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 pass_two(op_array);
624 zend_oparray_context_end(&original_oparray_context);
625 zend_file_context_end(&original_file_context);
626
627 CG(active_op_array) = original_active_op_array;
628 }
629
630 zend_ast_destroy(CG(ast));
631 zend_arena_destroy(CG(ast_arena));
632
633 CG(in_compilation) = original_in_compilation;
634
635 return op_array;
636 }
637
compile_file(zend_file_handle * file_handle,int type)638 ZEND_API zend_op_array *compile_file(zend_file_handle *file_handle, int type)
639 {
640 zend_lex_state original_lex_state;
641 zend_op_array *op_array = NULL;
642 zend_save_lexical_state(&original_lex_state);
643
644 if (open_file_for_scanning(file_handle)==FAILURE) {
645 if (!EG(exception)) {
646 if (type==ZEND_REQUIRE) {
647 zend_message_dispatcher(ZMSG_FAILED_REQUIRE_FOPEN, ZSTR_VAL(file_handle->filename));
648 } else {
649 zend_message_dispatcher(ZMSG_FAILED_INCLUDE_FOPEN, ZSTR_VAL(file_handle->filename));
650 }
651 }
652 } else {
653 op_array = zend_compile(ZEND_USER_FUNCTION);
654 }
655
656 zend_restore_lexical_state(&original_lex_state);
657 return op_array;
658 }
659
zend_compile_string_to_ast(zend_string * code,zend_arena ** ast_arena,zend_string * filename)660 ZEND_API zend_ast *zend_compile_string_to_ast(
661 zend_string *code, zend_arena **ast_arena, zend_string *filename) {
662 zval code_zv;
663 bool original_in_compilation;
664 zend_lex_state original_lex_state;
665 zend_ast *ast;
666
667 ZVAL_STR_COPY(&code_zv, code);
668
669 original_in_compilation = CG(in_compilation);
670 CG(in_compilation) = 1;
671
672 zend_save_lexical_state(&original_lex_state);
673 zend_prepare_string_for_scanning(&code_zv, filename);
674 CG(ast) = NULL;
675 CG(ast_arena) = zend_arena_create(1024 * 32);
676 LANG_SCNG(yy_state) = yycINITIAL;
677
678 if (zendparse() != 0) {
679 zend_ast_destroy(CG(ast));
680 zend_arena_destroy(CG(ast_arena));
681 CG(ast) = NULL;
682 }
683
684 /* restore_lexical_state changes CG(ast) and CG(ast_arena) */
685 ast = CG(ast);
686 *ast_arena = CG(ast_arena);
687
688 zend_restore_lexical_state(&original_lex_state);
689 CG(in_compilation) = original_in_compilation;
690
691 zval_ptr_dtor_str(&code_zv);
692
693 return ast;
694 }
695
compile_filename(int type,zend_string * filename)696 zend_op_array *compile_filename(int type, zend_string *filename)
697 {
698 zend_file_handle file_handle;
699 zend_op_array *retval;
700 zend_string *opened_path = NULL;
701
702 zend_stream_init_filename_ex(&file_handle, filename);
703
704 retval = zend_compile_file(&file_handle, type);
705 if (retval && file_handle.handle.stream.handle) {
706 if (!file_handle.opened_path) {
707 file_handle.opened_path = opened_path = zend_string_copy(filename);
708 }
709
710 zend_hash_add_empty_element(&EG(included_files), file_handle.opened_path);
711
712 if (opened_path) {
713 zend_string_release_ex(opened_path, 0);
714 }
715 }
716 zend_destroy_file_handle(&file_handle);
717
718 return retval;
719 }
720
zend_prepare_string_for_scanning(zval * str,zend_string * filename)721 ZEND_API void zend_prepare_string_for_scanning(zval *str, zend_string *filename)
722 {
723 char *buf;
724 size_t size, old_len;
725
726 /* enforce ZEND_MMAP_AHEAD trailing NULLs for flex... */
727 old_len = Z_STRLEN_P(str);
728 Z_STR_P(str) = zend_string_extend(Z_STR_P(str), old_len + ZEND_MMAP_AHEAD, 0);
729 Z_TYPE_INFO_P(str) = IS_STRING_EX;
730 memset(Z_STRVAL_P(str) + old_len, 0, ZEND_MMAP_AHEAD + 1);
731
732 SCNG(yy_in) = NULL;
733 SCNG(yy_start) = NULL;
734
735 buf = Z_STRVAL_P(str);
736 size = old_len;
737
738 if (CG(multibyte)) {
739 SCNG(script_org) = (unsigned char*)buf;
740 SCNG(script_org_size) = size;
741 SCNG(script_filtered) = NULL;
742
743 zend_multibyte_set_filter(zend_multibyte_get_internal_encoding());
744
745 if (SCNG(input_filter)) {
746 if ((size_t)-1 == SCNG(input_filter)(&SCNG(script_filtered), &SCNG(script_filtered_size), SCNG(script_org), SCNG(script_org_size))) {
747 zend_error_noreturn(E_COMPILE_ERROR, "Could not convert the script from the detected "
748 "encoding \"%s\" to a compatible encoding", zend_multibyte_get_encoding_name(LANG_SCNG(script_encoding)));
749 }
750 buf = (char*)SCNG(script_filtered);
751 size = SCNG(script_filtered_size);
752 }
753 }
754
755 yy_scan_buffer(buf, size);
756 zend_set_compiled_filename(filename);
757 CG(zend_lineno) = 1;
758 CG(increment_lineno) = 0;
759 RESET_DOC_COMMENT();
760 }
761
762
zend_get_scanned_file_offset(void)763 ZEND_API size_t zend_get_scanned_file_offset(void)
764 {
765 size_t offset = SCNG(yy_cursor) - SCNG(yy_start);
766 if (SCNG(input_filter)) {
