(global_names, (void **)global_list );
+
+    /* Iterate over the temporary list, bind names in each function */
+    for (uint64_t g = 0; g < no_globals; g++ )
+    {
+        if (global_list[g]->type == SYM_FUNCTION)
+            bind_names(global_list[g], global_list[g]->node);
+    }
+    // Free the temp list
+    free(global_list);
+}
+
+
+/**
+ * Prints the symbol table and the string array
+ * 
+ * Used in vslc.c
+ */
+void
+print_symbol_table ( void )
+{
+    /* Get the number of symbols, size up a temporary list and fill it */
+    uint64_t no_globals = tlhash_size(global_names);
+    symbol_t **global_list = malloc(no_globals * sizeof(symbol_t));
+    tlhash_values(global_names, (void **)global_list );
+
+    /* Iterate over the temporary list, printing entries */
+    for (uint64_t g = 0; g < no_globals; g++ )
+        // Print the tree structure for each global
+        print_global_tree(global_list[g]);
+    
+    free(global_list);
+
+    // Print strings
+    print_string_list();
+}
+
+
+/**
+ * Prins the tree of a global
+ * 
+ * @param global pointer to the global to be printed
+ */
+static void 
+print_global_tree(symbol_t* global)
+{
+    // Check if null ptr
+    if (!global)
+        return;
+    
+    // Print global root
+    printf("─%s: %-16s [nparams=%2ld, seq=%2ld, node=%p]\n", 
+        symbol_names[global->type], 
+        global->name, 
+        global->nparms,
+        global->seq,
+        global->node
+    );
+    
+    // If the global does not have params or locals, return
+    if (!global->nparms && !global->locals)
+        {putchar('\n');return;}
+    
+    // Need to fetch the whole size, since nparams 
+    // only count the params, not all locals
+    uint64_t no_locals = tlhash_size(global->locals);
+    symbol_t **locals_list = malloc(no_locals * sizeof(symbol_t));
+    tlhash_values(global->locals, (void **)locals_list );
+    // Go through all locals
+    for (int l = 0; l < no_locals; l++)
+    {   // Do some simple sorting, so seq num is in order
+        for (int ll = 0; ll < no_locals; ll++)
+        {
+            if (locals_list[ll]->seq == l)
+            {
+                printf(" %s─[%s]: %-22s\t[seq=%2ld, node=%p]\n",
+                    (l < (no_locals - 1)) ? "├" : "└", 
+                    symbol_names[locals_list[ll]->type],
+                    locals_list[ll]->name,
+                    locals_list[ll]->seq,
+                    locals_list[ll]->node
+                );
+                break;
+            }
+        }
+    }
+    putchar('\n');
+    free(locals_list);
+}
+
+
+/**
+ * Prints the array of strings
+ * 
+ */
+static void
+print_string_list(void)
+{   // Print out all the collected strings
+    printf("─STRINGS [%ld]\n", stringc);
+    for (uint64_t i = 0; i < stringc; i++)
+        printf(" %s─[%ld]: %s\n", 
+            (i < (stringc - 1)) ? "├" : "└", 
+            i, 
+            string_list[i]
+        );
+}
+
+
+/**
+ * Destroys all the dynamicly allocated memory and all the hash tables.
+ * Frees up the array of strings as well.
+ * 
+ * Used in vslc.c
+ */
+void
+destroy_symbol_table ( void )
+{
+    // FREE STRINGS
+    // Free all strings that are kept in the array
+    for (uint64_t c = 0; c < stringc; c++)
+        free(string_list[c]);
+    // Free the actual list
+    free(string_list);
+
+    // FREE SCOPES
+    // At the end of program, all scopes have to be popped
+    // Therefore only free the list
+    free(scopes);
+
+    // FREE GLOBAL NAMES
+    if (!global_names)
+        return;
+
+    // Fetch list of globals
+    uint64_t no_globals = tlhash_size(global_names);
+    symbol_t **global_list = malloc(no_globals * sizeof(symbol_t));
+    tlhash_values(global_names, (void **)global_list );
+
+    // Destroy all global elements
+    for (uint64_t g = 0; g < no_globals; g++)
+        destroy_global(global_list[g]);
+    
+    // Destory the global hash table
+    tlhash_finalize(global_names);
+    // Free the global hash table
+    free(global_names);
+
+    // Free the temp list
+    free(global_list);
+}
+
+
+/**
+ * Destroys the supplied global symbol by
+ * finalizing each of the local tables
+ * 
+ * @param global pointer to the global symbol to be destroyed
+ */
+static void 
+destroy_global(symbol_t* global)
+{
+    if (!global)
+        return;
+    if (!global->locals)
+    {
+        free(global);
+        return;
+    }
+    
+    uint64_t no_locals = tlhash_size(global->locals);
+    symbol_t **locals_list = malloc(no_locals * sizeof(symbol_t));
+    tlhash_values(global->locals, (void **)locals_list );
+
+    for (int l = 0; l < no_locals; l++)
+        free(locals_list[l]);
+    
+    tlhash_finalize(global->locals);
+    free(global->locals);
+    free(global);
+
+    free(locals_list);
+}
+
+
+/**
+ * Goes trough the root node and finds all global variables and functions
+ * 
+ * @return Returns the number of globals found (functions + variables)
+ */
+static uint64_t
+find_globals ( void )
+{
+    tlhash_init(global_names = malloc(sizeof(tlhash_t)), GLOBAL_BUCKET_SIZE);
+
+    uint64_t no_functions = 0, no_global_vars = 0;
+    node_t *global_list = root;
+
+    // Check if not nullptr
+    if (!global_list)
+        return 0;
+
+    symbol_t* global_symbol;
+    for (uint64_t global_i = 0; global_i < global_list->n_children; global_i++)
+    {
+        node_t *current_global = global_list->children[global_i];
+        switch (current_global->type)
+        {
+        case VARIABLE_LIST:
+
+            // Go through the variable list and get all the global variables
+            for (uint64_t var_i = 0; var_i < current_global->n_children; var_i++)
+            {
+                global_symbol = malloc(sizeof(symbol_t));
+                *global_symbol = (symbol_t){
+                    .type   = SYM_GLOBAL_VAR,
+                    .name   = current_global->children[var_i]->data,
+                    .node   = current_global->children[var_i],
+                    .seq    = 0,
+                    .nparms = 0,
+                    .locals = NULL
+                };
+                insert_symbol(global_names, global_symbol);
+                no_global_vars++;
+            }
+            break;
+        case FUNCTION:
+            node_t *function = current_global;
+
+            // Function node allways have the same structure,
+            // [0] are the identifier
+            // [1] are the variable list, within a paramerer_list
+            // [2] are the actual block
+            if (!function->children[0])
+                break;
+
+            // Create the function symbol
+            global_symbol = malloc(sizeof(symbol_t));
+            *global_symbol = (symbol_t){
+                .type   = SYM_FUNCTION,
+                .name   = current_global->children[0]->data,
+                .node   = current_global->children[2],
+                .seq    = no_functions++,
+                .nparms = 0,
+                .locals = malloc(sizeof(tlhash_t))
+            };
+
+            // Initialize the local variable table
+            tlhash_init(global_symbol->locals, LOCAL_BUCKET_SIZE);
+
+            // Insert the pointer to the newly created symbol
+            insert_symbol(global_names, global_symbol);
+
+            // If there are no parameters in function, break.
+            if (!current_global->children[1]->n_children)
+                break;
+
+
+            // Find all params and insert into hash table in global_symbol
+            symbol_t *param_sym;
+            node_t *param_list = current_global->children[1]->children[0];
+            global_symbol->nparms = param_list->n_children;
+
+            for (uint64_t param_i = 0; param_i < param_list->n_children; param_i++)
+            {
+                param_sym = malloc(sizeof(symbol_t));
