Coverage Report

Created: 2026-06-09 06:53

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/src/cpython/Parser/tokenizer/helpers.c
Line
Count
Source
1
#include "Python.h"
2
#include "errcode.h"
3
#include "pycore_token.h"
4
5
#include "../lexer/state.h"
6
7
8
/* ############## ERRORS ############## */
9
10
static int
11
_syntaxerror_range(struct tok_state *tok, const char *format,
12
                   int col_offset, int end_col_offset,
13
                   va_list vargs)
14
3.12k
{
15
    // In release builds, we don't want to overwrite a previous error, but in debug builds we
16
    // want to fail if we are not doing it so we can fix it.
17
3.12k
    assert(tok->done != E_ERROR);
18
3.12k
    if (tok->done == E_ERROR) {
19
0
        return ERRORTOKEN;
20
0
    }
21
3.12k
    PyObject *errmsg, *errtext, *args;
22
3.12k
    errmsg = PyUnicode_FromFormatV(format, vargs);
23
3.12k
    if (!errmsg) {
24
0
        goto error;
25
0
    }
26
27
3.12k
    errtext = PyUnicode_DecodeUTF8(tok->line_start, tok->cur - tok->line_start,
28
3.12k
                                   "replace");
29
3.12k
    if (!errtext) {
30
0
        goto error;
31
0
    }
32
33
3.12k
    if (col_offset == -1) {
34
2.46k
        col_offset = (int)PyUnicode_GET_LENGTH(errtext);
35
2.46k
    }
36
3.12k
    if (end_col_offset == -1) {
37
2.46k
        end_col_offset = col_offset;
38
2.46k
    }
39
40
3.12k
    Py_ssize_t line_len = strcspn(tok->line_start, "\n");
41
3.12k
    if (line_len != tok->cur - tok->line_start) {
42
2.35k
        Py_DECREF(errtext);
43
2.35k
        errtext = PyUnicode_DecodeUTF8(tok->line_start, line_len,
44
2.35k
                                       "replace");
45
2.35k
    }
46
3.12k
    if (!errtext) {
47
0
        goto error;
48
0
    }
49
50
3.12k
    args = Py_BuildValue("(O(OiiNii))", errmsg,
51
3.12k
                         tok->filename ? tok->filename : Py_None,
52
3.12k
                         tok->lineno, col_offset, errtext,
53
3.12k
                         tok->lineno, end_col_offset);
54
3.12k
    if (args) {
55
3.12k
        PyErr_SetObject(PyExc_SyntaxError, args);
56
3.12k
        Py_DECREF(args);
57
3.12k
    }
58
59
3.12k
error:
60
3.12k
    Py_XDECREF(errmsg);
61
3.12k
    tok->done = E_ERROR;
62
3.12k
    return ERRORTOKEN;
63
3.12k
}
64
65
int
66
_PyTokenizer_syntaxerror(struct tok_state *tok, const char *format, ...)
67
2.46k
{
68
    // These errors are cleaned on startup. Todo: Fix it.
69
2.46k
    va_list vargs;
70
2.46k
    va_start(vargs, format);
71
2.46k
    int ret = _syntaxerror_range(tok, format, -1, -1, vargs);
72
2.46k
    va_end(vargs);
73
2.46k
    return ret;
74
2.46k
}
75
76
int
77
_PyTokenizer_syntaxerror_known_range(struct tok_state *tok,
78
                        int col_offset, int end_col_offset,
79
                        const char *format, ...)