767 size_t original_offset = offset, length = 0;
768 do {
769 unsigned char *p = NULL;
770 if ((size_t)-1 == SCNG(input_filter)(&p, &length, SCNG(script_org), offset)) {
771 return (size_t)-1;
772 }
773 efree(p);
774 if (length > original_offset) {
775 offset--;
776 } else if (length < original_offset) {
777 offset++;
778 }
779 } while (original_offset != length);
780 }
781 return offset;
782 }
783
compile_string(zend_string * source_string,const char * filename,zend_compile_position position)784 zend_op_array *compile_string(zend_string *source_string, const char *filename, zend_compile_position position)
785 {
786 zend_lex_state original_lex_state;
787 zend_op_array *op_array = NULL;
788 zval tmp;
789 zend_string *filename_str;
790
791 if (ZSTR_LEN(source_string) == 0) {
792 return NULL;
793 }
794
795 ZVAL_STR_COPY(&tmp, source_string);
796
797 zend_save_lexical_state(&original_lex_state);
798 filename_str = zend_string_init(filename, strlen(filename), 0);
799 zend_prepare_string_for_scanning(&tmp, filename_str);
800 zend_string_release(filename_str);
801
802 switch (position) {
803 case ZEND_COMPILE_POSITION_AT_SHEBANG:
804 BEGIN(SHEBANG);
805 break;
806 case ZEND_COMPILE_POSITION_AT_OPEN_TAG:
807 BEGIN(INITIAL);
808 break;
809 case ZEND_COMPILE_POSITION_AFTER_OPEN_TAG:
810 BEGIN(ST_IN_SCRIPTING);
811 break;
812 }
813
814 op_array = zend_compile(ZEND_EVAL_CODE);
815
816 zend_restore_lexical_state(&original_lex_state);
817 zval_ptr_dtor(&tmp);
818
819 return op_array;
820 }
821
822
highlight_file(const char * filename,zend_syntax_highlighter_ini * syntax_highlighter_ini)823 zend_result highlight_file(const char *filename, zend_syntax_highlighter_ini *syntax_highlighter_ini)
824 {
825 zend_lex_state original_lex_state;
826 zend_file_handle file_handle;
827
828 zend_stream_init_filename(&file_handle, filename);
829 zend_save_lexical_state(&original_lex_state);
830 if (open_file_for_scanning(&file_handle)==FAILURE) {
831 zend_message_dispatcher(ZMSG_FAILED_HIGHLIGHT_FOPEN, filename);
832 zend_destroy_file_handle(&file_handle);
833 zend_restore_lexical_state(&original_lex_state);
834 return FAILURE;
835 }
836 zend_highlight(syntax_highlighter_ini);
837 if (SCNG(script_filtered)) {
838 efree(SCNG(script_filtered));
839 SCNG(script_filtered) = NULL;
840 }
841 zend_destroy_file_handle(&file_handle);
842 zend_restore_lexical_state(&original_lex_state);
843 return SUCCESS;
844 }
845
highlight_string(zend_string * str,zend_syntax_highlighter_ini * syntax_highlighter_ini,const char * filename)846 void highlight_string(zend_string *str, zend_syntax_highlighter_ini *syntax_highlighter_ini, const char *filename)
847 {
848 zend_lex_state original_lex_state;
849 zval str_zv;
850 zend_string *filename_str = zend_string_init(filename, strlen(filename), 0);
851 ZVAL_STR_COPY(&str_zv, str);
852 zend_save_lexical_state(&original_lex_state);
853 zend_prepare_string_for_scanning(&str_zv, filename_str);
854 zend_string_release(filename_str);
855 BEGIN(INITIAL);
856 zend_highlight(syntax_highlighter_ini);
857 if (SCNG(script_filtered)) {
858 efree(SCNG(script_filtered));
859 SCNG(script_filtered) = NULL;
860 }
861 zend_restore_lexical_state(&original_lex_state);
862 zval_ptr_dtor(&str_zv);
863 }
864
zend_multibyte_yyinput_again(zend_encoding_filter old_input_filter,const zend_encoding * old_encoding)865 ZEND_API void zend_multibyte_yyinput_again(zend_encoding_filter old_input_filter, const zend_encoding *old_encoding)
866 {
867 size_t length;
868 unsigned char *new_yy_start;
869
870 /* convert and set */
871 if (!SCNG(input_filter)) {
872 if (SCNG(script_filtered)) {
873 efree(SCNG(script_filtered));
874 SCNG(script_filtered) = NULL;
875 }
876 SCNG(script_filtered_size) = 0;
877 length = SCNG(script_org_size);
878 new_yy_start = SCNG(script_org);
879 } else {
880 if ((size_t)-1 == SCNG(input_filter)(&new_yy_start, &length, SCNG(script_org), SCNG(script_org_size))) {
881 zend_error_noreturn(E_COMPILE_ERROR, "Could not convert the script from the detected "
882 "encoding \"%s\" to a compatible encoding", zend_multibyte_get_encoding_name(LANG_SCNG(script_encoding)));
883 }
884 if (SCNG(script_filtered)) {
885 efree(SCNG(script_filtered));
886 }
887 SCNG(script_filtered) = new_yy_start;
888 SCNG(script_filtered_size) = length;
889 }
890
891 SCNG(yy_cursor) = new_yy_start + (SCNG(yy_cursor) - SCNG(yy_start));
892 SCNG(yy_marker) = new_yy_start + (SCNG(yy_marker) - SCNG(yy_start));
893 SCNG(yy_text) = new_yy_start + (SCNG(yy_text) - SCNG(yy_start));
894 SCNG(yy_limit) = new_yy_start + length;
895
896 SCNG(yy_start) = new_yy_start;
897 }
898
899
900 // TODO: avoid reallocation ???
