Files
cig/scanner.c
T
roodletoof af356a0d71
C Tests / Build and Test (push) Successful in 10m52s
Squash merge feature/scanner_v2 into main
2026-10-01 22:19:09 +02:00

356 lines
7.7 KiB
C

#include "cig.h"
#include <ctype.h>
#include <string.h>
#include <errno.h>
scanner_t make_scanner(char const *name, char const *buffer, allocator_t allocator) {
scanner_t s = {
.name = name,
.start = buffer,
.cur = buffer,
.allocator = allocator
};
return s;
}
void scanner_recover(scanner_t *s) {
while (*s->cur && *s->cur != '\n') {
s->cur++;
}
if (*s->cur == '\n')
s->cur++; // move *past* the newline
}
scan_result_t scan_error(scanner_t *s, char const * message) {
char const *error_pos = s->cur; // use current scanner position
// Compute line + column (1-based)
int line = 1, column = 1;
for (char const *p = s->start; p < error_pos; p++) {
if (*p == '\n') { line++; column = 1; }
else column++;
}
return (scan_result_t) {
.ok=false,
.error=(scan_error_t) {
.message=message,
.buffer_name=s->name,
.line=line,
.column=column
}
};
}
scan_result_t scan_value_func(scan_value_t value) {
return (scan_result_t) {
.ok=true,
.value=value,
};
}
scan_result_t scan_error_and_recover(scanner_t *s, char const *message) {
scan_result_t result = scan_error(s, message);
scanner_recover(s);
return result;
}
bool scan_eof(scanner_t *s) {
return *s->cur == '\0';
}
bool scan_literal(scanner_t *s, char const *lit) {
char const *save = s->cur;
while (*lit && *lit == *s->cur) {
lit++;
s->cur++;
}
if (*lit == '\0') {
return true;
}
s->cur = save;
return false;
}
int scan_repeat_literal(scanner_t *s, char const *lit) {
int n_repeats = 0;
while (scan_literal(s, lit)) {
n_repeats++;
}
return n_repeats;
}
bool scan_whitespace(scanner_t *s) {
char const *save = s->cur;
while (isspace((unsigned char)*s->cur)) s->cur++;
return save != s->cur;
}
scan_result_t scan_digit(scanner_t *s) {
if (!isdigit((unsigned char)*s->cur)) {
return scan_error(s, "not a digit");
}
scan_result_t ret = scan_value(.digit=(*s->cur) - '0');
s->cur++;
return ret;
}
scan_result_t scan_i64(scanner_t *s) {
char const *save = s->cur;
if (*s->cur == '-' || *s->cur == '+') s->cur++;
if (!isdigit((unsigned char)*s->cur)) {
s->cur = save;
return scan_error(s, "not an i64");
}
s->cur = save;
char *end;
errno = 0;
int64_t val = strtoll(s->cur, &end, 10);
if (end == s->cur) {
s->cur = save;
return scan_error(s, "not an i64");
}
if (errno == ERANGE) {
return scan_error(s, "integer does not fit in i64 value");
}
s->cur = end;
return scan_value(.i64=val);
}
scan_result_t scan_i32(scanner_t *s) {
scan_result_t result = scan_i64(s);
if (!result.ok) {
return result;
}
int32_t val = (int32_t)result.value.i64;
int64_t back = (int64_t)val;
if (back != result.value.i64) {
return scan_error(s, "int does not fit in i32 value");
}
return scan_value(val);
}
scan_result_t scan_i16(scanner_t *s) {
scan_result_t result = scan_i64(s);
if (!result.ok) {
return result;
}
int16_t val = (int16_t)result.value.i64;
int64_t back = (int64_t)val;
if (back != result.value.i64) {
return scan_error(s, "int does not fit in i16 value");
}
return scan_value(val);
}
scan_result_t scan_i8(scanner_t *s) {
scan_result_t result = scan_i64(s);
if (!result.ok) {
return result;
}
int8_t val = (int8_t)result.value.i64;
int64_t back = (int64_t)val;
if (back != result.value.i64) {
return scan_error(s, "int does not fit in i8 value");
}
return scan_value(val);
}
scan_result_t scan_u64(scanner_t *s) {
char const *save = s->cur;
if (*s->cur == '-') {
return scan_error(s, "- is not allowed for unsigned integers");
}
if (*s->cur == '+') s->cur++;
if (!isdigit((unsigned char)*s->cur)) {
s->cur = save;
return scan_error(s, "is not number");
}
s->cur = save;
char *end;
errno = 0;
uint64_t val = strtoull(s->cur, &end, 10);
if (end == s->cur) {
s->cur = save;
return scan_error(s, "is not number");