80
661
{
81
661
    va_list vargs;
82
661
    va_start(vargs, format);
83
661
    int ret = _syntaxerror_range(tok, format, col_offset, end_col_offset, vargs);
84
661
    va_end(vargs);
85
661
    return ret;
86
661
}
87
88
int
89
_PyTokenizer_indenterror(struct tok_state *tok)
90
4
{
91
4
    tok->done = E_TABSPACE;
92
4
    tok->cur = tok->inp;
93
4
    return ERRORTOKEN;
94
4
}
95
96
char *
97
_PyTokenizer_error_ret(struct tok_state *tok) /* XXX */
98
2.60k
{
99
2.60k
    tok->decoding_erred = 1;
100
2.60k
    if ((tok->fp != NULL || tok->readline != NULL) && tok->buf != NULL) {/* see _PyTokenizer_Free */
101
0
        PyMem_Free(tok->buf);
102
0
    }
103
2.60k
    tok->buf = tok->cur = tok->inp = NULL;
104
2.60k
    tok->start = NULL;
105
2.60k
    tok->end = NULL;
106
2.60k
    tok->done = E_DECODE;
107
2.60k
    return NULL;                /* as if it were EOF */
108
2.60k
}
109
110
int
111
_PyTokenizer_warn_invalid_escape_sequence(struct tok_state *tok, int first_invalid_escape_char)
112
624
{
113
624
    if (!tok->report_warnings) {
114
0
        return 0;
115
0
    }
116
117
624
    PyObject *msg = PyUnicode_FromFormat(
118
624
        "\"\\%c\" is an invalid escape sequence. "
119
624
        "Such sequences will not work in the future. "
120
624
        "Did you mean \"\\\\%c\"? A raw string is also an option.",
121
624
        (char) first_invalid_escape_char,
122
624
        (char) first_invalid_escape_char
123
624
    );
124
125
624
    if (msg == NULL) {
126
0
        return -1;
127
0
    }
128
129
624
    if (PyErr_WarnExplicitObject(PyExc_SyntaxWarning, msg, tok->filename,
130
624
                                 tok->lineno, tok->module, NULL) < 0) {
131
0
        Py_DECREF(msg);
132
133
0
        if (PyErr_ExceptionMatches(PyExc_SyntaxWarning)) {
134
            /* Replace the SyntaxWarning exception with a SyntaxError
135
               to get a more accurate error report */
136
0
            PyErr_Clear();
137
138
0
            return _PyTokenizer_syntaxerror(tok,
139
0
                "\"\\%c\" is an invalid escape sequence. "
140
0
                "Did you mean \"\\\\%c\"? A raw string is also an option.",
141
0
                (char) first_invalid_escape_char,
142
0
                (char) first_invalid_escape_char);
143
0
        }
144
145
0
        return -1;
146
0
    }
147
148
624
    Py_DECREF(msg);
149
624
    return 0;
150
624
}
151
152
int
153
_PyTokenizer_parser_warn(struct tok_state *tok, PyObject *category, const char *format, ...)
154
11.2k
{
155
11.2k
    if (!tok->report_warnings) {
156
0
        return 0;
157
0
    }
158
159
11.2k
    PyObject *errmsg;
160
11.2k
    va_list vargs;
161
11.2k
    va_start(vargs, format);
162
11.2k
    errmsg = PyUnicode_FromFormatV(format, vargs);
163
11.2k
    va_end(vargs);
164
11.2k
    if (!errmsg) {
165
0
        goto error;
166
0
    }
167
168
11.2k
    if (PyErr_WarnExplicitObject(category, errmsg, tok->filename,
169
11.2k
                                 tok->lineno, tok->module, NULL) < 0) {
170
0
        if (PyErr_ExceptionMatches(category)) {
171
            /* Replace the DeprecationWarning exception with a SyntaxError
172
               to get a more accurate error report */
173
0
            PyErr_Clear();
174
0
            _PyTokenizer_syntaxerror(tok, "%U", errmsg);
175
0
        }
176
0
        goto error;
177
0
    }
178
11.2k
    Py_DECREF(errmsg);
179
11.2k
    return 0;
180
181
0
error:
182
0
    Py_XDECREF(errmsg);
183
0
    tok->done = E_ERROR;
184
0
    return -1;
185
11.2k
}
186
187
188
/* ############## STRING MANIPULATION ############## */
189