901 # define zend_copy_value(zendlval, yytext, yyleng) \
902 if (SCNG(output_filter)) { \
903 size_t sz = 0; \
904 char *s = NULL; \
905 SCNG(output_filter)((unsigned char **)&s, &sz, (unsigned char *)yytext, (size_t)yyleng); \
906 ZVAL_STRINGL(zendlval, s, sz); \
907 efree(s); \
908 } else if (yyleng == 1) { \
909 ZVAL_INTERNED_STR(zendlval, ZSTR_CHAR((zend_uchar)*(yytext))); \
910 } else { \
911 ZVAL_STRINGL(zendlval, yytext, yyleng); \
912 }
913
zend_scan_escape_string(zval * zendlval,char * str,int len,char quote_type)914 static zend_result zend_scan_escape_string(zval *zendlval, char *str, int len, char quote_type)
915 {
916 char *s, *t;
917 char *end;
918
919 if (len <= 1) {
920 if (len < 1) {
921 ZVAL_EMPTY_STRING(zendlval);
922 } else {
923 zend_uchar c = (zend_uchar)*str;
924 if (c == '\n' || c == '\r') {
925 CG(zend_lineno)++;
926 }
927 ZVAL_INTERNED_STR(zendlval, ZSTR_CHAR(c));
928 }
929 goto skip_escape_conversion;
930 }
931
932 ZVAL_STRINGL(zendlval, str, len);
933
934 /* convert escape sequences */
935 s = Z_STRVAL_P(zendlval);
936 end = s+Z_STRLEN_P(zendlval);
937 while (1) {
938 if (UNEXPECTED(*s=='\\')) {
939 break;
940 }
941 if (*s == '\n' || (*s == '\r' && (*(s+1) != '\n'))) {
942 CG(zend_lineno)++;
943 }
944 s++;
945 if (s == end) {
946 goto skip_escape_conversion;
947 }
948 }
949
950 t = s;
951 while (s<end) {
952 if (*s=='\\') {
953 s++;
954 if (s >= end) {
955 *t++ = '\\';
956 break;
957 }
958
959 switch(*s) {
960 case 'n':
961 *t++ = '\n';
962 break;
963 case 'r':
964 *t++ = '\r';
965 break;
966 case 't':
967 *t++ = '\t';
968 break;
969 case 'f':
970 *t++ = '\f';
971 break;
972 case 'v':
973 *t++ = '\v';
974 break;
975 case 'e':
976 #ifdef ZEND_WIN32
977 *t++ = VK_ESCAPE;
978 #else
979 *t++ = '\e';
980 #endif
981 break;
982 case '"':
983 case '`':
984 if (*s != quote_type) {
985 *t++ = '\\';
986 *t++ = *s;
987 break;
988 }
989 ZEND_FALLTHROUGH;
990 case '\\':
991 case '$':
992 *t++ = *s;
993 break;
994 case 'x':
995 case 'X':
996 if (ZEND_IS_HEX(*(s+1))) {
997 char hex_buf[3] = { 0, 0, 0 };
998
999 hex_buf[0] = *(++s);
1000 if (ZEND_IS_HEX(*(s+1))) {
1001 hex_buf[1] = *(++s);
1002 }
1003 *t++ = (char) ZEND_STRTOL(hex_buf, NULL, 16);
1004 } else {
1005 *t++ = '\\';
1006 *t++ = *s;
1007 }
1008 break;
1009 /* UTF-8 codepoint escape, format: /\\u\{\x+\}/ */
1010 case 'u':
1011 {
1012 /* cache where we started so we can parse after validating */
1013 char *start = s + 1;
1014 size_t len = 0;
1015 bool valid = 1;
1016 unsigned long codepoint;
1017
1018 if (*start != '{') {
1019 /* we silently let this pass to avoid breaking code
1020 * with JSON in string literals (e.g. "\"\u202e\""
1021 */
1022 *t++ = '\\';
1023 *t++ = 'u';
1024 break;
1025 } else {
1026 /* on the other hand, invalid \u{blah} errors */
1027 s++;
1028 len++;
1029 s++;
1030 while (*s != '}') {
1031 if (!ZEND_IS_HEX(*s)) {
1032 valid = 0;
1033 break;
1034 } else {
1035 len++;
1036 }
1037 s++;
1038 }
1039 if (*s == '}') {
1040 valid = 1;
1041 len++;
1042 }
1043 }
1044
1045 /* \u{} is invalid */
1046 if (len <= 2) {
1047 valid = 0;
1048 }
1049
1050 if (!valid) {
1051 zend_throw_exception(zend_ce_parse_error,
1052 "Invalid UTF-8 codepoint escape sequence", 0);
1053 zval_ptr_dtor(zendlval);
1054 ZVAL_UNDEF(zendlval);
1055 return FAILURE;
1056 }
1057
1058 errno = 0;
1059 codepoint = strtoul(start + 1, NULL, 16);
1060
1061 /* per RFC 3629, UTF-8 can only represent 21 bits */
1062 if (codepoint > 0x10FFFF || errno) {
1063 zend_throw_exception(zend_ce_parse_error,
1064 "Invalid UTF-8 codepoint escape sequence: Codepoint too large", 0);
1065 zval_ptr_dtor(zendlval);
1066 ZVAL_UNDEF(zendlval);
1067 return FAILURE;
1068 }
1069
1070 /* based on https://en.wikipedia.org/wiki/UTF-8#Sample_code */
1071 if (codepoint < 0x80) {
1072 *t++ = codepoint;
1073 } else if (codepoint <= 0x7FF) {
1074 *t++ = (codepoint >> 6) + 0xC0;
1075 *t++ = (codepoint & 0x3F) + 0x80;
1076 } else if (codepoint <= 0xFFFF) {
1077 *t++ = (codepoint >> 12) + 0xE0;
1078 *t++ = ((codepoint >> 6) & 0x3F) + 0x80;
1079 *t++ = (codepoint & 0x3F) + 0x80;
1080 } else if (codepoint <= 0x10FFFF) {
1081 *t++ = (codepoint >> 18) + 0xF0;
1082 *t++ = ((codepoint >> 12) & 0x3F) + 0x80;
1083 *t++ = ((codepoint >> 6) & 0x3F) + 0x80;
1084 *t++ = (codepoint & 0x3F) + 0x80;
1085 }
1086 }
1087 break;
1088 default:
1089 /* check for an octal */
1090 if (ZEND_IS_OCT(*s)) {
1091 char octal_buf[4] = { 0, 0, 0, 0 };
1092
1093 octal_buf[0] = *s;
1094 if (ZEND_IS_OCT(*(s+1))) {
1095 octal_buf[1] = *(++s);
1096 if (ZEND_IS_OCT(*(s+1))) {
1097 octal_buf[2] = *(++s);
1098 }
1099 }
1100 if (octal_buf[2] && (octal_buf[0] > '3') && !SCNG(heredoc_scan_ahead)) {
1101 /* 3 octit values must not overflow 0xFF (\377) */
1102 zend_error(E_COMPILE_WARNING, "Octal escape sequence overflow \\%s is greater than \\377", octal_buf);
1103 }
1104
1105 *t++ = (char) ZEND_STRTOL(octal_buf, NULL, 8);
1106 } else {
1107 *t++ = '\\';
1108 *t++ = *s;
1109 }
1110 break;
1111 }
1112 } else {
1113 *t++ = *s;
1114 }
1115
1116 if (*s == '\n' || (*s == '\r' && (*(s+1) != '\n'))) {
1117 CG(zend_lineno)++;
1118 }
1119 s++;
1120 }
1121 *t = 0;
1122 Z_STRLEN_P(zendlval) = t - Z_STRVAL_P(zendlval);
1123
1124 skip_escape_conversion:
1125 if (SCNG(output_filter)) {
1126 size_t sz = 0;
1127 unsigned char *str;
1128 // TODO: avoid realocation ???
