TDT4205/exercises/05/vslc/src/ir.c

#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;
}