190
char *
191
_PyTokenizer_new_string(const char *s, Py_ssize_t len, struct tok_state *tok)
192
86.2k
{
193
86.2k
    char* result = (char *)PyMem_Malloc(len + 1);
194
86.2k
    if (!result) {
195
0
        tok->done = E_NOMEM;
196
0
        PyErr_NoMemory();
197
0
        return NULL;
198
0
    }
199
86.2k
    memcpy(result, s, len);
200
86.2k
    result[len] = '\0';
201
86.2k
    return result;
202
86.2k
}
203
204
PyObject *
205
4.08k
_PyTokenizer_translate_into_utf8(const char* str, const char* enc) {
206
4.08k
    PyObject *utf8;
207
4.08k
    PyObject* buf = PyUnicode_Decode(str, strlen(str), enc, NULL);
208
4.08k
    if (buf == NULL)
209
1.93k
        return NULL;
210
2.14k
    utf8 = PyUnicode_AsUTF8String(buf);
211
2.14k
    Py_DECREF(buf);
212
2.14k
    return utf8;
213
4.08k
}
214
215
char *
216
_PyTokenizer_translate_newlines(const char *s, int exec_input, int preserve_crlf,
217
105k
                   struct tok_state *tok) {
218
105k
    int skip_next_lf = 0;
219
105k
    size_t needed_length = strlen(s) + 2, final_length;
220
105k
    char *buf, *current;
221
105k
    char c = '\0';
222
105k
    buf = PyMem_Malloc(needed_length);
223
105k
    if (buf == NULL) {
224
0
        tok->done = E_NOMEM;
225
0
        PyErr_NoMemory();
226
0
        return NULL;
227
0
    }
228
7.93M
    for (current = buf; *s; s++, current++) {
229
7.82M
        c = *s;
230
7.82M
        if (skip_next_lf) {
231
10.7k
            skip_next_lf = 0;
232
10.7k
            if (c == '\n') {
233
524
                c = *++s;
234
524
                if (!c)
235
8
                    break;
236
524
            }
237
10.7k
        }
238
7.82M
        if (!preserve_crlf && c == '\r') {
239
10.8k
            skip_next_lf = 1;
240
10.8k
            c = '\n';
241
10.8k
        }
242
7.82M
        *current = c;
243
7.82M
    }
244
    /* If this is exec input, add a newline to the end of the string if
245
       there isn't one already. */
246
105k
    if (exec_input && c != '\n' && c != '\0') {
247
102k
        *current = '\n';
248
102k
        current++;
249
102k
    }
250
105k
    *current = '\0';
251
105k
    final_length = current - buf + 1;
252
105k
    if (final_length < needed_length && final_length) {
253
        /* should never fail */
254
3.07k
        char* result = PyMem_Realloc(buf, final_length);
255
3.07k
        if (result == NULL) {
256
0
            PyMem_Free(buf);
257
0
        }
258
3.07k
        buf = result;
259
3.07k
    }
260
105k
    return buf;
261
105k
}
262
263
/* ############## ENCODING STUFF ############## */
264
265
266
/* See whether the file starts with a BOM. If it does,
267
   invoke the set_readline function with the new encoding.
268
   Return 1 on success, 0 on failure.  */
269
int
270
_PyTokenizer_check_bom(int get_char(struct tok_state *),
271
          void unget_char(int, struct tok_state *),
272
          int set_readline(struct tok_state *, const char *),
273
          struct tok_state *tok)
274
21.2k
{
275
21.2k
    int ch1, ch2, ch3;
276
21.2k
    ch1 = get_char(tok);
277
21.2k
    tok->decoding_state = STATE_SEEK_CODING;
278
21.2k
    if (ch1 == EOF) {
279
0
        return 1;
280
21.2k
    } else if (ch1 == 0xEF) {
281
64
        ch2 = get_char(tok);
282
64
        if (ch2 != 0xBB) {
283
56
            unget_char(ch2, tok);
284
56
            unget_char(ch1, tok);
285
56
            return 1;
286
56
        }
287
8
        ch3 = get_char(tok);
288
8
        if (ch3 != 0xBF) {
289
3
            unget_char(ch3, tok);
290
3
            unget_char(ch2, tok);
291
3
            unget_char(ch1, tok);
292
3
            return 1;