+                *param_sym = (symbol_t){
+                    .type   = SYM_PARAMETER,
+                    .name   = param_list->children[param_i]->data,
+                    .node   = param_list->children[param_i],
+                    .seq    = param_i,
+                    .nparms = 0,
+                    .locals = NULL
+                };
+
+                insert_symbol(global_symbol->locals, param_sym);
+            }
+            break;
+        }
+    }
+    return no_functions + no_global_vars;
+}
+
+
+/**
+ * Inserts a symbol into a hash table, key is defined in the name field in the symbol supplied.
+ * 
+ * @param hash_table    pointer to the hash table the symbol is inserted into
+ * @param symbol        pointer to the symbol to be inserted
+ */
+void 
+insert_symbol(tlhash_t *hash_table, symbol_t* symbol)
+{
+    tlhash_insert(hash_table, symbol->name, strlen(symbol->name), symbol);
+}
+
+
+/**
+ * @brief Traverse a node root, and find all variables and strings
+ * 
+ * @param function  pointer to the current function
+ * @param root      pointer to the root node
+ */
+static void 
+bind_names ( symbol_t *function, node_t *root )
+{   // NULL check
+    if (!function)
+        return;
+    if (!root)
+        return;
+
+    // Can't declare variables inside switch unless 
+    // it is in a new scope
+    node_t *declarations;
+    // We want do top to bottom traverse, so do not 
+    // call recusivly unless we need to go deeper
+    switch (root->type)
+    {
+    // If new BLOCK start, push the scope and recurse from here.
+    case BLOCK:
+        push_scope();
+        for (uint64_t i = 0; i < root->n_children; i++)
+            bind_names(function, root->children[i]);
+        pop_scope();
+        break;
+
+    // If DECLARATION_LIST, find all the identifiers 
+    // and insert local into scope and function 
+    case DECLARATION_LIST:
+        if (!root->children[0])
+            break;
+
+        declarations = root->children[0];
+        for (uint64_t i = 0; i < declarations->n_children; i++)
+            // Insert each of the local variables in the declaration
+            insert_local_var(function, declarations->children[i]);
+
+        break;
+    
+    // If IDENTIFIER_DATA, look up the identifier in all the scopes. 
+    // If not found (NULL), crash the compiler with a somewhat helpful message.
+    case IDENTIFIER_DATA:
+        if (!root->data)
+            break;
+
+        if (!(root->entry = lookup_var(function, root->data)))
+        {
+            ERRPRT("Could not find %s in scope!\n", (char*)root->data)
+            exit(EXIT_FAILURE);
+        }
+        break;
+
+    // If STRING_DATA, collect the string and point the 
+    // data in the corresponding node to the array index
+    case STRING_DATA:
+        collect_string(root);
+        break;
+    
+    // If none of the above, go deeper if possible.
+    default:
+        for (uint64_t i = 0; i < root->n_children; i++)
+            bind_names(function, root->children[i]);
+        break;
+    }
+}
+
+/**
+ * Pushes a new hash table to the scope stack.
+ * 
+ * Increases the size of the stack if too small.
+ * 
+ */
+static void
+push_scope(void)
+{
+    // Allocate memory for the hash table and initialize
+    scopes[cur_scope_depth] = malloc(sizeof(tlhash_t));
+    tlhash_init(scopes[cur_scope_depth++], LOCAL_BUCKET_SIZE);
+
+    // Grow the amount of scopes if not enough
+    if (cur_scope_depth >= no_scopes)
+    {
+        no_scopes *= 2;
+        tlhash_t **new_scopes = realloc(scopes, no_scopes * sizeof(tlhash_t));
+        if (!new_scopes)
+        {
+            ERRPRT("Could not realloc scopes!\n");
+            exit(EXIT_FAILURE);
+        }
+        scopes = new_scopes;
+    }
+}
+
+
+/**
+ * Pops the dynamicy allocated hash table for the current scope depth
+ * 
+ */
+static void
+pop_scope(void)
+{
+    tlhash_finalize(scopes[--cur_scope_depth]);
+    free(scopes[cur_scope_depth]);
+}
+
+
+/**
+ * Allocates and inserts a local symbol into 
+ * the scope stack and into the function
+ * 
+ * @param function  pointer to the current function
+ * @param root      pointer to the root node for the symbol
+ */
+static void
+insert_local_var(symbol_t *function, node_t *root)
+{   // Null ptr check
+    if (!root->data)
+        return;
+
+    // Get the sequence num, is the size
+    size_t sequence = tlhash_size(function->locals);
+
+    symbol_t *variable = malloc(sizeof(symbol_t));
+    *variable = (symbol_t){
+        .type   = SYM_LOCAL_VAR,
+        .name   = root->data,
+        .node   = root,            
+        .seq    = sequence,     //! Use sequence as name in var list of function, strictly growing
+        .nparms = 0,
+        .locals = NULL
+    };
+    insert_local_to_scope(variable);
+    insert_local_to_func(function, variable);
+}
+
+
+/**
+ * Inserts a symbol to the top most scope in stack
+ * 
+ * @param local pointer to the local to be inserted
+ */
+static void 
+insert_local_to_scope(symbol_t *local)
+{
+    insert_symbol(scopes[cur_scope_depth - 1], local);
+}
+
+
+/**
+ * Insert local symbol to the functions table of local variables
+ * uses the seq num as key as this is strictly growing
+ * 
+ * @param function  pointer to the function to insert the symbol
+ * @param local     pointer to the symbol to be inserted in the table 
+ */
+void 
+insert_local_to_func(symbol_t *function, symbol_t *local)
+{
+    tlhash_insert(
+        function->locals,   //! Insert local to the function var table
+        &local->seq,         //! The key is a number, unique, strictly growing
+        sizeof(local->seq),  //! Size of key
+        local                //! The local symbol
+    );
+}
+
+
+/**
+ * Collects strings to the string array and 
+ * points the data in the associated node 
+ * to the array position
+ * 
+ * @param root  pointer to the root node of the string
+ */
+static void
+collect_string(node_t *root)
+{   // Null ptr check
+    if (!root->data)
+        return;
+    
+    // Get the string and allocate room for array index of string
+    string_list[stringc] = root->data;
+    root->data = malloc(sizeof(size_t));
+    // Set the data ptr
+    *((size_t*)root->data) = stringc++;
+
+    // Grow string array if nessecary
+    if (stringc >= n_string_list)
+    {
+        n_string_list *= 2;
+        char **new_string_list = realloc(string_list, n_string_list * sizeof(char*));
+        if (!new_string_list)
+        {
+            ERRPRT("Could not realloc string list!\n");
+            exit(EXIT_FAILURE);
+        }
+        string_list = new_string_list;
+    }
+}
+
+
+/**
+ * Looks up a variable identifier in all the scopes. 
+ * Start with the scopes, then the parameters and 
+ * the the globals
+ * 
+ * @param function  pointer to the function
+ * @param var       identifier to the variable
+ * @return Returns the pointer to the "closest" matched identifier. NULL if not found.
+ */
+static symbol_t*
+lookup_var(symbol_t *function, char* var)
+{
+    // Symbol to store the stymbol to be found
+    symbol_t* symbol = NULL;
+    // Result stores the result of the hash lookups
+    int result;
+
+    // Try the local scopes first
+    for (int64_t d = cur_scope_depth - 1; d >= 0; d--)
+    {
+        result = tlhash_lookup(scopes[d], var, strlen(var), (void**)&symbol);
+        if (result == TLHASH_SUCCESS)
+            return symbol;
+    }
+
+    // Then move to parameters
+    result = tlhash_lookup(function->locals, var, strlen(var), (void**)&symbol);
+    if (result == TLHASH_SUCCESS)
+        return symbol;
+    
+    // Last try global parameters
+    result = tlhash_lookup(global_names, var, strlen(var), (void**)&symbol);
+    if (result == TLHASH_SUCCESS)
+        return symbol;
+
+    // If nothing is found, return NULL
+    return NULL;
+}

