1 /*
2 +----------------------------------------------------------------------+
3 | Zend Engine |
4 +----------------------------------------------------------------------+
5 | Copyright (c) Zend Technologies Ltd. (http://www.zend.com) |
6 +----------------------------------------------------------------------+
7 | This source file is subject to version 2.00 of the Zend license, |
8 | that is bundled with this package in the file LICENSE, and is |
9 | available through the world-wide-web at the following url: |
10 | http://www.zend.com/license/2_00.txt. |
11 | If you did not receive a copy of the Zend license and are unable to |
12 | obtain it through the world-wide-web, please send a note to |
13 | license@zend.com so we can mail you a copy immediately. |
14 +----------------------------------------------------------------------+
15 | Authors: Marcus Boerger <helly@php.net> |
16 | Nuno Lopes <nlopess@php.net> |
17 | Scott MacVicar <scottmac@php.net> |
18 | Flex version authors: |
19 | Andi Gutmans <andi@php.net> |
20 | Zeev Suraski <zeev@php.net> |
21 +----------------------------------------------------------------------+
22 */
23
24 #if 0
25 # define YYDEBUG(s, c) printf("state: %d char: %c\n", s, c)
26 #else
27 # define YYDEBUG(s, c)
28 #endif
29
30 #include "zend_language_scanner_defs.h"
31
32 #include <errno.h>
33 #include "zend.h"
34 #ifdef ZEND_WIN32
35 # include <Winuser.h>
36 #endif
37 #include "zend_alloc.h"
38 #include <zend_language_parser.h>
39 #include "zend_compile.h"
40 #include "zend_language_scanner.h"
41 #include "zend_highlight.h"
42 #include "zend_constants.h"
43 #include "zend_variables.h"
44 #include "zend_operators.h"
45 #include "zend_API.h"
46 #include "zend_strtod.h"
47 #include "zend_exceptions.h"
48 #include "zend_virtual_cwd.h"
49
50 #define YYCTYPE unsigned char
51 #define YYFILL(n) { if ((YYCURSOR + n) >= (YYLIMIT + ZEND_MMAP_AHEAD)) { return 0; } }
52 #define YYCURSOR SCNG(yy_cursor)
53 #define YYLIMIT SCNG(yy_limit)
54 #define YYMARKER SCNG(yy_marker)
55
56 #define YYGETCONDITION() SCNG(yy_state)
57 #define YYSETCONDITION(s) SCNG(yy_state) = s
58
59 #define STATE(name) yyc##name
60
61 /* emulate flex constructs */
62 #define BEGIN(state) YYSETCONDITION(STATE(state))
63 #define YYSTATE YYGETCONDITION()
64 #define yytext ((char*)SCNG(yy_text))
65 #define yyleng SCNG(yy_leng)
66 #define yyless(x) do { YYCURSOR = (unsigned char*)yytext + x; \
67 yyleng = (unsigned int)x; } while(0)
68 #define yymore() goto yymore_restart
69
70 /* perform sanity check. If this message is triggered you should
71 increase the ZEND_MMAP_AHEAD value in the zend_streams.h file */
72 /*!max:re2c */
73 #if ZEND_MMAP_AHEAD < YYMAXFILL
74 # error ZEND_MMAP_AHEAD should be greater than or equal to YYMAXFILL
75 #endif
76
77 #include <stdarg.h>
78
79 #ifdef HAVE_UNISTD_H
80 # include <unistd.h>
81 #endif
82
83 /* Globals Macros */
84 #define SCNG LANG_SCNG
85 #ifdef ZTS
86 ZEND_API ts_rsrc_id language_scanner_globals_id;
87 ZEND_API size_t language_scanner_globals_offset;
88 #else
89 ZEND_API zend_php_scanner_globals language_scanner_globals;
90 #endif
91
92 #define HANDLE_NEWLINES(s, l) \
93 do { \
94 char *p = (s), *boundary = p+(l); \
95 \
96 while (p<boundary) { \
97 if (*p == '\n' || (*p == '\r' && (*(p+1) != '\n'))) { \
98 CG(zend_lineno)++; \
99 } \
100 p++; \
101 } \
102 } while (0)
103
104 #define HANDLE_NEWLINE(c) \
105 { \
106 if (c == '\n' || c == '\r') { \
107 CG(zend_lineno)++; \
108 } \
109 }
110
111 /* To save initial string length after scanning to first variable */
112 #define SET_DOUBLE_QUOTES_SCANNED_LENGTH(len) SCNG(scanned_string_len) = (len)
113 #define GET_DOUBLE_QUOTES_SCANNED_LENGTH() SCNG(scanned_string_len)
114
115 #define IS_LABEL_START(c) (((c) >= 'a' && (c) <= 'z') || ((c) >= 'A' && (c) <= 'Z') || (c) == '_' || (c) >= 0x80)
116 #define IS_LABEL_SUCCESSOR(c) (((c) >= 'a' && (c) <= 'z') || ((c) >= 'A' && (c) <= 'Z') || ((c) >= '0' && (c) <= '9') || (c) == '_' || (c) >= 0x80)
117
118 #define ZEND_IS_OCT(c) ((c)>='0' && (c)<='7')
119 #define ZEND_IS_HEX(c) (((c)>='0' && (c)<='9') || ((c)>='a' && (c)<='f') || ((c)>='A' && (c)<='F'))
120
121
strip_underscores(char * str,size_t * len)122 static void strip_underscores(char *str, size_t *len)
123 {
124 char *src = str, *dest = str;
125 while (*src != '\0') {
126 if (*src != '_') {
127 *dest = *src;
128 dest++;
129 } else {
130 --(*len);
131 }
132 src++;
133 }
134 *dest = '\0';
135 }
136
encoding_filter_script_to_internal(unsigned char ** to,size_t * to_length,const unsigned char * from,size_t from_length)137 static size_t encoding_filter_script_to_internal(unsigned char **to, size_t *to_length, const unsigned char *from, size_t from_length)
138 {
139 const zend_encoding *internal_encoding = zend_multibyte_get_internal_encoding();
140 ZEND_ASSERT(internal_encoding);
141 return zend_multibyte_encoding_converter(to, to_length, from, from_length, internal_encoding, LANG_SCNG(script_encoding));
142 }
143
encoding_filter_script_to_intermediate(unsigned char ** to,size_t * to_length,const unsigned char * from,size_t from_length)144 static size_t encoding_filter_script_to_intermediate(unsigned char **to, size_t *to_length, const unsigned char *from, size_t from_length)
145 {
146 return zend_multibyte_encoding_converter(to, to_length, from, from_length, zend_multibyte_encoding_utf8, LANG_SCNG(script_encoding));
147 }
148
encoding_filter_intermediate_to_script(unsigned char ** to,size_t * to_length,const unsigned char * from,size_t from_length)149 static size_t encoding_filter_intermediate_to_script(unsigned char **to, size_t *to_length, const unsigned char *from, size_t from_length)
150 {
151 return zend_multibyte_encoding_converter(to, to_length, from, from_length,
152 LANG_SCNG(script_encoding), zend_multibyte_encoding_utf8);
153 }
154
encoding_filter_intermediate_to_internal(unsigned char ** to,size_t * to_length,const unsigned char * from,size_t from_length)155 static size_t encoding_filter_intermediate_to_internal(unsigned char **to, size_t *to_length, const unsigned char *from, size_t from_length)
156 {
157 const zend_encoding *internal_encoding = zend_multibyte_get_internal_encoding();
158 ZEND_ASSERT(internal_encoding);
159 return zend_multibyte_encoding_converter(to, to_length, from, from_length,
160 internal_encoding, zend_multibyte_encoding_utf8);
161 }
162
163
_yy_push_state(int new_state)164 static void _yy_push_state(int new_state)
165 {
166 zend_stack_push(&SCNG(state_stack), (void *) &YYGETCONDITION());
167 YYSETCONDITION(new_state);
168 }
169
170 #define yy_push_state(state_and_tsrm) _yy_push_state(yyc##state_and_tsrm)
171
yy_pop_state(void)172 static void yy_pop_state(void)
173 {
174 int *stack_state = zend_stack_top(&SCNG(state_stack));
175 YYSETCONDITION(*stack_state);
176 zend_stack_del_top(&SCNG(state_stack));
177 }
178
yy_scan_buffer(char * str,size_t len)179 static void yy_scan_buffer(char *str, size_t len)
180 {
181 YYCURSOR = (YYCTYPE*)str;
182 YYLIMIT = YYCURSOR + len;
183 if (!SCNG(yy_start)) {
184 SCNG(yy_start) = YYCURSOR;
185 }
186 }
187
startup_scanner(void)188 void startup_scanner(void)
189 {
190 CG(parse_error) = 0;
191 CG(doc_comment) = NULL;
192 CG(extra_fn_flags) = 0;
193 zend_stack_init(&SCNG(state_stack), sizeof(int));
194 zend_stack_init(&SCNG(nest_location_stack), sizeof(zend_nest_location));
195 zend_ptr_stack_init(&SCNG(heredoc_label_stack));
196 SCNG(heredoc_scan_ahead) = 0;
197 }
198
heredoc_label_dtor(zend_heredoc_label * heredoc_label)199 static void heredoc_label_dtor(zend_heredoc_label *heredoc_label) {
200 efree(heredoc_label->label);
201 }
202
shutdown_scanner(void)203 void shutdown_scanner(void)
204 {
205 CG(parse_error) = 0;
206 RESET_DOC_COMMENT();
207 zend_stack_destroy(&SCNG(state_stack));
208 zend_stack_destroy(&SCNG(nest_location_stack));
209 zend_ptr_stack_clean(&SCNG(heredoc_label_stack), (void (*)(void *)) &heredoc_label_dtor, 1);
210 zend_ptr_stack_destroy(&SCNG(heredoc_label_stack));
211 SCNG(heredoc_scan_ahead) = 0;
212 SCNG(on_event) = NULL;
213 }
214
zend_save_lexical_state(zend_lex_state * lex_state)215 ZEND_API void zend_save_lexical_state(zend_lex_state *lex_state)
216 {
217 lex_state->yy_leng = SCNG(yy_leng);
218 lex_state->yy_start = SCNG(yy_start);
219 lex_state->yy_text = SCNG(yy_text);
220 lex_state->yy_cursor = SCNG(yy_cursor);
221 lex_state->yy_marker = SCNG(yy_marker);
222 lex_state->yy_limit = SCNG(yy_limit);
223
224 lex_state->state_stack = SCNG(state_stack);
225 zend_stack_init(&SCNG(state_stack), sizeof(int));
226
227 lex_state->nest_location_stack = SCNG(nest_location_stack);
228 zend_stack_init(&SCNG(nest_location_stack), sizeof(zend_nest_location));
229
230 lex_state->heredoc_label_stack = SCNG(heredoc_label_stack);
231 zend_ptr_stack_init(&SCNG(heredoc_label_stack));
232
233 lex_state->in = SCNG(yy_in);
234 lex_state->yy_state = YYSTATE;
235 lex_state->filename = CG(compiled_filename);
236 lex_state->lineno = CG(zend_lineno);
237 CG(compiled_filename) = NULL;
238
239 lex_state->script_org = SCNG(script_org);
240 lex_state->script_org_size = SCNG(script_org_size);
241 lex_state->script_filtered = SCNG(script_filtered);
242 lex_state->script_filtered_size = SCNG(script_filtered_size);
243 lex_state->input_filter = SCNG(input_filter);
244 lex_state->output_filter = SCNG(output_filter);
245 lex_state->script_encoding = SCNG(script_encoding);
246
247 lex_state->on_event = SCNG(on_event);
248 lex_state->on_event_context = SCNG(on_event_context);
249
250 lex_state->ast = CG(ast);
251 lex_state->ast_arena = CG(ast_arena);
252 }
253
zend_restore_lexical_state(zend_lex_state * lex_state)254 ZEND_API void zend_restore_lexical_state(zend_lex_state *lex_state)
255 {
256 SCNG(yy_leng) = lex_state->yy_leng;
257 SCNG(yy_start) = lex_state->yy_start;
258 SCNG(yy_text) = lex_state->yy_text;
259 SCNG(yy_cursor) = lex_state->yy_cursor;
260 SCNG(yy_marker) = lex_state->yy_marker;
261 SCNG(yy_limit) = lex_state->yy_limit;
262
263 zend_stack_destroy(&SCNG(state_stack));
264 SCNG(state_stack) = lex_state->state_stack;
265
266 zend_stack_destroy(&SCNG(nest_location_stack));
267 SCNG(nest_location_stack) = lex_state->nest_location_stack;
268
269 zend_ptr_stack_clean(&SCNG(heredoc_label_stack), (void (*)(void *)) &heredoc_label_dtor, 1);
270 zend_ptr_stack_destroy(&SCNG(heredoc_label_stack));
271 SCNG(heredoc_label_stack) = lex_state->heredoc_label_stack;
272
273 SCNG(yy_in) = lex_state->in;
274 YYSETCONDITION(lex_state->yy_state);
275 CG(zend_lineno) = lex_state->lineno;
276 zend_restore_compiled_filename(lex_state->filename);
277
278 if (SCNG(script_filtered)) {
279 efree(SCNG(script_filtered));
280 SCNG(script_filtered) = NULL;
281 }
282 SCNG(script_org) = lex_state->script_org;
283 SCNG(script_org_size) = lex_state->script_org_size;
284 SCNG(script_filtered) = lex_state->script_filtered;
285 SCNG(script_filtered_size) = lex_state->script_filtered_size;
286 SCNG(input_filter) = lex_state->input_filter;
287 SCNG(output_filter) = lex_state->output_filter;
288 SCNG(script_encoding) = lex_state->script_encoding;
289
290 SCNG(on_event) = lex_state->on_event;
291 SCNG(on_event_context) = lex_state->on_event_context;
292
293 CG(ast) = lex_state->ast;
294 CG(ast_arena) = lex_state->ast_arena;
295
296 RESET_DOC_COMMENT();
297 }
298
zend_lex_tstring(zval * zv,unsigned char * ident)299 ZEND_API zend_result zend_lex_tstring(zval *zv, unsigned char *ident)
300 {
301 unsigned char *end = ident;
302 while ((*end >= 'a' && *end <= 'z') || (*end >= 'A' && *end <= 'Z') || *end == '_') {
303 end++;
304 }
305
306 size_t length = end - ident;
307 if (length == 0) {
308 ZEND_ASSERT(ident[0] == '<' && ident[1] == '?' && ident[2] == '=');
309 zend_throw_exception(zend_ce_parse_error, "Cannot use \"<?=\" as an identifier", 0);
310 return FAILURE;
311 }
312
313 if (SCNG(on_event)) {
314 SCNG(on_event)(ON_FEEDBACK, T_STRING, 0, (char *) ident, length, SCNG(on_event_context));
315 }
316
317 ZVAL_STRINGL(zv, (char *) ident, length);
318 return SUCCESS;
319 }
320
321 #define BOM_UTF32_BE "\x00\x00\xfe\xff"
