1
2 /* Tokenizer implementation */
3
4 #include "Python.h"
5 #include "pgenheaders.h"
6
7 #include <ctype.h>
8 #include <assert.h>
9
10 #include "tokenizer.h"
11 #include "errcode.h"
12
13 #ifndef PGEN
14 #include "unicodeobject.h"
15 #include "bytesobject.h"
16 #include "fileobject.h"
17 #include "codecs.h"
18 #include "abstract.h"
19 #endif /* PGEN */
20
21 #define is_potential_identifier_start(c) (\
22 (c >= 'a' && c <= 'z')\
23 || (c >= 'A' && c <= 'Z')\
24 || c == '_'\
25 || (c >= 128))
26
27 #define is_potential_identifier_char(c) (\
28 (c >= 'a' && c <= 'z')\
29 || (c >= 'A' && c <= 'Z')\
30 || (c >= '0' && c <= '9')\
31 || c == '_'\
32 || (c >= 128))
33
34 extern char *PyOS_Readline(FILE *, FILE *, const char *);
35 /* Return malloc'ed string including trailing \n;
36 empty malloc'ed string for EOF;
37 NULL if interrupted */
38
39 /* Don't ever change this -- it would break the portability of Python code */
40 #define TABSIZE 8
41
42 /* Forward */
43 static struct tok_state *tok_new(void);
44 static int tok_nextc(struct tok_state *tok);
45 static void tok_backup(struct tok_state *tok, int c);
46
47
48 /* Token names */
49
50 const char *_PyParser_TokenNames[] = {
51 "ENDMARKER",
52 "NAME",
53 "NUMBER",
54 "STRING",
55 "NEWLINE",
56 "INDENT",
57 "DEDENT",
58 "LPAR",
59 "RPAR",
60 "LSQB",
61 "RSQB",
62 "COLON",
63 "COMMA",
64 "SEMI",
65 "PLUS",
66 "MINUS",
67 "STAR",
68 "SLASH",
69 "VBAR",
70 "AMPER",
71 "LESS",
72 "GREATER",
73 "EQUAL",
74 "DOT",
75 "PERCENT",
76 "LBRACE",
77 "RBRACE",
78 "EQEQUAL",
79 "NOTEQUAL",
80 "LESSEQUAL",
81 "GREATEREQUAL",
82 "TILDE",
83 "CIRCUMFLEX",
84 "LEFTSHIFT",
85 "RIGHTSHIFT",
86 "DOUBLESTAR",
87 "PLUSEQUAL",
88 "MINEQUAL",
89 "STAREQUAL",
90 "SLASHEQUAL",
91 "PERCENTEQUAL",
92 "AMPEREQUAL",
93 "VBAREQUAL",
94 "CIRCUMFLEXEQUAL",
95 "LEFTSHIFTEQUAL",
96 "RIGHTSHIFTEQUAL",
97 "DOUBLESTAREQUAL",
98 "DOUBLESLASH",
99 "DOUBLESLASHEQUAL",
100 "AT",
101 "ATEQUAL",
102 "RARROW",
103 "ELLIPSIS",
104 /* This table must match the #defines in token.h! */
105 "OP",
106 "AWAIT",
107 "ASYNC",
108 "<ERRORTOKEN>",
109 "<N_TOKENS>"
110 };
111
112
113 /* Create and initialize a new tok_state structure */
114
115 static struct tok_state *
tok_new(void)116 tok_new(void)
117 {
118 struct tok_state *tok = (struct tok_state *)PyMem_MALLOC(
119 sizeof(struct tok_state));
120 if (tok == NULL)
121 return NULL;
122 tok->buf = tok->cur = tok->end = tok->inp = tok->start = NULL;
123 tok->done = E_OK;
124 tok->fp = NULL;
125 tok->input = NULL;
126 tok->tabsize = TABSIZE;
127 tok->indent = 0;
128 tok->indstack[0] = 0;
129
130 tok->atbol = 1;
131 tok->pendin = 0;
132 tok->prompt = tok->nextprompt = NULL;
133 tok->lineno = 0;
134 tok->level = 0;
135 tok->altwarning = 1;
136 tok->alterror = 1;
137 tok->alttabsize = 1;
138 tok->altindstack[0] = 0;
139 tok->decoding_state = STATE_INIT;
140 tok->decoding_erred = 0;
141 tok->read_coding_spec = 0;
142 tok->enc = NULL;
143 tok->encoding = NULL;
144 tok->cont_line = 0;
145 #ifndef PGEN
146 tok->filename = NULL;
147 tok->decoding_readline = NULL;
148 tok->decoding_buffer = NULL;
149 #endif
150
151 tok->async_def = 0;
152 tok->async_def_indent = 0;
153 tok->async_def_nl = 0;
154
155 return tok;
156 }
157
158 static char *
new_string(const char * s,Py_ssize_t len,struct tok_state * tok)159 new_string(const char *s, Py_ssize_t len, struct tok_state *tok)
160 {
161 char* result = (char *)PyMem_MALLOC(len + 1);
162 if (!result) {
163 tok->done = E_NOMEM;
164 return NULL;
165 }
166 memcpy(result, s, len);
167 result[len] = '\0';
168 return result;
169 }
170
171 #ifdef PGEN
172
173 static char *
decoding_fgets(char * s,int size,struct tok_state * tok)174 decoding_fgets(char *s, int size, struct tok_state *tok)
175 {
176 return fgets(s, size, tok->fp);
177 }
178
179 static int
decoding_feof(struct tok_state * tok)180 decoding_feof(struct tok_state *tok)
181 {
182 return feof(tok->fp);
183 }
184
185 static char *
decode_str(const char * str,int exec_input,struct tok_state * tok)186 decode_str(const char *str, int exec_input, struct tok_state *tok)
187 {
188 return new_string(str, strlen(str), tok);
189 }
190
191 #else /* PGEN */
192
193 static char *
error_ret(struct tok_state * tok)194 error_ret(struct tok_state *tok) /* XXX */
195 {
196 tok->decoding_erred = 1;
197 if (tok->fp != NULL && tok->buf != NULL) /* see PyTokenizer_Free */
198 PyMem_FREE(tok->buf);
199 tok->buf = tok->cur = tok->end = tok->inp = tok->start = NULL;
200 tok->done = E_DECODE;
201 return NULL; /* as if it were EOF */
202 }
203
204
205 static const char *
get_normal_name(const char * s)206 get_normal_name(const char *s) /* for utf-8 and latin-1 */
207 {
208 char buf[13];
209 int i;
210 for (i = 0; i < 12; i++) {
211 int c = s[i];
212 if (c == '\0')
213 break;
214 else if (c == '_')
215 buf[i] = '-';
216 else
217 buf[i] = tolower(c);
218 }
219 buf[i] = '\0';
220 if (strcmp(buf, "utf-8") == 0 ||
221 strncmp(buf, "utf-8-", 6) == 0)
222 return "utf-8";
223 else if (strcmp(buf, "latin-1") == 0 ||
224 strcmp(buf, "iso-8859-1") == 0 ||
225 strcmp(buf, "iso-latin-1") == 0 ||
226 strncmp(buf, "latin-1-", 8) == 0 ||
227 strncmp(buf, "iso-8859-1-", 11) == 0 ||
228 strncmp(buf, "iso-latin-1-", 12) == 0)
229 return "iso-8859-1";
230 else
231 return s;
232 }
233
234 /* Return the coding spec in S, or NULL if none is found. */
235
236 static int
get_coding_spec(const char * s,char ** spec,Py_ssize_t size,struct tok_state * tok)237 get_coding_spec(const char *s, char **spec, Py_ssize_t size, struct tok_state *tok)
238 {
239 Py_ssize_t i;
240 *spec = NULL;
241 /* Coding spec must be in a comment, and that comment must be
242 * the only statement on the source code line. */
243 for (i = 0; i < size - 6; i++) {
244 if (s[i] == '#')
245 break;
246 if (s[i] != ' ' && s[i] != '\t' && s[i] != '\014')
247 return 1;
248 }
249 for (; i < size - 6; i++) { /* XXX inefficient search */
250 const char* t = s + i;
251 if (strncmp(t, "coding", 6) == 0) {
252 const char* begin = NULL;
253 t += 6;
254 if (t[0] != ':' && t[0] != '=')
255 continue;
256 do {
257 t++;
258 } while (t[0] == '\x20' || t[0] == '\t');
259
260 begin = t;
261 while (Py_ISALNUM(t[0]) ||
262 t[0] == '-' || t[0] == '_' || t[0] == '.')