1129 s = Z_STRVAL_P(zendlval);
1130 SCNG(output_filter)(&str, &sz, (unsigned char *)s, (size_t)Z_STRLEN_P(zendlval));
1131 zval_ptr_dtor(zendlval);
1132 ZVAL_STRINGL(zendlval, (char *) str, sz);
1133 efree(str);
1134 }
1135 return SUCCESS;
1136 }
1137
1138 #define HEREDOC_USING_SPACES 1
1139 #define HEREDOC_USING_TABS 2
1140
next_newline(const char * str,const char * end,size_t * newline_len)1141 static const char *next_newline(const char *str, const char *end, size_t *newline_len) {
1142 for (; str < end; str++) {
1143 if (*str == '\r') {
1144 *newline_len = str + 1 < end && *(str + 1) == '\n' ? 2 : 1;
1145 return str;
1146 } else if (*str == '\n') {
1147 *newline_len = 1;
1148 return str;
1149 }
1150 }
1151 *newline_len = 0;
1152 return NULL;
1153 }
1154
strip_multiline_string_indentation(zval * zendlval,int indentation,bool using_spaces,bool newline_at_start,bool newline_at_end)1155 static bool strip_multiline_string_indentation(
1156 zval *zendlval, int indentation, bool using_spaces,
1157 bool newline_at_start, bool newline_at_end)
1158 {
1159 const char *str = Z_STRVAL_P(zendlval), *end = str + Z_STRLEN_P(zendlval);
1160 char *copy = Z_STRVAL_P(zendlval);
1161
1162 int newline_count = 0;
1163 size_t newline_len;
1164 const char *nl;
1165
1166 if (!newline_at_start) {
1167 nl = next_newline(str, end, &newline_len);
1168 if (!nl) {
1169 return 1;
1170 }
1171
1172 str = nl + newline_len;
1173 copy = (char *) nl + newline_len;
1174 newline_count++;
1175 } else {
1176 nl = str;
1177 }
1178
1179 /* <= intentional */
1180 while (str <= end && nl) {
1181 size_t skip;
1182 nl = next_newline(str, end, &newline_len);
1183 if (!nl && newline_at_end) {
1184 nl = end;
1185 }
1186
1187 /* Try to skip indentation */
1188 for (skip = 0; skip < indentation; skip++, str++) {
1189 if (str == nl) {
1190 /* Don't require full indentation on whitespace-only lines */
1191 break;
1192 }
1193
1194 if (str == end || (*str != ' ' && *str != '\t')) {
1195 CG(zend_lineno) += newline_count;
1196 zend_throw_exception_ex(zend_ce_parse_error, 0,
1197 "Invalid body indentation level (expecting an indentation level of at least %d)", indentation);
1198 goto error;
1199 }
1200
1201 if ((!using_spaces && *str == ' ') || (using_spaces && *str == '\t')) {
1202 CG(zend_lineno) += newline_count;
1203 zend_throw_exception(zend_ce_parse_error,
1204 "Invalid indentation - tabs and spaces cannot be mixed", 0);
1205 goto error;
1206 }
1207 }
1208
1209 if (str == end) {
1210 break;
1211 }
1212
1213 size_t len = nl ? (nl - str + newline_len) : (end - str);
1214 memmove(copy, str, len);
1215 str += len;
1216 copy += len;
1217 newline_count++;
1218 }
1219
1220 *copy = '\0';
1221 Z_STRLEN_P(zendlval) = copy - Z_STRVAL_P(zendlval);
1222 return 1;
1223
1224 error:
1225 zval_ptr_dtor_str(zendlval);
1226 ZVAL_UNDEF(zendlval);
1227
1228 return 0;
1229 }
1230
copy_heredoc_label_stack(void * void_heredoc_label)1231 static void copy_heredoc_label_stack(void *void_heredoc_label)
1232 {
1233 zend_heredoc_label *heredoc_label = void_heredoc_label;
1234 zend_heredoc_label *new_heredoc_label = emalloc(sizeof(zend_heredoc_label));
1235
1236 *new_heredoc_label = *heredoc_label;
1237 new_heredoc_label->label = estrndup(heredoc_label->label, heredoc_label->length);
1238
1239 zend_ptr_stack_push(&SCNG(heredoc_label_stack), (void *) new_heredoc_label);
1240 }
1241
1242 /* Check that { }, [ ], ( ) are nested correctly */
report_bad_nesting(char opening,int opening_lineno,char closing)1243 static void report_bad_nesting(char opening, int opening_lineno, char closing)
1244 {
1245 char buf[256];
1246 size_t used = 0;
1247
1248 used = snprintf(buf, sizeof(buf), "Unclosed '%c'", opening);
1249
1250 if (opening_lineno != CG(zend_lineno)) {
1251 used += snprintf(buf + used, sizeof(buf) - used, " on line %d", opening_lineno);
1252 }
1253
1254 if (closing) { /* 'closing' will be 0 if at end of file */