322 #define BOM_UTF32_LE "\xff\xfe\x00\x00"
323 #define BOM_UTF16_BE "\xfe\xff"
324 #define BOM_UTF16_LE "\xff\xfe"
325 #define BOM_UTF8 "\xef\xbb\xbf"
326
zend_multibyte_detect_utf_encoding(const unsigned char * script,size_t script_size)327 static const zend_encoding *zend_multibyte_detect_utf_encoding(const unsigned char *script, size_t script_size)
328 {
329 const unsigned char *p;
330 int wchar_size = 2;
331 int le = 0;
332
333 /* utf-16 or utf-32? */
334 p = script;
335 assert(p >= script);
336 while ((size_t)(p-script) < script_size) {
337 p = memchr(p, 0, script_size-(p-script)-2);
338 if (!p) {
339 break;
340 }
341 if (*(p+1) == '\0' && *(p+2) == '\0') {
342 wchar_size = 4;
343 break;
344 }
345
346 /* searching for UTF-32 specific byte orders, so this will do */
347 p += 4;
348 }
349
350 /* BE or LE? */
351 p = script;
352 assert(p >= script);
353 while ((size_t)(p-script) < script_size) {
354 if (*p == '\0' && *(p+wchar_size-1) != '\0') {
355 /* BE */
356 le = 0;
357 break;
358 } else if (*p != '\0' && *(p+wchar_size-1) == '\0') {
359 /* LE* */
360 le = 1;
361 break;
362 }
363 p += wchar_size;
364 }
365
366 if (wchar_size == 2) {
367 return le ? zend_multibyte_encoding_utf16le : zend_multibyte_encoding_utf16be;
368 } else {
369 return le ? zend_multibyte_encoding_utf32le : zend_multibyte_encoding_utf32be;
370 }
371
372 return NULL;
373 }
374
zend_multibyte_detect_unicode(void)375 static const zend_encoding* zend_multibyte_detect_unicode(void)
376 {
377 const zend_encoding *script_encoding = NULL;
378 int bom_size;
379 unsigned char *pos1, *pos2;
380
381 if (LANG_SCNG(script_org_size) < sizeof(BOM_UTF32_LE)-1) {
382 return NULL;
383 }
384
385 /* check out BOM */
386 if (!memcmp(LANG_SCNG(script_org), BOM_UTF32_BE, sizeof(BOM_UTF32_BE)-1)) {
387 script_encoding = zend_multibyte_encoding_utf32be;
388 bom_size = sizeof(BOM_UTF32_BE)-1;
389 } else if (!memcmp(LANG_SCNG(script_org), BOM_UTF32_LE, sizeof(BOM_UTF32_LE)-1)) {
390 script_encoding = zend_multibyte_encoding_utf32le;
391 bom_size = sizeof(BOM_UTF32_LE)-1;
392 } else if (!memcmp(LANG_SCNG(script_org), BOM_UTF16_BE, sizeof(BOM_UTF16_BE)-1)) {
393 script_encoding = zend_multibyte_encoding_utf16be;
394 bom_size = sizeof(BOM_UTF16_BE)-1;
395 } else if (!memcmp(LANG_SCNG(script_org), BOM_UTF16_LE, sizeof(BOM_UTF16_LE)-1)) {
396 script_encoding = zend_multibyte_encoding_utf16le;
397 bom_size = sizeof(BOM_UTF16_LE)-1;
398 } else if (!memcmp(LANG_SCNG(script_org), BOM_UTF8, sizeof(BOM_UTF8)-1)) {
399 script_encoding = zend_multibyte_encoding_utf8;
400 bom_size = sizeof(BOM_UTF8)-1;
401 }
402
403 if (script_encoding) {
404 /* remove BOM */
405 LANG_SCNG(script_org) += bom_size;
406 LANG_SCNG(script_org_size) -= bom_size;
407
408 return script_encoding;
409 }
410
411 /* script contains NULL bytes -> auto-detection */
412 if ((pos1 = memchr(LANG_SCNG(script_org), 0, LANG_SCNG(script_org_size)))) {
413 /* check if the NULL byte is after the __HALT_COMPILER(); */
414 pos2 = LANG_SCNG(script_org);
415
416 while ((size_t)(pos1 - pos2) >= sizeof("__HALT_COMPILER();")-1) {
417 pos2 = memchr(pos2, '_', pos1 - pos2);
418 if (!pos2) break;
419 pos2++;
420 if (strncasecmp((char*)pos2, "_HALT_COMPILER", sizeof("_HALT_COMPILER")-1) == 0) {
421 pos2 += sizeof("_HALT_COMPILER")-1;
422 while (*pos2 == ' ' ||
423 *pos2 == '\t' ||
424 *pos2 == '\r' ||
425 *pos2 == '\n') {
426 pos2++;
427 }
428 if (*pos2 == '(') {
429 pos2++;
430 while (*pos2 == ' ' ||
431 *pos2 == '\t' ||
432 *pos2 == '\r' ||
433 *pos2 == '\n') {
434 pos2++;
435 }
436 if (*pos2 == ')') {
437 pos2++;
438 while (*pos2 == ' ' ||
439 *pos2 == '\t' ||
440 *pos2 == '\r' ||
441 *pos2 == '\n') {
442 pos2++;
443 }
444 if (*pos2 == ';') {
445 return NULL;
446 }
447 }
448 }
449 }
450 }
451 /* make best effort if BOM is missing */
452 return zend_multibyte_detect_utf_encoding(LANG_SCNG(script_org), LANG_SCNG(script_org_size));
453 }
454
455 return NULL;
456 }
457
zend_multibyte_find_script_encoding(void)458 static const zend_encoding* zend_multibyte_find_script_encoding(void)
459 {
460 const zend_encoding *script_encoding;
461
462 if (CG(detect_unicode)) {
463 /* check out bom(byte order mark) and see if containing wchars */
464 script_encoding = zend_multibyte_detect_unicode();
465 if (script_encoding != NULL) {
466 /* bom or wchar detection is prior to 'script_encoding' option */
467 return script_encoding;
468 }
469 }
470
471 /* if no script_encoding specified, just leave alone */
472 if (!CG(script_encoding_list) || !CG(script_encoding_list_size)) {
473 return NULL;
474 }
475
476 /* if multiple encodings specified, detect automagically */
477 if (CG(script_encoding_list_size) > 1) {
478 return zend_multibyte_encoding_detector(LANG_SCNG(script_org), LANG_SCNG(script_org_size), CG(script_encoding_list), CG(script_encoding_list_size));
479 }
480
481 return CG(script_encoding_list)[0];
482 }
483
zend_multibyte_set_filter(const zend_encoding * onetime_encoding)484 ZEND_API zend_result zend_multibyte_set_filter(const zend_encoding *onetime_encoding)
485 {
486 const zend_encoding *internal_encoding = zend_multibyte_get_internal_encoding();
487 const zend_encoding *script_encoding = onetime_encoding ? onetime_encoding: zend_multibyte_find_script_encoding();
488
489 if (!script_encoding) {
490 return FAILURE;
491 }
492
493 /* judge input/output filter */
494 LANG_SCNG(script_encoding) = script_encoding;
495 LANG_SCNG(input_filter) = NULL;
496 LANG_SCNG(output_filter) = NULL;
497
498 if (!internal_encoding || LANG_SCNG(script_encoding) == internal_encoding) {
499 if (!zend_multibyte_check_lexer_compatibility(LANG_SCNG(script_encoding))) {
500 /* and if not, work around w/ script_encoding -> utf-8 -> script_encoding conversion */
501 LANG_SCNG(input_filter) = encoding_filter_script_to_intermediate;
502 LANG_SCNG(output_filter) = encoding_filter_intermediate_to_script;
503 } else {
504 LANG_SCNG(input_filter) = NULL;
505 LANG_SCNG(output_filter) = NULL;
506 }
507 return SUCCESS;
508 }
509
510 if (zend_multibyte_check_lexer_compatibility(internal_encoding)) {
511 LANG_SCNG(input_filter) = encoding_filter_script_to_internal;
512 LANG_SCNG(output_filter) = NULL;
513 } else if (zend_multibyte_check_lexer_compatibility(LANG_SCNG(script_encoding))) {
514 LANG_SCNG(input_filter) = NULL;
515 LANG_SCNG(output_filter) = encoding_filter_script_to_internal;
516 } else {
517 /* both script and internal encodings are incompatible w/ flex */
518 LANG_SCNG(input_filter) = encoding_filter_script_to_intermediate;
519 LANG_SCNG(output_filter) = encoding_filter_intermediate_to_internal;
520 }
521
522 return SUCCESS;
523 }
524
open_file_for_scanning(zend_file_handle * file_handle)525 ZEND_API zend_result open_file_for_scanning(zend_file_handle *file_handle)
526 {
527 char *buf;
528 size_t size;
529 zend_string *compiled_filename;
530
531 if (zend_stream_fixup(file_handle, &buf, &size) == FAILURE) {
532 /* Still add it to open_files to make destroy_file_handle work */
533 zend_llist_add_element(&CG(open_files), file_handle);
534 file_handle->in_list = 1;
535 return FAILURE;
536 }
537
538 ZEND_ASSERT(!EG(exception) && "stream_fixup() should have failed");
539 zend_llist_add_element(&CG(open_files), file_handle);
540 file_handle->in_list = 1;
541
542 /* Reset the scanner for scanning the new file */
543 SCNG(yy_in) = file_handle;
544 SCNG(yy_start) = NULL;
545
546 if (size != (size_t)-1) {
547 if (CG(multibyte)) {
548 SCNG(script_org) = (unsigned char*)buf;
549 SCNG(script_org_size) = size;
550 SCNG(script_filtered) = NULL;
551
552 zend_multibyte_set_filter(NULL);
553
554 if (SCNG(input_filter)) {
555 if ((size_t)-1 == SCNG(input_filter)(&SCNG(script_filtered), &SCNG(script_filtered_size), SCNG(script_org), SCNG(script_org_size))) {
556 zend_error_noreturn(E_COMPILE_ERROR, "Could not convert the script from the detected "
557 "encoding \"%s\" to a compatible encoding", zend_multibyte_get_encoding_name(LANG_SCNG(script_encoding)));
558 }
559 buf = (char*)SCNG(script_filtered);
560 size = SCNG(script_filtered_size);
561 }
562 }
563 SCNG(yy_start) = (unsigned char *)buf;
564 yy_scan_buffer(buf, size);
565 } else {
566 zend_error_noreturn(E_COMPILE_ERROR, "zend_stream_mmap() failed");
567 }
568
569 if (CG(skip_shebang)) {
570 BEGIN(SHEBANG);
571 } else {
572 BEGIN(INITIAL);
573 }
574
575 if (file_handle->opened_path) {
576 compiled_filename = zend_string_copy(file_handle->opened_path);
577 } else {
578 compiled_filename = zend_string_copy(file_handle->filename);
579 }
580
581 zend_set_compiled_filename(compiled_filename);
582 zend_string_release_ex(compiled_filename, 0);
583
584 RESET_DOC_COMMENT();
585 CG(zend_lineno) = 1;
586 CG(increment_lineno) = 0;
587 return SUCCESS;
588 }
589
zend_compile(int type)590 static zend_op_array *zend_compile(int type)
591 {
592 zend_op_array *op_array = NULL;
593 bool original_in_compilation = CG(in_compilation);
594
595 CG(in_compilation) = 1;
596 CG(ast) = NULL;
597 CG(ast_arena) = zend_arena_create(1024 * 32);
598
599 if (!zendparse()) {
600 int last_lineno = CG(zend_lineno);
601 zend_file_context original_file_context;
602 zend_oparray_context original_oparray_context;
603 zend_op_array *original_active_op_array = CG(active_op_array);
604
605 op_array = emalloc(sizeof(zend_op_array));
606 init_op_array(op_array, type, INITIAL_OP_ARRAY_SIZE);
607 CG(active_op_array) = op_array;
608
609 /* Use heap to not waste arena memory */
610 op_array->fn_flags |= ZEND_ACC_HEAP_RT_CACHE;
611
612 if (zend_ast_process) {
613 zend_ast_process(CG(ast));
614 }
615
616 zend_file_context_begin(&original_file_context);
617 zend_oparray_context_begin(&original_oparray_context);
618 zend_compile_top_stmt(CG(ast));
619 CG(zend_lineno) = last_lineno;
620 zend_emit_final_return(type == ZEND_USER_FUNCTION);
621 op_array->line_start = 1;
622 op_array->line_end = last_lineno;
623 zend_init_static_variables_map_ptr(op_array);
624 pass_two(op_array);
625 zend_oparray_context_end(&original_oparray_context);
626 zend_file_context_end(&original_file_context);
627
628 CG(active_op_array) = original_active_op_array;
629 }
630
631 zend_ast_destroy(CG(ast));
632 zend_arena_destroy(CG(ast_arena));
633
634 CG(in_compilation) = original_in_compilation;
635
636 return op_array;
637 }
638
compile_file(zend_file_handle * file_handle,int type)639 ZEND_API zend_op_array *compile_file(zend_file_handle *file_handle, int type)
640 {
641 zend_lex_state original_lex_state;
642 zend_op_array *op_array = NULL;
643 zend_save_lexical_state(&original_lex_state);
644
645 if (open_file_for_scanning(file_handle)==FAILURE) {
646 if (!EG(exception)) {
647 if (type==ZEND_REQUIRE) {
648 zend_message_dispatcher(ZMSG_FAILED_REQUIRE_FOPEN, ZSTR_VAL(file_handle->filename));
649 } else {
650 zend_message_dispatcher(ZMSG_FAILED_INCLUDE_FOPEN, ZSTR_VAL(file_handle->filename));
651 }
652 }
653 } else {
654 op_array = zend_compile(ZEND_USER_FUNCTION);
655 }
656
657 zend_restore_lexical_state(&original_lex_state);
658 return op_array;
659 }
660
zend_compile_string_to_ast(zend_string * code,zend_arena ** ast_arena,zend_string * filename)661 ZEND_API zend_ast *zend_compile_string_to_ast(
662 zend_string *code, zend_arena **ast_arena, zend_string *filename) {
663 zval code_zv;
664 bool original_in_compilation;
665 zend_lex_state original_lex_state;
666 zend_ast *ast;
667
668 ZVAL_STR_COPY(&code_zv, code);
669
670 original_in_compilation = CG(in_compilation);
671 CG(in_compilation) = 1;
672
673 zend_save_lexical_state(&original_lex_state);
674 zend_prepare_string_for_scanning(&code_zv, filename);
675 CG(ast) = NULL;
676 CG(ast_arena) = zend_arena_create(1024 * 32);
677 LANG_SCNG(yy_state) = yycINITIAL;
678
679 if (zendparse() != 0) {
680 zend_ast_destroy(CG(ast));
681 zend_arena_destroy(CG(ast_arena));
682 CG(ast) = NULL;
683 }
684
685 /* restore_lexical_state changes CG(ast) and CG(ast_arena) */
686 ast = CG(ast);
687 *ast_arena = CG(ast_arena);
688
689 zend_restore_lexical_state(&original_lex_state);
690 CG(in_compilation) = original_in_compilation;
691
692 zval_ptr_dtor_str(&code_zv);
693