293
3
        }
294
21.1k
    } else {
295
21.1k
        unget_char(ch1, tok);
296
21.1k
        return 1;
297
21.1k
    }
298
5
    if (tok->encoding != NULL)
299
0
        PyMem_Free(tok->encoding);
300
5
    tok->encoding = _PyTokenizer_new_string("utf-8", 5, tok);
301
5
    if (!tok->encoding)
302
0
        return 0;
303
    /* No need to set_readline: input is already utf-8 */
304
5
    return 1;
305
5
}
306
307
static const char *
308
get_normal_name(const char *s)  /* for utf-8 and latin-1 */
309
4.08k
{
310
4.08k
    char buf[13];
311
4.08k
    int i;
312
26.7k
    for (i = 0; i < 12; i++) {
313
26.4k
        int c = s[i];
314
26.4k
        if (c == '\0')
315
3.86k
            break;
316
22.6k
        else if (c == '_')
317
610
            buf[i] = '-';
318
22.0k
        else
319
22.0k
            buf[i] = Py_TOLOWER(c);
320
26.4k
    }
321
4.08k
    buf[i] = '\0';
322
4.08k
    if (strcmp(buf, "utf-8") == 0 ||
323
4.08k
        strncmp(buf, "utf-8-", 6) == 0)
324
3
        return "utf-8";
325
4.08k
    else if (strcmp(buf, "latin-1") == 0 ||
326
4.07k
             strcmp(buf, "iso-8859-1") == 0 ||
327
4.07k
             strcmp(buf, "iso-latin-1") == 0 ||
328
4.07k
             strncmp(buf, "latin-1-", 8) == 0 ||
329
4.07k
             strncmp(buf, "iso-8859-1-", 11) == 0 ||
330
4.07k
             strncmp(buf, "iso-latin-1-", 12) == 0)
331
16
        return "iso-8859-1";
332
4.06k
    else
333
4.06k
        return s;
334
4.08k
}
335
336
/* Return the coding spec in S, or NULL if none is found.  */
337
static int
338
get_coding_spec(const char *s, char **spec, Py_ssize_t size, struct tok_state *tok)
339
21.7k
{
340
21.7k
    Py_ssize_t i;
341
21.7k
    *spec = NULL;
342
    /* Coding spec must be in a comment, and that comment must be
343
     * the only statement on the source code line. */
344
24.5k
    for (i = 0; i < size - 6; i++) {
345
20.0k
        if (s[i] == '#')
346
4.28k
            break;
347
15.7k
        if (s[i] != ' ' && s[i] != '\t' && s[i] != '\014')
348
12.9k
            return 1;
349
15.7k
    }
350
250k
    for (; i < size - 6; i++) { /* XXX inefficient search */
351
245k
        const char* t = s + i;
352
245k
        if (memcmp(t, "coding", 6) == 0) {
353
4.84k
            const char* begin = NULL;
354
4.84k
            t += 6;
355
4.84k
            if (t[0] != ':' && t[0] != '=')
356
353
                continue;
357
5.22k
            do {
358
5.22k
                t++;
359
5.22k
            } while (t[0] == ' ' || t[0] == '\t');
360
361
4.49k
            begin = t;
362
32.6k
            while (Py_ISALNUM(t[0]) ||
363
7.36k
                   t[0] == '-' || t[0] == '_' || t[0] == '.')
364
28.1k
                t++;
365
366
4.49k
            if (begin < t) {
367
4.08k
                char* r = _PyTokenizer_new_string(begin, t - begin, tok);
368
4.08k
                const char* q;
369
4.08k
                if (!r)
370
0
                    return 0;
371
4.08k
                q = get_normal_name(r);
372
4.08k
                if (r != q) {
373
19
                    PyMem_Free(r);
374
19
                    r = _PyTokenizer_new_string(q, strlen(q), tok);
375
19
                    if (!r)
376
0
                        return 0;
377
19
                }
378
4.08k
                *spec = r;
379
4.08k
                break;
380
4.08k
            }
381
4.49k
        }
382
245k
    }
383
8.86k
    return 1;
384
8.86k
}
385
386
/* Check whether the line contains a coding spec. If it does,
387
   invoke the set_readline function for the new encoding.
388
   This function receives the tok_state and the new encoding.
