- FEAT: add source code length field in `fzl_reader_t` struct - FIX: check name, lpair (lparen) and rpair (rparen) when lexing call - FIX: check string literal termination - FIX: increment position even on empty string to prevent infinite lexing loop - FIX: add string terminator in `fzl_mktoken` when allocating space for string value - FIX: add check for errors in recursevly called `walk` - FIX: change source code holder from stack-stored `char[8192]` to dynamically allocated with actual file size holder - FIX: correct constant hasing util to hash actual string value instead of link
193 lines
4.0 KiB
C
193 lines
4.0 KiB
C
#include <compiler/fzl_reader.h>
|
|
#include <stdbool.h>
|
|
#include <stdio.h>
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
|
|
// Last error from reader
|
|
static char last_err[256] = {0};
|
|
|
|
fzl_reader_t *fzl_mkreader(char *source)
|
|
{
|
|
fzl_reader_ll_t *first_node = malloc(sizeof(fzl_reader_ll_t));
|
|
first_node->token = NULL;
|
|
first_node->next = NULL;
|
|
|
|
fzl_reader_t *reader = malloc(sizeof(fzl_reader_t));
|
|
reader->first = first_node;
|
|
reader->last = first_node;
|
|
reader->source = source;
|
|
reader->size = strlen(source);
|
|
reader->position = 0;
|
|
reader->at_end = true;
|
|
|
|
return reader;
|
|
}
|
|
|
|
// --- Read zone ---
|
|
|
|
// Try parse name
|
|
fzl_token_t *parse_name(fzl_reader_t *reader)
|
|
{
|
|
size_t lpos = reader->position;
|
|
size_t size = 0;
|
|
|
|
do
|
|
{
|
|
lpos++;
|
|
size++;
|
|
} while ((reader->source[lpos] >= 'A' && reader->source[lpos] <= 'Z') ||
|
|
(reader->source[lpos] >= 'a' && reader->source[lpos] <= 'z'));
|
|
|
|
char *value = malloc(size);
|
|
memcpy(value, reader->source + reader->position, size);
|
|
|
|
size_t start = reader->position;
|
|
reader->position += size;
|
|
|
|
fzl_token_t *tok =
|
|
fzl_mktoken(FZL_TOKTYPE_NAME, start, reader->position - 1, value, size);
|
|
free(value);
|
|
|
|
return tok;
|
|
}
|
|
|
|
// Try parse string literal
|
|
fzl_token_t *parse_string_literal(fzl_reader_t *reader)
|
|
{
|
|
size_t lpos = reader->position;
|
|
size_t size = 0;
|
|
|
|
if (reader->source[lpos] != '\'' && reader->source[lpos] != '"')
|
|
return NULL;
|
|
|
|
char open_symbol = reader->source[lpos];
|
|
char close_symbol = '\0';
|
|
|
|
do
|
|
{
|
|
lpos++;
|
|
size++;
|
|
|
|
if (reader->source[lpos] == open_symbol)
|
|
close_symbol = open_symbol;
|
|
|
|
} while (reader->source[lpos] != open_symbol && lpos < reader->size);
|
|
size++;
|
|
|
|
if (close_symbol != open_symbol)
|
|
return NULL;
|
|
|
|
if (size == 2)
|
|
{
|
|
fzl_token_t *tok = fzl_mktoken(FZL_TOKTYPE_STR_LITERAL, reader->position,
|
|
reader->position + 1, "", 0);
|
|
reader->position += 2;
|
|
return tok;
|
|
}
|
|
|
|
char *value = malloc(size - 2);
|
|
memcpy(value, reader->source + reader->position + 1, size - 2);
|
|
fzl_token_t *tok = fzl_mktoken(FZL_TOKTYPE_STR_LITERAL, reader->position,
|
|
lpos, value, size - 2);
|
|
free(value);
|
|
|
|
reader->position += size;
|
|
|
|
return tok;
|
|
}
|
|
|
|
int fzl_read(fzl_reader_t *reader)
|
|
{
|
|
if (reader->position >= reader->size)
|
|
return 0;
|
|
|
|
do
|
|
{
|
|
size_t p = reader->position;
|
|
fzl_token_t *token;
|
|
char curr = reader->source[p];
|
|
|
|
if (curr == '\n' || curr == ' ' || curr == '\r')
|
|
{
|
|
reader->position++;
|
|
continue;
|
|
}
|
|
|
|
if (curr == '\'' || curr == '"')
|
|
token = parse_string_literal(reader);
|
|
else if (curr == '(')
|
|
{
|
|
token = fzl_mktoken(FZL_TOKTYPE_LPAIR, p, p, NULL, 0);
|
|
reader->position++;
|
|
}
|
|
else if (curr == ')')
|
|
{
|
|
token = fzl_mktoken(FZL_TOKTYPE_RPAIR, p, p, NULL, 0);
|
|
reader->position++;
|
|
}
|
|
else if (curr >= 'A' && curr <= 'Z' || curr >= 'a' && curr <= 'z')
|
|
token = parse_name(reader);
|
|
else
|
|
{
|
|
sprintf(last_err, "undefined token '%c'", curr);
|
|
return -1;
|
|
}
|
|
|
|
if (!token)
|
|
{
|
|
size_t line = 1, linepos = 1;
|
|
for (size_t i = 0; i < p; i++)
|
|
{
|
|
char sym = reader->source[i];
|
|
if (sym == '\n')
|
|
{
|
|
linepos = 1;
|
|
line++;
|
|
}
|
|
else
|
|
linepos++;
|
|
}
|
|
|
|
sprintf(last_err, "cannot parse token (%ld:%ld)", line, linepos);
|
|
return -1;
|
|
}
|
|
|
|
fzl_reader_ll_t *tok_node = malloc(sizeof(fzl_reader_ll_t));
|
|
tok_node->token = token;
|
|
|
|
reader->last->next = tok_node;
|
|
reader->last = tok_node;
|
|
} while (reader->position < reader->size);
|
|
|
|
reader->at_end = true;
|
|
return 0;
|
|
}
|
|
|
|
// --- Read zone end ---
|
|
|
|
char *fzl_reader_geterror() { return last_err; }
|
|
|
|
void free_node_and_next(fzl_reader_ll_t *node)
|
|
{
|
|
if (!node)
|
|
return;
|
|
if (node->next)
|
|
free_node_and_next(node->next);
|
|
|
|
if (node->token)
|
|
fzl_freetoken(node->token);
|
|
|
|
free(node);
|
|
}
|
|
|
|
void fzl_freereader(fzl_reader_t *reader)
|
|
{
|
|
if (!reader)
|
|
return;
|
|
if (reader->first)
|
|
free_node_and_next(reader->first);
|
|
|
|
free(reader);
|
|
}
|