694 return ast;
695 }
696
compile_filename(int type,zend_string * filename)697 zend_op_array *compile_filename(int type, zend_string *filename)
698 {
699 zend_file_handle file_handle;
700 zend_op_array *retval;
701 zend_string *opened_path = NULL;
702
703 zend_stream_init_filename_ex(&file_handle, filename);
704
705 retval = zend_compile_file(&file_handle, type);
706 if (retval && file_handle.handle.stream.handle) {
707 if (!file_handle.opened_path) {
708 file_handle.opened_path = opened_path = zend_string_copy(filename);
709 }
710
711 zend_hash_add_empty_element(&EG(included_files), file_handle.opened_path);
712
713 if (opened_path) {
714 zend_string_release_ex(opened_path, 0);
715 }
716 }
717 zend_destroy_file_handle(&file_handle);
718
719 return retval;
720 }
721
zend_prepare_string_for_scanning(zval * str,zend_string * filename)722 ZEND_API void zend_prepare_string_for_scanning(zval *str, zend_string *filename)
723 {
724 char *buf;
725 size_t size, old_len;
726
727 /* enforce ZEND_MMAP_AHEAD trailing NULLs for flex... */
728 old_len = Z_STRLEN_P(str);
729 Z_STR_P(str) = zend_string_extend(Z_STR_P(str), old_len + ZEND_MMAP_AHEAD, 0);
730 Z_TYPE_INFO_P(str) = IS_STRING_EX;
731 memset(Z_STRVAL_P(str) + old_len, 0, ZEND_MMAP_AHEAD + 1);
732
733 SCNG(yy_in) = NULL;
734 SCNG(yy_start) = NULL;
735
736 buf = Z_STRVAL_P(str);
737 size = old_len;
738
739 if (CG(multibyte)) {
740 SCNG(script_org) = (unsigned char*)buf;
741 SCNG(script_org_size) = size;
742 SCNG(script_filtered) = NULL;
743
744 zend_multibyte_set_filter(zend_multibyte_get_internal_encoding());
745
746 if (SCNG(input_filter)) {
747 if ((size_t)-1 == SCNG(input_filter)(&SCNG(script_filtered), &SCNG(script_filtered_size), SCNG(script_org), SCNG(script_org_size))) {
748 zend_error_noreturn(E_COMPILE_ERROR, "Could not convert the script from the detected "
749 "encoding \"%s\" to a compatible encoding", zend_multibyte_get_encoding_name(LANG_SCNG(script_encoding)));
750 }
751 buf = (char*)SCNG(script_filtered);
752 size = SCNG(script_filtered_size);
753 }
754 }
755
756 yy_scan_buffer(buf, size);
757 zend_set_compiled_filename(filename);
758 CG(zend_lineno) = 1;
759 CG(increment_lineno) = 0;
760 RESET_DOC_COMMENT();
761 }
762
763
zend_get_scanned_file_offset(void)764 ZEND_API size_t zend_get_scanned_file_offset(void)
765 {
766 size_t offset = SCNG(yy_cursor) - SCNG(yy_start);
767 if (SCNG(input_filter)) {
768 size_t original_offset = offset, length = 0;
769 do {
770 unsigned char *p = NULL;
771 if ((size_t)-1 == SCNG(input_filter)(&p, &length, SCNG(script_org), offset)) {
772 return (size_t)-1;
773 }
774 efree(p);
775 if (length > original_offset) {
776 offset--;
777 } else if (length < original_offset) {
778 offset++;
779 }
780 } while (original_offset != length);
781 }
782 return offset;
783 }
784
compile_string(zend_string * source_string,const char * filename)785 zend_op_array *compile_string(zend_string *source_string, const char *filename)
786 {
787 zend_lex_state original_lex_state;
788 zend_op_array *op_array = NULL;
789 zval tmp;
790 zend_string *filename_str;
791
792 if (ZSTR_LEN(source_string) == 0) {
793 return NULL;
794 }
795
796 ZVAL_STR_COPY(&tmp, source_string);
797
798 zend_save_lexical_state(&original_lex_state);
799 filename_str = zend_string_init(filename, strlen(filename), 0);
800 zend_prepare_string_for_scanning(&tmp, filename_str);
801 zend_string_release(filename_str);
802 BEGIN(ST_IN_SCRIPTING);
803 op_array = zend_compile(ZEND_EVAL_CODE);
804
805 zend_restore_lexical_state(&original_lex_state);
806 zval_ptr_dtor(&tmp);
807
808 return op_array;
809 }
810
811
highlight_file(const char * filename,zend_syntax_highlighter_ini * syntax_highlighter_ini)812 zend_result highlight_file(const char *filename, zend_syntax_highlighter_ini *syntax_highlighter_ini)
813 {
814 zend_lex_state original_lex_state;
815 zend_file_handle file_handle;
816
817 zend_stream_init_filename(&file_handle, filename);
818 zend_save_lexical_state(&original_lex_state);
819 if (open_file_for_scanning(&file_handle)==FAILURE) {
820 zend_message_dispatcher(ZMSG_FAILED_HIGHLIGHT_FOPEN, filename);
821 zend_destroy_file_handle(&file_handle);
822 zend_restore_lexical_state(&original_lex_state);
823 return FAILURE;
824 }
825 zend_highlight(syntax_highlighter_ini);
826 if (SCNG(script_filtered)) {
827 efree(SCNG(script_filtered));
828 SCNG(script_filtered) = NULL;
829 }
830 zend_destroy_file_handle(&file_handle);
831 zend_restore_lexical_state(&original_lex_state);
832 return SUCCESS;
833 }
834
highlight_string(zend_string * str,zend_syntax_highlighter_ini * syntax_highlighter_ini,const char * filename)835 void highlight_string(zend_string *str, zend_syntax_highlighter_ini *syntax_highlighter_ini, const char *filename)
836 {
837 zend_lex_state original_lex_state;
838 zval str_zv;
839 zend_string *filename_str = zend_string_init(filename, strlen(filename), 0);
840 ZVAL_STR_COPY(&str_zv, str);
841 zend_save_lexical_state(&original_lex_state);
842 zend_prepare_string_for_scanning(&str_zv, filename_str);
843 zend_string_release(filename_str);
844 BEGIN(INITIAL);
845 zend_highlight(syntax_highlighter_ini);
846 if (SCNG(script_filtered)) {
847 efree(SCNG(script_filtered));
848 SCNG(script_filtered) = NULL;
849 }
850 zend_restore_lexical_state(&original_lex_state);
851 zval_ptr_dtor(&str_zv);
852 }
853
zend_multibyte_yyinput_again(zend_encoding_filter old_input_filter,const zend_encoding * old_encoding)854 ZEND_API void zend_multibyte_yyinput_again(zend_encoding_filter old_input_filter, const zend_encoding *old_encoding)
855 {
856 size_t length;
857 unsigned char *new_yy_start;
858
859 /* convert and set */
860 if (!SCNG(input_filter)) {
861 if (SCNG(script_filtered)) {
862 efree(SCNG(script_filtered));
863 SCNG(script_filtered) = NULL;
864 }
865 SCNG(script_filtered_size) = 0;
866 length = SCNG(script_org_size);
867 new_yy_start = SCNG(script_org);
868 } else {
869 if ((size_t)-1 == SCNG(input_filter)(&new_yy_start, &length, SCNG(script_org), SCNG(script_org_size))) {
870 zend_error_noreturn(E_COMPILE_ERROR, "Could not convert the script from the detected "
871 "encoding \"%s\" to a compatible encoding", zend_multibyte_get_encoding_name(LANG_SCNG(script_encoding)));
872 }
873 if (SCNG(script_filtered)) {
874 efree(SCNG(script_filtered));
875 }
876 SCNG(script_filtered) = new_yy_start;
877 SCNG(script_filtered_size) = length;
878 }
879
880 SCNG(yy_cursor) = new_yy_start + (SCNG(yy_cursor) - SCNG(yy_start));
881 SCNG(yy_marker) = new_yy_start + (SCNG(yy_marker) - SCNG(yy_start));
882 SCNG(yy_text) = new_yy_start + (SCNG(yy_text) - SCNG(yy_start));
883 SCNG(yy_limit) = new_yy_start + length;
884
885 SCNG(yy_start) = new_yy_start;
886 }
887
888
889 // TODO: avoid reallocation ???
890 # define zend_copy_value(zendlval, yytext, yyleng) \
891 if (SCNG(output_filter)) { \
892 size_t sz = 0; \
893 char *s = NULL; \
894 SCNG(output_filter)((unsigned char **)&s, &sz, (unsigned char *)yytext, (size_t)yyleng); \
895 ZVAL_STRINGL(zendlval, s, sz); \
896 efree(s); \
897 } else if (yyleng == 1) { \
898 ZVAL_INTERNED_STR(zendlval, ZSTR_CHAR((zend_uchar)*(yytext))); \
899 } else { \
900 ZVAL_STRINGL(zendlval, yytext, yyleng); \
901 }
902
zend_scan_escape_string(zval * zendlval,char * str,int len,char quote_type)903 static zend_result zend_scan_escape_string(zval *zendlval, char *str, int len, char quote_type)
904 {
905 char *s, *t;
906 char *end;
907
908 if (len <= 1) {
909 if (len < 1) {
910 ZVAL_EMPTY_STRING(zendlval);
911 } else {
912 zend_uchar c = (zend_uchar)*str;
913 if (c == '\n' || c == '\r') {
914 CG(zend_lineno)++;
915 }
916 ZVAL_INTERNED_STR(zendlval, ZSTR_CHAR(c));
917 }
918 goto skip_escape_conversion;
919 }
920
921 ZVAL_STRINGL(zendlval, str, len);
922
923 /* convert escape sequences */
924 s = Z_STRVAL_P(zendlval);
925 end = s+Z_STRLEN_P(zendlval);
926 while (1) {
927 if (UNEXPECTED(*s=='\\')) {
928 break;
929 }
930 if (*s == '\n' || (*s == '\r' && (*(s+1) != '\n'))) {
931 CG(zend_lineno)++;
932 }
933 s++;
934 if (s == end) {
935 goto skip_escape_conversion;
936 }
937 }
938
939 t = s;
940 while (s<end) {
941 if (*s=='\\') {
942 s++;
943 if (s >= end) {
944 *t++ = '\\';
945 break;
946 }
947
948 switch(*s) {
949 case 'n':
950 *t++ = '\n';
951 break;
952 case 'r':
953 *t++ = '\r';
954 break;
955 case 't':
956 *t++ = '\t';
957 break;
958 case 'f':
959 *t++ = '\f';
960 break;
961 case 'v':
962 *t++ = '\v';
963 break;
964 case 'e':
965 #ifdef ZEND_WIN32
966 *t++ = VK_ESCAPE;
967 #else
968 *t++ = '\e';
969 #endif
970 break;
971 case '"':
972 case '`':
973 if (*s != quote_type) {
974 *t++ = '\\';
975 *t++ = *s;
976 break;
977 }
978 ZEND_FALLTHROUGH;
979 case '\\':
980 case '$':
981 *t++ = *s;
982 break;
983 case 'x':
984 case 'X':
985 if (ZEND_IS_HEX(*(s+1))) {
986 char hex_buf[3] = { 0, 0, 0 };
987
988 hex_buf[0] = *(++s);
989 if (ZEND_IS_HEX(*(s+1))) {
990 hex_buf[1] = *(++s);
991 }
992 *t++ = (char) ZEND_STRTOL(hex_buf, NULL, 16);
993 } else {
994 *t++ = '\\';
995 *t++ = *s;
996 }
997 break;
998 /* UTF-8 codepoint escape, format: /\\u\{\x+\}/ */
999 case 'u':
1000 {
1001 /* cache where we started so we can parse after validating */
1002 char *start = s + 1;
1003 size_t len = 0;
1004 bool valid = 1;
1005 unsigned long codepoint;
1006
1007 if (*start != '{') {
1008 /* we silently let this pass to avoid breaking code
1009 * with JSON in string literals (e.g. "\"\u202e\""
1010 */
1011 *t++ = '\\';
1012 *t++ = 'u';
1013 break;
1014 } else {
1015 /* on the other hand, invalid \u{blah} errors */
1016 s++;
1017 len++;
1018 s++;
1019 while (*s != '}') {
1020 if (!ZEND_IS_HEX(*s)) {
1021 valid = 0;
1022 break;
1023 } else {
1024 len++;
1025 }
1026 s++;
1027 }
1028 if (*s == '}') {
1029 valid = 1;
1030 len++;
1031 }
1032 }
1033
1034 /* \u{} is invalid */
1035 if (len <= 2) {
1036 valid = 0;
1037 }
1038
1039 if (!valid) {
1040 zend_throw_exception(zend_ce_parse_error,
1041 "Invalid UTF-8 codepoint escape sequence", 0);
1042 zval_ptr_dtor(zendlval);
1043 ZVAL_UNDEF(zendlval);
1044 return FAILURE;
1045 }
1046
1047 errno = 0;
1048 codepoint = strtoul(start + 1, NULL, 16);
1049
1050 /* per RFC 3629, UTF-8 can only represent 21 bits */
1051 if (codepoint > 0x10FFFF || errno) {
1052 zend_throw_exception(zend_ce_parse_error,
1053 "Invalid UTF-8 codepoint escape sequence: Codepoint too large", 0);
1054 zval_ptr_dtor(zendlval);
1055 ZVAL_UNDEF(zendlval);
1056 return FAILURE;
1057 }
1058
1059 /* based on https://en.wikipedia.org/wiki/UTF-8#Sample_code */
1060 if (codepoint < 0x80) {
1061 *t++ = codepoint;
1062 } else if (codepoint <= 0x7FF) {
1063 *t++ = (codepoint >> 6) + 0xC0;
1064 *t++ = (codepoint & 0x3F) + 0x80;
1065 } else if (codepoint <= 0xFFFF) {
1066 *t++ = (codepoint >> 12) + 0xE0;
1067 *t++ = ((codepoint >> 6) & 0x3F) + 0x80;
1068 *t++ = (codepoint & 0x3F) + 0x80;
1069 } else if (codepoint <= 0x10FFFF) {
1070 *t++ = (codepoint >> 18) + 0xF0;
1071 *t++ = ((codepoint >> 12) & 0x3F) + 0x80;
1072 *t++ = ((codepoint >> 6) & 0x3F) + 0x80;
1073 *t++ = (codepoint & 0x3F) + 0x80;
1074 }
1075 }
1076 break;
1077 default:
1078 /* check for an octal */
1079 if (ZEND_IS_OCT(*s)) {
1080 char octal_buf[4] = { 0, 0, 0, 0 };
1081
1082 octal_buf[0] = *s;
1083 if (ZEND_IS_OCT(*(s+1))) {
1084 octal_buf[1] = *(++s);
1085 if (ZEND_IS_OCT(*(s+1))) {
1086 octal_buf[2] = *(++s);
1087 }
1088 }
1089 if (octal_buf[2] && (octal_buf[0] > '3') && !SCNG(heredoc_scan_ahead)) {
1090 /* 3 octit values must not overflow 0xFF (\377) */
1091 zend_error(E_COMPILE_WARNING, "Octal escape sequence overflow \\%s is greater than \\377", octal_buf);
1092 }
1093
1094 *t++ = (char) ZEND_STRTOL(octal_buf, NULL, 8);
1095 } else {
1096 *t++ = '\\';
1097 *t++ = *s;
1098 }
1099 break;
1100 }
1101 } else {
1102 *t++ = *s;
1103 }
1104
1105 if (*s == '\n' || (*s == '\r' && (*(s+1) != '\n'))) {
1106 CG(zend_lineno)++;
1107 }
1108 s++;
1109 }
1110 *t = 0;
1111 Z_STRLEN_P(zendlval) = t - Z_STRVAL_P(zendlval);
1112
1113 skip_escape_conversion:
1114 if (SCNG(output_filter)) {
1115 size_t sz = 0;
1116 unsigned char *str;
1117 // TODO: avoid realocation ???