1255 used += snprintf(buf + used, sizeof(buf) - used, " does not match '%c'", closing);
1256 }
1257
1258 zend_throw_exception(zend_ce_parse_error, buf, 0);
1259 }
1260
enter_nesting(char opening)1261 static void enter_nesting(char opening)
1262 {
1263 zend_nest_location nest_loc = {opening, CG(zend_lineno)};
1264 zend_stack_push(&SCNG(nest_location_stack), &nest_loc);
1265 }
1266
exit_nesting(char closing)1267 static zend_result exit_nesting(char closing)
1268 {
1269 if (zend_stack_is_empty(&SCNG(nest_location_stack))) {
1270 zend_throw_exception_ex(zend_ce_parse_error, 0, "Unmatched '%c'", closing);
1271 return FAILURE;
1272 }
1273
1274 zend_nest_location *nest_loc = zend_stack_top(&SCNG(nest_location_stack));
1275 char opening = nest_loc->text;
1276
1277 if ((opening == '{' && closing != '}') ||
1278 (opening == '[' && closing != ']') ||
1279 (opening == '(' && closing != ')')) {
1280 report_bad_nesting(opening, nest_loc->lineno, closing);
1281 return FAILURE;
1282 }
1283
1284 zend_stack_del_top(&SCNG(nest_location_stack));
1285 return SUCCESS;
1286 }
1287
check_nesting_at_end(void)1288 static zend_result check_nesting_at_end(void)
1289 {
1290 if (!zend_stack_is_empty(&SCNG(nest_location_stack))) {
1291 zend_nest_location *nest_loc = zend_stack_top(&SCNG(nest_location_stack));
1292 report_bad_nesting(nest_loc->text, nest_loc->lineno, 0);
1293 return FAILURE;
1294 }
1295
1296 return SUCCESS;
1297 }
1298
1299 #define PARSER_MODE() \
1300 EXPECTED(elem != NULL)
1301
1302 #define RETURN_TOKEN(_token) do { \
1303 token = _token; \
1304 goto emit_token; \
1305 } while (0)
1306
1307 #define RETURN_TOKEN_WITH_VAL(_token) do { \
1308 token = _token; \
1309 goto emit_token_with_val; \
1310 } while (0)
1311
1312 #define RETURN_TOKEN_WITH_STR(_token, _offset) do { \
1313 token = _token; \
1314 offset = _offset; \
1315 goto emit_token_with_str; \
1316 } while (0)
1317
1318 #define RETURN_TOKEN_WITH_IDENT(_token) do { \
1319 token = _token; \
1320 goto emit_token_with_ident; \
1321 } while (0)
1322
1323 #define RETURN_OR_SKIP_TOKEN(_token) do { \
1324 token = _token; \
1325 if (PARSER_MODE()) { \
1326 goto skip_token; \
1327 } \
1328 goto emit_token; \
1329 } while (0)
1330
1331 #define RETURN_EXIT_NESTING_TOKEN(_token) do { \
1332 if (exit_nesting(_token) && PARSER_MODE()) { \
1333 RETURN_TOKEN(T_ERROR); \
1334 } else { \
1335 RETURN_TOKEN(_token); \
1336 } \
1337 } while(0)
1338
1339 #define RETURN_END_TOKEN do { \
1340 if (check_nesting_at_end() && PARSER_MODE()) { \
1341 RETURN_TOKEN(T_ERROR); \
1342 } else { \
1343 RETURN_TOKEN(END); \
1344 } \
1345 } while (0)
1346
lex_scan(zval * zendlval,zend_parser_stack_elem * elem)1347 int ZEND_FASTCALL lex_scan(zval *zendlval, zend_parser_stack_elem *elem)
1348 {
1349 int token;
1350 int offset;
1351 int start_line = CG(zend_lineno);
1352
1353 ZVAL_UNDEF(zendlval);
1354 restart:
1355 SCNG(yy_text) = YYCURSOR;
1356
1357 /*!re2c
1358 re2c:yyfill:check = 0;
1359 LNUM [0-9]+(_[0-9]+)*
1360 DNUM ({LNUM}?"."{LNUM})|({LNUM}"."{LNUM}?)
1361 EXPONENT_DNUM (({LNUM}|{DNUM})[eE][+-]?{LNUM})
1362 HNUM "0x"[0-9a-fA-F]+(_[0-9a-fA-F]+)*
1363 BNUM "0b"[01]+(_[01]+)*
1364 ONUM "0o"[0-7]+(_[0-7]+)*
1365 LABEL [a-zA-Z_\x80-\xff][a-zA-Z0-9_\x80-\xff]*
1366 WHITESPACE [ \n\r\t]+
1367 TABS_AND_SPACES [ \t]*
1368 TOKENS [;:,.|^&+-/*=%!~$<>?@]
1369 ANY_CHAR [^]
1370 NEWLINE ("\r"|"\n"|"\r\n")
1371
1372 /* compute yyleng before each rule */
1373 <!*> := yyleng = YYCURSOR - SCNG(yy_text);
1374
1375 <ST_IN_SCRIPTING>"exit" {
1376 RETURN_TOKEN_WITH_IDENT(T_EXIT);
1377 }
1378
1379 <ST_IN_SCRIPTING>"die" {
1380 RETURN_TOKEN_WITH_IDENT(T_EXIT);
1381 }
1382
1383 <ST_IN_SCRIPTING>"fn" {
1384 RETURN_TOKEN_WITH_IDENT(T_FN);
1385 }
1386
1387 <ST_IN_SCRIPTING>"function" {
1388 RETURN_TOKEN_WITH_IDENT(T_FUNCTION);
1389 }
1390
1391 <ST_IN_SCRIPTING>"const" {
1392 RETURN_TOKEN_WITH_IDENT(T_CONST);
1393 }
1394
1395 <ST_IN_SCRIPTING>"return" {
1396 RETURN_TOKEN_WITH_IDENT(T_RETURN);
1397 }
1398
1399 <ST_IN_SCRIPTING>"#[" {