exercises/06/vslc/src/nodetypes.c

@@ -0,0 +1,34 @@
+#define STRING(x) #x
+char *node_string[30] = {
+    STRING(PROGRAM),
+    STRING(GLOBAL_LIST),
+    STRING(GLOBAL),
+    STRING(STATEMENT_LIST),
+    STRING(PRINT_LIST),
+    STRING(EXPRESSION_LIST),
+    STRING(VARIABLE_LIST),
+    STRING(ARGUMENT_LIST),
+    STRING(PARAMETER_LIST),
+    STRING(DECLARATION_LIST),
+    STRING(FUNCTION),
+    STRING(STATEMENT),
+    STRING(BLOCK),
+    STRING(ASSIGNMENT_STATEMENT),
+    STRING(ADD_STATEMENT),
+    STRING(SUBTRACT_STATEMENT),
+    STRING(MULTIPLY_STATEMENT),
+    STRING(DIVIDE_STATEMENT),
+    STRING(RETURN_STATEMENT),
+    STRING(PRINT_STATEMENT),
+    STRING(NULL_STATEMENT),
+    STRING(IF_STATEMENT),
+    STRING(WHILE_STATEMENT),
+    STRING(EXPRESSION),
+    STRING(RELATION),
+    STRING(DECLARATION),
+    STRING(PRINT_ITEM),
+    STRING(IDENTIFIER_DATA),
+    STRING(NUMBER_DATA),
+    STRING(STRING_DATA)
+};
+#undef STRING