263 t++;
264
265 if (begin < t) {
266 char* r = new_string(begin, t - begin, tok);
267 const char* q;
268 if (!r)
269 return 0;
270 q = get_normal_name(r);
271 if (r != q) {
272 PyMem_FREE(r);
273 r = new_string(q, strlen(q), tok);
274 if (!r)
275 return 0;
276 }
277 *spec = r;
278 break;
279 }
280 }
281 }
282 return 1;
283 }
284
285 /* Check whether the line contains a coding spec. If it does,
286 invoke the set_readline function for the new encoding.
287 This function receives the tok_state and the new encoding.
288 Return 1 on success, 0 on failure. */
289
290 static int
check_coding_spec(const char * line,Py_ssize_t size,struct tok_state * tok,int set_readline (struct tok_state *,const char *))291 check_coding_spec(const char* line, Py_ssize_t size, struct tok_state *tok,
292 int set_readline(struct tok_state *, const char *))
293 {
294 char *cs;
295 int r = 1;
296
297 if (tok->cont_line) {
298 /* It's a continuation line, so it can't be a coding spec. */
299 tok->read_coding_spec = 1;
300 return 1;
301 }
302 if (!get_coding_spec(line, &cs, size, tok))
303 return 0;
304 if (!cs) {
305 Py_ssize_t i;
306 for (i = 0; i < size; i++) {
307 if (line[i] == '#' || line[i] == '\n' || line[i] == '\r')
308 break;
309 if (line[i] != ' ' && line[i] != '\t' && line[i] != '\014') {
310 /* Stop checking coding spec after a line containing
311 * anything except a comment. */
312 tok->read_coding_spec = 1;
313 break;
314 }
315 }
316 return 1;
317 }
318 tok->read_coding_spec = 1;
319 if (tok->encoding == NULL) {
320 assert(tok->decoding_state == STATE_RAW);
321 if (strcmp(cs, "utf-8") == 0) {
322 tok->encoding = cs;
323 } else {
324 r = set_readline(tok, cs);
325 if (r) {
326 tok->encoding = cs;
327 tok->decoding_state = STATE_NORMAL;
328 }
329 else {
330 PyErr_Format(PyExc_SyntaxError,
331 "encoding problem: %s", cs);
332 PyMem_FREE(cs);
333 }
334 }
335 } else { /* then, compare cs with BOM */
336 r = (strcmp(tok->encoding, cs) == 0);
337 if (!r)
338 PyErr_Format(PyExc_SyntaxError,
339 "encoding problem: %s with BOM", cs);
340 PyMem_FREE(cs);
341 }
342 return r;
343 }
344
345 /* See whether the file starts with a BOM. If it does,
346 invoke the set_readline function with the new encoding.
347 Return 1 on success, 0 on failure. */
348
349 static int
check_bom(int get_char (struct tok_state *),void unget_char (int,struct tok_state *),int set_readline (struct tok_state *,const char *),struct tok_state * tok)350 check_bom(int get_char(struct tok_state *),
351 void unget_char(int, struct tok_state *),
352 int set_readline(struct tok_state *, const char *),
353 struct tok_state *tok)
354 {
355 int ch1, ch2, ch3;
356 ch1 = get_char(tok);
357 tok->decoding_state = STATE_RAW;
358 if (ch1 == EOF) {
359 return 1;
360 } else if (ch1 == 0xEF) {
361 ch2 = get_char(tok);
362 if (ch2 != 0xBB) {
363 unget_char(ch2, tok);
364 unget_char(ch1, tok);
365 return 1;
366 }
367 ch3 = get_char(tok);
368 if (ch3 != 0xBF) {
369 unget_char(ch3, tok);
370 unget_char(ch2, tok);
371 unget_char(ch1, tok);
372 return 1;
373 }
374 #if 0
375 /* Disable support for UTF-16 BOMs until a decision
376 is made whether this needs to be supported. */
377 } else if (ch1 == 0xFE) {
378 ch2 = get_char(tok);
379 if (ch2 != 0xFF) {
380 unget_char(ch2, tok);
381 unget_char(ch1, tok);
382 return 1;
383 }
384 if (!set_readline(tok, "utf-16-be"))
385 return 0;
386 tok->decoding_state = STATE_NORMAL;
387 } else if (ch1 == 0xFF) {
388 ch2 = get_char(tok);
389 if (ch2 != 0xFE) {
390 unget_char(ch2, tok);
391 unget_char(ch1, tok);
392 return 1;
393 }
394 if (!set_readline(tok, "utf-16-le"))
395 return 0;
396 tok->decoding_state = STATE_NORMAL;
397 #endif
398 } else {
399 unget_char(ch1, tok);
400 return 1;
401 }
402 if (tok->encoding != NULL)
403 PyMem_FREE(tok->encoding);
404 tok->encoding = new_string("utf-8", 5, tok);
405 if (!tok->encoding)
406 return 0;
407 /* No need to set_readline: input is already utf-8 */
408 return 1;
409 }
410
411 /* Read a line of text from TOK into S, using the stream in TOK.
412 Return NULL on failure, else S.
413
414 On entry, tok->decoding_buffer will be one of:
415 1) NULL: need to call tok->decoding_readline to get a new line
416 2) PyUnicodeObject *: decoding_feof has called tok->decoding_readline and
417 stored the result in tok->decoding_buffer
418 3) PyByteArrayObject *: previous call to fp_readl did not have enough room
419 (in the s buffer) to copy entire contents of the line read
420 by tok->decoding_readline. tok->decoding_buffer has the overflow.
421 In this case, fp_readl is called in a loop (with an expanded buffer)
422 until the buffer ends with a '\n' (or until the end of the file is
423 reached): see tok_nextc and its calls to decoding_fgets.
424 */
425
426 static char *
fp_readl(char * s,int size,struct tok_state * tok)427 fp_readl(char *s, int size, struct tok_state *tok)
428 {
429 PyObject* bufobj;
430 const char *buf;
431 Py_ssize_t buflen;
432
433 /* Ask for one less byte so we can terminate it */
434 assert(size > 0);
435 size--;
436
437 if (tok->decoding_buffer) {
438 bufobj = tok->decoding_buffer;
439 Py_INCREF(bufobj);
440 }
441 else
442 {
443 bufobj = PyObject_CallObject(tok->decoding_readline, NULL);
444 if (bufobj == NULL)
445 goto error;
446 }
447 if (PyUnicode_CheckExact(bufobj))
448 {
449 buf = PyUnicode_AsUTF8AndSize(bufobj, &buflen);
450 if (buf == NULL) {
451 goto error;
452 }
453 }
454 else
455 {
456 buf = PyByteArray_AsString(bufobj);
457 if (buf == NULL) {
458 goto error;
459 }
460 buflen = PyByteArray_GET_SIZE(bufobj);
461 }
462
463 Py_XDECREF(tok->decoding_buffer);
464 if (buflen > size) {
465 /* Too many chars, the rest goes into tok->decoding_buffer */
466 tok->decoding_buffer = PyByteArray_FromStringAndSize(buf+size,
467 buflen-size);
468 if (tok->decoding_buffer == NULL)
469 goto error;
470 buflen = size;
471 }
472 else
473 tok->decoding_buffer = NULL;
474
475 memcpy(s, buf, buflen);
476 s[buflen] = '\0';
477 if (buflen == 0) /* EOF */
478 s = NULL;
479 Py_DECREF(bufobj);
480 return s;
481
482 error:
483 Py_XDECREF(bufobj);
484 return error_ret(tok);
485 }
486
487 /* Set the readline function for TOK to a StreamReader's
488 readline function. The StreamReader is named ENC.