1400 enter_nesting('[');
1401 RETURN_TOKEN(T_ATTRIBUTE);
1402 }
1403
1404 <ST_IN_SCRIPTING>"yield"{WHITESPACE}"from"[^a-zA-Z0-9_\x80-\xff] {
1405 yyless(yyleng - 1);
1406 HANDLE_NEWLINES(yytext, yyleng);
1407 RETURN_TOKEN_WITH_IDENT(T_YIELD_FROM);
1408 }
1409
1410 <ST_IN_SCRIPTING>"yield" {
1411 RETURN_TOKEN_WITH_IDENT(T_YIELD);
1412 }
1413
1414 <ST_IN_SCRIPTING>"try" {
1415 RETURN_TOKEN_WITH_IDENT(T_TRY);
1416 }
1417
1418 <ST_IN_SCRIPTING>"catch" {
1419 RETURN_TOKEN_WITH_IDENT(T_CATCH);
1420 }
1421
1422 <ST_IN_SCRIPTING>"finally" {
1423 RETURN_TOKEN_WITH_IDENT(T_FINALLY);
1424 }
1425
1426 <ST_IN_SCRIPTING>"throw" {
1427 RETURN_TOKEN_WITH_IDENT(T_THROW);
1428 }
1429
1430 <ST_IN_SCRIPTING>"if" {
1431 RETURN_TOKEN_WITH_IDENT(T_IF);
1432 }
1433
1434 <ST_IN_SCRIPTING>"elseif" {
1435 RETURN_TOKEN_WITH_IDENT(T_ELSEIF);
1436 }
1437
1438 <ST_IN_SCRIPTING>"endif" {
1439 RETURN_TOKEN_WITH_IDENT(T_ENDIF);
1440 }
1441
1442 <ST_IN_SCRIPTING>"else" {
1443 RETURN_TOKEN_WITH_IDENT(T_ELSE);
1444 }
1445
1446 <ST_IN_SCRIPTING>"while" {
1447 RETURN_TOKEN_WITH_IDENT(T_WHILE);
1448 }
1449
1450 <ST_IN_SCRIPTING>"endwhile" {
1451 RETURN_TOKEN_WITH_IDENT(T_ENDWHILE);
1452 }
1453
1454 <ST_IN_SCRIPTING>"do" {
1455 RETURN_TOKEN_WITH_IDENT(T_DO);
1456 }
1457
1458 <ST_IN_SCRIPTING>"for" {
1459 RETURN_TOKEN_WITH_IDENT(T_FOR);
1460 }
1461
1462 <ST_IN_SCRIPTING>"endfor" {
1463 RETURN_TOKEN_WITH_IDENT(T_ENDFOR);
1464 }
1465
1466 <ST_IN_SCRIPTING>"foreach" {
1467 RETURN_TOKEN_WITH_IDENT(T_FOREACH);
1468 }
1469
1470 <ST_IN_SCRIPTING>"endforeach" {
1471 RETURN_TOKEN_WITH_IDENT(T_ENDFOREACH);
1472 }
1473
1474 <ST_IN_SCRIPTING>"declare" {
1475 RETURN_TOKEN_WITH_IDENT(T_DECLARE);
1476 }
1477
1478 <ST_IN_SCRIPTING>"enddeclare" {
1479 RETURN_TOKEN_WITH_IDENT(T_ENDDECLARE);
1480 }
1481
1482 <ST_IN_SCRIPTING>"instanceof" {
1483 RETURN_TOKEN_WITH_IDENT(T_INSTANCEOF);
1484 }
1485
1486 <ST_IN_SCRIPTING>"as" {
1487 RETURN_TOKEN_WITH_IDENT(T_AS);
1488 }
1489
1490 <ST_IN_SCRIPTING>"switch" {
1491 RETURN_TOKEN_WITH_IDENT(T_SWITCH);
1492 }
1493
1494 <ST_IN_SCRIPTING>"match" {
1495 RETURN_TOKEN_WITH_IDENT(T_MATCH);
1496 }
1497
1498 <ST_IN_SCRIPTING>"endswitch" {
1499 RETURN_TOKEN_WITH_IDENT(T_ENDSWITCH);
1500 }
1501
1502 <ST_IN_SCRIPTING>"case" {
1503 RETURN_TOKEN_WITH_IDENT(T_CASE);
1504 }
1505
1506 <ST_IN_SCRIPTING>"default" {
1507 RETURN_TOKEN_WITH_IDENT(T_DEFAULT);
1508 }
1509
1510 <ST_IN_SCRIPTING>"break" {
1511 RETURN_TOKEN_WITH_IDENT(T_BREAK);
1512 }
1513
1514 <ST_IN_SCRIPTING>"continue" {
1515 RETURN_TOKEN_WITH_IDENT(T_CONTINUE);
1516 }
1517
1518 <ST_IN_SCRIPTING>"goto" {
1519 RETURN_TOKEN_WITH_IDENT(T_GOTO);
1520 }
1521
1522 <ST_IN_SCRIPTING>"echo" {
1523 RETURN_TOKEN_WITH_IDENT(T_ECHO);
1524 }
1525
1526 <ST_IN_SCRIPTING>"print" {
1527 RETURN_TOKEN_WITH_IDENT(T_PRINT);
1528 }
1529
1530 <ST_IN_SCRIPTING>"class" {
1531 RETURN_TOKEN_WITH_IDENT(T_CLASS);
1532 }
1533
1534 <ST_IN_SCRIPTING>"interface" {
1535 RETURN_TOKEN_WITH_IDENT(T_INTERFACE);
1536 }
1537
1538 <ST_IN_SCRIPTING>"trait" {
1539 RETURN_TOKEN_WITH_IDENT(T_TRAIT);
1540 }
1541
1542 /*
1543 * The enum keyword must be followed by whitespace and another identifier.
1544 * This avoids the BC break of using enum in classes, namespaces, functions and constants.
1545 */
1546 <ST_IN_SCRIPTING>"enum"{WHITESPACE}("extends"|"implements") {
1547 yyless(4);
1548 RETURN_TOKEN_WITH_STR(T_STRING, 0);
1549 }
1550 <ST_IN_SCRIPTING>"enum"{WHITESPACE}[a-zA-Z_\x80-\xff] {
1551 yyless(4);
1552 RETURN_TOKEN_WITH_IDENT(T_ENUM);
1553 }
1554
1555 <ST_IN_SCRIPTING>"extends" {
1556 RETURN_TOKEN_WITH_IDENT(T_EXTENDS);
1557 }
1558
1559 <ST_IN_SCRIPTING>"implements" {
1560 RETURN_TOKEN_WITH_IDENT(T_IMPLEMENTS);
1561 }
1562
1563 <ST_IN_SCRIPTING>"->" {
1564 yy_push_state(ST_LOOKING_FOR_PROPERTY);
1565 RETURN_TOKEN(T_OBJECT_OPERATOR);
1566 }
1567
1568 <ST_IN_SCRIPTING>"?->" {
1569 yy_push_state(ST_LOOKING_FOR_PROPERTY);
1570 RETURN_TOKEN(T_NULLSAFE_OBJECT_OPERATOR);
1571 }
1572