1118 s = Z_STRVAL_P(zendlval);
1119 SCNG(output_filter)(&str, &sz, (unsigned char *)s, (size_t)Z_STRLEN_P(zendlval));
1120 zval_ptr_dtor(zendlval);
1121 ZVAL_STRINGL(zendlval, (char *) str, sz);
1122 efree(str);
1123 }
1124 return SUCCESS;
1125 }
1126
1127 #define HEREDOC_USING_SPACES 1
1128 #define HEREDOC_USING_TABS 2
1129
next_newline(const char * str,const char * end,size_t * newline_len)1130 static const char *next_newline(const char *str, const char *end, size_t *newline_len) {
1131 for (; str < end; str++) {
1132 if (*str == '\r') {
1133 *newline_len = str + 1 < end && *(str + 1) == '\n' ? 2 : 1;
1134 return str;
1135 } else if (*str == '\n') {
1136 *newline_len = 1;
1137 return str;
1138 }
1139 }
1140 *newline_len = 0;
1141 return NULL;
1142 }
1143
strip_multiline_string_indentation(zval * zendlval,int indentation,bool using_spaces,bool newline_at_start,bool newline_at_end)1144 static bool strip_multiline_string_indentation(
1145 zval *zendlval, int indentation, bool using_spaces,
1146 bool newline_at_start, bool newline_at_end)
1147 {
1148 const char *str = Z_STRVAL_P(zendlval), *end = str + Z_STRLEN_P(zendlval);
1149 char *copy = Z_STRVAL_P(zendlval);
1150
1151 int newline_count = 0;
1152 size_t newline_len;
1153 const char *nl;
1154
1155 if (!newline_at_start) {
1156 nl = next_newline(str, end, &newline_len);
1157 if (!nl) {
1158 return 1;
1159 }
1160
1161 str = nl + newline_len;
1162 copy = (char *) nl + newline_len;
1163 newline_count++;
1164 } else {
1165 nl = str;
1166 }
1167
1168 /* <= intentional */
1169 while (str <= end && nl) {
1170 size_t skip;
1171 nl = next_newline(str, end, &newline_len);
1172 if (!nl && newline_at_end) {
1173 nl = end;
1174 }
1175
1176 /* Try to skip indentation */
1177 for (skip = 0; skip < indentation; skip++, str++) {
1178 if (str == nl) {
1179 /* Don't require full indentation on whitespace-only lines */
1180 break;
1181 }
1182
1183 if (str == end || (*str != ' ' && *str != '\t')) {
1184 CG(zend_lineno) += newline_count;
1185 zend_throw_exception_ex(zend_ce_parse_error, 0,
1186 "Invalid body indentation level (expecting an indentation level of at least %d)", indentation);
1187 goto error;
1188 }
1189
1190 if ((!using_spaces && *str == ' ') || (using_spaces && *str == '\t')) {
1191 CG(zend_lineno) += newline_count;
1192 zend_throw_exception(zend_ce_parse_error,
1193 "Invalid indentation - tabs and spaces cannot be mixed", 0);
1194 goto error;
1195 }
1196 }
1197
1198 if (str == end) {
1199 break;
1200 }
1201
1202 size_t len = nl ? (nl - str + newline_len) : (end - str);
1203 memmove(copy, str, len);
1204 str += len;
1205 copy += len;
1206 newline_count++;
1207 }
1208
1209 *copy = '\0';
1210 Z_STRLEN_P(zendlval) = copy - Z_STRVAL_P(zendlval);
1211 return 1;
1212
1213 error:
1214 zval_ptr_dtor_str(zendlval);
1215 ZVAL_UNDEF(zendlval);
1216
1217 return 0;
1218 }
1219
copy_heredoc_label_stack(void * void_heredoc_label)1220 static void copy_heredoc_label_stack(void *void_heredoc_label)
1221 {
1222 zend_heredoc_label *heredoc_label = void_heredoc_label;
1223 zend_heredoc_label *new_heredoc_label = emalloc(sizeof(zend_heredoc_label));
1224
1225 *new_heredoc_label = *heredoc_label;
1226 new_heredoc_label->label = estrndup(heredoc_label->label, heredoc_label->length);
1227
1228 zend_ptr_stack_push(&SCNG(heredoc_label_stack), (void *) new_heredoc_label);
1229 }
1230
1231 /* Check that { }, [ ], ( ) are nested correctly */
report_bad_nesting(char opening,int opening_lineno,char closing)1232 static void report_bad_nesting(char opening, int opening_lineno, char closing)
1233 {
1234 char buf[256];
1235 size_t used = 0;
1236
1237 used = snprintf(buf, sizeof(buf), "Unclosed '%c'", opening);
1238
1239 if (opening_lineno != CG(zend_lineno)) {
1240 used += snprintf(buf + used, sizeof(buf) - used, " on line %d", opening_lineno);
1241 }
1242
1243 if (closing) { /* 'closing' will be 0 if at end of file */
1244 used += snprintf(buf + used, sizeof(buf) - used, " does not match '%c'", closing);
1245 }
1246
1247 zend_throw_exception(zend_ce_parse_error, buf, 0);
1248 }
1249
enter_nesting(char opening)1250 static void enter_nesting(char opening)
1251 {
1252 zend_nest_location nest_loc = {opening, CG(zend_lineno)};
1253 zend_stack_push(&SCNG(nest_location_stack), &nest_loc);
1254 }
1255
exit_nesting(char closing)1256 static zend_result exit_nesting(char closing)
1257 {
1258 if (zend_stack_is_empty(&SCNG(nest_location_stack))) {
1259 zend_throw_exception_ex(zend_ce_parse_error, 0, "Unmatched '%c'", closing);
1260 return FAILURE;
1261 }
1262
1263 zend_nest_location *nest_loc = zend_stack_top(&SCNG(nest_location_stack));
1264 char opening = nest_loc->text;
1265
1266 if ((opening == '{' && closing != '}') ||
1267 (opening == '[' && closing != ']') ||
1268 (opening == '(' && closing != ')')) {
1269 report_bad_nesting(opening, nest_loc->lineno, closing);
1270 return FAILURE;
1271 }
1272
1273 zend_stack_del_top(&SCNG(nest_location_stack));
1274 return SUCCESS;
1275 }
1276
check_nesting_at_end(void)1277 static zend_result check_nesting_at_end(void)
1278 {
1279 if (!zend_stack_is_empty(&SCNG(nest_location_stack))) {
1280 zend_nest_location *nest_loc = zend_stack_top(&SCNG(nest_location_stack));
1281 report_bad_nesting(nest_loc->text, nest_loc->lineno, 0);
1282 return FAILURE;
1283 }
1284
1285 return SUCCESS;
1286 }
1287
1288 #define PARSER_MODE() \
1289 EXPECTED(elem != NULL)
1290
1291 #define RETURN_TOKEN(_token) do { \
1292 token = _token; \
1293 goto emit_token; \
1294 } while (0)
1295
1296 #define RETURN_TOKEN_WITH_VAL(_token) do { \
1297 token = _token; \
1298 goto emit_token_with_val; \
1299 } while (0)
1300
1301 #define RETURN_TOKEN_WITH_STR(_token, _offset) do { \
1302 token = _token; \
1303 offset = _offset; \
1304 goto emit_token_with_str; \
1305 } while (0)
1306
1307 #define RETURN_TOKEN_WITH_IDENT(_token) do { \
1308 token = _token; \
1309 goto emit_token_with_ident; \
1310 } while (0)
1311
1312 #define RETURN_OR_SKIP_TOKEN(_token) do { \
1313 token = _token; \
1314 if (PARSER_MODE()) { \
1315 goto skip_token; \
1316 } \
1317 goto emit_token; \
1318 } while (0)
1319
1320 #define RETURN_EXIT_NESTING_TOKEN(_token) do { \
1321 if (exit_nesting(_token) && PARSER_MODE()) { \
1322 RETURN_TOKEN(T_ERROR); \
1323 } else { \
1324 RETURN_TOKEN(_token); \
1325 } \
1326 } while(0)
1327
1328 #define RETURN_END_TOKEN do { \
1329 if (check_nesting_at_end() && PARSER_MODE()) { \
1330 RETURN_TOKEN(T_ERROR); \
1331 } else { \
1332 RETURN_TOKEN(END); \
1333 } \
1334 } while (0)
1335
lex_scan(zval * zendlval,zend_parser_stack_elem * elem)1336 int ZEND_FASTCALL lex_scan(zval *zendlval, zend_parser_stack_elem *elem)
1337 {
1338 int token;
1339 int offset;
1340 int start_line = CG(zend_lineno);
1341
1342 ZVAL_UNDEF(zendlval);
1343 restart:
1344 SCNG(yy_text) = YYCURSOR;
1345
1346 /*!re2c
1347 re2c:yyfill:check = 0;
1348 LNUM [0-9]+(_[0-9]+)*
1349 DNUM ({LNUM}?"."{LNUM})|({LNUM}"."{LNUM}?)
1350 EXPONENT_DNUM (({LNUM}|{DNUM})[eE][+-]?{LNUM})
1351 HNUM "0x"[0-9a-fA-F]+(_[0-9a-fA-F]+)*
1352 BNUM "0b"[01]+(_[01]+)*
1353 ONUM "0o"[0-7]+(_[0-7]+)*
1354 LABEL [a-zA-Z_\x80-\xff][a-zA-Z0-9_\x80-\xff]*
1355 WHITESPACE [ \n\r\t]+
1356 TABS_AND_SPACES [ \t]*
1357 TOKENS [;:,.|^&+-/*=%!~$<>?@]
1358 ANY_CHAR [^]
1359 NEWLINE ("\r"|"\n"|"\r\n")
1360
1361 /* compute yyleng before each rule */
1362 <!*> := yyleng = YYCURSOR - SCNG(yy_text);
1363
1364 <ST_IN_SCRIPTING>"exit" {
1365 RETURN_TOKEN_WITH_IDENT(T_EXIT);
1366 }
1367
1368 <ST_IN_SCRIPTING>"die" {
1369 RETURN_TOKEN_WITH_IDENT(T_EXIT);
1370 }
1371
1372 <ST_IN_SCRIPTING>"fn" {
1373 RETURN_TOKEN_WITH_IDENT(T_FN);
1374 }
1375
1376 <ST_IN_SCRIPTING>"function" {
1377 RETURN_TOKEN_WITH_IDENT(T_FUNCTION);
1378 }
1379
1380 <ST_IN_SCRIPTING>"const" {
1381 RETURN_TOKEN_WITH_IDENT(T_CONST);
1382 }
1383
1384 <ST_IN_SCRIPTING>"return" {
1385 RETURN_TOKEN_WITH_IDENT(T_RETURN);
1386 }
1387
1388 <ST_IN_SCRIPTING>"#[" {
1389 enter_nesting('[');
1390 RETURN_TOKEN(T_ATTRIBUTE);
1391 }
1392
1393 <ST_IN_SCRIPTING>"yield"{WHITESPACE}"from"[^a-zA-Z0-9_\x80-\xff] {
1394 yyless(yyleng - 1);
1395 HANDLE_NEWLINES(yytext, yyleng);
1396 RETURN_TOKEN_WITH_IDENT(T_YIELD_FROM);
1397 }
1398
1399 <ST_IN_SCRIPTING>"yield" {
1400 RETURN_TOKEN_WITH_IDENT(T_YIELD);
1401 }
1402
1403 <ST_IN_SCRIPTING>"try" {
1404 RETURN_TOKEN_WITH_IDENT(T_TRY);
1405 }
1406
1407 <ST_IN_SCRIPTING>"catch" {
1408 RETURN_TOKEN_WITH_IDENT(T_CATCH);
1409 }
1410
1411 <ST_IN_SCRIPTING>"finally" {
1412 RETURN_TOKEN_WITH_IDENT(T_FINALLY);
1413 }
1414
1415 <ST_IN_SCRIPTING>"throw" {
1416 RETURN_TOKEN_WITH_IDENT(T_THROW);
1417 }
1418
1419 <ST_IN_SCRIPTING>"if" {
1420 RETURN_TOKEN_WITH_IDENT(T_IF);
1421 }
1422
1423 <ST_IN_SCRIPTING>"elseif" {
1424 RETURN_TOKEN_WITH_IDENT(T_ELSEIF);
1425 }
1426
1427 <ST_IN_SCRIPTING>"endif" {
1428 RETURN_TOKEN_WITH_IDENT(T_ENDIF);
1429 }
1430
1431 <ST_IN_SCRIPTING>"else" {
1432 RETURN_TOKEN_WITH_IDENT(T_ELSE);
1433 }
1434
1435 <ST_IN_SCRIPTING>"while" {
1436 RETURN_TOKEN_WITH_IDENT(T_WHILE);
1437 }
1438
1439 <ST_IN_SCRIPTING>"endwhile" {
1440 RETURN_TOKEN_WITH_IDENT(T_ENDWHILE);
1441 }
1442
1443 <ST_IN_SCRIPTING>"do" {
1444 RETURN_TOKEN_WITH_IDENT(T_DO);
1445 }
1446
1447 <ST_IN_SCRIPTING>"for" {
1448 RETURN_TOKEN_WITH_IDENT(T_FOR);
1449 }
1450
1451 <ST_IN_SCRIPTING>"endfor" {
1452 RETURN_TOKEN_WITH_IDENT(T_ENDFOR);
1453 }
1454
1455 <ST_IN_SCRIPTING>"foreach" {
1456 RETURN_TOKEN_WITH_IDENT(T_FOREACH);
1457 }
1458
1459 <ST_IN_SCRIPTING>"endforeach" {
1460 RETURN_TOKEN_WITH_IDENT(T_ENDFOREACH);
1461 }
1462
1463 <ST_IN_SCRIPTING>"declare" {
1464 RETURN_TOKEN_WITH_IDENT(T_DECLARE);
1465 }
1466
1467 <ST_IN_SCRIPTING>"enddeclare" {
1468 RETURN_TOKEN_WITH_IDENT(T_ENDDECLARE);
1469 }
1470
1471 <ST_IN_SCRIPTING>"instanceof" {
1472 RETURN_TOKEN_WITH_IDENT(T_INSTANCEOF);
1473 }
1474
1475 <ST_IN_SCRIPTING>"as" {
1476 RETURN_TOKEN_WITH_IDENT(T_AS);
1477 }
1478
1479 <ST_IN_SCRIPTING>"switch" {
1480 RETURN_TOKEN_WITH_IDENT(T_SWITCH);
1481 }
1482
1483 <ST_IN_SCRIPTING>"match" {
1484 RETURN_TOKEN_WITH_IDENT(T_MATCH);
1485 }
1486
1487 <ST_IN_SCRIPTING>"endswitch" {
1488 RETURN_TOKEN_WITH_IDENT(T_ENDSWITCH);
1489 }
1490
1491 <ST_IN_SCRIPTING>"case" {
1492 RETURN_TOKEN_WITH_IDENT(T_CASE);
1493 }
1494
1495 <ST_IN_SCRIPTING>"default" {
1496 RETURN_TOKEN_WITH_IDENT(T_DEFAULT);
1497 }
1498
1499 <ST_IN_SCRIPTING>"break" {
1500 RETURN_TOKEN_WITH_IDENT(T_BREAK);
1501 }
1502
1503 <ST_IN_SCRIPTING>"continue" {
1504 RETURN_TOKEN_WITH_IDENT(T_CONTINUE);
1505 }
1506
1507 <ST_IN_SCRIPTING>"goto" {
1508 RETURN_TOKEN_WITH_IDENT(T_GOTO);
1509 }
1510
1511 <ST_IN_SCRIPTING>"echo" {
1512 RETURN_TOKEN_WITH_IDENT(T_ECHO);
1513 }
1514
1515 <ST_IN_SCRIPTING>"print" {
1516 RETURN_TOKEN_WITH_IDENT(T_PRINT);
1517 }
1518
1519 <ST_IN_SCRIPTING>"class" {
1520 RETURN_TOKEN_WITH_IDENT(T_CLASS);
1521 }
1522
1523 <ST_IN_SCRIPTING>"interface" {
1524 RETURN_TOKEN_WITH_IDENT(T_INTERFACE);
1525 }
1526
1527 <ST_IN_SCRIPTING>"trait" {
1528 RETURN_TOKEN_WITH_IDENT(T_TRAIT);
1529 }
1530
1531 /*
1532 * The enum keyword must be followed by whitespace and another identifier.