489
490 This function is called from check_bom and check_coding_spec.
491
492 ENC is usually identical to the future value of tok->encoding,
493 except for the (currently unsupported) case of UTF-16.
494
495 Return 1 on success, 0 on failure. */
496
497 static int
fp_setreadl(struct tok_state * tok,const char * enc)498 fp_setreadl(struct tok_state *tok, const char* enc)
499 {
500 PyObject *readline, *io, *stream;
501 _Py_IDENTIFIER(open);
502 _Py_IDENTIFIER(readline);
503 int fd;
504 long pos;
505
506 fd = fileno(tok->fp);
507 /* Due to buffering the file offset for fd can be different from the file
508 * position of tok->fp. If tok->fp was opened in text mode on Windows,
509 * its file position counts CRLF as one char and can't be directly mapped
510 * to the file offset for fd. Instead we step back one byte and read to
511 * the end of line.*/
512 pos = ftell(tok->fp);
513 if (pos == -1 ||
514 lseek(fd, (off_t)(pos > 0 ? pos - 1 : pos), SEEK_SET) == (off_t)-1) {
515 PyErr_SetFromErrnoWithFilename(PyExc_OSError, NULL);
516 return 0;
517 }
518
519 io = PyImport_ImportModuleNoBlock("io");
520 if (io == NULL)
521 return 0;
522
523 stream = _PyObject_CallMethodId(io, &PyId_open, "isisOOO",
524 fd, "r", -1, enc, Py_None, Py_None, Py_False);
525 Py_DECREF(io);
526 if (stream == NULL)
527 return 0;
528
529 readline = _PyObject_GetAttrId(stream, &PyId_readline);
530 Py_DECREF(stream);
531 if (readline == NULL)
532 return 0;
533 Py_XSETREF(tok->decoding_readline, readline);
534
535 if (pos > 0) {
536 PyObject *bufobj = PyObject_CallObject(readline, NULL);
537 if (bufobj == NULL)
538 return 0;
539 Py_DECREF(bufobj);
540 }
541
542 return 1;
543 }
544
545 /* Fetch the next byte from TOK. */
546
fp_getc(struct tok_state * tok)547 static int fp_getc(struct tok_state *tok) {
548 return getc(tok->fp);
549 }
550
551 /* Unfetch the last byte back into TOK. */
552
fp_ungetc(int c,struct tok_state * tok)553 static void fp_ungetc(int c, struct tok_state *tok) {
554 ungetc(c, tok->fp);
555 }
556
557 /* Check whether the characters at s start a valid
558 UTF-8 sequence. Return the number of characters forming
559 the sequence if yes, 0 if not. */
valid_utf8(const unsigned char * s)560 static int valid_utf8(const unsigned char* s)
561 {
562 int expected = 0;
563 int length;
564 if (*s < 0x80)
565 /* single-byte code */
566 return 1;
567 if (*s < 0xc0)
568 /* following byte */
569 return 0;
570 if (*s < 0xE0)
571 expected = 1;
572 else if (*s < 0xF0)
573 expected = 2;
574 else if (*s < 0xF8)
575 expected = 3;
576 else
577 return 0;
578 length = expected + 1;
579 for (; expected; expected--)
580 if (s[expected] < 0x80 || s[expected] >= 0xC0)
581 return 0;
582 return length;
583 }
584
585 /* Read a line of input from TOK. Determine encoding
586 if necessary. */
587
588 static char *
decoding_fgets(char * s,int size,struct tok_state * tok)589 decoding_fgets(char *s, int size, struct tok_state *tok)
590 {
591 char *line = NULL;
592 int badchar = 0;
593 for (;;) {
594 if (tok->decoding_state == STATE_NORMAL) {
595 /* We already have a codec associated with
596 this input. */
597 line = fp_readl(s, size, tok);
598 break;
599 } else if (tok->decoding_state == STATE_RAW) {
600 /* We want a 'raw' read. */
601 line = Py_UniversalNewlineFgets(s, size,
602 tok->fp, NULL);
603 break;
604 } else {
605 /* We have not yet determined the encoding.
606 If an encoding is found, use the file-pointer
607 reader functions from now on. */
608 if (!check_bom(fp_getc, fp_ungetc, fp_setreadl, tok))
609 return error_ret(tok);
610 assert(tok->decoding_state != STATE_INIT);
611 }
612 }
613 if (line != NULL && tok->lineno < 2 && !tok->read_coding_spec) {
614 if (!check_coding_spec(line, strlen(line), tok, fp_setreadl)) {
615 return error_ret(tok);
616 }
617 }
618 #ifndef PGEN
619 /* The default encoding is UTF-8, so make sure we don't have any
620 non-UTF-8 sequences in it. */
621 if (line && !tok->encoding) {
622 unsigned char *c;
623 int length;
624 for (c = (unsigned char *)line; *c; c += length)
625 if (!(length = valid_utf8(c))) {
626 badchar = *c;
627 break;
628 }
629 }
630 if (badchar) {
631 /* Need to add 1 to the line number, since this line
632 has not been counted, yet. */
633 PyErr_Format(PyExc_SyntaxError,
634 "Non-UTF-8 code starting with '\\x%.2x' "
635 "in file %U on line %i, "
636 "but no encoding declared; "
637 "see http://python.org/dev/peps/pep-0263/ for details",
638 badchar, tok->filename, tok->lineno + 1);
639 return error_ret(tok);
640 }
641 #endif
642 return line;
643 }
644
645 static int
decoding_feof(struct tok_state * tok)646 decoding_feof(struct tok_state *tok)
647 {
648 if (tok->decoding_state != STATE_NORMAL) {
649 return feof(tok->fp);
650 } else {
651 PyObject* buf = tok->decoding_buffer;
652 if (buf == NULL) {
653 buf = PyObject_CallObject(tok->decoding_readline, NULL);
654 if (buf == NULL) {
655 error_ret(tok);
656 return 1;
657 } else {
658 tok->decoding_buffer = buf;
659 }
660 }
661 return PyObject_Length(buf) == 0;
662 }
663 }
664
665 /* Fetch a byte from TOK, using the string buffer. */
666
667 static int
buf_getc(struct tok_state * tok)668 buf_getc(struct tok_state *tok) {
669 return Py_CHARMASK(*tok->str++);
670 }
671
672 /* Unfetch a byte from TOK, using the string buffer. */
673
674 static void
buf_ungetc(int c,struct tok_state * tok)675 buf_ungetc(int c, struct tok_state *tok) {
676 tok->str--;
677 assert(Py_CHARMASK(*tok->str) == c); /* tok->cur may point to read-only segment */
678 }
679
680 /* Set the readline function for TOK to ENC. For the string-based
681 tokenizer, this means to just record the encoding. */
682
683 static int
buf_setreadl(struct tok_state * tok,const char * enc)684 buf_setreadl(struct tok_state *tok, const char* enc) {
685 tok->enc = enc;
686 return 1;
687 }
688
689 /* Return a UTF-8 encoding Python string object from the
690 C byte string STR, which is encoded with ENC. */
691
692 static PyObject *
translate_into_utf8(const char * str,const char * enc)693 translate_into_utf8(const char* str, const char* enc) {
694 PyObject *utf8;
695 PyObject* buf = PyUnicode_Decode(str, strlen(str), enc, NULL);
696 if (buf == NULL)
697 return NULL;
698 utf8 = PyUnicode_AsUTF8String(buf);
699 Py_DECREF(buf);
700 return utf8;
701 }
702
703
704 static char *
translate_newlines(const char * s,int exec_input,struct tok_state * tok)705 translate_newlines(const char *s, int exec_input, struct tok_state *tok) {
706 int skip_next_lf = 0;
707 size_t needed_length = strlen(s) + 2, final_length;
708 char *buf, *current;
709 char c = '\0';
710 buf = PyMem_MALLOC(needed_length);
711 if (buf == NULL) {
712 tok->done = E_NOMEM;
713 return NULL;
714 }
715 for (current = buf; *s; s++, current++) {
716 c = *s;
717 if (skip_next_lf) {
718 skip_next_lf = 0;
719 if (c == '\n') {
720 c = *++s;
721 if (!c)
722 break;
723 }
724 }
725 if (c == '\r') {
726 skip_next_lf = 1;
727 c = '\n';
728 }
729 *current = c;
730 }
731 /* If this is exec input, add a newline to the end of the string if
732 there isn't one already. */
733 if (exec_input && c != '\n') {
734 *current = '\n';
735 current++;
736 }
737 *current = '\0';
738 final_length = current - buf + 1;
739 if (final_length < needed_length && final_length)
740 /* should never fail */
741 buf = PyMem_REALLOC(buf, final_length);
742 return buf;
743 }
744
745 /* Decode a byte string STR for use as the buffer of TOK.