1573 <ST_IN_SCRIPTING,ST_LOOKING_FOR_PROPERTY>{WHITESPACE}+ {
1574 goto return_whitespace;
1575 }
1576
1577 <ST_LOOKING_FOR_PROPERTY>"->" {
1578 RETURN_TOKEN(T_OBJECT_OPERATOR);
1579 }
1580
1581 <ST_LOOKING_FOR_PROPERTY>"?->" {
1582 RETURN_TOKEN(T_NULLSAFE_OBJECT_OPERATOR);
1583 }
1584
1585 <ST_LOOKING_FOR_PROPERTY>{LABEL} {
1586 yy_pop_state();
1587 RETURN_TOKEN_WITH_STR(T_STRING, 0);
1588 }
1589
1590 <ST_IN_SCRIPTING>"::" {
1591 RETURN_TOKEN(T_PAAMAYIM_NEKUDOTAYIM);
1592 }
1593
1594 <ST_IN_SCRIPTING>"..." {
1595 RETURN_TOKEN(T_ELLIPSIS);
1596 }
1597
1598 <ST_IN_SCRIPTING>"??" {
1599 RETURN_TOKEN(T_COALESCE);
1600 }
1601
1602 <ST_IN_SCRIPTING>"new" {
1603 RETURN_TOKEN_WITH_IDENT(T_NEW);
1604 }
1605
1606 <ST_IN_SCRIPTING>"clone" {
1607 RETURN_TOKEN_WITH_IDENT(T_CLONE);
1608 }
1609
1610 <ST_IN_SCRIPTING>"var" {
1611 RETURN_TOKEN_WITH_IDENT(T_VAR);
1612 }
1613
1614 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("int"|"integer"){TABS_AND_SPACES}")" {
1615 RETURN_TOKEN(T_INT_CAST);
1616 }
1617
1618 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("double"|"float"){TABS_AND_SPACES}")" {
1619 RETURN_TOKEN(T_DOUBLE_CAST);
1620 }
1621
1622 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}"real"{TABS_AND_SPACES}")" {
1623 if (PARSER_MODE()) {
1624 zend_throw_exception(zend_ce_parse_error, "The (real) cast has been removed, use (float) instead", 0);
1625 RETURN_TOKEN(T_ERROR);
1626 }
1627 RETURN_TOKEN(T_DOUBLE_CAST);
1628 }
1629
1630 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("string"|"binary"){TABS_AND_SPACES}")" {
1631 RETURN_TOKEN(T_STRING_CAST);
1632 }
1633
1634 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}"array"{TABS_AND_SPACES}")" {
1635 RETURN_TOKEN(T_ARRAY_CAST);
1636 }
1637
1638 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}"object"{TABS_AND_SPACES}")" {
1639 RETURN_TOKEN(T_OBJECT_CAST);
1640 }
1641
1642 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("bool"|"boolean"){TABS_AND_SPACES}")" {
1643 RETURN_TOKEN(T_BOOL_CAST);
1644 }
1645
1646 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("unset"){TABS_AND_SPACES}")" {
1647 RETURN_TOKEN(T_UNSET_CAST);
1648 }
1649
1650 <ST_IN_SCRIPTING>"eval" {
1651 RETURN_TOKEN_WITH_IDENT(T_EVAL);
1652 }
1653
1654 <ST_IN_SCRIPTING>"include" {
1655 RETURN_TOKEN_WITH_IDENT(T_INCLUDE);
1656 }
1657
1658 <ST_IN_SCRIPTING>"include_once" {
1659 RETURN_TOKEN_WITH_IDENT(T_INCLUDE_ONCE);
1660 }
1661
1662 <ST_IN_SCRIPTING>"require" {
1663 RETURN_TOKEN_WITH_IDENT(T_REQUIRE);
1664 }
1665
1666 <ST_IN_SCRIPTING>"require_once" {
1667 RETURN_TOKEN_WITH_IDENT(T_REQUIRE_ONCE);
1668 }
1669
1670 <ST_IN_SCRIPTING>"namespace" {
1671 RETURN_TOKEN_WITH_IDENT(T_NAMESPACE);
1672 }
1673
1674 <ST_IN_SCRIPTING>"use" {
1675 RETURN_TOKEN_WITH_IDENT(T_USE);
1676 }
1677
1678 <ST_IN_SCRIPTING>"insteadof" {
1679 RETURN_TOKEN_WITH_IDENT(T_INSTEADOF);
1680 }
1681
1682 <ST_IN_SCRIPTING>"global" {
1683 RETURN_TOKEN_WITH_IDENT(T_GLOBAL);
1684 }
1685
1686 <ST_IN_SCRIPTING>"isset" {
1687 RETURN_TOKEN_WITH_IDENT(T_ISSET);
1688 }
1689
1690 <ST_IN_SCRIPTING>"empty" {
1691 RETURN_TOKEN_WITH_IDENT(T_EMPTY);
1692 }
1693
1694 <ST_IN_SCRIPTING>"__halt_compiler" {
1695 RETURN_TOKEN_WITH_IDENT(T_HALT_COMPILER);
1696 }
1697
1698 <ST_IN_SCRIPTING>"static" {
1699 RETURN_TOKEN_WITH_IDENT(T_STATIC);
1700 }
1701
1702 <ST_IN_SCRIPTING>"abstract" {
1703 RETURN_TOKEN_WITH_IDENT(T_ABSTRACT);
1704 }
1705
1706 <ST_IN_SCRIPTING>"final" {
1707 RETURN_TOKEN_WITH_IDENT(T_FINAL);
1708 }
1709
1710 <ST_IN_SCRIPTING>"private" {
1711 RETURN_TOKEN_WITH_IDENT(T_PRIVATE);
1712 }
1713
1714 <ST_IN_SCRIPTING>"protected" {
1715 RETURN_TOKEN_WITH_IDENT(T_PROTECTED);
1716 }
1717
1718 <ST_IN_SCRIPTING>"public" {
1719 RETURN_TOKEN_WITH_IDENT(T_PUBLIC);
1720 }
1721
1722 <ST_IN_SCRIPTING>"readonly" {
1723 RETURN_TOKEN_WITH_IDENT(T_READONLY);
1724 }
1725
1726 <ST_IN_SCRIPTING>"unset" {
1727 RETURN_TOKEN_WITH_IDENT(T_UNSET);
1728 }
1729
1730 <ST_IN_SCRIPTING>"=>" {