exercises/06/vslc/src/parser.y

@@ -0,0 +1,313 @@
+%{
+#include <vslc.h>
+
+#define NODE(type, data, n_children, children...) node_init(malloc(sizeof(node_t)), type, data, n_children,  ##children)
+%}
+
+%define api.value.type {node_t}
+%token FUNC PRINT RETURN CONTINUE IF THEN ELSE WHILE DO OPENBLOCK CLOSEBLOCK
+%token VAR NUMBER IDENTIFIER STRING
+
+%left '|' '&' '^'
+%left '+' '-'
+%left '*' '/'
+%nonassoc UMINUS
+%right '~'
+//%expect 1
+
+%nonassoc IF THEN
+%nonassoc ELSE
+
+/* Tried fixing vscode complaining about the type for the non-terminals, didn't work
+%union {
+  node_t* node;
+}
+
+%type <node> global_list global 
+%type <node> statement_list print_list expression_list variable_list argument_list parameter_list declaration_list
+%type <node> function statement block 
+%type <node> assignment_statement return_statement print_statement null_statement if_statement while_statement
+%type <node> relation expression declaration print_item identifier number string
+*/ 
+
+%%
+
+program:
+    global_list {
+      root = NODE(PROGRAM, NULL, 1, $1);
+    }
+  ;
+
+global_list:
+    global {
+      $$ = NODE(GLOBAL_LIST, NULL, 1, $1);
+    }
+  | global_list global {
+      $$ = NODE(GLOBAL_LIST, NULL, 2, $1, $2);
+    }
+  ;
+
+global:
+    function {
+      $$ = NODE(GLOBAL, NULL, 1, $1);
+    }
+  | declaration {
+      $$ = NODE(GLOBAL, NULL, 1, $1);
+    }
+  ;
+
+statement_list:
+    statement {
+      $$ = NODE(STATEMENT_LIST, NULL, 1, $1);
+    }
+  | statement_list statement {
+      $$ = NODE(STATEMENT_LIST, NULL, 2, $1, $2);
+    }
+  ;
+
+print_list: 
+    print_item {
+      $$ = NODE(PRINT_LIST, NULL, 1, $1);
+    }
+  | print_list ',' print_item {
+      $$ = NODE(PRINT_LIST, NULL, 2, $1, $3);
+    }
+  ;
+
+expression_list:
+    expression {
+      $$ = NODE(EXPRESSION_LIST, NULL, 1, $1);
+    }
+  | expression_list ',' expression {
+      $$ = NODE(EXPRESSION_LIST, NULL, 2, $1, $3);
+    }
+  ;
+
+variable_list:
+    identifier {
+      $$ = NODE(VARIABLE_LIST, NULL, 1, $1);
+    }
+  | variable_list ',' identifier {
+      $$ = NODE(VARIABLE_LIST, NULL, 2, $1, $3);
+    }
+  ;
+
+argument_list:
+    expression_list {
+      $$ = NODE(ARGUMENT_LIST, NULL, 1, $1);
+    }
+  | /* epsilon */ {
+      $$ = NODE(ARGUMENT_LIST, NULL, 0);
+    }
+  ;
+
+parameter_list:
+    variable_list {
+      $$ = NODE(PARAMETER_LIST, NULL, 1, $1);
+    }
+  | /* epsilon */ {
+      $$ = NODE(PARAMETER_LIST, NULL, 0);
+    }
+  ;
+
+declaration_list:
+    declaration {
+      $$ = NODE(DECLARATION_LIST, NULL, 1, $1);
+    }
+  | declaration_list declaration {
+      $$ = NODE(DECLARATION_LIST, NULL, 2, $1, $2);
+    }
+  ;
+
+function:
+    FUNC identifier '(' parameter_list ')' statement  {
+      $$ = NODE(FUNCTION, NULL, 3, $2, $4, $6);
+    }
+  ;
+
+statement:
+    assignment_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | return_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | print_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | if_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | while_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | null_statement {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  | block {
+      $$ = NODE(STATEMENT, NULL, 1, $1);
+    }
+  ;
+
+block:
+    OPENBLOCK declaration_list statement_list CLOSEBLOCK {
+      $$ = NODE(BLOCK, NULL, 2, $2, $3);
+    }
+  | OPENBLOCK statement_list CLOSEBLOCK {
+      $$ = NODE(BLOCK, NULL, 1, $2);
+    }
+  ;
+
+assignment_statement:
+    identifier ':' '=' expression {
+      $$ = NODE(ASSIGNMENT_STATEMENT, NULL, 2, $1, $4);
+    }
+  | identifier '+' '=' expression {
+      $$ = NODE(ADD_STATEMENT, NULL, 2, $1, $4);
+    }
+  | identifier '-' '=' expression {
+      $$ = NODE(SUBTRACT_STATEMENT, NULL, 2, $1, $4);
+    }
+  | identifier '*' '=' expression {
+      $$ = NODE(MULTIPLY_STATEMENT, NULL, 2, $1, $4);
+    }
+  | identifier '/' '=' expression {
+      $$ = NODE(DIVIDE_STATEMENT, NULL, 2, $1, $4);
+    }
+  ;
+
+return_statement:
+    RETURN expression {
+      $$ = NODE(RETURN_STATEMENT, NULL, 1, $2);
+    }
+  ;
+
+print_statement:
+    PRINT print_list {
+      $$ = NODE(PRINT_STATEMENT, NULL, 1, $2);
+    }
+  ;
+
+null_statement:
+    CONTINUE {
+      $$ = NODE(NULL_STATEMENT, NULL, 0);
+    }
+  ;
+
+if_statement:
+    IF relation THEN statement {
+      $$ = NODE(IF_STATEMENT, NULL, 2, $2, $4);
+    }
+  | IF relation THEN statement ELSE statement {
+      $$ = NODE(IF_STATEMENT, NULL, 3, $2, $4, $6);
+    }
+  ;
+
+while_statement:
+    WHILE relation DO statement {
+      $$ = NODE(WHILE_STATEMENT, NULL, 2, $2, $4);
+    }
+  ;
+
+relation:
+    expression '=' expression {
+      $$ = NODE(RELATION, strdup("="), 2, $1, $3);
+    }
+  | expression '<' expression {
+      $$ = NODE(RELATION, strdup("<"), 2, $1, $3);
+    }
+  | expression '>' expression {
+      $$ = NODE(RELATION, strdup(">"), 2, $1, $3);
+    }
+  /* This can actually be extented with the following (with some minor tweaks to the generator.c)
+  | expression {
+      $$ = NODE(RELATION, NULL, 1, $1);
+    }
+    That will allow to have "if 1 then" or "while 1 do" */
+  ;
+
+expression:
+    expression '|' expression {
+      $$ = NODE(EXPRESSION, strdup("|"), 2, $1, $3);
+    }
+  | expression '^' expression {
+      $$ = NODE(EXPRESSION, strdup("^"), 2, $1, $3);
+    }
+  | expression '&' expression {
+      $$ = NODE(EXPRESSION, strdup("&"), 2, $1, $3);
+    }
+  | expression '+' expression {
+      $$ = NODE(EXPRESSION, strdup("+"), 2, $1, $3);
+    }
+  | expression '-' expression {
+      $$ = NODE(EXPRESSION, strdup("-"), 2, $1, $3);
+    }
+  | expression '*' expression {
+      $$ = NODE(EXPRESSION, strdup("*"), 2, $1, $3);
+    }
+  | expression '/' expression {
+      $$ = NODE(EXPRESSION, strdup("/"), 2, $1, $3);
+    }
+  | '-' expression %prec UMINUS {
+      $$ = NODE(EXPRESSION, strdup("-"), 1, $2);
+    }
+  | '~' expression {
+      $$ = NODE(EXPRESSION, strdup("~"), 1, $2);
+    }
+  | '(' expression ')' {
+      $$ = NODE(EXPRESSION, NULL /*strdup("group")*/, 1, $2);
+    }
+  | number {
+      $$ = NODE(EXPRESSION, NULL /*strdup("number")*/, 1, $1);
+    }
+  | identifier {
+      $$ = NODE(EXPRESSION, NULL /*strdup("identifier")*/, 1, $1);
+    }
+  | identifier '(' argument_list ')' {
+      $$ = NODE(EXPRESSION, /*NULL*/ strdup("function_call"), 2, $1, $3);
+    }
+  ;
+
+declaration: 
+    VAR variable_list {
+      $$ = NODE(DECLARATION, NULL, 1, $2);
+    }
+  ;
+
+print_item:
+    expression {
+      $$ = NODE(PRINT_ITEM, NULL, 1, $1);
+    }
+  | string {
+      $$ = NODE(PRINT_ITEM, NULL, 1, $1);
+    }
+  ;
+
+identifier:
+    IDENTIFIER {
+      $$ = NODE(IDENTIFIER_DATA, strdup(yytext), 0); // Zero children
+    }
+  ;
+
+number:
+    NUMBER {
+      uint64_t* p_number = malloc(sizeof(uint64_t));
+      *p_number = strtol(yytext, NULL, 10);
+      $$ = NODE(NUMBER_DATA, p_number, 0); // Zero children
+    }
+  ;
+
+string:
+    STRING {
+      $$ = NODE(STRING_DATA, strdup(yytext), 0); // Zero children
+    }
+  ;
+
+%%
+
+int
+yyerror ( const char *error )
+{
+    fprintf ( stderr, "%s on line %d\n", error, yylineno );
+    exit ( EXIT_FAILURE );
+}