389
   Return 1 on success, 0 on failure.  */
390
int
391
_PyTokenizer_check_coding_spec(const char* line, Py_ssize_t size, struct tok_state *tok,
392
                  int set_readline(struct tok_state *, const char *))
393
21.7k
{
394
21.7k
    char *cs;
395
21.7k
    if (tok->cont_line) {
396
        /* It's a continuation line, so it can't be a coding spec. */
397
0
        tok->decoding_state = STATE_NORMAL;
398
0
        return 1;
399
0
    }
400
21.7k
    if (!get_coding_spec(line, &cs, size, tok)) {
401
0
        return 0;
402
0
    }
403
21.7k
    if (!cs) {
404
17.7k
        Py_ssize_t i;
405
20.6k
        for (i = 0; i < size; i++) {
406
20.3k
            if (line[i] == '#' || line[i] == '\n' || line[i] == '\r')
407
469
                break;
408
19.8k
            if (line[i] != ' ' && line[i] != '\t' && line[i] != '\014') {
409
                /* Stop checking coding spec after a line containing
410
                 * anything except a comment. */
411
16.9k
                tok->decoding_state = STATE_NORMAL;
412
16.9k
                break;
413
16.9k
            }
414
19.8k
        }
415
17.7k
        return 1;
416
17.7k
    }
417
4.08k
    tok->decoding_state = STATE_NORMAL;
418
4.08k
    if (tok->encoding == NULL) {
419
4.08k
        assert(tok->decoding_readline == NULL);
420
4.08k
        if (strcmp(cs, "utf-8") != 0 && !set_readline(tok, cs)) {
421
0
            _PyTokenizer_error_ret(tok);
422
0
            PyErr_Format(PyExc_SyntaxError, "encoding problem: %s", cs);
423
0
            PyMem_Free(cs);
424
0
            return 0;
425
0
        }
426
4.08k
        tok->encoding = cs;
427
4.08k
    } else {                /* then, compare cs with BOM */
428
3
        if (strcmp(tok->encoding, cs) != 0) {
429
2
            tok->line_start = line;
430
2
            tok->cur = (char *)line;
431
2
            assert(size <= INT_MAX);
432
2
            _PyTokenizer_syntaxerror_known_range(tok, 0, (int)size,
433
2
                        "encoding problem: %s with BOM", cs);
434
2
            PyMem_Free(cs);
435
2
            _PyTokenizer_error_ret(tok);
436
2
            return 0;
437
2
        }
438
1
        PyMem_Free(cs);
439
1
    }
440
4.08k
    return 1;
441
4.08k
}
442
443
/* Check whether the characters at s start a valid
444
   UTF-8 sequence. Return the number of characters forming
445
   the sequence if yes, 0 if not.  The special cases match
446
   those in stringlib/codecs.h:utf8_decode.
447
*/
448
static int
449
valid_utf8(const unsigned char* s)
450
4.70M
{
451
4.70M
    int expected = 0;
452
4.70M
    int length;
453
4.70M
    if (*s < 0x80) {
454
        /* single-byte code */
455
4.69M
        return 1;
456
4.69M
    }
457
9.34k
    else if (*s < 0xE0) {
458
        /* \xC2\x80-\xDF\xBF -- 0080-07FF */
459
3.89k
        if (*s < 0xC2) {
460
            /* invalid sequence
461
               \x80-\xBF -- continuation byte
462
               \xC0-\xC1 -- fake 0000-007F */
463
147
            return 0;
464
147
        }
465
3.75k
        expected = 1;
466
3.75k
    }
467
5.44k
    else if (*s < 0xF0) {
468
        /* \xE0\xA0\x80-\xEF\xBF\xBF -- 0800-FFFF */
469
2.28k
        if (*s == 0xE0 && *(s + 1) < 0xA0) {
470
            /* invalid sequence
471
               \xE0\x80\x80-\xE0\x9F\xBF -- fake 0000-0800 */
472
20
            return 0;
473
20
        }
474
2.26k
        else if (*s == 0xED && *(s + 1) >= 0xA0) {
475
            /* Decoding UTF-8 sequences in range \xED\xA0\x80-\xED\xBF\xBF
476
               will result in surrogates in range D800-DFFF. Surrogates are
477
               not valid UTF-8 so they are rejected.