1533 * This avoids the BC break of using enum in classes, namespaces, functions and constants.
1534 */
1535 <ST_IN_SCRIPTING>"enum"{WHITESPACE}("extends"|"implements") {
1536 yyless(4);
1537 RETURN_TOKEN_WITH_STR(T_STRING, 0);
1538 }
1539 <ST_IN_SCRIPTING>"enum"{WHITESPACE}[a-zA-Z_\x80-\xff] {
1540 yyless(4);
1541 RETURN_TOKEN_WITH_IDENT(T_ENUM);
1542 }
1543
1544 <ST_IN_SCRIPTING>"extends" {
1545 RETURN_TOKEN_WITH_IDENT(T_EXTENDS);
1546 }
1547
1548 <ST_IN_SCRIPTING>"implements" {
1549 RETURN_TOKEN_WITH_IDENT(T_IMPLEMENTS);
1550 }
1551
1552 <ST_IN_SCRIPTING>"->" {
1553 yy_push_state(ST_LOOKING_FOR_PROPERTY);
1554 RETURN_TOKEN(T_OBJECT_OPERATOR);
1555 }
1556
1557 <ST_IN_SCRIPTING>"?->" {
1558 yy_push_state(ST_LOOKING_FOR_PROPERTY);
1559 RETURN_TOKEN(T_NULLSAFE_OBJECT_OPERATOR);
1560 }
1561
1562 <ST_IN_SCRIPTING,ST_LOOKING_FOR_PROPERTY>{WHITESPACE}+ {
1563 goto return_whitespace;
1564 }
1565
1566 <ST_LOOKING_FOR_PROPERTY>"->" {
1567 RETURN_TOKEN(T_OBJECT_OPERATOR);
1568 }
1569
1570 <ST_LOOKING_FOR_PROPERTY>"?->" {
1571 RETURN_TOKEN(T_NULLSAFE_OBJECT_OPERATOR);
1572 }
1573
1574 <ST_LOOKING_FOR_PROPERTY>{LABEL} {
1575 yy_pop_state();
1576 RETURN_TOKEN_WITH_STR(T_STRING, 0);
1577 }
1578
1579 <ST_LOOKING_FOR_PROPERTY>{ANY_CHAR} {
1580 yyless(0);
1581 yy_pop_state();
1582 goto restart;
1583 }
1584
1585 <ST_IN_SCRIPTING>"::" {
1586 RETURN_TOKEN(T_PAAMAYIM_NEKUDOTAYIM);
1587 }
1588
1589 <ST_IN_SCRIPTING>"..." {
1590 RETURN_TOKEN(T_ELLIPSIS);
1591 }
1592
1593 <ST_IN_SCRIPTING>"??" {
1594 RETURN_TOKEN(T_COALESCE);
1595 }
1596
1597 <ST_IN_SCRIPTING>"new" {
1598 RETURN_TOKEN_WITH_IDENT(T_NEW);
1599 }
1600
1601 <ST_IN_SCRIPTING>"clone" {
1602 RETURN_TOKEN_WITH_IDENT(T_CLONE);
1603 }
1604
1605 <ST_IN_SCRIPTING>"var" {
1606 RETURN_TOKEN_WITH_IDENT(T_VAR);
1607 }
1608
1609 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("int"|"integer"){TABS_AND_SPACES}")" {
1610 RETURN_TOKEN(T_INT_CAST);
1611 }
1612
1613 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("double"|"float"){TABS_AND_SPACES}")" {
1614 RETURN_TOKEN(T_DOUBLE_CAST);
1615 }
1616
1617 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}"real"{TABS_AND_SPACES}")" {
1618 if (PARSER_MODE()) {
1619 zend_throw_exception(zend_ce_parse_error, "The (real) cast has been removed, use (float) instead", 0);
1620 RETURN_TOKEN(T_ERROR);
1621 }
1622 RETURN_TOKEN(T_DOUBLE_CAST);
1623 }
1624
1625 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("string"|"binary"){TABS_AND_SPACES}")" {
1626 RETURN_TOKEN(T_STRING_CAST);
1627 }
1628
1629 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}"array"{TABS_AND_SPACES}")" {
1630 RETURN_TOKEN(T_ARRAY_CAST);
1631 }
1632
1633 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}"object"{TABS_AND_SPACES}")" {
1634 RETURN_TOKEN(T_OBJECT_CAST);
1635 }
1636
1637 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("bool"|"boolean"){TABS_AND_SPACES}")" {
1638 RETURN_TOKEN(T_BOOL_CAST);
1639 }
1640
1641 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("unset"){TABS_AND_SPACES}")" {
1642 RETURN_TOKEN(T_UNSET_CAST);
1643 }
1644
1645 <ST_IN_SCRIPTING>"eval" {
1646 RETURN_TOKEN_WITH_IDENT(T_EVAL);
1647 }
1648
1649 <ST_IN_SCRIPTING>"include" {
1650 RETURN_TOKEN_WITH_IDENT(T_INCLUDE);
1651 }
1652
1653 <ST_IN_SCRIPTING>"include_once" {
1654 RETURN_TOKEN_WITH_IDENT(T_INCLUDE_ONCE);
1655 }
1656
1657 <ST_IN_SCRIPTING>"require" {
1658 RETURN_TOKEN_WITH_IDENT(T_REQUIRE);
1659 }
1660
1661 <ST_IN_SCRIPTING>"require_once" {
1662 RETURN_TOKEN_WITH_IDENT(T_REQUIRE_ONCE);
1663 }
1664
1665 <ST_IN_SCRIPTING>"namespace" {
1666 RETURN_TOKEN_WITH_IDENT(T_NAMESPACE);
1667 }
1668
1669 <ST_IN_SCRIPTING>"use" {
1670 RETURN_TOKEN_WITH_IDENT(T_USE);
1671 }
1672
1673 <ST_IN_SCRIPTING>"insteadof" {
1674 RETURN_TOKEN_WITH_IDENT(T_INSTEADOF);
1675 }
1676
1677 <ST_IN_SCRIPTING>"global" {
1678 RETURN_TOKEN_WITH_IDENT(T_GLOBAL);
1679 }
1680
1681 <ST_IN_SCRIPTING>"isset" {
1682 RETURN_TOKEN_WITH_IDENT(T_ISSET);
1683 }
1684
1685 <ST_IN_SCRIPTING>"empty" {
1686 RETURN_TOKEN_WITH_IDENT(T_EMPTY);
1687 }
1688
1689 <ST_IN_SCRIPTING>"__halt_compiler" {
1690 RETURN_TOKEN_WITH_IDENT(T_HALT_COMPILER);
1691 }
1692
1693 <ST_IN_SCRIPTING>"static" {
1694 RETURN_TOKEN_WITH_IDENT(T_STATIC);
1695 }
1696
1697 <ST_IN_SCRIPTING>"abstract" {
1698 RETURN_TOKEN_WITH_IDENT(T_ABSTRACT);
1699 }
1700
1701 <ST_IN_SCRIPTING>"final" {
1702 RETURN_TOKEN_WITH_IDENT(T_FINAL);
1703 }
1704
1705 <ST_IN_SCRIPTING>"private" {
1706 RETURN_TOKEN_WITH_IDENT(T_PRIVATE);
1707 }
1708
1709 <ST_IN_SCRIPTING>"protected" {
1710 RETURN_TOKEN_WITH_IDENT(T_PROTECTED);
1711 }
1712
1713 <ST_IN_SCRIPTING>"public" {
1714 RETURN_TOKEN_WITH_IDENT(T_PUBLIC);
1715 }
1716
1717 <ST_IN_SCRIPTING>"readonly" {
1718 RETURN_TOKEN_WITH_IDENT(T_READONLY);
1719 }
1720
1721 /* Don't treat "readonly(" as a keyword, to allow using it as a function name. */
1722 <ST_IN_SCRIPTING>"readonly"[ \n\r\t]*"(" {
1723 yyless(strlen("readonly"));
1724 RETURN_TOKEN_WITH_STR(T_STRING, 0);
1725 }
1726
1727 <ST_IN_SCRIPTING>"unset" {
1728 RETURN_TOKEN_WITH_IDENT(T_UNSET);
1729 }
1730
1731 <ST_IN_SCRIPTING>"=>" {
1732 RETURN_TOKEN(T_DOUBLE_ARROW);
1733 }
1734
1735 <ST_IN_SCRIPTING>"list" {
1736 RETURN_TOKEN_WITH_IDENT(T_LIST);
1737 }
1738
1739 <ST_IN_SCRIPTING>"array" {
1740 RETURN_TOKEN_WITH_IDENT(T_ARRAY);
1741 }
1742
1743 <ST_IN_SCRIPTING>"callable" {
1744 RETURN_TOKEN_WITH_IDENT(T_CALLABLE);
1745 }
1746
1747 <ST_IN_SCRIPTING>"++" {
1748 RETURN_TOKEN(T_INC);
1749 }
1750
1751 <ST_IN_SCRIPTING>"--" {
1752 RETURN_TOKEN(T_DEC);
1753 }
1754
1755 <ST_IN_SCRIPTING>"===" {
1756 RETURN_TOKEN(T_IS_IDENTICAL);
1757 }
1758
1759 <ST_IN_SCRIPTING>"!==" {
1760 RETURN_TOKEN(T_IS_NOT_IDENTICAL);
1761 }
1762
1763 <ST_IN_SCRIPTING>"==" {
1764 RETURN_TOKEN(T_IS_EQUAL);
1765 }
1766
1767 <ST_IN_SCRIPTING>"!="|"<>" {
1768 RETURN_TOKEN(T_IS_NOT_EQUAL);
1769 }
1770
1771 <ST_IN_SCRIPTING>"<=>" {
1772 RETURN_TOKEN(T_SPACESHIP);
1773 }
1774
1775 <ST_IN_SCRIPTING>"<=" {
1776 RETURN_TOKEN(T_IS_SMALLER_OR_EQUAL);
1777 }
1778
1779 <ST_IN_SCRIPTING>">=" {
1780 RETURN_TOKEN(T_IS_GREATER_OR_EQUAL);
1781 }
1782
1783 <ST_IN_SCRIPTING>"+=" {
1784 RETURN_TOKEN(T_PLUS_EQUAL);
1785 }
1786
1787 <ST_IN_SCRIPTING>"-=" {
1788 RETURN_TOKEN(T_MINUS_EQUAL);
1789 }
1790
1791 <ST_IN_SCRIPTING>"*=" {
1792 RETURN_TOKEN(T_MUL_EQUAL);
1793 }
1794
1795 <ST_IN_SCRIPTING>"*\*" {
1796 RETURN_TOKEN(T_POW);
1797 }
1798
1799 <ST_IN_SCRIPTING>"*\*=" {
1800 RETURN_TOKEN(T_POW_EQUAL);
1801 }
1802
1803 <ST_IN_SCRIPTING>"/=" {
1804 RETURN_TOKEN(T_DIV_EQUAL);
1805 }
1806
1807 <ST_IN_SCRIPTING>".=" {
1808 RETURN_TOKEN(T_CONCAT_EQUAL);
1809 }
1810
1811 <ST_IN_SCRIPTING>"%=" {
1812 RETURN_TOKEN(T_MOD_EQUAL);
1813 }
1814
1815 <ST_IN_SCRIPTING>"<<=" {
1816 RETURN_TOKEN(T_SL_EQUAL);
1817 }
1818
1819 <ST_IN_SCRIPTING>">>=" {
1820 RETURN_TOKEN(T_SR_EQUAL);
1821 }
1822
1823 <ST_IN_SCRIPTING>"&=" {
1824 RETURN_TOKEN(T_AND_EQUAL);
1825 }
1826
1827 <ST_IN_SCRIPTING>"|=" {
1828 RETURN_TOKEN(T_OR_EQUAL);
1829 }
1830
1831 <ST_IN_SCRIPTING>"^=" {
1832 RETURN_TOKEN(T_XOR_EQUAL);
1833 }
1834
1835 <ST_IN_SCRIPTING>"??=" {
1836 RETURN_TOKEN(T_COALESCE_EQUAL);
1837 }
1838
1839 <ST_IN_SCRIPTING>"||" {
1840 RETURN_TOKEN(T_BOOLEAN_OR);
1841 }
1842
1843 <ST_IN_SCRIPTING>"&&" {
1844 RETURN_TOKEN(T_BOOLEAN_AND);
1845 }
1846
1847 <ST_IN_SCRIPTING>"OR" {
1848 RETURN_TOKEN_WITH_IDENT(T_LOGICAL_OR);
1849 }
1850
1851 <ST_IN_SCRIPTING>"AND" {
1852 RETURN_TOKEN_WITH_IDENT(T_LOGICAL_AND);
1853 }
1854
1855 <ST_IN_SCRIPTING>"XOR" {
1856 RETURN_TOKEN_WITH_IDENT(T_LOGICAL_XOR);
1857 }
1858
1859 <ST_IN_SCRIPTING>"<<" {
1860 RETURN_TOKEN(T_SL);
1861 }
1862
1863 <ST_IN_SCRIPTING>">>" {
1864 RETURN_TOKEN(T_SR);
1865 }
1866
1867 <ST_IN_SCRIPTING>"&"[ \t\r\n]*("$"|"...") {
1868 yyless(1);
1869 RETURN_TOKEN(T_AMPERSAND_FOLLOWED_BY_VAR_OR_VARARG);
1870 }
1871
1872 <ST_IN_SCRIPTING>"&" {
1873 RETURN_TOKEN(T_AMPERSAND_NOT_FOLLOWED_BY_VAR_OR_VARARG);
1874 }
1875
1876 <ST_IN_SCRIPTING>"]"|")" {
1877 /* Check that ] and ) match up properly with a preceding [ or ( */
1878 RETURN_EXIT_NESTING_TOKEN(yytext[0]);
1879 }
1880
1881 <ST_IN_SCRIPTING>"["|"(" {
1882 enter_nesting(yytext[0]);
1883 RETURN_TOKEN(yytext[0]);
1884 }
1885
1886 <ST_IN_SCRIPTING>{TOKENS} {
1887 RETURN_TOKEN(yytext[0]);
1888 }
1889
1890
1891 <ST_IN_SCRIPTING>"{" {
1892 yy_push_state(ST_IN_SCRIPTING);
1893 enter_nesting('{');
1894 RETURN_TOKEN('{');
1895 }
1896
1897
1898 <ST_DOUBLE_QUOTES,ST_BACKQUOTE,ST_HEREDOC>"${" {
1899 yy_push_state(ST_LOOKING_FOR_VARNAME);
1900 enter_nesting('{');
1901 RETURN_TOKEN(T_DOLLAR_OPEN_CURLY_BRACES);
1902 }
1903
1904 <ST_IN_SCRIPTING>"}" {
1905 RESET_DOC_COMMENT();
1906 if (!zend_stack_is_empty(&SCNG(state_stack))) {
1907 yy_pop_state();
1908 }
1909 RETURN_EXIT_NESTING_TOKEN('}');
1910 }
1911
1912
1913 <ST_LOOKING_FOR_VARNAME>{LABEL}[[}] {
1914 yyless(yyleng - 1);
1915 yy_pop_state();
1916 yy_push_state(ST_IN_SCRIPTING);
1917 RETURN_TOKEN_WITH_STR(T_STRING_VARNAME, 0);
1918 }
1919
1920
1921 <ST_LOOKING_FOR_VARNAME>{ANY_CHAR} {
1922 yyless(0);
1923 yy_pop_state();
1924 yy_push_state(ST_IN_SCRIPTING);
1925 goto restart;
1926 }
1927
1928 <ST_IN_SCRIPTING>{BNUM} {
1929 /* The +/- 2 skips "0b" */
1930 size_t len = yyleng - 2;
1931 char *end, *bin = yytext + 2;
1932 bool contains_underscores;
1933
1934 /* Skip any leading 0s */
1935 while (len > 0 && (*bin == '0' || *bin == '_')) {
1936 ++bin;