746 Look for encoding declarations inside STR, and record them
747 inside TOK. */
748
749 static const char *
decode_str(const char * input,int single,struct tok_state * tok)750 decode_str(const char *input, int single, struct tok_state *tok)
751 {
752 PyObject* utf8 = NULL;
753 const char *str;
754 const char *s;
755 const char *newl[2] = {NULL, NULL};
756 int lineno = 0;
757 tok->input = str = translate_newlines(input, single, tok);
758 if (str == NULL)
759 return NULL;
760 tok->enc = NULL;
761 tok->str = str;
762 if (!check_bom(buf_getc, buf_ungetc, buf_setreadl, tok))
763 return error_ret(tok);
764 str = tok->str; /* string after BOM if any */
765 assert(str);
766 if (tok->enc != NULL) {
767 utf8 = translate_into_utf8(str, tok->enc);
768 if (utf8 == NULL)
769 return error_ret(tok);
770 str = PyBytes_AsString(utf8);
771 }
772 for (s = str;; s++) {
773 if (*s == '\0') break;
774 else if (*s == '\n') {
775 assert(lineno < 2);
776 newl[lineno] = s;
777 lineno++;
778 if (lineno == 2) break;
779 }
780 }
781 tok->enc = NULL;
782 /* need to check line 1 and 2 separately since check_coding_spec
783 assumes a single line as input */
784 if (newl[0]) {
785 if (!check_coding_spec(str, newl[0] - str, tok, buf_setreadl))
786 return error_ret(tok);
787 if (tok->enc == NULL && !tok->read_coding_spec && newl[1]) {
788 if (!check_coding_spec(newl[0]+1, newl[1] - newl[0],
789 tok, buf_setreadl))
790 return error_ret(tok);
791 }
792 }
793 if (tok->enc != NULL) {
794 assert(utf8 == NULL);
795 utf8 = translate_into_utf8(str, tok->enc);
796 if (utf8 == NULL)
797 return error_ret(tok);
798 str = PyBytes_AS_STRING(utf8);
799 }
800 assert(tok->decoding_buffer == NULL);
801 tok->decoding_buffer = utf8; /* CAUTION */
802 return str;
803 }
804
805 #endif /* PGEN */
806
807 /* Set up tokenizer for string */
808
809 struct tok_state *
PyTokenizer_FromString(const char * str,int exec_input)810 PyTokenizer_FromString(const char *str, int exec_input)
811 {
812 struct tok_state *tok = tok_new();
813 if (tok == NULL)
814 return NULL;
815 str = decode_str(str, exec_input, tok);
816 if (str == NULL) {
817 PyTokenizer_Free(tok);
818 return NULL;
819 }
820
821 /* XXX: constify members. */
822 tok->buf = tok->cur = tok->end = tok->inp = (char*)str;
823 return tok;
824 }
825
826 struct tok_state *
PyTokenizer_FromUTF8(const char * str,int exec_input)827 PyTokenizer_FromUTF8(const char *str, int exec_input)
828 {
829 struct tok_state *tok = tok_new();
830 if (tok == NULL)
831 return NULL;
832 #ifndef PGEN
833 tok->input = str = translate_newlines(str, exec_input, tok);
834 #endif
835 if (str == NULL) {
836 PyTokenizer_Free(tok);
837 return NULL;
838 }
839 tok->decoding_state = STATE_RAW;
840 tok->read_coding_spec = 1;
841 tok->enc = NULL;
842 tok->str = str;
843 tok->encoding = (char *)PyMem_MALLOC(6);
844 if (!tok->encoding) {
845 PyTokenizer_Free(tok);
846 return NULL;
847 }
848 strcpy(tok->encoding, "utf-8");
849
850 /* XXX: constify members. */
851 tok->buf = tok->cur = tok->end = tok->inp = (char*)str;
852 return tok;
853 }
854
855 /* Set up tokenizer for file */
856
857 struct tok_state *
PyTokenizer_FromFile(FILE * fp,const char * enc,const char * ps1,const char * ps2)858 PyTokenizer_FromFile(FILE *fp, const char* enc,
859 const char *ps1, const char *ps2)
860 {
861 struct tok_state *tok = tok_new();
862 if (tok == NULL)
863 return NULL;
864 if ((tok->buf = (char *)PyMem_MALLOC(BUFSIZ)) == NULL) {
865 PyTokenizer_Free(tok);
866 return NULL;
867 }
868 tok->cur = tok->inp = tok->buf;
869 tok->end = tok->buf + BUFSIZ;
870 tok->fp = fp;
871 tok->prompt = ps1;
872 tok->nextprompt = ps2;
873 if (enc != NULL) {
874 /* Must copy encoding declaration since it
875 gets copied into the parse tree. */
876 tok->encoding = PyMem_MALLOC(strlen(enc)+1);
877 if (!tok->encoding) {
878 PyTokenizer_Free(tok);
879 return NULL;
880 }
881 strcpy(tok->encoding, enc);
882 tok->decoding_state = STATE_NORMAL;
883 }
884 return tok;
885 }
886
887
888 /* Free a tok_state structure */
889
890 void
PyTokenizer_Free(struct tok_state * tok)891 PyTokenizer_Free(struct tok_state *tok)
892 {
893 if (tok->encoding != NULL)
894 PyMem_FREE(tok->encoding);
895 #ifndef PGEN
896 Py_XDECREF(tok->decoding_readline);
897 Py_XDECREF(tok->decoding_buffer);
898 Py_XDECREF(tok->filename);
899 #endif
900 if (tok->fp != NULL && tok->buf != NULL)
901 PyMem_FREE(tok->buf);
902 if (tok->input)
903 PyMem_FREE((char *)tok->input);
904 PyMem_FREE(tok);
905 }
906
907 /* Get next char, updating state; error code goes into tok->done */
908
909 static int
tok_nextc(struct tok_state * tok)910 tok_nextc(struct tok_state *tok)
911 {
912 for (;;) {
913 if (tok->cur != tok->inp) {
914 return Py_CHARMASK(*tok->cur++); /* Fast path */
915 }
916 if (tok->done != E_OK)
917 return EOF;
918 if (tok->fp == NULL) {
919 char *end = strchr(tok->inp, '\n');
920 if (end != NULL)
921 end++;
922 else {
923 end = strchr(tok->inp, '\0');
924 if (end == tok->inp) {
925 tok->done = E_EOF;
926 return EOF;
927 }
928 }
929 if (tok->start == NULL)
930 tok->buf = tok->cur;
931 tok->line_start = tok->cur;
932 tok->lineno++;
933 tok->inp = end;
934 return Py_CHARMASK(*tok->cur++);
935 }
936 if (tok->prompt != NULL) {
937 char *newtok = PyOS_Readline(stdin, stdout, tok->prompt);
938 #ifndef PGEN
939 if (newtok != NULL) {
940 char *translated = translate_newlines(newtok, 0, tok);
941 PyMem_FREE(newtok);
942 if (translated == NULL)
943 return EOF;
944 newtok = translated;
945 }
946 if (tok->encoding && newtok && *newtok) {
947 /* Recode to UTF-8 */
948 Py_ssize_t buflen;
949 const char* buf;
950 PyObject *u = translate_into_utf8(newtok, tok->encoding);
951 PyMem_FREE(newtok);
952 if (!u) {
953 tok->done = E_DECODE;
954 return EOF;
955 }
956 buflen = PyBytes_GET_SIZE(u);
957 buf = PyBytes_AS_STRING(u);
958 newtok = PyMem_MALLOC(buflen+1);
959 strcpy(newtok, buf);
960 Py_DECREF(u);
961 }
962 #endif
963 if (tok->nextprompt != NULL)
