42sh/src/lexer/lexer.c

#include "lexer.h"

#include <ctype.h>
#include <stdbool.h>
#include <stddef.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#include "io_backend/io_backend.h"
#include "utils/string_utils/string_utils.h"

static char *end_last_token;
static ssize_t remaining_chars;

/* @brief: saves state for the next call the the lexer.
 *
 */
static void save_state(char *stream, ssize_t i)
{
    remaining_chars -= i;
    end_last_token = stream + i;
    return;
}

/* @return: true if a special character from the grammar was found,
 *          false otherwise.
 *
 */
static bool is_special_char(char c)
{
    return c == '\'' || c == '\n' || c == ';' || c == EOF;
}

/* @brief: if a special character is found at [begin],
 * [tok->token_type] is set accordingly
 *
 */
static void set_token_spechar(struct token *tok, char *begin, ssize_t size)
{
    if (size != 1)
        return;
    if (begin[0] == EOF)
    {
        tok->type = TOKEN_EOF;
    }
    else if (begin[0] == ';')
    {
        tok->type = TOKEN_NEWLINE;
    }
    else if (begin[0] == '\'')
    {
        tok->type = TOKEN_QUOTE;
    }
    else if (begin[0] == ';')
    {
        tok->type = TOKEN_SEMICOLON;
    }
}

/* @brief: if a keyword is found at [begin],
 * [tok->token_type] is set accordingly
 *
 */
static void set_token_keyword(struct token *tok, char *begin, ssize_t size)
{
    if (tok->type != TOKEN_NULL)
        return;
    if (strncmp(begin, "if", size) == 0)
    {
        tok->type = TOKEN_IF;
    }
    else if (strncmp(begin, "fi", size) == 0)
    {
        tok->type = TOKEN_FI;
    }
    else if (strncmp(begin, "then", size) == 0)
    {
        tok->type = TOKEN_THEN;
    }
    else if (strncmp(begin, "else", size) == 0)
    {
        tok->type = TOKEN_ELSE;
    }
    else if (strncmp(begin, "elif", size) == 0)
    {
        tok->type = TOKEN_ELIF;
    }

    tok->data = calloc(size + 1, sizeof(char));
    if (tok->data == NULL)
        return;
    strncpy(tok->data, begin, size);
}

/* @brief: if token_type has not yet been set, then it is a TOKEN_WORD
 *         Also allocates the data and fills it.
 */
static void set_token_word(struct token *tok, char *begin, ssize_t size)
{
    if (tok->type == TOKEN_NULL)
    {
        tok->data = calloc(size + 1, sizeof(char));
        if (tok->data == NULL)
            return;
        strncpy(tok->data, begin, size);
    }
}

struct token *new_token(char *begin, ssize_t size)
{
    struct token *tok = calloc(1, sizeof(struct token));
    if (tok == NULL)
        return NULL;

    set_token_spechar(tok, begin, size);
    set_token_keyword(tok, begin, size);
    set_token_word(tok, begin, size);

    return tok;
}

void free_token(struct token *tok)
{
    if (tok == NULL)
        return;
    if (tok->data != NULL)
        free(tok->data);
    free(tok);
}

char *stream_init(void)
{
    char *stream;

    if (remaining_chars == 0)
    {
        remaining_chars = stream_read(&stream);
    }
    else
    {
        stream = end_last_token;
    }

    char *trimed_stream = trim_blank_left(stream);
    remaining_chars -= trimed_stream - stream;
    stream = trimed_stream;

    return stream;
}

struct token *peek_token(void)
{
    char *stream = stream_init();

    ssize_t i = 0;

    while (i < remaining_chars)
    {
        if (is_special_char(stream[i]))
        {
            if (i == 0) // where we create spe_char token
                i++;
            break;
        }
        if (isblank(stream[i]))
        {
            break;
        }
        i++;
    }

    return new_token(stream, i);
}

struct token *pop_token(void)
{
    char *stream = stream_init();

    ssize_t i = 0;

    while (i < remaining_chars)
    {
        if (is_special_char(stream[i]))
        {
            if (i == 0) // where we create spe_char token
                i++;
            break;
        }
        if (isblank(stream[i]))
        {
            break;
        }
        i++;
    }

    save_state(stream, i);

    return new_token(stream, i);
}