1937 --len;
1938 }
1939
1940 contains_underscores = (memchr(bin, '_', len) != NULL);
1941
1942 if (contains_underscores) {
1943 bin = estrndup(bin, len);
1944 strip_underscores(bin, &len);
1945 }
1946
1947 if (len < SIZEOF_ZEND_LONG * 8) {
1948 if (len == 0) {
1949 ZVAL_LONG(zendlval, 0);
1950 } else {
1951 errno = 0;
1952 ZVAL_LONG(zendlval, ZEND_STRTOL(bin, &end, 2));
1953 ZEND_ASSERT(!errno && end == bin + len);
1954 }
1955 if (contains_underscores) {
1956 efree(bin);
1957 }
1958 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
1959 } else {
1960 ZVAL_DOUBLE(zendlval, zend_bin_strtod(bin, (const char **)&end));
1961 /* errno isn't checked since we allow HUGE_VAL/INF overflow */
1962 ZEND_ASSERT(end == bin + len);
1963 if (contains_underscores) {
1964 efree(bin);
1965 }
1966 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
1967 }
1968 }
1969
1970 <ST_IN_SCRIPTING>{ONUM} {
1971 /* The +/- 2 skips "0o" */
1972 size_t len = yyleng - 2;
1973 char *end, *octal = yytext + 2;
1974 bool contains_underscores = (memchr(octal, '_', len) != NULL);
1975
1976 /* Skip any leading 0s */
1977 while (len > 0 && (*octal == '0' || *octal == '_')) {
1978 ++octal;
1979 --len;
1980 }
1981
1982 if (len == 0) {
1983 ZVAL_LONG(zendlval, 0);
1984 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
1985 }
1986
1987 if (contains_underscores) {
1988 octal = estrndup(octal, len);
1989 strip_underscores(octal, &len);
1990 }
1991
1992 errno = 0;
1993
1994 ZVAL_LONG(zendlval, ZEND_STRTOL(octal, &end, 8));
1995
1996 ZEND_ASSERT(end == octal + len);
1997
1998 if (!errno) {
1999 if (contains_underscores) {
2000 efree(octal);
2001 }
2002 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
2003 }
2004
2005 /* Overflow */
2006 ZEND_ASSERT(errno == ERANGE);
2007 /* Reset errno */
2008 errno = 0;
2009
2010 /* zend_oct_strtod skips leading '0' */
2011 ZVAL_DOUBLE(zendlval, zend_oct_strtod(octal, (const char **)&end));
2012 ZEND_ASSERT(!errno);
2013 ZEND_ASSERT(end == octal + len);
2014 if (contains_underscores) {
2015 efree(octal);
2016 }
2017 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
2018 }
2019
2020 <ST_IN_SCRIPTING>{LNUM} {
2021 size_t len = yyleng;
2022 char *end, *lnum = yytext;
2023 bool is_octal = lnum[0] == '0';
2024 bool contains_underscores = (memchr(lnum, '_', len) != NULL);
2025
2026 if (contains_underscores) {
2027 lnum = estrndup(lnum, len);
2028 strip_underscores(lnum, &len);
2029 }
2030
2031 /* Digits 8 and 9 are illegal in octal literals. */
2032 if (is_octal) {
2033 size_t i;
2034 for (i = 0; i < len; i++) {
2035 if (lnum[i] == '8' || lnum[i] == '9') {
2036 zend_throw_exception(zend_ce_parse_error, "Invalid numeric literal", 0);
2037 if (PARSER_MODE()) {
2038 if (contains_underscores) {
2039 efree(lnum);
2040 }
2041 ZVAL_UNDEF(zendlval);
2042 RETURN_TOKEN(T_ERROR);
2043 }
2044
2045 /* Continue in order to determine if this is T_LNUMBER or T_DNUMBER. */
2046 len = i;
2047 break;
2048 }
2049 }
2050 }
2051
2052
2053 if (len < MAX_LENGTH_OF_LONG - 1) { /* Won't overflow */
2054 errno = 0;
2055 /* base must be passed explicitly for correct parse error on Windows */
2056 ZVAL_LONG(zendlval, ZEND_STRTOL(lnum, &end, is_octal ? 8 : 10));
2057 ZEND_ASSERT(end == lnum + len);
2058 } else {
2059 errno = 0;
2060 ZVAL_LONG(zendlval, ZEND_STRTOL(lnum, &end, 0));
2061 if (errno == ERANGE) { /* Overflow */
2062 errno = 0;
2063 if (is_octal) { /* octal overflow */
2064 ZVAL_DOUBLE(zendlval, zend_oct_strtod(lnum, (const char **)&end));
2065 } else {
2066 ZVAL_DOUBLE(zendlval, zend_strtod(lnum, (const char **)&end));
2067 }
2068 ZEND_ASSERT(end == lnum + len);
2069 if (contains_underscores) {
2070 efree(lnum);
2071 }
2072 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
2073 }
2074 ZEND_ASSERT(end == lnum + len);
2075 }
2076 ZEND_ASSERT(!errno);
2077 if (contains_underscores) {
2078 efree(lnum);
2079 }
2080 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
2081 }
2082
2083 <ST_IN_SCRIPTING>{HNUM} {
2084 /* The +/- 2 skips "0x" */
2085 size_t len = yyleng - 2;
2086 char *end, *hex = yytext + 2;
2087 bool contains_underscores;
2088
2089 /* Skip any leading 0s */
2090 while (len > 0 && (*hex == '0' || *hex == '_')) {
2091 ++hex;
2092 --len;
2093 }
2094
2095 contains_underscores = (memchr(hex, '_', len) != NULL);
2096
2097 if (contains_underscores) {
2098 hex = estrndup(hex, len);
2099 strip_underscores(hex, &len);
2100 }
2101
2102 if (len < SIZEOF_ZEND_LONG * 2 || (len == SIZEOF_ZEND_LONG * 2 && *hex <= '7')) {
2103 if (len == 0) {
2104 ZVAL_LONG(zendlval, 0);
2105 } else {
2106 errno = 0;
2107 ZVAL_LONG(zendlval, ZEND_STRTOL(hex, &end, 16));
2108 ZEND_ASSERT(!errno && end == hex + len);
2109 }
2110 if (contains_underscores) {
2111 efree(hex);
2112 }
2113 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
2114 } else {
2115 ZVAL_DOUBLE(zendlval, zend_hex_strtod(hex, (const char **)&end));
2116 /* errno isn't checked since we allow HUGE_VAL/INF overflow */
2117 ZEND_ASSERT(end == hex + len);
2118 if (contains_underscores) {
2119 efree(hex);
2120 }
2121 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
2122 }
2123 }
2124
2125 <ST_VAR_OFFSET>[0]|([1-9][0-9]*) { /* Offset could be treated as a long */
2126 if (yyleng < MAX_LENGTH_OF_LONG - 1 || (yyleng == MAX_LENGTH_OF_LONG - 1 && strcmp(yytext, long_min_digits) < 0)) {
2127 char *end;
2128 errno = 0;
2129 ZVAL_LONG(zendlval, ZEND_STRTOL(yytext, &end, 10));
2130 if (errno == ERANGE) {
2131 goto string;
2132 }
2133 ZEND_ASSERT(end == yytext + yyleng);
2134 } else {
2135 string:
2136 ZVAL_STRINGL(zendlval, yytext, yyleng);
2137 }
2138 RETURN_TOKEN_WITH_VAL(T_NUM_STRING);
2139 }
2140
2141 <ST_VAR_OFFSET>{LNUM}|{HNUM}|{BNUM}|{ONUM} { /* Offset must be treated as a string */
2142 if (yyleng == 1) {
2143 ZVAL_INTERNED_STR(zendlval, ZSTR_CHAR((zend_uchar)*(yytext)));
2144 } else {
2145 ZVAL_STRINGL(zendlval, yytext, yyleng);
2146 }
2147 RETURN_TOKEN_WITH_VAL(T_NUM_STRING);
2148 }
2149
2150 <ST_IN_SCRIPTING>{DNUM}|{EXPONENT_DNUM} {
2151 const char *end;
2152 size_t len = yyleng;
2153 char *dnum = yytext;
2154 bool contains_underscores = (memchr(dnum, '_', len) != NULL);
2155
2156 if (contains_underscores) {
2157 dnum = estrndup(dnum, len);
2158 strip_underscores(dnum, &len);
2159 }
2160
2161 ZVAL_DOUBLE(zendlval, zend_strtod(dnum, &end));
2162 /* errno isn't checked since we allow HUGE_VAL/INF overflow */
2163 ZEND_ASSERT(end == dnum + len);
2164 if (contains_underscores) {
2165 efree(dnum);
2166 }
2167 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
2168 }
2169
2170 <ST_IN_SCRIPTING>"__CLASS__" {
2171 RETURN_TOKEN_WITH_IDENT(T_CLASS_C);
2172 }
2173
2174 <ST_IN_SCRIPTING>"__TRAIT__" {
2175 RETURN_TOKEN_WITH_IDENT(T_TRAIT_C);
2176 }
2177
2178 <ST_IN_SCRIPTING>"__FUNCTION__" {
2179 RETURN_TOKEN_WITH_IDENT(T_FUNC_C);
2180 }
2181
2182 <ST_IN_SCRIPTING>"__METHOD__" {
2183 RETURN_TOKEN_WITH_IDENT(T_METHOD_C);
2184 }
2185
2186 <ST_IN_SCRIPTING>"__LINE__" {
2187 RETURN_TOKEN_WITH_IDENT(T_LINE);
2188 }
2189
2190 <ST_IN_SCRIPTING>"__FILE__" {
2191 RETURN_TOKEN_WITH_IDENT(T_FILE);
2192 }
2193
2194 <ST_IN_SCRIPTING>"__DIR__" {
2195 RETURN_TOKEN_WITH_IDENT(T_DIR);
2196 }
2197
2198 <ST_IN_SCRIPTING>"__NAMESPACE__" {
2199 RETURN_TOKEN_WITH_IDENT(T_NS_C);
2200 }
2201
2202 <SHEBANG>"#!" .* {NEWLINE} {
2203 CG(zend_lineno)++;
2204 BEGIN(INITIAL);
2205 goto restart;
2206 }
2207
2208 <SHEBANG>{ANY_CHAR} {
2209 yyless(0);
2210 BEGIN(INITIAL);
2211 goto restart;
2212 }
2213
2214 <INITIAL>"<?=" {
2215 BEGIN(ST_IN_SCRIPTING);
2216 if (PARSER_MODE()) {
2217 /* We'll reject this as an identifier in zend_lex_tstring. */
2218 RETURN_TOKEN_WITH_IDENT(T_ECHO);
2219 }
2220 RETURN_TOKEN(T_OPEN_TAG_WITH_ECHO);
2221 }
2222
2223
2224 <INITIAL>"<?php"([ \t]|{NEWLINE}) {
2225 HANDLE_NEWLINE(yytext[yyleng-1]);
2226 BEGIN(ST_IN_SCRIPTING);
2227 RETURN_OR_SKIP_TOKEN(T_OPEN_TAG);
2228 }
2229
2230 <INITIAL>"<?php" {
2231 /* Allow <?php followed by end of file. */
2232 if (YYCURSOR == YYLIMIT) {
2233 BEGIN(ST_IN_SCRIPTING);
2234 RETURN_OR_SKIP_TOKEN(T_OPEN_TAG);
2235 }
2236 /* Degenerate case: <?phpX is interpreted as <? phpX with short tags. */
2237 if (CG(short_tags)) {
2238 yyless(2);
2239 BEGIN(ST_IN_SCRIPTING);
2240 RETURN_OR_SKIP_TOKEN(T_OPEN_TAG);
2241 }
2242 goto inline_char_handler;
2243 }
2244
2245 <INITIAL>"<?" {
2246 if (CG(short_tags)) {
2247 BEGIN(ST_IN_SCRIPTING);
2248 RETURN_OR_SKIP_TOKEN(T_OPEN_TAG);
2249 } else {
2250 goto inline_char_handler;
2251 }
2252 }
2253
2254 <INITIAL>{ANY_CHAR} {
2255 if (YYCURSOR > YYLIMIT) {
2256 RETURN_END_TOKEN;
2257 }
2258
2259 inline_char_handler:
2260
2261 while (1) {
2262 YYCTYPE *ptr = memchr(YYCURSOR, '<', YYLIMIT - YYCURSOR);
2263
2264 YYCURSOR = ptr ? ptr + 1 : YYLIMIT;
2265
2266 if (YYCURSOR >= YYLIMIT) {
2267 break;
2268 }
2269
2270 if (*YYCURSOR == '?') {
2271 if (CG(short_tags) /* <? */
2272 || (*(YYCURSOR + 1) == '=') /* <?= */
2273 || (!strncasecmp((char*)YYCURSOR + 1, "php", 3) && /* <?php[ \t\r\n] */
2274 (YYCURSOR + 4 == YYLIMIT ||
2275 YYCURSOR[4] == ' ' || YYCURSOR[4] == '\t' ||
2276 YYCURSOR[4] == '\n' || YYCURSOR[4] == '\r'))
2277 ) {
2278 YYCURSOR--;
2279 break;
2280 }
2281 }
2282 }
2283
2284 yyleng = YYCURSOR - SCNG(yy_text);
2285
2286 if (SCNG(output_filter)) {
2287 size_t readsize;
2288 char *s = NULL;
2289 size_t sz = 0;
2290 // TODO: avoid reallocation ???