964 tok->prompt = tok->nextprompt;
965 if (newtok == NULL)
966 tok->done = E_INTR;
967 else if (*newtok == '\0') {
968 PyMem_FREE(newtok);
969 tok->done = E_EOF;
970 }
971 else if (tok->start != NULL) {
972 size_t start = tok->start - tok->buf;
973 size_t oldlen = tok->cur - tok->buf;
974 size_t newlen = oldlen + strlen(newtok);
975 char *buf = tok->buf;
976 buf = (char *)PyMem_REALLOC(buf, newlen+1);
977 tok->lineno++;
978 if (buf == NULL) {
979 PyMem_FREE(tok->buf);
980 tok->buf = NULL;
981 PyMem_FREE(newtok);
982 tok->done = E_NOMEM;
983 return EOF;
984 }
985 tok->buf = buf;
986 tok->cur = tok->buf + oldlen;
987 tok->line_start = tok->cur;
988 strcpy(tok->buf + oldlen, newtok);
989 PyMem_FREE(newtok);
990 tok->inp = tok->buf + newlen;
991 tok->end = tok->inp + 1;
992 tok->start = tok->buf + start;
993 }
994 else {
995 tok->lineno++;
996 if (tok->buf != NULL)
997 PyMem_FREE(tok->buf);
998 tok->buf = newtok;
999 tok->cur = tok->buf;
1000 tok->line_start = tok->buf;
1001 tok->inp = strchr(tok->buf, '\0');
1002 tok->end = tok->inp + 1;
1003 }
1004 }
1005 else {
1006 int done = 0;
1007 Py_ssize_t cur = 0;
1008 char *pt;
1009 if (tok->start == NULL) {
1010 if (tok->buf == NULL) {
1011 tok->buf = (char *)
1012 PyMem_MALLOC(BUFSIZ);
1013 if (tok->buf == NULL) {
1014 tok->done = E_NOMEM;
1015 return EOF;
1016 }
1017 tok->end = tok->buf + BUFSIZ;
1018 }
1019 if (decoding_fgets(tok->buf, (int)(tok->end - tok->buf),
1020 tok) == NULL) {
1021 if (!tok->decoding_erred)
1022 tok->done = E_EOF;
1023 done = 1;
1024 }
1025 else {
1026 tok->done = E_OK;
1027 tok->inp = strchr(tok->buf, '\0');
1028 done = tok->inp == tok->buf || tok->inp[-1] == '\n';
1029 }
1030 }
1031 else {
1032 cur = tok->cur - tok->buf;
1033 if (decoding_feof(tok)) {
1034 tok->done = E_EOF;
1035 done = 1;
1036 }
1037 else
1038 tok->done = E_OK;
1039 }
1040 tok->lineno++;
1041 /* Read until '\n' or EOF */
1042 while (!done) {
1043 Py_ssize_t curstart = tok->start == NULL ? -1 :
1044 tok->start - tok->buf;
1045 Py_ssize_t curvalid = tok->inp - tok->buf;
1046 Py_ssize_t newsize = curvalid + BUFSIZ;
1047 char *newbuf = tok->buf;
1048 newbuf = (char *)PyMem_REALLOC(newbuf,
1049 newsize);
1050 if (newbuf == NULL) {
1051 tok->done = E_NOMEM;
1052 tok->cur = tok->inp;
1053 return EOF;
1054 }
1055 tok->buf = newbuf;
1056 tok->cur = tok->buf + cur;
1057 tok->line_start = tok->cur;
1058 tok->inp = tok->buf + curvalid;
1059 tok->end = tok->buf + newsize;
1060 tok->start = curstart < 0 ? NULL :
1061 tok->buf + curstart;
1062 if (decoding_fgets(tok->inp,
1063 (int)(tok->end - tok->inp),
1064 tok) == NULL) {
1065 /* Break out early on decoding
1066 errors, as tok->buf will be NULL
1067 */
1068 if (tok->decoding_erred)
1069 return EOF;
1070 /* Last line does not end in \n,
1071 fake one */
1072 strcpy(tok->inp, "\n");
1073 }
1074 tok->inp = strchr(tok->inp, '\0');
1075 done = tok->inp[-1] == '\n';
1076 }
1077 if (tok->buf != NULL) {
1078 tok->cur = tok->buf + cur;
1079 tok->line_start = tok->cur;
1080 /* replace "\r\n" with "\n" */
1081 /* For Mac leave the \r, giving a syntax error */
1082 pt = tok->inp - 2;
1083 if (pt >= tok->buf && *pt == '\r') {
1084 *pt++ = '\n';
1085 *pt = '\0';
1086 tok->inp = pt;
1087 }
1088 }
1089 }
1090 if (tok->done != E_OK) {
1091 if (tok->prompt != NULL)
1092 PySys_WriteStderr("\n");
1093 tok->cur = tok->inp;
1094 return EOF;
1095 }
1096 }
1097 /*NOTREACHED*/
1098 }
1099
1100
1101 /* Back-up one character */
1102
1103 static void
tok_backup(struct tok_state * tok,int c)1104 tok_backup(struct tok_state *tok, int c)
1105 {
1106 if (c != EOF) {
1107 if (--tok->cur < tok->buf)
1108 Py_FatalError("tok_backup: beginning of buffer");
1109 if (*tok->cur != c)
1110 *tok->cur = c;
1111 }
1112 }
1113
1114
1115 /* Return the token corresponding to a single character */
1116
1117 int
PyToken_OneChar(int c)1118 PyToken_OneChar(int c)
1119 {
1120 switch (c) {
1121 case '(': return LPAR;
1122 case ')': return RPAR;
1123 case '[': return LSQB;
1124 case ']': return RSQB;
1125 case ':': return COLON;
1126 case ',': return COMMA;
1127 case ';': return SEMI;
1128 case '+': return PLUS;
1129 case '-': return MINUS;
1130 case '*': return STAR;
1131 case '/': return SLASH;
1132 case '|': return VBAR;
1133 case '&': return AMPER;
1134 case '<': return LESS;
1135 case '>': return GREATER;
1136 case '=': return EQUAL;
1137 case '.': return DOT;
1138 case '%': return PERCENT;
1139 case '{': return LBRACE;
1140 case '}': return RBRACE;
1141 case '^': return CIRCUMFLEX;
1142 case '~': return TILDE;
1143 case '@': return AT;
1144 default: return OP;
1145 }
1146 }
1147
1148
1149 int
PyToken_TwoChars(int c1,int c2)1150 PyToken_TwoChars(int c1, int c2)
1151 {
1152 switch (c1) {
1153 case '=':
1154 switch (c2) {
1155 case '=': return EQEQUAL;
1156 }
1157 break;
1158 case '!':
1159 switch (c2) {
1160 case '=': return NOTEQUAL;
1161 }
1162 break;
1163 case '<':
1164 switch (c2) {
1165 case '>': return NOTEQUAL;
1166 case '=': return LESSEQUAL;
1167 case '<': return LEFTSHIFT;
1168 }
1169 break;
1170 case '>':
1171 switch (c2) {
1172 case '=': return GREATEREQUAL;
1173 case '>': return RIGHTSHIFT;
1174 }
1175 break;
1176 case '+':
1177 switch (c2) {
1178 case '=': return PLUSEQUAL;
1179 }
1180 break;
1181 case '-':
1182 switch (c2) {
1183 case '=': return MINEQUAL;
1184 case '>': return RARROW;
1185 }
1186 break;
1187 case '*':
1188 switch (c2) {
1189 case '*': return DOUBLESTAR;
1190 case '=': return STAREQUAL;
1191 }
1192 break;
1193 case '/':
1194 switch (c2) {
1195 case '/': return DOUBLESLASH;
1196 case '=': return SLASHEQUAL;
1197 }
1198 break;
1199 case '|':
1200 switch (c2) {
1201 case '=': return VBAREQUAL;
1202 }
1203 break;
1204 case '%':
1205 switch (c2) {
1206 case '=': return PERCENTEQUAL;
1207 }
1208 break;
1209 case '&':
1210 switch (c2) {
1211 case '=': return AMPEREQUAL;
1212 }
1213 break;
1214 case '^':
1215 switch (c2) {
1216 case '=': return CIRCUMFLEXEQUAL;