exercises/06/vslc/src/scanner.l

@@ -0,0 +1,30 @@
+%{
+#include <vslc.h>
+%}
+%option noyywrap
+%option array
+%option yylineno
+
+WHITESPACE [\ \t\v\r\n]
+COMMENT \/\/[^\n]+
+QUOTED \"([^\"\n]|\\\")*\"
+%%
+{WHITESPACE}+           { /* Eliminate whitespace */ }
+{COMMENT}               { /* Eliminate comments */ }
+func                    { return FUNC; }
+print                   { return PRINT; }
+return                  { return RETURN; }
+continue                { return CONTINUE; }
+if                      { return IF; }
+then                    { return THEN; }
+else                    { return ELSE; }
+while                   { return WHILE; }
+do                      { return DO; }
+begin                   { return OPENBLOCK; }
+end                     { return CLOSEBLOCK; }
+var                     { return VAR; }
+[0-9]+                  { return NUMBER; }
+[A-Za-z_][0-9A-Za-z_]*  { return IDENTIFIER; }
+{QUOTED}                { return STRING; }
+.                       { return yytext[0]; }
+%%

exercises/06/vslc/src/tlhash.c

@@ -0,0 +1,281 @@
+#include <string.h>
+#include <stdlib.h>
+#include <stdint.h>
+
+#include <tlhash.h>
+
+/*********************************************************************
+ * Declarations of the utility functions for obtaining hashes, found *
+ * at the bottom of this file.                                       *
+ *********************************************************************/
+
+/* Little-endian, for x86-s */
+#define CRC32_IEEE802_3 (0xedb88320)
+static const uint32_t crc32_ieee802_3[256];
+static const uint32_t *crc_table = (uint32_t *)crc32_ieee802_3;
+
+static uint32_t crc32 ( void *input, size_t length );
+
+
+/********************************
+ * External interface functions *
+ ********************************/
+
+
+/* Initializer 
+ * Returns
+ *  ENOMEM - if allocation of table entries fails.
+ */
+int
+tlhash_init ( tlhash_t *tab, size_t n_buckets )
+{
+    size_t i;
+    tab->n_buckets = n_buckets;
+    tab->size = 0;
+    tab->buckets = (tlhash_element_t **) calloc (
+        n_buckets, sizeof(tlhash_element_t *)
+    );
+    if ( tab->buckets == NULL )
+        return TLHASH_ENOMEM;
+    for ( i=0; i<n_buckets; i++ )
+        tab->buckets[i] = NULL;
+    return TLHASH_SUCCESS;
+}
+
+
+/* Finalizer
+ * Returns
+ *  ENOENT - if there is no table to free.
+ */
+int
+tlhash_finalize ( tlhash_t *tab )
+{
+    size_t i;
+    if ( tab == NULL )
+        return TLHASH_ENOENT;
+    for ( i=0; i<tab->n_buckets; i++ )
+    {
+        tlhash_element_t *element = tab->buckets[i], *next;
+        while ( element != NULL )
+        {
+            next = element->next;
+            free ( element->key );
+            free ( element );
+            tab->size -= 1;
+            element = next;
+        }
+    }
+
+    free ( tab->buckets );
+    return TLHASH_SUCCESS;
+}
+
+
+/* Insert - find hash value, modulate over buckets, append to linked list
+ * Returns
+ *  EEXIST - if an element is already indexed by this key
+ *  ENOMEM - if allocation of element or key copy fails
+ */
+int
+tlhash_insert (
+    tlhash_t *tab, void *key, size_t key_length, void *value
+)
+{
+    void *test_entry;
+    int test = tlhash_lookup ( tab, key, key_length, &test_entry );
+    if ( test != TLHASH_ENOENT )
+        return TLHASH_EEXIST;
+    uint32_t hash = crc32 ( key, key_length );
+    size_t bucket = hash % tab->n_buckets;
+    tlhash_element_t *element = malloc ( sizeof(tlhash_element_t) );
+    if ( element == NULL )
+        return TLHASH_ENOMEM;
+    void *key_copy = malloc ( key_length );
+    if ( key_copy == NULL )
+    {
+        free ( element );
+        return TLHASH_ENOMEM;
+    }
+    memcpy ( key_copy, key, key_length );
+    element->key         = key_copy;
+    element->key_length  = key_length;
+    element->value       = value;
+    element->next        = tab->buckets[bucket];
+    tab->buckets[bucket] = element;
+    tab->size += 1;
+    return TLHASH_SUCCESS;
+}
+
+
+/* Lookup - find hash value, modulate over buckets, search linked list
+ * Returns
+ *  ENOENT - if no element is indexed by this key
+ */
+int
+tlhash_lookup (
+    tlhash_t *tab, void *key, size_t key_length, void **value
+)
+{
+    uint32_t hash = crc32 ( key, key_length );
+    size_t bucket = hash % tab->n_buckets;
+    tlhash_element_t *el = tab->buckets[bucket];
+
+    *value = NULL;
+    while ( el != NULL )
+    {
+        if ( el->key_length == key_length && ! memcmp(el->key,key,key_length) )
+        {
+            *value = el->value;
+            break;
+        }
+        el = el->next;
+    }
+    if ( el != NULL )
+        return TLHASH_SUCCESS;
+    else
+        return TLHASH_ENOENT;
+}
+
+
+/* Removal - find hash value, modulate over buckets, delete entry
+ * Returns
+ *  ENOENT - no such element to remove was found.
+ */
+int
+tlhash_remove ( tlhash_t *tab, void *key, size_t key_length )
+{
+    uint32_t hash = crc32 ( key, key_length );
+    size_t bucket = hash % tab->n_buckets;
+    tlhash_element_t *el = tab->buckets[bucket], *prev = NULL;
+
+    while ( el != NULL )
+    {
+        if ( el->key_length == key_length && ! memcmp(el->key,key,key_length) )
+        {
+            /* We have a match. */
+            if ( prev != NULL ) /* Remove from list if it's not the head */
+                prev->next = (void *)el->next;
+            else                /* Substitute it if it IS the head */
+                tab->buckets[bucket] = el->next;
+            /* Free the container and key copy allocated by this lib */
+            free ( el->key );
+            free ( el );
+            break;
+        }
+        prev = el;
+        el = el->next;
+    }
+    if ( el == NULL )
+        return TLHASH_ENOENT;
+    else
+    {
+        tab->size -= 1;
+        return TLHASH_SUCCESS;
+    }
+}
+
+
+size_t
+tlhash_size ( tlhash_t *tab )
+{
+    return tab->size;
+}
+
+
+void
+tlhash_keys ( tlhash_t *tab, void **keys )
+{
+    size_t b, i = 0;
+    for ( b=0; b<tab->n_buckets; b++ )
+    {
+        tlhash_element_t *el = tab->buckets[b];
+        while ( el != NULL )
+        {
+            keys[i] = el->key;
+            i += 1;
+            el = el->next;
+        }
+    }
+}
+
+
+void
+tlhash_values ( tlhash_t *tab, void **values )
+{
+    size_t b, i = 0;
+    for ( b=0; b<tab->n_buckets; b++ )
+    {
+        tlhash_element_t *el = tab->buckets[b];
+        while ( el != NULL )
+        {
+            values[i] = el->value;
+            i += 1;
+            el = el->next;
+        }
+    }
+}
+
+
+/***************************************
+ * Hashing function and IEEE data blob *
+ ***************************************/
+
+
+static uint32_t
+crc32 ( void *input, size_t length )
+{
+    const uint8_t *data = (uint8_t *)input;
+    size_t i = 0;
+    uint32_t hash = 0xFFFFFFFF;
+    for ( i = 0; i<length; i++ )
+        hash = (hash>>8) ^ crc_table [ data[i] ^ (uint8_t)hash ];
+    return (hash^0xFFFFFFFF);
+}
+
+
+static const uint32_t
+crc32_ieee802_3[256] = {
+    0x00000000, 0x77073096, 0xee0e612c, 0x990951ba, 0x076dc419, 0x706af48f,
+    0xe963a535, 0x9e6495a3, 0x0edb8832, 0x79dcb8a4, 0xe0d5e91e, 0x97d2d988,
+    0x09b64c2b, 0x7eb17cbd, 0xe7b82d07, 0x90bf1d91, 0x1db71064, 0x6ab020f2,
+    0xf3b97148, 0x84be41de, 0x1adad47d, 0x6ddde4eb, 0xf4d4b551, 0x83d385c7,
+    0x136c9856, 0x646ba8c0, 0xfd62f97a, 0x8a65c9ec, 0x14015c4f, 0x63066cd9,
+    0xfa0f3d63, 0x8d080df5, 0x3b6e20c8, 0x4c69105e, 0xd56041e4, 0xa2677172,
+    0x3c03e4d1, 0x4b04d447, 0xd20d85fd, 0xa50ab56b, 0x35b5a8fa, 0x42b2986c,
+    0xdbbbc9d6, 0xacbcf940, 0x32d86ce3, 0x45df5c75, 0xdcd60dcf, 0xabd13d59,
+    0x26d930ac, 0x51de003a, 0xc8d75180, 0xbfd06116, 0x21b4f4b5, 0x56b3c423,
+    0xcfba9599, 0xb8bda50f, 0x2802b89e, 0x5f058808, 0xc60cd9b2, 0xb10be924,
+    0x2f6f7c87, 0x58684c11, 0xc1611dab, 0xb6662d3d, 0x76dc4190, 0x01db7106,
+    0x98d220bc, 0xefd5102a, 0x71b18589, 0x06b6b51f, 0x9fbfe4a5, 0xe8b8d433,
+    0x7807c9a2, 0x0f00f934, 0x9609a88e, 0xe10e9818, 0x7f6a0dbb, 0x086d3d2d,
+    0x91646c97, 0xe6635c01, 0x6b6b51f4, 0x1c6c6162, 0x856530d8, 0xf262004e,
+    0x6c0695ed, 0x1b01a57b, 0x8208f4c1, 0xf50fc457, 0x65b0d9c6, 0x12b7e950,
+    0x8bbeb8ea, 0xfcb9887c, 0x62dd1ddf, 0x15da2d49, 0x8cd37cf3, 0xfbd44c65,
+    0x4db26158, 0x3ab551ce, 0xa3bc0074, 0xd4bb30e2, 0x4adfa541, 0x3dd895d7,
+    0xa4d1c46d, 0xd3d6f4fb, 0x4369e96a, 0x346ed9fc, 0xad678846, 0xda60b8d0,
+    0x44042d73, 0x33031de5, 0xaa0a4c5f, 0xdd0d7cc9, 0x5005713c, 0x270241aa,
+    0xbe0b1010, 0xc90c2086, 0x5768b525, 0x206f85b3, 0xb966d409, 0xce61e49f,
+    0x5edef90e, 0x29d9c998, 0xb0d09822, 0xc7d7a8b4, 0x59b33d17, 0x2eb40d81,
+    0xb7bd5c3b, 0xc0ba6cad, 0xedb88320, 0x9abfb3b6, 0x03b6e20c, 0x74b1d29a,
+    0xead54739, 0x9dd277af, 0x04db2615, 0x73dc1683, 0xe3630b12, 0x94643b84,
+    0x0d6d6a3e, 0x7a6a5aa8, 0xe40ecf0b, 0x9309ff9d, 0x0a00ae27, 0x7d079eb1,
+    0xf00f9344, 0x8708a3d2, 0x1e01f268, 0x6906c2fe, 0xf762575d, 0x806567cb,
+    0x196c3671, 0x6e6b06e7, 0xfed41b76, 0x89d32be0, 0x10da7a5a, 0x67dd4acc,
+    0xf9b9df6f, 0x8ebeeff9, 0x17b7be43, 0x60b08ed5, 0xd6d6a3e8, 0xa1d1937e,
+    0x38d8c2c4, 0x4fdff252, 0xd1bb67f1, 0xa6bc5767, 0x3fb506dd, 0x48b2364b,
+    0xd80d2bda, 0xaf0a1b4c, 0x36034af6, 0x41047a60, 0xdf60efc3, 0xa867df55,
+    0x316e8eef, 0x4669be79, 0xcb61b38c, 0xbc66831a, 0x256fd2a0, 0x5268e236,
+    0xcc0c7795, 0xbb0b4703, 0x220216b9, 0x5505262f, 0xc5ba3bbe, 0xb2bd0b28,
+    0x2bb45a92, 0x5cb36a04, 0xc2d7ffa7, 0xb5d0cf31, 0x2cd99e8b, 0x5bdeae1d,
+    0x9b64c2b0, 0xec63f226, 0x756aa39c, 0x026d930a, 0x9c0906a9, 0xeb0e363f,
+    0x72076785, 0x05005713, 0x95bf4a82, 0xe2b87a14, 0x7bb12bae, 0x0cb61b38,
+    0x92d28e9b, 0xe5d5be0d, 0x7cdcefb7, 0x0bdbdf21, 0x86d3d2d4, 0xf1d4e242,
+    0x68ddb3f8, 0x1fda836e, 0x81be16cd, 0xf6b9265b, 0x6fb077e1, 0x18b74777,
+    0x88085ae6, 0xff0f6a70, 0x66063bca, 0x11010b5c, 0x8f659eff, 0xf862ae69,
+    0x616bffd3, 0x166ccf45, 0xa00ae278, 0xd70dd2ee, 0x4e048354, 0x3903b3c2,
+    0xa7672661, 0xd06016f7, 0x4969474d, 0x3e6e77db, 0xaed16a4a, 0xd9d65adc,
+    0x40df0b66, 0x37d83bf0, 0xa9bcae53, 0xdebb9ec5, 0x47b2cf7f, 0x30b5ffe9,
+    0xbdbdf21c, 0xcabac28a, 0x53b39330, 0x24b4a3a6, 0xbad03605, 0xcdd70693,
+    0x54de5729, 0x23d967bf, 0xb3667a2e, 0xc4614ab8, 0x5d681b02, 0x2a6f2b94,
+    0xb40bbe37, 0xc30c8ea1, 0x5a05df1b, 0x2d02ef8d
+};