2291 readsize = SCNG(output_filter)((unsigned char **)&s, &sz, (unsigned char *)yytext, (size_t)yyleng);
2292 ZVAL_STRINGL(zendlval, s, sz);
2293 efree(s);
2294 if (readsize < yyleng) {
2295 yyless(readsize);
2296 }
2297 } else if (yyleng == 1) {
2298 ZVAL_INTERNED_STR(zendlval, ZSTR_CHAR((zend_uchar)*yytext));
2299 } else {
2300 ZVAL_STRINGL(zendlval, yytext, yyleng);
2301 }
2302 HANDLE_NEWLINES(yytext, yyleng);
2303 RETURN_TOKEN_WITH_VAL(T_INLINE_HTML);
2304 }
2305
2306
2307 /* Make sure a label character follows "->" or "?->", otherwise there is no property
2308 * and "->"/"?->" will be taken literally
2309 */
2310 <ST_DOUBLE_QUOTES,ST_HEREDOC,ST_BACKQUOTE>"$"{LABEL}"->"[a-zA-Z_\x80-\xff] {
2311 yyless(yyleng - 3);
2312 yy_push_state(ST_LOOKING_FOR_PROPERTY);
2313 RETURN_TOKEN_WITH_STR(T_VARIABLE, 1);
2314 }
2315
2316 <ST_DOUBLE_QUOTES,ST_HEREDOC,ST_BACKQUOTE>"$"{LABEL}"?->"[a-zA-Z_\x80-\xff] {
2317 yyless(yyleng - 4);
2318 yy_push_state(ST_LOOKING_FOR_PROPERTY);
2319 RETURN_TOKEN_WITH_STR(T_VARIABLE, 1);
2320 }
2321
2322 /* A [ always designates a variable offset, regardless of what follows
2323 */
2324 <ST_DOUBLE_QUOTES,ST_HEREDOC,ST_BACKQUOTE>"$"{LABEL}"[" {
2325 yyless(yyleng - 1);
2326 yy_push_state(ST_VAR_OFFSET);
2327 RETURN_TOKEN_WITH_STR(T_VARIABLE, 1);
2328 }
2329
2330 <ST_IN_SCRIPTING,ST_DOUBLE_QUOTES,ST_HEREDOC,ST_BACKQUOTE,ST_VAR_OFFSET>"$"{LABEL} {
2331 RETURN_TOKEN_WITH_STR(T_VARIABLE, 1);
2332 }
2333
2334 <ST_VAR_OFFSET>"]" {
2335 yy_pop_state();
2336 RETURN_TOKEN(']');
2337 }
2338
2339 <ST_VAR_OFFSET>{TOKENS}|[[(){}"`] {
2340 /* Only '[' or '-' can be valid, but returning other tokens will allow a more explicit parse error */
2341 RETURN_TOKEN(yytext[0]);
2342 }
2343
2344 <ST_VAR_OFFSET>[ \n\r\t\\'#] {
2345 /* Invalid rule to return a more explicit parse error with proper line number */
2346 yyless(0);
2347 yy_pop_state();
2348 ZVAL_NULL(zendlval);
2349 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2350 }
2351
2352 <ST_IN_SCRIPTING>"namespace"("\\"{LABEL})+ {
2353 RETURN_TOKEN_WITH_STR(T_NAME_RELATIVE, sizeof("namespace\\") - 1);
2354 }
2355
2356 <ST_IN_SCRIPTING>{LABEL}("\\"{LABEL})+ {
2357 RETURN_TOKEN_WITH_STR(T_NAME_QUALIFIED, 0);
2358 }
2359
2360 <ST_IN_SCRIPTING>"\\"{LABEL}("\\"{LABEL})* {
2361 RETURN_TOKEN_WITH_STR(T_NAME_FULLY_QUALIFIED, 1);
2362 }
2363
2364 <ST_IN_SCRIPTING>"\\" {
2365 RETURN_TOKEN(T_NS_SEPARATOR);
2366 }
2367
2368 <ST_IN_SCRIPTING,ST_VAR_OFFSET>{LABEL} {
2369 RETURN_TOKEN_WITH_STR(T_STRING, 0);
2370 }
2371
2372
2373 <ST_IN_SCRIPTING>"#"|"//" {
2374 while (YYCURSOR < YYLIMIT) {
2375 switch (*YYCURSOR++) {
2376 case '\r':
2377 case '\n':
2378 YYCURSOR--;
2379 break;
2380 case '?':
2381 if (*YYCURSOR == '>') {
2382 YYCURSOR--;
2383 break;
2384 }
2385 ZEND_FALLTHROUGH;
2386 default:
2387 continue;
2388 }
2389
2390 break;
2391 }
2392
2393 yyleng = YYCURSOR - SCNG(yy_text);
2394 RETURN_OR_SKIP_TOKEN(T_COMMENT);
2395 }
2396
2397 <ST_IN_SCRIPTING>"/*"|"/**"{WHITESPACE} {
2398 int doc_com;
2399
2400 if (yyleng > 2) {
2401 doc_com = 1;
2402 RESET_DOC_COMMENT();
2403 } else {
2404 doc_com = 0;
2405 }
2406
2407 while (YYCURSOR < YYLIMIT) {
2408 if (*YYCURSOR++ == '*' && *YYCURSOR == '/') {
2409 break;
2410 }
2411 }
2412
2413 if (YYCURSOR < YYLIMIT) {
2414 YYCURSOR++;
2415 } else {
2416 zend_throw_exception_ex(zend_ce_parse_error, 0, "Unterminated comment starting line %d", CG(zend_lineno));
2417 if (PARSER_MODE()) {
2418 RETURN_TOKEN(T_ERROR);
2419 }
2420 }
2421
2422 yyleng = YYCURSOR - SCNG(yy_text);
2423 HANDLE_NEWLINES(yytext, yyleng);
2424
2425 if (doc_com) {
2426 CG(doc_comment) = zend_string_init(yytext, yyleng, 0);
2427 RETURN_OR_SKIP_TOKEN(T_DOC_COMMENT);
2428 }
2429
2430 RETURN_OR_SKIP_TOKEN(T_COMMENT);
2431 }
2432
2433 <ST_IN_SCRIPTING>"?>"{NEWLINE}? {
2434 BEGIN(INITIAL);
2435 if (yytext[yyleng-1] != '>') {
2436 CG(increment_lineno) = 1;
2437 }
2438 if (PARSER_MODE()) {
2439 RETURN_TOKEN(';'); /* implicit ';' at php-end tag */
2440 }
2441 RETURN_TOKEN(T_CLOSE_TAG);
2442 }
2443
2444
2445 <ST_IN_SCRIPTING>b?['] {
2446 char *s, *t;
2447 char *end;
2448 int bprefix = (yytext[0] != '\'') ? 1 : 0;
2449
2450 while (1) {
2451 if (YYCURSOR < YYLIMIT) {
2452 if (*YYCURSOR == '\'') {
2453 YYCURSOR++;
2454 yyleng = YYCURSOR - SCNG(yy_text);
2455
2456 break;
2457 } else if (*YYCURSOR++ == '\\' && YYCURSOR < YYLIMIT) {
2458 YYCURSOR++;
2459 }
2460 } else {
2461 yyleng = YYLIMIT - SCNG(yy_text);
2462
2463 /* Unclosed single quotes; treat similar to double quotes, but without a separate token
2464 * for ' (unrecognized by parser), instead of old flex fallback to "Unexpected character..."
2465 * rule, which continued in ST_IN_SCRIPTING state after the quote */
2466 ZVAL_NULL(zendlval);
2467 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2468 }
2469 }
2470
2471 if (yyleng-bprefix-2 <= 1) {
2472 if (yyleng-bprefix-2 < 1) {
2473 ZVAL_EMPTY_STRING(zendlval);
2474 } else {
2475 zend_uchar c = (zend_uchar)*(yytext+bprefix+1);
2476 if (c == '\n' || c == '\r') {
2477 CG(zend_lineno)++;
2478 }
2479 ZVAL_INTERNED_STR(zendlval, ZSTR_CHAR(c));
2480 }
2481 goto skip_escape_conversion;
2482 }
2483 ZVAL_STRINGL(zendlval, yytext+bprefix+1, yyleng-bprefix-2);
2484
2485 /* convert escape sequences */
2486 s = Z_STRVAL_P(zendlval);
2487 end = s+Z_STRLEN_P(zendlval);
2488 while (1) {
2489 if (UNEXPECTED(*s=='\\')) {
2490 break;
2491 }
2492 if (*s == '\n' || (*s == '\r' && (*(s+1) != '\n'))) {
2493 CG(zend_lineno)++;
2494 }
2495 s++;
2496 if (s == end) {
2497 goto skip_escape_conversion;
2498 }
2499 }
2500
2501 t = s;
2502 while (s<end) {
2503 if (*s=='\\') {
2504 s++;
2505 if (*s == '\\' || *s == '\'') {
2506 *t++ = *s;
2507 } else {
2508 *t++ = '\\';
2509 *t++ = *s;
2510 }
2511 } else {
2512 *t++ = *s;
2513 }
2514 if (*s == '\n' || (*s == '\r' && (*(s+1) != '\n'))) {
2515 CG(zend_lineno)++;
2516 }
2517 s++;
2518 }
2519 *t = 0;
2520 Z_STRLEN_P(zendlval) = t - Z_STRVAL_P(zendlval);
2521
2522 skip_escape_conversion:
2523 if (SCNG(output_filter)) {
2524 size_t sz = 0;
2525 char *str = NULL;
2526 zend_string *new_str;
2527 s = Z_STRVAL_P(zendlval);
2528 // TODO: avoid reallocation ???
2529 SCNG(output_filter)((unsigned char **)&str, &sz, (unsigned char *)s, (size_t)Z_STRLEN_P(zendlval));
2530 new_str = zend_string_init(str, sz, 0);
2531 if (str != s) {
2532 efree(str);
2533 }
2534 zend_string_release_ex(Z_STR_P(zendlval), 0);
2535 ZVAL_STR(zendlval, new_str);
2536 }
2537 RETURN_TOKEN_WITH_VAL(T_CONSTANT_ENCAPSED_STRING);
2538 }
2539
2540
2541 <ST_IN_SCRIPTING>b?["] {
2542 int bprefix = (yytext[0] != '"') ? 1 : 0;
2543
2544 while (YYCURSOR < YYLIMIT) {
2545 switch (*YYCURSOR++) {
2546 case '"':
2547 yyleng = YYCURSOR - SCNG(yy_text);
2548 if (EXPECTED(zend_scan_escape_string(zendlval, yytext+bprefix+1, yyleng-bprefix-2, '"') == SUCCESS)
2549 || !PARSER_MODE()) {
2550 RETURN_TOKEN_WITH_VAL(T_CONSTANT_ENCAPSED_STRING);
2551 } else {
2552 RETURN_TOKEN(T_ERROR);
2553 }
2554 case '$':
2555 if (IS_LABEL_START(*YYCURSOR) || *YYCURSOR == '{') {
2556 break;
2557 }
2558 continue;
2559 case '{':
2560 if (*YYCURSOR == '$') {
2561 break;
2562 }
2563 continue;
2564 case '\\':
2565 if (YYCURSOR < YYLIMIT) {
2566 YYCURSOR++;
2567 }
2568 ZEND_FALLTHROUGH;
2569 default:
2570 continue;
2571 }
2572
2573 YYCURSOR--;
2574 break;
2575 }
2576
2577 /* Remember how much was scanned to save rescanning */
2578 SET_DOUBLE_QUOTES_SCANNED_LENGTH(YYCURSOR - SCNG(yy_text) - yyleng);
2579
2580 YYCURSOR = SCNG(yy_text) + yyleng;
2581
2582 BEGIN(ST_DOUBLE_QUOTES);
2583 RETURN_TOKEN('"');
2584 }
2585
2586
2587 <ST_IN_SCRIPTING>b?"<<<"{TABS_AND_SPACES}({LABEL}|([']{LABEL}['])|(["]{LABEL}["])){NEWLINE} {
2588 char *s;
2589 unsigned char *saved_cursor;
2590 int bprefix = (yytext[0] != '<') ? 1 : 0, spacing = 0, indentation = 0;
2591 zend_heredoc_label *heredoc_label = emalloc(sizeof(zend_heredoc_label));
2592 bool is_heredoc = 1;
2593
2594 CG(zend_lineno)++;
2595 heredoc_label->length = yyleng-bprefix-3-1-(yytext[yyleng-2]=='\r'?1:0);
2596 s = yytext+bprefix+3;
2597 while ((*s == ' ') || (*s == '\t')) {
2598 s++;
2599 heredoc_label->length--;
2600 }
2601
2602 if (*s == '\'') {
2603 s++;
2604 heredoc_label->length -= 2;
2605 is_heredoc = 0;
2606
2607 BEGIN(ST_NOWDOC);
2608 } else {
2609 if (*s == '"') {
2610 s++;
2611 heredoc_label->length -= 2;
2612 }
2613
2614 BEGIN(ST_HEREDOC);
2615 }
2616
2617 heredoc_label->label = estrndup(s, heredoc_label->length);
2618 heredoc_label->indentation_uses_spaces = 0;
2619 heredoc_label->indentation = 0;
2620 saved_cursor = YYCURSOR;
2621
2622 zend_ptr_stack_push(&SCNG(heredoc_label_stack), (void *) heredoc_label);
2623
2624 while (YYCURSOR < YYLIMIT && (*YYCURSOR == ' ' || *YYCURSOR == '\t')) {
2625 if (*YYCURSOR == '\t') {
2626 spacing |= HEREDOC_USING_TABS;
2627 } else {
2628 spacing |= HEREDOC_USING_SPACES;
2629 }
2630 ++YYCURSOR;
2631 ++indentation;
2632 }
2633
2634 if (YYCURSOR == YYLIMIT) {
2635 YYCURSOR = saved_cursor;
2636 RETURN_TOKEN(T_START_HEREDOC);
2637 }
2638
2639 /* Check for ending label on the next line */
2640 if (heredoc_label->length < YYLIMIT - YYCURSOR && !memcmp(YYCURSOR, s, heredoc_label->length)) {
2641 if (!IS_LABEL_SUCCESSOR(YYCURSOR[heredoc_label->length])) {
2642 if (spacing == (HEREDOC_USING_SPACES | HEREDOC_USING_TABS)) {
2643 zend_throw_exception(zend_ce_parse_error, "Invalid indentation - tabs and spaces cannot be mixed", 0);
2644 if (PARSER_MODE()) {
2645 RETURN_TOKEN(T_ERROR);
2646 }
2647 }
2648
2649 YYCURSOR = saved_cursor;
2650 heredoc_label->indentation = indentation;
2651
2652 BEGIN(ST_END_HEREDOC);
2653 RETURN_TOKEN(T_START_HEREDOC);
2654 }
2655 }
2656
2657 YYCURSOR = saved_cursor;
2658
2659 if (is_heredoc && !SCNG(heredoc_scan_ahead)) {
2660 zend_lex_state current_state;
2661 zend_string *saved_doc_comment = CG(doc_comment);
2662 int heredoc_nesting_level = 1;
2663 int first_token = 0;
2664 int error = 0;
2665
2666 zend_save_lexical_state(¤t_state);
2667
2668 SCNG(heredoc_scan_ahead) = 1;
2669 SCNG(heredoc_indentation) = 0;
2670 SCNG(heredoc_indentation_uses_spaces) = 0;
2671 LANG_SCNG(on_event) = NULL;
2672 CG(doc_comment) = NULL;
2673
2674 zend_ptr_stack_reverse_apply(¤t_state.heredoc_label_stack, copy_heredoc_label_stack);
2675
2676 zend_exception_save();
2677 while (heredoc_nesting_level) {
2678 zval zv;
2679 int retval;
2680
2681 ZVAL_UNDEF(&zv);
2682 retval = lex_scan(&zv, NULL);
2683 zval_ptr_dtor_nogc(&zv);
2684
2685 if (EG(exception)) {
2686 zend_clear_exception();
2687 break;
2688 }
2689
2690 if (!first_token) {
2691 first_token = retval;
2692 }
2693
2694 switch (retval) {
2695 case T_START_HEREDOC:
2696 ++heredoc_nesting_level;
2697 break;
2698 case T_END_HEREDOC:
2699 --heredoc_nesting_level;
2700 break;
2701 case END:
2702 heredoc_nesting_level = 0;
2703 }
2704 }
2705 zend_exception_restore();
2706
2707 if (
2708 (first_token == T_VARIABLE
2709 || first_token == T_DOLLAR_OPEN_CURLY_BRACES
2710 || first_token == T_CURLY_OPEN
2711 ) && SCNG(heredoc_indentation)) {