1217 }
1218 break;
1219 case '@':
1220 switch (c2) {
1221 case '=': return ATEQUAL;
1222 }
1223 break;
1224 }
1225 return OP;
1226 }
1227
1228 int
PyToken_ThreeChars(int c1,int c2,int c3)1229 PyToken_ThreeChars(int c1, int c2, int c3)
1230 {
1231 switch (c1) {
1232 case '<':
1233 switch (c2) {
1234 case '<':
1235 switch (c3) {
1236 case '=':
1237 return LEFTSHIFTEQUAL;
1238 }
1239 break;
1240 }
1241 break;
1242 case '>':
1243 switch (c2) {
1244 case '>':
1245 switch (c3) {
1246 case '=':
1247 return RIGHTSHIFTEQUAL;
1248 }
1249 break;
1250 }
1251 break;
1252 case '*':
1253 switch (c2) {
1254 case '*':
1255 switch (c3) {
1256 case '=':
1257 return DOUBLESTAREQUAL;
1258 }
1259 break;
1260 }
1261 break;
1262 case '/':
1263 switch (c2) {
1264 case '/':
1265 switch (c3) {
1266 case '=':
1267 return DOUBLESLASHEQUAL;
1268 }
1269 break;
1270 }
1271 break;
1272 case '.':
1273 switch (c2) {
1274 case '.':
1275 switch (c3) {
1276 case '.':
1277 return ELLIPSIS;
1278 }
1279 break;
1280 }
1281 break;
1282 }
1283 return OP;
1284 }
1285
1286 static int
indenterror(struct tok_state * tok)1287 indenterror(struct tok_state *tok)
1288 {
1289 if (tok->alterror) {
1290 tok->done = E_TABSPACE;
1291 tok->cur = tok->inp;
1292 return 1;
1293 }
1294 if (tok->altwarning) {
1295 #ifdef PGEN
1296 PySys_WriteStderr("inconsistent use of tabs and spaces "
1297 "in indentation\n");
1298 #else
1299 PySys_FormatStderr("%U: inconsistent use of tabs and spaces "
1300 "in indentation\n", tok->filename);
1301 #endif
1302 tok->altwarning = 0;
1303 }
1304 return 0;
1305 }
1306
1307 #ifdef PGEN
1308 #define verify_identifier(tok) 1
1309 #else
1310 /* Verify that the identifier follows PEP 3131.
1311 All identifier strings are guaranteed to be "ready" unicode objects.
1312 */
1313 static int
verify_identifier(struct tok_state * tok)1314 verify_identifier(struct tok_state *tok)
1315 {
1316 PyObject *s;
1317 int result;
1318 if (tok->decoding_erred)
1319 return 0;
1320 s = PyUnicode_DecodeUTF8(tok->start, tok->cur - tok->start, NULL);
1321 if (s == NULL || PyUnicode_READY(s) == -1) {
1322 if (PyErr_ExceptionMatches(PyExc_UnicodeDecodeError)) {
1323 PyErr_Clear();
1324 tok->done = E_IDENTIFIER;
1325 } else {
1326 tok->done = E_ERROR;
1327 }
1328 return 0;
1329 }
1330 result = PyUnicode_IsIdentifier(s);
1331 Py_DECREF(s);
1332 if (result == 0)
1333 tok->done = E_IDENTIFIER;
1334 return result;
1335 }
1336 #endif
1337
1338 static int
tok_decimal_tail(struct tok_state * tok)1339 tok_decimal_tail(struct tok_state *tok)
1340 {
1341 int c;
1342
1343 while (1) {
1344 do {
1345 c = tok_nextc(tok);
1346 } while (isdigit(c));
1347 if (c != '_') {
1348 break;
1349 }
1350 c = tok_nextc(tok);
1351 if (!isdigit(c)) {
1352 tok->done = E_TOKEN;
1353 tok_backup(tok, c);
1354 return 0;
1355 }
1356 }
1357 return c;
1358 }
1359
1360 /* Get next token, after space stripping etc. */
1361
1362 static int
tok_get(struct tok_state * tok,char ** p_start,char ** p_end)1363 tok_get(struct tok_state *tok, char **p_start, char **p_end)
1364 {
1365 int c;
1366 int blankline, nonascii;
1367
1368 *p_start = *p_end = NULL;
1369 nextline:
1370 tok->start = NULL;
1371 blankline = 0;
1372
1373 /* Get indentation level */
1374 if (tok->atbol) {
1375 int col = 0;
1376 int altcol = 0;
1377 tok->atbol = 0;
1378 for (;;) {
1379 c = tok_nextc(tok);
1380 if (c == ' ') {
1381 col++, altcol++;
1382 }
1383 else if (c == '\t') {
1384 col = (col/tok->tabsize + 1) * tok->tabsize;
1385 altcol = (altcol/tok->alttabsize + 1)
1386 * tok->alttabsize;
1387 }
1388 else if (c == '\014') {/* Control-L (formfeed) */
1389 col = altcol = 0; /* For Emacs users */
1390 }
1391 else {
1392 break;
1393 }
1394 }
1395 tok_backup(tok, c);
1396 if (c == '#' || c == '\n') {
1397 /* Lines with only whitespace and/or comments
1398 shouldn't affect the indentation and are
1399 not passed to the parser as NEWLINE tokens,
1400 except *totally* empty lines in interactive
1401 mode, which signal the end of a command group. */
1402 if (col == 0 && c == '\n' && tok->prompt != NULL) {
1403 blankline = 0; /* Let it through */
1404 }
1405 else {
1406 blankline = 1; /* Ignore completely */
1407 }
1408 /* We can't jump back right here since we still
1409 may need to skip to the end of a comment */
1410 }
1411 if (!blankline && tok->level == 0) {
1412 if (col == tok->indstack[tok->indent]) {
1413 /* No change */
1414 if (altcol != tok->altindstack[tok->indent]) {
1415 if (indenterror(tok)) {
1416 return ERRORTOKEN;
1417 }
1418 }
1419 }
1420 else if (col > tok->indstack[tok->indent]) {
1421 /* Indent -- always one */
1422 if (tok->indent+1 >= MAXINDENT) {
1423 tok->done = E_TOODEEP;
1424 tok->cur = tok->inp;
1425 return ERRORTOKEN;
1426 }
1427 if (altcol <= tok->altindstack[tok->indent]) {
1428 if (indenterror(tok)) {
1429 return ERRORTOKEN;
1430 }
1431 }
1432 tok->pendin++;
1433 tok->indstack[++tok->indent] = col;
1434 tok->altindstack[tok->indent] = altcol;
1435 }
1436 else /* col < tok->indstack[tok->indent] */ {
1437 /* Dedent -- any number, must be consistent */
1438 while (tok->indent > 0 &&
1439 col < tok->indstack[tok->indent]) {
1440 tok->pendin--;
1441 tok->indent--;
1442 }
1443 if (col != tok->indstack[tok->indent]) {
1444 tok->done = E_DEDENT;
1445 tok->cur = tok->inp;
1446 return ERRORTOKEN;
1447 }
1448 if (altcol != tok->altindstack[tok->indent]) {
1449 if (indenterror(tok)) {
1450 return ERRORTOKEN;
1451 }
1452 }
1453 }
1454 }
1455 }
1456
1457 tok->start = tok->cur;
1458
1459 /* Return pending indents/dedents */
1460 if (tok->pendin != 0) {
1461 if (tok->pendin < 0) {
1462 tok->pendin++;
1463 return DEDENT;
1464 }
1465 else {
1466 tok->pendin--;
1467 return INDENT;
1468 }
1469 }
1470
1471 if (tok->async_def
1472 && !blankline
1473 && tok->level == 0
1474 /* There was a NEWLINE after ASYNC DEF,
1475 so we're past the signature. */
1476 && tok->async_def_nl
1477 /* Current indentation level is less than where