exercises/06/vslc/src/tree.c

@@ -0,0 +1,400 @@
+#include <vslc.h>
+
+static void node_print ( node_t *root, int nesting );
+static void simplify_tree ( node_t **simplified, node_t *root );
+static void node_finalize ( node_t *discard );
+
+typedef struct stem_t *stem;
+struct stem_t { const char *str; stem next; };
+static void
+tree_print(node_t* root, stem head);
+
+
+static void destroy_subtree ( node_t *discard );
+
+static void prune_children(node_t **simplified, node_t *root);
+static void resolve_constant_expressions(node_t **simplified, node_t *root);
+static void flatten(node_t **simplified, node_t *root);
+
+
+/* External interface */
+void
+destroy_syntax_tree ( void )
+{
+    destroy_subtree ( root );
+}
+
+
+void
+simplify_syntax_tree ( void )
+{
+    simplify_tree ( &root, root );
+}
+
+
+extern bool new_print_style;
+void
+print_syntax_tree ( void )
+{
+    if (new_print_style)
+        tree_print ( root, 0 );
+    // Old tree printing
+    else
+        node_print ( root, 0 );
+}
+
+
+// Changed so it returns the pointer to the new node, can be used as before, but makes the parser file cleaner
+node_t*
+node_init (node_t *nd, node_index_t type, void *data, uint64_t n_children, ...)
+{
+    va_list child_list;
+    *nd = (node_t) {
+        .type = type,
+        .data = data,
+        .entry = NULL,
+        .n_children = n_children,
+        .children = (node_t **) malloc ( n_children * sizeof(node_t *) )
+    };
+    va_start ( child_list, n_children );
+    for ( uint64_t i=0; i<n_children; i++ )
+        nd->children[i] = va_arg ( child_list, node_t * );
+    va_end ( child_list );
+
+    return nd;
+}
+
+
+static void
+tree_print(node_t* root, stem head) 
+{
+    static const char *sdown = " │", *slast = " └", *snone = "  ";
+    struct stem_t col = {0, 0}, *tail;
+
+    // Print stems of branches coming further down
+    for (tail = head; tail; tail = tail->next) {
+        if (!tail->next) {
+            if (!strcmp(sdown, tail->str))
+                printf(" ├");
+            else 
+                printf("%s", tail->str);
+            break;
+        }
+        printf("%s", tail->str);
+    }
+    
+    if (root == NULL) {
+        // Secure against null pointers sent as root
+        printf("─(nil)\n");
+        return;
+    }
+    printf("─%s", node_string[root->type]);
+    if ( root->type == IDENTIFIER_DATA  ||
+         root->type == STRING_DATA      ||
+         root->type == EXPRESSION       ||
+         root->type == RELATION)  
+        printf("(%s)", (char *) root->data);
+    else if (root->type == NUMBER_DATA)
+        printf("(%ld)", *((int64_t *)root->data));
+    putchar('\n');
+ 
+    if (!root->n_children) return;
+ 
+    if (tail && tail->str == slast)
+        tail->str = snone;
+ 
+    if (!tail)  tail = head = &col;
+    else        tail->next = &col;
+
+    for ( int64_t i=0; i < root->n_children; i++ ) {
+        col.str = root->n_children - i - 1 ? sdown : slast;
+        tree_print(root->children[i], head);
+    }
+    tail->next = 0;
+}
+
+/* Internal choices */
+static void
+node_print ( node_t *root, int nesting )
+{
+    if ( root != NULL )
+    {
+        printf ( "%*c%s", nesting, ' ', node_string[root->type] );
+        if ( root->type == IDENTIFIER_DATA ||
+             root->type == STRING_DATA ||
+             root->type == EXPRESSION ) 
+            printf ( "(%s)", (char *) root->data );
+        else if ( root->type == NUMBER_DATA )
+            printf ( "(%ld)", *((int64_t *)root->data) );
+        putchar ( '\n' );
+        for ( int64_t i=0; i<root->n_children; i++ )
+            node_print ( root->children[i], nesting+1 );
+    }
+    else
+        printf ( "%*c%p\n", nesting, ' ', root );
+}
+
+
+static void
+node_finalize ( node_t *discard )
+{
+    if ( discard != NULL )
+    {
+        free ( discard->data );
+        free ( discard->children );
+        free ( discard );
+    }
+}
+
+
+static void
+destroy_subtree ( node_t *discard )
+{
+    if ( discard != NULL )
+    {
+        for ( uint64_t i=0; i<discard->n_children; i++ )
+            destroy_subtree ( discard->children[i] );
+        node_finalize ( discard );
+    }
+}
+
+
+static void
+flatten(node_t **simplified, node_t *root)
+{
+    /* This will flatten left-expanded lists */
+    if (!root)
+        return;
+
+    /* Do this recursivly */
+    for (int i = 0; i < root->n_children; i++)
+        flatten(&root->children[i], root->children[i]);
+
+    node_t **new_children, *result = root;
+    switch (root->type)
+    {
+    case GLOBAL_LIST:
+    case STATEMENT_LIST:
+    case PRINT_LIST:
+    case EXPRESSION_LIST:
+    case VARIABLE_LIST:
+    case DECLARATION_LIST:
+        // Check if node have more than two children
+        if (root->n_children < 2)
+            break;
+        
+        result = root->children[0];
+        result->n_children++;
+
+        // Realloc the array of children to the new size 
+        if (!(new_children = realloc(result->children, result->n_children * sizeof(node_t*))))
+            break;
+        // if successs, insert the new array
+        result->children = new_children;
+
+        // Insert child at the end
+        result->children[result->n_children - 1] = root->children[1];
+
+        node_finalize(root);
+        break;
+    }
+
+    *simplified = result;
+}
+
+
+static void
+prune_children(node_t **simplified, node_t *root) 
+{
+    if (!root)
+        return;
+
+    /* Do this recursivly */
+    for (int i = 0; i < root->n_children; i++)
+        prune_children(&root->children[i], root->children[i]);
+
+    node_t *result = root;
+    switch (root->type)
+    {
+    case PROGRAM:
+    case GLOBAL:
+    //case ARGUMENT_LIST:     // For this to work, need to change order of operations
+    //case PARAMETER_LIST:    // For this to work, need to change order of operations
+    //case VARIABLE_LIST:
+    //case EXPRESSION_LIST:
+    case DECLARATION:
+    case STATEMENT:
+    case PRINT_ITEM:
+    case PRINT_STATEMENT:
+        result = root->children[0];
+
+        // The print_statement only contains a print_list, still need a print_statement.
+        if (root->type == PRINT_STATEMENT)
+            result->type = PRINT_STATEMENT;
+        
+        node_finalize(root);
+        break;
+    }
+
+    *simplified = result;
+}
+
+
+static void
+resolve_constant_expressions(node_t **simplified, node_t *root)
+{
+    if (!root)
+        return;
+
+    /* Do this recursivly */
+    for (int i = 0; i < root->n_children; i++)
+        resolve_constant_expressions(&root->children[i], root->children[i]);
+    
+    if (root->type != EXPRESSION)
+        return;
+    
+    node_t *result = root;
+    
+    switch (root->n_children)
+    {
+    case 1:
+        result = root->children[0];
+        
+        if (root->data && 
+            result->type == NUMBER_DATA && 
+            result->data)
+        {
+            switch (*((char*)root->data))
+            {
+            case '-':
+                *((int64_t*)result->data) *= -1;
+                break;
+            case '~':
+                *((int64_t*)result->data) = ~*((int64_t*)result->data);
+                break;
+            }
+            node_finalize(root);
+        }
+        else if (!root->data)
+            node_finalize(root);
+        else
+            result = root;
+
+        break;
+    
+    case 2:
+        // Both children needs to be numbers to resolve constants
+        if (root->children[0]->type == NUMBER_DATA && 
+            root->children[1]->type == NUMBER_DATA)
+        {
+            // Check if children does not contain null pointers
+            if (!root->children[0]->data)
+                break;
+            if (!root->children[1]->data)
+                break;
+
+            // Check if data field is not null pointer
+            if (!root->data)
+                break;
+            
+            result = root->children[0];
+            int64_t
+                *lhs = result->data,
+                *rhs = root->children[1]->data;
+
+            switch (*(char*)root->data)
+            {
+            /* Assignments */
+            case '|': *lhs |= *rhs; break;
+            case '^': *lhs ^= *rhs; break;
+            case '&': *lhs &= *rhs; break;
+            case '+': *lhs += *rhs; break;
+            case '-': *lhs -= *rhs; break;
+            case '*': *lhs *= *rhs; break;
+            case '/': *lhs /= *rhs; break;
+            }
+
+            node_finalize(root->children[1]);
+            node_finalize(root);
+        }            
+        
+        
+        break;
+    }
+    *simplified = result;
+}
+
+
+static void
+resolve_constant_relations( node_t** simplified, node_t* root)
+{
+    if (!root)
+        return;
+
+    /* Do this recursivly */
+    for (int i = 0; i < root->n_children; i++)
+        resolve_constant_relations(&root->children[i], root->children[i]);
+    
+    if (root->type != RELATION)//|| root->type != RELATION)
+        return;
+
+    node_t *result = root;
+
+    if (root->n_children != 2)
+        return;
+
+    // Both children must be constant numbers
+    if (root->children[0]->type != NUMBER_DATA ||
+        root->children[1]->type != NUMBER_DATA)
+        return;
+
+    // Check if children does not contain null pointers
+    if (!root->children[0]->data)
+        return;
+    if (!root->children[1]->data)
+        return;
+
+    // Check if data field is not null pointer
+    if (!root->data)
+        return;
+
+    result = root->children[0];
+    int64_t
+        *lhs = result->data,
+        *rhs = root->children[1]->data;
+
+    switch (*(char*)root->data)
+    {
+    /* Relations */
+    case '=': *lhs = (*lhs == *rhs); break;
+    case '<': *lhs = (*lhs <  *rhs); break;
+    case '>': *lhs = (*lhs >  *rhs); break;
+    }
+
+    node_finalize(root->children[1]);
+    node_finalize(root);
+
+    *simplified = result;
+}
+
+
+static void
+simplify_tree ( node_t **simplified, node_t *root )
+{
+    if (!root) 
+        return;
+
+    /*  
+        Each of  the functions do their operations recursivly.
+        This opens up for a lot more flexibility, like removing 
+        variable list after it is flatten
+    */
+    flatten(&root, root);
+    prune_children(&root, root);
+    resolve_constant_expressions(&root, root);
+
+    // The following is experimental, will resolve the constant relations
+    resolve_constant_relations(&root, root);
+
+    *simplified = root;
+}