// TODO: standardize error messages
}
if (errno == ERANGE) {
return scan_error(s, "integer does not fit in u64 value");
}
s->cur = end;
return scan_value(.u64=val);
}
scan_result_t scan_u32(scanner_t *s) {
scan_result_t result = scan_u64(s);
if (!result.ok) {
return result;
}
uint32_t val = (uint32_t)result.value.u64;
uint64_t back = (uint64_t)val;
if (back != result.value.u64) {
return scan_error(s, "int does not fit in u32 value");
}
return scan_value(val);
}
scan_result_t scan_u16(scanner_t *s) {
scan_result_t result = scan_u64(s);
if (!result.ok) {
return result;
}
uint16_t val = (uint16_t)result.value.u64;
uint64_t back = (uint64_t)val;
if (back != result.value.u64) {
return scan_error(s, "int does not fit in u16 value");
}
return scan_value(val);
}
scan_result_t scan_u8(scanner_t *s) {
scan_result_t result = scan_u64(s);
if (!result.ok) {
return result;
}
uint8_t val = (uint8_t)result.value.u64;
uint64_t back = (uint64_t)val;
if (back != result.value.u64) {
return scan_error(s, "int does not fit in u8 value");
}
return scan_value(val);
}
scan_result_t scan_f64(scanner_t *s) {
char const *save = s->cur;
char *end;
errno = 0;
double val = strtod(s->cur, &end);
if (end == s->cur) {
s->cur = save;
return scan_error(s, "not a f64");
}
if (errno == ERANGE) {
return scan_error(s, "float does not fit in f64 value");
}
s->cur = end;
return scan_value(.f64=val);
}
scan_result_t scan_f32(scanner_t *s) {
scan_result_t result = scan_f64(s);
if (!result.ok) {
return result;
}
float val = (float)result.value.f64;
double back = (double)val;
if (back != result.value.f64) {
return scan_error(s, "float does not fit in f32 value");
}
return scan_value(.f32=val);
}
scan_result_t scan_identifier(scanner_t *s) {
if (!isalpha((unsigned char)*s->cur) && *s->cur != '_') {
return scan_error(s, "not an identifier");
}
char const *start = s->cur++;
while (isalnum((unsigned char)*s->cur) || *s->cur == '_') {
s->cur++;
}
size_t len = s->cur - start;
char *buf = allocator_alloc(s->allocator, len + 1);
memcpy(buf, start, len);
buf[len] = '\0';
return scan_value(.identifier=buf);
}
scan_result_t scan_string_literal(scanner_t *s) {
char const *save = s->cur;
if (*s->cur != '"')
return scan_error(s, "not a string");
s->cur++; // skip opening quote
char *buf = allocator_alloc(s->allocator, 256);
size_t cap = 256;
size_t len = 0;
while (*s->cur && *s->cur != '"') {
char c = *s->cur++;
scan_result_t ret;
if (c == '\\') {
char esc = *s->cur++;
switch (esc) {
case 'n': c = '\n'; break;
case 't': c = '\t'; break;
case 'r': c = '\r'; break;
case '\\': c = '\\'; break;
case '"': c = '"'; break;
case '0': c = '\0'; break;
default:
ret = scan_error(s, "invalid escape sequence");
s->cur = save;
return ret;
}
}
if (len + 2 >= cap) {
cap *= 2;
buf = allocator_resize(s->allocator, buf, cap);
}
buf[len++] = c;
}
if (*s->cur != '"') {
scan_result_t ret = scan_error(s, "unterminated string literal");
s->cur = save;
return ret;
}
s->cur++; // skip closing quote
buf[len] = '\0';
return scan_value(.string_literal=buf);
}
bool scanner_print_errors_func(FILE *fp, scan_result_t const *results, int results_len) {
bool did_print_errors = false;
for (int i = 0; i < results_len; i++) {
scan_result_t res = results[i];
if (!res.ok) {
scan_error_t err = res.error;
fprintf(fp, "%s:%d:%d: %s", err.buffer_name, err.line, err.column, err.message);
did_print_errors = true;
}
}
return did_print_errors;
}
bool looks_like_float(scanner_t *s) {
char const *cur = s->cur;
if (
((*cur) == '+') ||
((*cur) == '-')
) {
cur++;
}
if (!isdigit(((unsigned char)*cur))) {
return false;
} else {
cur++;
}
while (isdigit(((unsigned char)*cur))) {
cur++;
}
if ((*cur) != '.') {
return false;
} else {
cur++;
}
return isdigit(((unsigned char)*cur));
}
bool looks_like_int(scanner_t *s) {
char const *cur = s->cur;
if (*cur == '+' || *cur == '-') {
cur++;
}
return isdigit(*cur);
}