1478 the async function was defined */
1479 && tok->async_def_indent >= tok->indent)
1480 {
1481 tok->async_def = 0;
1482 tok->async_def_indent = 0;
1483 tok->async_def_nl = 0;
1484 }
1485
1486 again:
1487 tok->start = NULL;
1488 /* Skip spaces */
1489 do {
1490 c = tok_nextc(tok);
1491 } while (c == ' ' || c == '\t' || c == '\014');
1492
1493 /* Set start of current token */
1494 tok->start = tok->cur - 1;
1495
1496 /* Skip comment */
1497 if (c == '#') {
1498 while (c != EOF && c != '\n') {
1499 c = tok_nextc(tok);
1500 }
1501 }
1502
1503 /* Check for EOF and errors now */
1504 if (c == EOF) {
1505 return tok->done == E_EOF ? ENDMARKER : ERRORTOKEN;
1506 }
1507
1508 /* Identifier (most frequent token!) */
1509 nonascii = 0;
1510 if (is_potential_identifier_start(c)) {
1511 /* Process the various legal combinations of b"", r"", u"", and f"". */
1512 int saw_b = 0, saw_r = 0, saw_u = 0, saw_f = 0;
1513 while (1) {
1514 if (!(saw_b || saw_u || saw_f) && (c == 'b' || c == 'B'))
1515 saw_b = 1;
1516 /* Since this is a backwards compatibility support literal we don't
1517 want to support it in arbitrary order like byte literals. */
1518 else if (!(saw_b || saw_u || saw_r || saw_f)
1519 && (c == 'u'|| c == 'U')) {
1520 saw_u = 1;
1521 }
1522 /* ur"" and ru"" are not supported */
1523 else if (!(saw_r || saw_u) && (c == 'r' || c == 'R')) {
1524 saw_r = 1;
1525 }
1526 else if (!(saw_f || saw_b || saw_u) && (c == 'f' || c == 'F')) {
1527 saw_f = 1;
1528 }
1529 else {
1530 break;
1531 }
1532 c = tok_nextc(tok);
1533 if (c == '"' || c == '\'') {
1534 goto letter_quote;
1535 }
1536 }
1537 while (is_potential_identifier_char(c)) {
1538 if (c >= 128) {
1539 nonascii = 1;
1540 }
1541 c = tok_nextc(tok);
1542 }
1543 tok_backup(tok, c);
1544 if (nonascii && !verify_identifier(tok)) {
1545 return ERRORTOKEN;
1546 }
1547 *p_start = tok->start;
1548 *p_end = tok->cur;
1549
1550 /* async/await parsing block. */
1551 if (tok->cur - tok->start == 5) {
1552 /* Current token length is 5. */
1553 if (tok->async_def) {
1554 /* We're inside an 'async def' function. */
1555 if (memcmp(tok->start, "async", 5) == 0) {
1556 return ASYNC;
1557 }
1558 if (memcmp(tok->start, "await", 5) == 0) {
1559 return AWAIT;
1560 }
1561 }
1562 else if (memcmp(tok->start, "async", 5) == 0) {
1563 /* The current token is 'async'.
1564 Look ahead one token.*/
1565
1566 struct tok_state ahead_tok;
1567 char *ahead_tok_start = NULL, *ahead_tok_end = NULL;
1568 int ahead_tok_kind;
1569
1570 memcpy(&ahead_tok, tok, sizeof(ahead_tok));
1571 ahead_tok_kind = tok_get(&ahead_tok, &ahead_tok_start,
1572 &ahead_tok_end);
1573
1574 if (ahead_tok_kind == NAME
1575 && ahead_tok.cur - ahead_tok.start == 3
1576 && memcmp(ahead_tok.start, "def", 3) == 0)
1577 {
1578 /* The next token is going to be 'def', so instead of
1579 returning 'async' NAME token, we return ASYNC. */
1580 tok->async_def_indent = tok->indent;
1581 tok->async_def = 1;
1582 return ASYNC;
1583 }
1584 }
1585 }
1586
1587 return NAME;
1588 }
1589
1590 /* Newline */
1591 if (c == '\n') {
1592 tok->atbol = 1;
1593 if (blankline || tok->level > 0) {
1594 goto nextline;
1595 }
1596 *p_start = tok->start;
1597 *p_end = tok->cur - 1; /* Leave '\n' out of the string */
1598 tok->cont_line = 0;
1599 if (tok->async_def) {
1600 /* We're somewhere inside an 'async def' function, and
1601 we've encountered a NEWLINE after its signature. */
1602 tok->async_def_nl = 1;
1603 }
1604 return NEWLINE;
1605 }
1606
1607 /* Period or number starting with period? */
1608 if (c == '.') {
1609 c = tok_nextc(tok);
1610 if (isdigit(c)) {
1611 goto fraction;
1612 } else if (c == '.') {
1613 c = tok_nextc(tok);
1614 if (c == '.') {
1615 *p_start = tok->start;
1616 *p_end = tok->cur;
1617 return ELLIPSIS;
1618 }
1619 else {
1620 tok_backup(tok, c);
1621 }
1622 tok_backup(tok, '.');
1623 }
1624 else {
1625 tok_backup(tok, c);
1626 }
1627 *p_start = tok->start;
1628 *p_end = tok->cur;
1629 return DOT;
1630 }
1631
1632 /* Number */
1633 if (isdigit(c)) {
1634 if (c == '0') {
1635 /* Hex, octal or binary -- maybe. */
1636 c = tok_nextc(tok);
1637 if (c == 'x' || c == 'X') {
1638 /* Hex */
1639 c = tok_nextc(tok);
1640 do {
1641 if (c == '_') {
1642 c = tok_nextc(tok);
1643 }
1644 if (!isxdigit(c)) {
1645 tok->done = E_TOKEN;
1646 tok_backup(tok, c);
1647 return ERRORTOKEN;
1648 }
1649 do {
1650 c = tok_nextc(tok);
1651 } while (isxdigit(c));
1652 } while (c == '_');
1653 }
1654 else if (c == 'o' || c == 'O') {
1655 /* Octal */
1656 c = tok_nextc(tok);
1657 do {
1658 if (c == '_') {
1659 c = tok_nextc(tok);
1660 }
1661 if (c < '0' || c >= '8') {
1662 tok->done = E_TOKEN;
1663 tok_backup(tok, c);
1664 return ERRORTOKEN;
1665 }
1666 do {
1667 c = tok_nextc(tok);
1668 } while ('0' <= c && c < '8');
1669 } while (c == '_');
1670 }
1671 else if (c == 'b' || c == 'B') {
1672 /* Binary */
1673 c = tok_nextc(tok);
1674 do {
1675 if (c == '_') {
1676 c = tok_nextc(tok);
1677 }
1678 if (c != '0' && c != '1') {
1679 tok->done = E_TOKEN;
1680 tok_backup(tok, c);
1681 return ERRORTOKEN;
1682 }
1683 do {
1684 c = tok_nextc(tok);
1685 } while (c == '0' || c == '1');
1686 } while (c == '_');
1687 }
1688 else {
1689 int nonzero = 0;
1690 /* maybe old-style octal; c is first char of it */
1691 /* in any case, allow '0' as a literal */
1692 while (1) {
1693 if (c == '_') {
1694 c = tok_nextc(tok);
1695 if (!isdigit(c)) {
1696 tok->done = E_TOKEN;
1697 tok_backup(tok, c);
1698 return ERRORTOKEN;
1699 }
1700 }
1701 if (c != '0') {
1702 break;
1703 }
1704 c = tok_nextc(tok);
1705 }
1706 if (isdigit(c)) {
1707 nonzero = 1;
1708 c = tok_decimal_tail(tok);
1709 if (c == 0) {
1710 return ERRORTOKEN;
1711 }
1712 }
1713 if (c == '.') {
1714 c = tok_nextc(tok);
1715 goto fraction;
1716 }
1717 else if (c == 'e' || c == 'E') {
1718 goto exponent;
1719 }
1720 else if (c == 'j' || c == 'J') {
1721 goto imaginary;
1722 }
1723 else if (nonzero) {
1724 /* Old-style octal: now disallowed. */