exercises/06/vslc/src/vslc.c

@@ -0,0 +1,83 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <getopt.h>
+#include <vslc.h>
+
+
+/* Global state */
+
+node_t *root;                   // Syntax tree
+tlhash_t *global_names;         // Symbol table
+char **string_list;             // List of strings in the source
+size_t n_string_list    = 8;    // Initial string list capacity (grow on demand)
+size_t stringc          = 0;    // Initial string count
+
+/* Command line option parsing for the main function */
+static void options ( int argc, char **argv );
+bool
+    print_full_tree             = false,
+    print_simplified_tree       = false,
+    print_symbol_table_contents = false,
+    print_generated_program     = true,
+    new_print_style             = true;
+
+
+/* Entry point */
+int
+main ( int argc, char **argv )
+{
+    options ( argc, argv );
+
+    yyparse();                  // Generated from grammar/bison, constructs syntax tree
+    yylex_destroy();            // Free heap used by flex
+
+    if ( print_full_tree )
+        print_syntax_tree ();
+    simplify_syntax_tree ();    // In tree.c
+    if ( print_simplified_tree )
+        print_syntax_tree ();
+
+    create_symbol_table ();     // In ir.c
+    if ( print_symbol_table_contents )
+        print_symbol_table();
+
+    if ( print_generated_program )
+        generate_program ();    // In generator.c
+
+
+    destroy_syntax_tree ();     // In tree.c
+    destroy_symbol_table ();    // In ir.c
+}
+
+
+static const char *usage =
+"Command line options\n"
+"\t-h\tOutput this text and halt\n"
+"\t-t\tOutput the full syntax tree\n"
+"\t-T\tOutput the simplified syntax tree\n"
+"\t-s\tOutput the symbol table contents\n"
+"\t-q\tQuiet: suppress output from the code generator\n"
+"\t-u\tDo not use print style more like the tree command\n";
+
+
+static void
+options ( int argc, char **argv )
+{
+    int o;
+    while ( (o=getopt(argc,argv,"htTsqu")) != -1 )
+    {
+        switch ( o )
+        {
+            case 'h':
+                printf ( "%s:\n%s", argv[0], usage );
+                exit ( EXIT_FAILURE );
+                break;
+            case 't':   print_full_tree = true;             break;
+            case 'T':   print_simplified_tree = true;       break;
+            case 's':   print_symbol_table_contents = true; break;
+            case 'q':   print_generated_program = false;    break;
+            case 'u':   new_print_style = false;            break;
+        }
+    }
+}
+

exercises/06/vslc/vsl_programs/Makefile

@@ -0,0 +1,45 @@
+VSLC := ../src/vslc
+AS := gcc
+
+# This updated makefile calls the relevant print for each assignment
+# The added target compile will attempt to compile all vsl source files,
+# for ps5 and ps6. Other programs are mostly not interesting after the their
+# targeted assignment.
+
+# Call `vslc -h` to see the available flags, and call `vslc [flags] < file.vsl`
+# to compile a single file.
+
+PS2_EXAMPLES := $(patsubst ps2-parser/%.vsl, ps2-parser/%.ast, $(wildcard ps2-parser/*.vsl))
+PS3_EXAMPLES := $(patsubst ps3-simplify/%.vsl, ps3-simplify/%.sast, $(wildcard ps3-simplify/*.vsl))
+PS4_EXAMPLES := $(patsubst ps4-symtab/%.vsl, ps4-symtab/%.sym, $(wildcard ps4-symtab/*.vsl))
+PS5_EXAMPLES := $(patsubst ps5-codegen1/%.vsl, ps5-codegen1/%.S, $(wildcard ps5-codegen1/*.vsl))
+PS6_EXAMPLES := $(patsubst ps6-codegen2/%.vsl, ps6-codegen2/%.S, $(wildcard ps6-codegen2/*.vsl))
+
+PS5_OBJECTS := $(PS5_EXAMPLES:.S=.bin)
+PS6_OBJECTS := $(PS6_EXAMPLES:.S=.bin)
+# OBJECTS := $(PS5_OBJECTS) $(PS4_EXAMPLES:.sym=.bin) $(PS3_EXAMPLES:.sast=.bin) $(PS2_EXAMPLES:.ast=.bin)
+OBJECTS := $(PS5_OBJECTS) $(PS6_OBJECTS)
+all: clean $(PS2_EXAMPLES) $(PS3_EXAMPLES) $(PS4_EXAMPLES) $(PS5_EXAMPLES) $(PS6_EXAMPLES)
+ps2: $(PS2_EXAMPLES)
+ps3: $(PS3_EXAMPLES)
+ps4: $(PS4_EXAMPLES)
+ps5: $(PS5_EXAMPLES)
+
+%.ast: %.vsl
+	$(VSLC) -t -q < $^ > $@ 2> $@
+%.sast: %.vsl
+	$(VSLC) -T -q < $^ > $@ 2> $@
+%.sym: %.vsl
+	$(VSLC) -s -q < $^ > $@ 2> $@
+%.S: %.vsl
+	$(VSLC) < $^ > $@
+# This target is only tested on x86-linux
+%.bin: %.S
+	$(AS) -no-pie -o $@ $^
+
+ps5-compile: $(PS5_OBJECTS)
+ps6-compile: $(PS6_OBJECTS)
+compile: $(OBJECTS)
+
+clean:
+	-rm -r */*.ast */*.sast */*.sym */*.bin */*.S

exercises/06/vslc/vsl_programs/ps2-parser/assignments.vsl

@@ -0,0 +1,14 @@
+// checking that comments are ignored
+
+// This program checks the assignment operators
+
+func main()
+begin
+    var a
+    a := 3
+    a += 1
+    a /= 2
+    a *= 32
+    a -= 2
+    print a
+end

exercises/06/vslc/vsl_programs/ps2-parser/function_call.vsl

@@ -0,0 +1,8 @@
+func add(a, b) begin
+    return a + b
+end
+
+func main() 
+begin
+    print add(40, 2)
+end

exercises/06/vslc/vsl_programs/ps2-parser/helloworld.vsl

@@ -0,0 +1,3 @@
+func main() begin
+    print "Hello, World!"
+end

exercises/06/vslc/vsl_programs/ps2-parser/if_else.vsl

@@ -0,0 +1,25 @@
+func main()
+begin
+    var a, b, c, d
+    c := 1
+    a := 3
+    b := a + c // 4
+    d := a * 100 + 50
+    print "a", a
+    print "b", b
+    print "c", c
+    print "d", d
+    if a = 14 then
+        print 1, "N", d / 5 + a, "RPR", a, "TERS "
+    else
+        print "COMP", c, "L", "ERS "
+    
+    print b, "R", a, " "
+
+    if a < b then
+        if d > 42 then
+            print b, "W", d, "ME"
+        else
+            print "L", b, "M", c
+    // A dangling else, what could go wrong?
+end

exercises/06/vslc/vsl_programs/ps2-parser/variables.vsl

@@ -0,0 +1,19 @@
+
+var global_var
+
+func my_func(param)
+begin
+    var local_var, local_var2
+    local_var := 1
+end
+
+var glob1, glob2
+
+func main()
+begin
+    var main_local_var
+    begin
+        var main_local_nested_var
+        main_local_nested_var := main_local_var
+    end
+end

exercises/06/vslc/vsl_programs/ps2-parser/while.vsl

@@ -0,0 +1,10 @@
+// check parsing of do-while loop
+
+func main()
+begin   
+    var i
+    i := 2
+    while i < 9000 do
+        i := i * i
+    print i
+end

exercises/06/vslc/vsl_programs/ps3-simplify/constants.vsl

@@ -0,0 +1,9 @@
+
+func main() begin
+    var a, b
+    a := 1 + 2 + 4 + 5 + 6 + 7 + 8 + 9
+    b := (10 + 10 * 4) * (2 + 2 * (1 + 1)) / 10 + 2 * 5 + 6 / 3
+
+    if a = b then
+        print "The answer is", b
+end

exercises/06/vslc/vsl_programs/ps3-simplify/euclid.vsl

@@ -0,0 +1,20 @@
+func euclid ( a, b )
+begin
+    if a < 0 then a := -a
+    if b < 0 then b := -b
+    if gcd ( a, b ) > 1 then
+        print "Greatest common divisor of", a, "and", b, "is", gcd ( a, b )
+    else
+        print a, "and", b, "are relative primes"
+    return 0
+end
+
+func gcd( a, b )
+begin
+    var g
+    if b > 0 then
+        g := gcd ( b, a - ((a/b)*b) )
+    else
+        g := a
+    return g
+end

exercises/06/vslc/vsl_programs/ps3-simplify/lists.vsl

@@ -0,0 +1,33 @@
+
+func my_func(a, b, c, d, e, f, g, h) begin
+    var i, j, k, l, m
+
+    i := a + b + d
+
+    if i = f then begin
+        print "hmmm"
+    end
+
+    if 1 < 2 then begin
+        print "Whaa"
+    end
+    else begin
+        print "Whooo"
+    end
+
+
+
+    return i
+end
+
+func main() begin
+    var n, o, p, q, r, s, t, u, v, w
+    n := 5
+    n += my_func(1, 2, 3, 5, 8, 13, 21, 34)
+
+    if my_func(1, 2, 3, 5, 8, 13, 21, 34) > 3 then begin
+        print "True!"
+    end
+
+    return 0
+end