2712 zend_throw_exception_ex(zend_ce_parse_error, 0, "Invalid body indentation level (expecting an indentation level of at least %d)", SCNG(heredoc_indentation));
2713 error = 1;
2714 }
2715
2716 heredoc_label->indentation = SCNG(heredoc_indentation);
2717 heredoc_label->indentation_uses_spaces = SCNG(heredoc_indentation_uses_spaces);
2718
2719 zend_restore_lexical_state(¤t_state);
2720 SCNG(heredoc_scan_ahead) = 0;
2721 CG(increment_lineno) = 0;
2722 CG(doc_comment) = saved_doc_comment;
2723
2724 if (PARSER_MODE() && error) {
2725 RETURN_TOKEN(T_ERROR);
2726 }
2727 }
2728
2729 RETURN_TOKEN(T_START_HEREDOC);
2730 }
2731
2732
2733 <ST_IN_SCRIPTING>[`] {
2734 BEGIN(ST_BACKQUOTE);
2735 RETURN_TOKEN('`');
2736 }
2737
2738
2739 <ST_END_HEREDOC>{ANY_CHAR} {
2740 zend_heredoc_label *heredoc_label = zend_ptr_stack_pop(&SCNG(heredoc_label_stack));
2741
2742 yyleng = heredoc_label->indentation + heredoc_label->length;
2743 YYCURSOR += yyleng - 1;
2744
2745 heredoc_label_dtor(heredoc_label);
2746 efree(heredoc_label);
2747
2748 BEGIN(ST_IN_SCRIPTING);
2749 RETURN_TOKEN(T_END_HEREDOC);
2750 }
2751
2752
2753 <ST_DOUBLE_QUOTES,ST_BACKQUOTE,ST_HEREDOC>"{$" {
2754 yy_push_state(ST_IN_SCRIPTING);
2755 yyless(1);
2756 enter_nesting('{');
2757 RETURN_TOKEN(T_CURLY_OPEN);
2758 }
2759
2760
2761 <ST_DOUBLE_QUOTES>["] {
2762 BEGIN(ST_IN_SCRIPTING);
2763 RETURN_TOKEN('"');
2764 }
2765
2766 <ST_BACKQUOTE>[`] {
2767 BEGIN(ST_IN_SCRIPTING);
2768 RETURN_TOKEN('`');
2769 }
2770
2771
2772 <ST_DOUBLE_QUOTES>{ANY_CHAR} {
2773 if (GET_DOUBLE_QUOTES_SCANNED_LENGTH()) {
2774 YYCURSOR += GET_DOUBLE_QUOTES_SCANNED_LENGTH() - 1;
2775 SET_DOUBLE_QUOTES_SCANNED_LENGTH(0);
2776
2777 goto double_quotes_scan_done;
2778 }
2779
2780 if (YYCURSOR > YYLIMIT) {
2781 RETURN_END_TOKEN;
2782 }
2783 if (yytext[0] == '\\' && YYCURSOR < YYLIMIT) {
2784 YYCURSOR++;
2785 }
2786
2787 while (YYCURSOR < YYLIMIT) {
2788 switch (*YYCURSOR++) {
2789 case '"':
2790 break;
2791 case '$':
2792 if (IS_LABEL_START(*YYCURSOR) || *YYCURSOR == '{') {
2793 break;
2794 }
2795 continue;
2796 case '{':
2797 if (*YYCURSOR == '$') {
2798 break;
2799 }
2800 continue;
2801 case '\\':
2802 if (YYCURSOR < YYLIMIT) {
2803 YYCURSOR++;
2804 }
2805 ZEND_FALLTHROUGH;
2806 default:
2807 continue;
2808 }
2809
2810 YYCURSOR--;
2811 break;
2812 }
2813
2814 double_quotes_scan_done:
2815 yyleng = YYCURSOR - SCNG(yy_text);
2816
2817 if (EXPECTED(zend_scan_escape_string(zendlval, yytext, yyleng, '"') == SUCCESS)
2818 || !PARSER_MODE()) {
2819 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2820 } else {
2821 RETURN_TOKEN(T_ERROR);
2822 }
2823 }
2824
2825
2826 <ST_BACKQUOTE>{ANY_CHAR} {
2827 if (YYCURSOR > YYLIMIT) {
2828 RETURN_END_TOKEN;
2829 }
2830 if (yytext[0] == '\\' && YYCURSOR < YYLIMIT) {
2831 YYCURSOR++;
2832 }
2833
2834 while (YYCURSOR < YYLIMIT) {
2835 switch (*YYCURSOR++) {
2836 case '`':
2837 break;
2838 case '$':
2839 if (IS_LABEL_START(*YYCURSOR) || *YYCURSOR == '{') {
2840 break;
2841 }
2842 continue;
2843 case '{':
2844 if (*YYCURSOR == '$') {
2845 break;
2846 }
2847 continue;
2848 case '\\':
2849 if (YYCURSOR < YYLIMIT) {
2850 YYCURSOR++;
2851 }
2852 ZEND_FALLTHROUGH;
2853 default:
2854 continue;
2855 }
2856
2857 YYCURSOR--;
2858 break;
2859 }
2860
2861 yyleng = YYCURSOR - SCNG(yy_text);
2862
2863 if (EXPECTED(zend_scan_escape_string(zendlval, yytext, yyleng, '`') == SUCCESS)
2864 || !PARSER_MODE()) {
2865 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2866 } else {
2867 RETURN_TOKEN(T_ERROR);
2868 }
2869 }
2870
2871
2872 <ST_HEREDOC>{ANY_CHAR} {
2873 zend_heredoc_label *heredoc_label = zend_ptr_stack_top(&SCNG(heredoc_label_stack));
2874 int newline = 0, indentation = 0, spacing = 0;
2875
2876 if (YYCURSOR > YYLIMIT) {
2877 RETURN_END_TOKEN;
2878 }
2879
2880 YYCURSOR--;
2881
2882 while (YYCURSOR < YYLIMIT) {
2883 switch (*YYCURSOR++) {
2884 case '\r':
2885 if (*YYCURSOR == '\n') {
2886 YYCURSOR++;
2887 }
2888 ZEND_FALLTHROUGH;
2889 case '\n':
2890 indentation = spacing = 0;
2891
2892 while (YYCURSOR < YYLIMIT && (*YYCURSOR == ' ' || *YYCURSOR == '\t')) {
2893 if (*YYCURSOR == '\t') {
2894 spacing |= HEREDOC_USING_TABS;
2895 } else {
2896 spacing |= HEREDOC_USING_SPACES;
2897 }
2898 ++YYCURSOR;
2899 ++indentation;
2900 }
2901
2902 if (YYCURSOR == YYLIMIT) {
2903 yyleng = YYCURSOR - SCNG(yy_text);
2904 HANDLE_NEWLINES(yytext, yyleng);
2905 ZVAL_NULL(zendlval);
2906 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2907 }
2908
2909 /* Check for ending label on the next line */
2910 if (IS_LABEL_START(*YYCURSOR) && heredoc_label->length < YYLIMIT - YYCURSOR && !memcmp(YYCURSOR, heredoc_label->label, heredoc_label->length)) {
2911 if (IS_LABEL_SUCCESSOR(YYCURSOR[heredoc_label->length])) {
2912 continue;
2913 }
2914
2915 if (spacing == (HEREDOC_USING_SPACES | HEREDOC_USING_TABS)) {
2916 zend_throw_exception(zend_ce_parse_error, "Invalid indentation - tabs and spaces cannot be mixed", 0);
2917 if (PARSER_MODE()) {
2918 RETURN_TOKEN(T_ERROR);
2919 }
2920 }
2921
2922 /* newline before label will be subtracted from returned text, but
2923 * yyleng/yytext will include it, for zend_highlight/strip, tokenizer, etc. */
2924 if (YYCURSOR[-indentation - 2] == '\r' && YYCURSOR[-indentation - 1] == '\n') {
2925 newline = 2; /* Windows newline */
2926 } else {
2927 newline = 1;
2928 }
2929
2930 CG(increment_lineno) = 1; /* For newline before label */
2931
2932 if (SCNG(heredoc_scan_ahead)) {
2933 SCNG(heredoc_indentation) = indentation;
2934 SCNG(heredoc_indentation_uses_spaces) = (spacing == HEREDOC_USING_SPACES);
2935 } else {
2936 YYCURSOR -= indentation;
2937 }
2938
2939 BEGIN(ST_END_HEREDOC);
2940
2941 goto heredoc_scan_done;
2942 }
2943 continue;
2944 case '$':
2945 if (IS_LABEL_START(*YYCURSOR) || *YYCURSOR == '{') {
2946 break;
2947 }
2948 continue;
2949 case '{':
2950 if (*YYCURSOR == '$') {
2951 break;
2952 }
2953 continue;
2954 case '\\':
2955 if (YYCURSOR < YYLIMIT && *YYCURSOR != '\n' && *YYCURSOR != '\r') {
2956 YYCURSOR++;
2957 }
2958 ZEND_FALLTHROUGH;
2959 default:
2960 continue;
2961 }
2962
2963 YYCURSOR--;
2964 break;
2965 }
2966
2967 heredoc_scan_done:
2968
2969 yyleng = YYCURSOR - SCNG(yy_text);
2970 ZVAL_STRINGL(zendlval, yytext, yyleng - newline);
2971
2972 if (!SCNG(heredoc_scan_ahead) && !EG(exception) && PARSER_MODE()) {
2973 bool newline_at_start = *(yytext - 1) == '\n' || *(yytext - 1) == '\r';
2974 zend_string *copy = Z_STR_P(zendlval);
2975
2976 if (!strip_multiline_string_indentation(
2977 zendlval, heredoc_label->indentation, heredoc_label->indentation_uses_spaces,
2978 newline_at_start, newline != 0)) {
2979 RETURN_TOKEN(T_ERROR);
2980 }
2981
2982 if (UNEXPECTED(zend_scan_escape_string(zendlval, ZSTR_VAL(copy), ZSTR_LEN(copy), 0) != SUCCESS)) {
2983 zend_string_efree(copy);
2984 RETURN_TOKEN(T_ERROR);
2985 }
2986
2987 zend_string_efree(copy);
2988 } else {
2989 HANDLE_NEWLINES(yytext, yyleng - newline);
2990 }
2991
2992 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2993 }
2994
2995
2996 <ST_NOWDOC>{ANY_CHAR} {
2997 zend_heredoc_label *heredoc_label = zend_ptr_stack_top(&SCNG(heredoc_label_stack));
2998 int newline = 0, indentation = 0, spacing = -1;
2999
3000 if (YYCURSOR > YYLIMIT) {
3001 RETURN_END_TOKEN;
3002 }
3003
3004 YYCURSOR--;
3005
3006 while (YYCURSOR < YYLIMIT) {
3007 switch (*YYCURSOR++) {
3008 case '\r':
3009 if (*YYCURSOR == '\n') {
3010 YYCURSOR++;
3011 }
3012 ZEND_FALLTHROUGH;
3013 case '\n':
3014 indentation = spacing = 0;
3015
3016 while (YYCURSOR < YYLIMIT && (*YYCURSOR == ' ' || *YYCURSOR == '\t')) {
3017 if (*YYCURSOR == '\t') {
3018 spacing |= HEREDOC_USING_TABS;
3019 } else {
3020 spacing |= HEREDOC_USING_SPACES;
3021 }
3022 ++YYCURSOR;
3023 ++indentation;
3024 }
3025
3026 if (YYCURSOR == YYLIMIT) {
3027 yyleng = YYCURSOR - SCNG(yy_text);
3028 HANDLE_NEWLINES(yytext, yyleng);
3029 ZVAL_NULL(zendlval);
3030 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
3031 }
3032
3033 /* Check for ending label on the next line */
3034 if (IS_LABEL_START(*YYCURSOR) && heredoc_label->length < YYLIMIT - YYCURSOR && !memcmp(YYCURSOR, heredoc_label->label, heredoc_label->length)) {
3035 if (IS_LABEL_SUCCESSOR(YYCURSOR[heredoc_label->length])) {
3036 continue;
3037 }
3038
3039 if (spacing == (HEREDOC_USING_SPACES | HEREDOC_USING_TABS)) {
3040 zend_throw_exception(zend_ce_parse_error, "Invalid indentation - tabs and spaces cannot be mixed", 0);
3041 if (PARSER_MODE()) {
3042 RETURN_TOKEN(T_ERROR);
3043 }
3044 }
3045
3046 /* newline before label will be subtracted from returned text, but
3047 * yyleng/yytext will include it, for zend_highlight/strip, tokenizer, etc. */
3048 if (YYCURSOR[-indentation - 2] == '\r' && YYCURSOR[-indentation - 1] == '\n') {
3049 newline = 2; /* Windows newline */
3050 } else {
3051 newline = 1;
3052 }
3053
3054 CG(increment_lineno) = 1; /* For newline before label */
3055
3056 YYCURSOR -= indentation;
3057 heredoc_label->indentation = indentation;
3058
3059 BEGIN(ST_END_HEREDOC);
3060
3061 goto nowdoc_scan_done;
3062 }
3063 ZEND_FALLTHROUGH;
3064 default:
3065 continue;
3066 }
3067 }
3068
3069 nowdoc_scan_done:
3070 yyleng = YYCURSOR - SCNG(yy_text);
3071 ZVAL_STRINGL(zendlval, yytext, yyleng - newline);
3072
3073 if (!EG(exception) && spacing != -1 && PARSER_MODE()) {
3074 bool newline_at_start = *(yytext - 1) == '\n' || *(yytext - 1) == '\r';
3075 if (!strip_multiline_string_indentation(
3076 zendlval, indentation, spacing == HEREDOC_USING_SPACES,
3077 newline_at_start, newline != 0)) {
3078 RETURN_TOKEN(T_ERROR);
3079 }
3080 }
3081
3082 HANDLE_NEWLINES(yytext, yyleng - newline);
3083 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
3084 }
3085
3086
3087 <ST_IN_SCRIPTING,ST_VAR_OFFSET>{ANY_CHAR} {
3088 if (YYCURSOR > YYLIMIT) {
3089 RETURN_END_TOKEN;
3090 }
3091
3092 RETURN_TOKEN(T_BAD_CHARACTER);
3093 }
3094
3095 */
3096
3097 emit_token_with_str:
3098 zend_copy_value(zendlval, (yytext + offset), (yyleng - offset));
3099
3100 emit_token_with_val:
3101 if (PARSER_MODE()) {
3102 ZEND_ASSERT(Z_TYPE_P(zendlval) != IS_UNDEF);
3103 elem->ast = zend_ast_create_zval_with_lineno(zendlval, start_line);
3104 }
3105
3106 emit_token:
3107 if (SCNG(on_event)) {
3108 SCNG(on_event)(ON_TOKEN, token, start_line, yytext, yyleng, SCNG(on_event_context));
3109 }
3110 return token;
3111
3112 emit_token_with_ident:
3113 if (PARSER_MODE()) {
3114 elem->ident = SCNG(yy_text);
3115 }
3116 if (SCNG(on_event)) {
3117 SCNG(on_event)(ON_TOKEN, token, start_line, yytext, yyleng, SCNG(on_event_context));
3118 }
3119 return token;
3120
3121 return_whitespace:
3122 HANDLE_NEWLINES(yytext, yyleng);
3123 if (SCNG(on_event)) {
3124 SCNG(on_event)(ON_TOKEN, T_WHITESPACE, start_line, yytext, yyleng, SCNG(on_event_context));
3125 }
3126 if (PARSER_MODE()) {
3127 start_line = CG(zend_lineno);
3128 goto restart;
3129 } else {
3130 return T_WHITESPACE;
3131 }
3132
3133 skip_token:
3134 if (SCNG(on_event)) {
3135 SCNG(on_event)(ON_TOKEN, token, start_line, yytext, yyleng, SCNG(on_event_context));
3136 }
3137 start_line = CG(zend_lineno);
3138 goto restart;
3139 }
3140