478
               See https://www.unicode.org/versions/Unicode5.2.0/ch03.pdf
479
               (table 3-7) and http://www.rfc-editor.org/rfc/rfc3629.txt */
480
19
            return 0;
481
19
        }
482
2.24k
        expected = 2;
483
2.24k
    }
484
3.16k
    else if (*s < 0xF5) {
485
        /* \xF0\x90\x80\x80-\xF4\x8F\xBF\xBF -- 10000-10FFFF */
486
3.13k
        if (*(s + 1) < 0x90 ? *s == 0xF0 : *s == 0xF4) {
487
            /* invalid sequence -- one of:
488
               \xF0\x80\x80\x80-\xF0\x8F\xBF\xBF -- fake 0000-FFFF
489
               \xF4\x90\x80\x80- -- 110000- overflow */
490
41
            return 0;
491
41
        }
492
3.08k
        expected = 3;
493
3.08k
    }
494
38
    else {
495
        /* invalid start byte */
496
38
        return 0;
497
38
    }
498
9.08k
    length = expected + 1;
499
25.9k
    for (int i = 1; i <= expected; i++) {
500
17.2k
        if (s[i] < 0x80 || s[i] >= 0xC0) {
501
348
            return 0;
502
348
        }
503
17.2k
    }
504
8.73k
    return length;
505
9.08k
}
506
507
int
508
_PyTokenizer_ensure_utf8(const char *line, struct tok_state *tok, int lineno)
509
17.4k
{
510
17.4k
    const char *badchar = NULL;
511
17.4k
    const char *c;
512
17.4k
    int length;
513
17.4k
    int col_offset = 0;
514
17.4k
    const char *line_start = line;
515
4.71M
    for (c = line; *c; c += length) {
516
4.70M
        if (!(length = valid_utf8((const unsigned char *)c))) {
517
613
            badchar = c;
518
613
            break;
519
613
        }
520
4.69M
        col_offset++;
521
4.69M
        if (*c == '\n') {
522
194k
            lineno++;
523
194k
            col_offset = 0;
524
194k
            line_start = c + 1;
525
194k
        }
526
4.69M
    }
527
17.4k
    if (badchar) {
528
613
        tok->lineno = lineno;
529
613
        tok->line_start = line_start;
530
613
        tok->cur = (char *)badchar;
531
613
        _PyTokenizer_syntaxerror_known_range(tok,
532
613
                col_offset + 1, col_offset + 1,
533
613
                "Non-UTF-8 code starting with '\\x%.2x'"
534
613
                "%s%V on line %i, "
535
613
                "but no encoding declared; "
536
613
                "see https://peps.python.org/pep-0263/ for details",
537
613
                (unsigned char)*badchar,
538
613
                tok->filename ? " in file " : "", tok->filename, "",
539
613
                lineno);
540
613
        return 0;
541
613
    }
542
16.8k
    return 1;
543
17.4k
}
544
545
546
/* ############## DEBUGGING STUFF ############## */
547
548
#ifdef Py_DEBUG
549
void
550
_PyTokenizer_print_escape(FILE *f, const char *s, Py_ssize_t size)
551
{
552
    if (s == NULL) {
553
        fputs("NULL", f);
554
        return;
555
    }
556
    putc('"', f);
557
    while (size-- > 0) {
558
        unsigned char c = *s++;
559
        switch (c) {
560
            case '\n': fputs("\\n", f); break;
561
            case '\r': fputs("\\r", f); break;
562
            case '\t': fputs("\\t", f); break;
563
            case '\f': fputs("\\f", f); break;
564
            case '\'': fputs("\\'", f); break;
565
            case '"': fputs("\\\"", f); break;
566
            default:
567
                if (0x20 <= c && c <= 0x7f)
568
                    putc(c, f);
569
                else
570
                    fprintf(f, "\\x%02x", c);
571
        }
572
    }
573
    putc('"', f);
574
}
575
576
void
577
_PyTokenizer_tok_dump(int type, char *start, char *end)
578
{
579
    fprintf(stderr, "%s", _PyParser_TokenNames[type]);
580
    if (type == NAME || type == NUMBER || type == STRING || type == OP)
581
        fprintf(stderr, "(%.*s)", (int)(end - start), start);
582
}
583
#endif