exercises/06/vslc/vsl_programs/ps3-simplify/while_test.vsl

@@ -0,0 +1,57 @@
+// This program is a simple test of while loops, counting down from 19 to 0
+// and skipping 10 (if continue is implemented)
+
+func while_test ()
+begin
+    var a, b
+    a := 20
+    
+    b := test_while()
+    
+    print a
+
+
+    if a > 0 then print "foobar"
+    while a > 0 do
+    begin
+        if a = 10 then
+        begin
+            a -= 1
+            print "Skip..."
+            continue
+        end
+        else
+            a -= 1
+        print a
+    end
+    return 0
+end
+
+func test_while()
+begin
+    var n, m
+
+    n := 4
+    m := 21
+
+    while n > 0 do
+    begin
+        n -= 1
+        if n = 2 then
+            continue
+
+        while m > 0 do
+        begin
+            m -= 1
+
+            if m = 10 then
+                continue
+
+            print n, m
+        end
+
+
+    end
+
+    return 0
+end

exercises/06/vslc/vsl_programs/ps4-symtab/globals.vsl

@@ -0,0 +1,11 @@
+var global_var0, global_var1
+
+func my_func(param0, param1) begin
+    var a
+    return 0
+end
+
+func main() begin
+    var a
+    print "a string"
+end

exercises/06/vslc/vsl_programs/ps4-symtab/shadow.vsl

@@ -0,0 +1,24 @@
+
+func main() begin
+    var a, b
+    a := 1
+    begin
+        var a
+        a := 2
+        b := 40
+        begin
+            var a
+            a := b + 2
+            print a, b
+        end
+        print a
+        begin
+            var b
+            b := 38
+            a := b + 3
+            print a, b
+        end
+        print a
+    end
+    print b
+end

exercises/06/vslc/vsl_programs/ps5-codegen1/function_call.vsl

@@ -0,0 +1,10 @@
+func add(a, b) begin
+    print "adding", a, "and", b
+    return a + b
+end
+
+func main() 
+begin
+    print 2 + add(40, 2) + 2
+    return 0
+end

exercises/06/vslc/vsl_programs/ps5-codegen1/helloworld.vsl

@@ -0,0 +1,4 @@
+func main() begin
+    print "Hello, World!"
+    return 0
+end

exercises/06/vslc/vsl_programs/ps5-codegen1/ps5.vsl

@@ -0,0 +1,30 @@
+// This program tests activation records, function call and return
+func funcall ()
+begin
+    var x,y,z
+    x := 5
+    y := 10
+    print "Calling my_function with parameters", x, y
+    z := my_function ( x, y )
+    print "The returned result is", z
+    z := my_other_function ()
+    print "The other returned result is", z
+    return 0
+end
+
+func my_function ( s, t )
+begin
+    var u
+    u := s*s + t*t
+    print "Parameter s is", s
+    print "Parameter t is", t
+    print "The sum of their squares is", u
+    return u
+end
+
+func my_other_function ()
+begin
+    var x
+    x := 42
+    return x
+end

exercises/06/vslc/vsl_programs/ps5-codegen1/shadow.vsl

@@ -0,0 +1,25 @@
+
+func main() begin
+    var a, b
+    a := 1
+    begin
+        var a
+        a := 2
+        b := 40
+        begin
+            var a
+            a := b + 2
+            print a, b
+        end
+        print a
+        begin
+            var b
+            b := 38
+            a := b + 3
+            print a, b
+        end
+        print a
+    end
+    print b
+    return 0
+end

exercises/06/vslc/vsl_programs/ps6-codegen2/euclid.vsl

@@ -0,0 +1,20 @@
+func euclid ( a, b )
+begin
+    if a < 0 then a := -a
+    if b < 0 then b := -b
+    if gcd ( a, b ) > 1 then
+        print "Greatest common divisor of", a, "and", b, "is", gcd ( a, b )
+    else
+        print a, "and", b, "are relative primes"
+    return 0
+end
+
+func gcd( a, b )
+begin
+    var g
+    if b > 0 then
+        g := gcd ( b, a - ((a/b)*b) )
+    else
+        g := a
+    return g
+end

exercises/06/vslc/vsl_programs/ps6-codegen2/lists.vsl

@@ -0,0 +1,35 @@
+
+func my_func(a, b, c, d, e, f, g, h) begin
+    var i, j, k, l, m
+
+    i := a + b + d
+
+    print i, f
+
+    if i = f then begin
+        print "hmmm"
+    end
+
+    if 1 < 2 then begin
+        print "Whaa"
+    end
+    else begin
+        print "Whooo"
+    end
+
+
+
+    return i
+end
+
+func main() begin
+    var n, o, p, q, r, s, t, u, v, w
+    n := 5
+    n += my_func(1, 2, 3, 5, 8, 13, 21, 34)
+
+    if my_func(1, 2, 3, 5, 8, 13, 21, 34) > 3 then begin
+        print "True!"
+    end
+
+    return 0
+end

exercises/06/vslc/vsl_programs/ps6-codegen2/newton.vsl

@@ -0,0 +1,21 @@
+// Approximate square root by the Newton/Raphson method for f(x) = x^2 - n
+// f(x) = x^2 - n = 0
+// f'(x) = 2x
+// x{n+1} = x{n} - (x^2-n) / 2x 
+
+func newton ( n )
+begin
+    print "The square root of", n, "is", improve ( n, 1 )
+    return 0
+end
+
+func improve ( n, estimate )
+begin
+    var next
+    next := estimate - ( (estimate * estimate - n) / ( 2 * estimate ) )
+    if next - estimate = 0 then
+        // Integer precision converges at smallest int greater than the square
+        return next-1
+    else
+        return improve ( n, next )
+end

exercises/06/vslc/vsl_programs/ps6-codegen2/while_test.vsl

@@ -0,0 +1,83 @@
+// This program is a simple test of while loops, counting down from 19 to 0
+// and skipping 10 (if continue is implemented)
+
+func while_test ()
+begin
+    var a, b
+    a := 20
+
+    b := test_while()
+
+    print "Loops done in test_while:", b
+    
+    print a
+
+
+    if a > 0 then print "foobar"
+    while a > 0 do
+    begin
+        if a = 10 then
+        begin
+            a -= 1
+            print "Skip..."
+            continue
+        end
+        else
+            a -= 1
+        print a
+    end
+    return 0
+end
+
+func test_while()
+begin
+    var n, m, loops_done
+
+    m := 10
+    n := 5
+
+    loops_done := 0
+
+
+    while m > 0 do
+    begin
+        
+        if m = 7 then
+        begin
+            m -= 1
+            print "Skipping loop 7"
+            continue
+        end
+        
+        n := 5
+        while n > 0 do
+        begin
+            
+
+            if n = 3 then
+            begin
+                n -= 1
+                print "-------Skip 3 in inner loop"
+                continue
+            end
+
+
+            loops_done += 1
+            print "loop:", loops_done, "Values:", m, n
+            n -= 1
+        end
+
+        if m = 5 then
+        begin
+            m -= 1
+            print "<<<<<<Skip after exit of inner while"
+            continue
+        end
+
+        print "######## Loop", m, "done"
+        m -= 1
+
+    end
+
+    return loops_done
+end

exercises/vsl.py

@@ -0,0 +1,23 @@
+from pygments.lexer import RegexLexer, bygroups
+from pygments.token import *
+
+class VSLLexer(RegexLexer):
+    name        = "VSL"
+    aliases     = ["vsl"]
+    filenames   = ["*.vsl"]
+
+    tokens = {
+        "root": [
+            (r"[\ \t\v\r\n]+", Whitespace),
+            (r"\/\/[^\n]+", Comment.Single),
+            (r"var", Keyword.Declaration),
+            (r"func|print|return|continue|if|then|else|while|do|begin|end", Keyword),
+            (r"\^|\||:|=|\+|-|\*|\/|<|>|&", Operator),
+            (r"[0-9]+", Number.Integer),
+            (r"([A-Za-z_][0-9A-Za-z_]*)([\ \t\v\r\n]*)(\()", bygroups(Name.Function, Whitespace, Punctuation)),
+            (r"\(|\)|\[|\]|{|}", Punctuation),
+            (r"[A-Za-z_][0-9A-Za-z_]*", Name.Variable),
+            (r"\"([^\"\n]|\\\")*\"", String),
+            (r".", Text),
+        ]
+    }