1725 tok->done = E_TOKEN;
1726 tok_backup(tok, c);
1727 return ERRORTOKEN;
1728 }
1729 }
1730 }
1731 else {
1732 /* Decimal */
1733 c = tok_decimal_tail(tok);
1734 if (c == 0) {
1735 return ERRORTOKEN;
1736 }
1737 {
1738 /* Accept floating point numbers. */
1739 if (c == '.') {
1740 c = tok_nextc(tok);
1741 fraction:
1742 /* Fraction */
1743 if (isdigit(c)) {
1744 c = tok_decimal_tail(tok);
1745 if (c == 0) {
1746 return ERRORTOKEN;
1747 }
1748 }
1749 }
1750 if (c == 'e' || c == 'E') {
1751 int e;
1752 exponent:
1753 e = c;
1754 /* Exponent part */
1755 c = tok_nextc(tok);
1756 if (c == '+' || c == '-') {
1757 c = tok_nextc(tok);
1758 if (!isdigit(c)) {
1759 tok->done = E_TOKEN;
1760 tok_backup(tok, c);
1761 return ERRORTOKEN;
1762 }
1763 } else if (!isdigit(c)) {
1764 tok_backup(tok, c);
1765 tok_backup(tok, e);
1766 *p_start = tok->start;
1767 *p_end = tok->cur;
1768 return NUMBER;
1769 }
1770 c = tok_decimal_tail(tok);
1771 if (c == 0) {
1772 return ERRORTOKEN;
1773 }
1774 }
1775 if (c == 'j' || c == 'J') {
1776 /* Imaginary part */
1777 imaginary:
1778 c = tok_nextc(tok);
1779 }
1780 }
1781 }
1782 tok_backup(tok, c);
1783 *p_start = tok->start;
1784 *p_end = tok->cur;
1785 return NUMBER;
1786 }
1787
1788 letter_quote:
1789 /* String */
1790 if (c == '\'' || c == '"') {
1791 int quote = c;
1792 int quote_size = 1; /* 1 or 3 */
1793 int end_quote_size = 0;
1794
1795 /* Find the quote size and start of string */
1796 c = tok_nextc(tok);
1797 if (c == quote) {
1798 c = tok_nextc(tok);
1799 if (c == quote) {
1800 quote_size = 3;
1801 }
1802 else {
1803 end_quote_size = 1; /* empty string found */
1804 }
1805 }
1806 if (c != quote) {
1807 tok_backup(tok, c);
1808 }
1809
1810 /* Get rest of string */
1811 while (end_quote_size != quote_size) {
1812 c = tok_nextc(tok);
1813 if (c == EOF) {
1814 if (quote_size == 3) {
1815 tok->done = E_EOFS;
1816 }
1817 else {
1818 tok->done = E_EOLS;
1819 }
1820 tok->cur = tok->inp;
1821 return ERRORTOKEN;
1822 }
1823 if (quote_size == 1 && c == '\n') {
1824 tok->done = E_EOLS;
1825 tok->cur = tok->inp;
1826 return ERRORTOKEN;
1827 }
1828 if (c == quote) {
1829 end_quote_size += 1;
1830 }
1831 else {
1832 end_quote_size = 0;
1833 if (c == '\\') {
1834 tok_nextc(tok); /* skip escaped char */
1835 }
1836 }
1837 }
1838
1839 *p_start = tok->start;
1840 *p_end = tok->cur;
1841 return STRING;
1842 }
1843
1844 /* Line continuation */
1845 if (c == '\\') {
1846 c = tok_nextc(tok);
1847 if (c != '\n') {
1848 tok->done = E_LINECONT;
1849 tok->cur = tok->inp;
1850 return ERRORTOKEN;
1851 }
1852 tok->cont_line = 1;
1853 goto again; /* Read next line */
1854 }
1855
1856 /* Check for two-character token */
1857 {
1858 int c2 = tok_nextc(tok);
1859 int token = PyToken_TwoChars(c, c2);
1860 if (token != OP) {
1861 int c3 = tok_nextc(tok);
1862 int token3 = PyToken_ThreeChars(c, c2, c3);
1863 if (token3 != OP) {
1864 token = token3;
1865 }
1866 else {
1867 tok_backup(tok, c3);
1868 }
1869 *p_start = tok->start;
1870 *p_end = tok->cur;
1871 return token;
1872 }
1873 tok_backup(tok, c2);
1874 }
1875
1876 /* Keep track of parentheses nesting level */
1877 switch (c) {
1878 case '(':
1879 case '[':
1880 case '{':
1881 tok->level++;
1882 break;
1883 case ')':
1884 case ']':
1885 case '}':
1886 tok->level--;
1887 break;
1888 }
1889
1890 /* Punctuation character */
1891 *p_start = tok->start;
1892 *p_end = tok->cur;
1893 return PyToken_OneChar(c);
1894 }
1895
1896 int
PyTokenizer_Get(struct tok_state * tok,char ** p_start,char ** p_end)1897 PyTokenizer_Get(struct tok_state *tok, char **p_start, char **p_end)
1898 {
1899 int result = tok_get(tok, p_start, p_end);
1900 if (tok->decoding_erred) {
1901 result = ERRORTOKEN;
1902 tok->done = E_DECODE;
1903 }
1904 return result;
1905 }
1906
1907 /* Get the encoding of a Python file. Check for the coding cookie and check if
1908 the file starts with a BOM.
1909
1910 PyTokenizer_FindEncodingFilename() returns NULL when it can't find the
1911 encoding in the first or second line of the file (in which case the encoding
1912 should be assumed to be UTF-8).
1913
1914 The char* returned is malloc'ed via PyMem_MALLOC() and thus must be freed
1915 by the caller. */
1916
1917 char *
PyTokenizer_FindEncodingFilename(int fd,PyObject * filename)1918 PyTokenizer_FindEncodingFilename(int fd, PyObject *filename)
1919 {
1920 struct tok_state *tok;
1921 FILE *fp;
1922 char *p_start =NULL , *p_end =NULL , *encoding = NULL;
1923
1924 #ifndef PGEN
1925 fd = _Py_dup(fd);
1926 #else
1927 fd = dup(fd);
1928 #endif
1929 if (fd < 0) {
1930 return NULL;
1931 }
1932
1933 fp = fdopen(fd, "r");
1934 if (fp == NULL) {
1935 return NULL;
1936 }
1937 tok = PyTokenizer_FromFile(fp, NULL, NULL, NULL);
1938 if (tok == NULL) {
1939 fclose(fp);
1940 return NULL;
1941 }
1942 #ifndef PGEN
1943 if (filename != NULL) {
1944 Py_INCREF(filename);
1945 tok->filename = filename;
1946 }
1947 else {
1948 tok->filename = PyUnicode_FromString("<string>");
1949 if (tok->filename == NULL) {
1950 fclose(fp);
1951 PyTokenizer_Free(tok);
1952 return encoding;
1953 }
1954 }
1955 #endif
1956 while (tok->lineno < 2 && tok->done == E_OK) {
1957 PyTokenizer_Get(tok, &p_start, &p_end);
1958 }
1959 fclose(fp);
1960 if (tok->encoding) {
1961 encoding = (char *)PyMem_MALLOC(strlen(tok->encoding) + 1);
1962 if (encoding)
1963 strcpy(encoding, tok->encoding);
1964 }
1965 PyTokenizer_Free(tok);
1966 return encoding;
1967 }
1968
1969 char *
PyTokenizer_FindEncoding(int fd)1970 PyTokenizer_FindEncoding(int fd)
1971 {
1972 return PyTokenizer_FindEncodingFilename(fd, NULL);
1973 }
1974
1975 #ifdef Py_DEBUG
1976
1977 void
tok_dump(int type,char * start,char * end)1978 tok_dump(int type, char *start, char *end)
1979 {
1980 printf("%s", _PyParser_TokenNames[type]);
1981 if (type == NAME || type == NUMBER || type == STRING || type == OP)
1982 printf("(%.*s)", (int)(end - start), start);
1983 }
1984
1985 #endif
1986