1731 RETURN_TOKEN(T_DOUBLE_ARROW);
1732 }
1733
1734 <ST_IN_SCRIPTING>"list" {
1735 RETURN_TOKEN_WITH_IDENT(T_LIST);
1736 }
1737
1738 <ST_IN_SCRIPTING>"array" {
1739 RETURN_TOKEN_WITH_IDENT(T_ARRAY);
1740 }
1741
1742 <ST_IN_SCRIPTING>"callable" {
1743 RETURN_TOKEN_WITH_IDENT(T_CALLABLE);
1744 }
1745
1746 <ST_IN_SCRIPTING>"++" {
1747 RETURN_TOKEN(T_INC);
1748 }
1749
1750 <ST_IN_SCRIPTING>"--" {
1751 RETURN_TOKEN(T_DEC);
1752 }
1753
1754 <ST_IN_SCRIPTING>"===" {
1755 RETURN_TOKEN(T_IS_IDENTICAL);
1756 }
1757
1758 <ST_IN_SCRIPTING>"!==" {
1759 RETURN_TOKEN(T_IS_NOT_IDENTICAL);
1760 }
1761
1762 <ST_IN_SCRIPTING>"==" {
1763 RETURN_TOKEN(T_IS_EQUAL);
1764 }
1765
1766 <ST_IN_SCRIPTING>"!="|"<>" {
1767 RETURN_TOKEN(T_IS_NOT_EQUAL);
1768 }
1769
1770 <ST_IN_SCRIPTING>"<=>" {
1771 RETURN_TOKEN(T_SPACESHIP);
1772 }
1773
1774 <ST_IN_SCRIPTING>"<=" {
1775 RETURN_TOKEN(T_IS_SMALLER_OR_EQUAL);
1776 }
1777
1778 <ST_IN_SCRIPTING>">=" {
1779 RETURN_TOKEN(T_IS_GREATER_OR_EQUAL);
1780 }
1781
1782 <ST_IN_SCRIPTING>"+=" {
1783 RETURN_TOKEN(T_PLUS_EQUAL);
1784 }
1785
1786 <ST_IN_SCRIPTING>"-=" {
1787 RETURN_TOKEN(T_MINUS_EQUAL);
1788 }
1789
1790 <ST_IN_SCRIPTING>"*=" {
1791 RETURN_TOKEN(T_MUL_EQUAL);
1792 }
1793
1794 <ST_IN_SCRIPTING>"*\*" {
1795 RETURN_TOKEN(T_POW);
1796 }
1797
1798 <ST_IN_SCRIPTING>"*\*=" {
1799 RETURN_TOKEN(T_POW_EQUAL);
1800 }
1801
1802 <ST_IN_SCRIPTING>"/=" {
1803 RETURN_TOKEN(T_DIV_EQUAL);
1804 }
1805
1806 <ST_IN_SCRIPTING>".=" {
1807 RETURN_TOKEN(T_CONCAT_EQUAL);
1808 }
1809
1810 <ST_IN_SCRIPTING>"%=" {
1811 RETURN_TOKEN(T_MOD_EQUAL);
1812 }
1813
1814 <ST_IN_SCRIPTING>"<<=" {
1815 RETURN_TOKEN(T_SL_EQUAL);
1816 }
1817
1818 <ST_IN_SCRIPTING>">>=" {
1819 RETURN_TOKEN(T_SR_EQUAL);
1820 }
1821
1822 <ST_IN_SCRIPTING>"&=" {
1823 RETURN_TOKEN(T_AND_EQUAL);
1824 }
1825
1826 <ST_IN_SCRIPTING>"|=" {
1827 RETURN_TOKEN(T_OR_EQUAL);
1828 }
1829
1830 <ST_IN_SCRIPTING>"^=" {
1831 RETURN_TOKEN(T_XOR_EQUAL);
1832 }
1833
1834 <ST_IN_SCRIPTING>"??=" {
1835 RETURN_TOKEN(T_COALESCE_EQUAL);
1836 }
1837
1838 <ST_IN_SCRIPTING>"||" {
1839 RETURN_TOKEN(T_BOOLEAN_OR);
1840 }
1841
1842 <ST_IN_SCRIPTING>"&&" {
1843 RETURN_TOKEN(T_BOOLEAN_AND);
1844 }
1845
1846 <ST_IN_SCRIPTING>"OR" {
1847 RETURN_TOKEN_WITH_IDENT(T_LOGICAL_OR);
1848 }
1849
1850 <ST_IN_SCRIPTING>"AND" {
1851 RETURN_TOKEN_WITH_IDENT(T_LOGICAL_AND);
1852 }
1853
1854 <ST_IN_SCRIPTING>"XOR" {
1855 RETURN_TOKEN_WITH_IDENT(T_LOGICAL_XOR);
1856 }
1857
1858 <ST_IN_SCRIPTING>"<<" {
1859 RETURN_TOKEN(T_SL);
1860 }
1861
1862 <ST_IN_SCRIPTING>">>" {
1863 RETURN_TOKEN(T_SR);
1864 }
1865
1866 <ST_IN_SCRIPTING>"&"[ \t\r\n]*("$"|"...") {
1867 yyless(1);
1868 RETURN_TOKEN(T_AMPERSAND_FOLLOWED_BY_VAR_OR_VARARG);
1869 }
1870
1871 <ST_IN_SCRIPTING>"&" {
1872 RETURN_TOKEN(T_AMPERSAND_NOT_FOLLOWED_BY_VAR_OR_VARARG);
1873 }
1874
1875 <ST_IN_SCRIPTING>"]"|")" {
1876 /* Check that ] and ) match up properly with a preceding [ or ( */
1877 RETURN_EXIT_NESTING_TOKEN(yytext[0]);
1878 }
1879
1880 <ST_IN_SCRIPTING>"["|"(" {
1881 enter_nesting(yytext[0]);
1882 RETURN_TOKEN(yytext[0]);
1883 }
1884
1885 <ST_IN_SCRIPTING>{TOKENS} {
1886 RETURN_TOKEN(yytext[0]);
1887 }
1888
1889
1890 <ST_IN_SCRIPTING>"{" {
1891 yy_push_state(ST_IN_SCRIPTING);
1892 enter_nesting('{');
1893 RETURN_TOKEN('{');
1894 }
1895
1896
1897 <ST_DOUBLE_QUOTES,ST_BACKQUOTE,ST_HEREDOC>"${" {
1898 yy_push_state(ST_LOOKING_FOR_VARNAME);
1899 enter_nesting('{');
1900 RETURN_TOKEN(T_DOLLAR_OPEN_CURLY_BRACES);
1901 }
1902
1903 <ST_IN_SCRIPTING>"}" {
1904 RESET_DOC_COMMENT();
1905 if (!zend_stack_is_empty(&SCNG(state_stack))) {
1906 yy_pop_state();
1907 }
1908 RETURN_EXIT_NESTING_TOKEN('}');
1909 }
1910
1911
1912 <ST_LOOKING_FOR_VARNAME>{LABEL}[[}] {
1913 yyless(yyleng - 1);
1914 yy_pop_state();
1915 yy_push_state(ST_IN_SCRIPTING);
1916 RETURN_TOKEN_WITH_STR(T_STRING_VARNAME, 0);
1917 }
1918
1919
1920 <ST_LOOKING_FOR_VARNAME>{ANY_CHAR} {
1921 yyless(0);
1922 yy_pop_state();
1923 yy_push_state(ST_IN_SCRIPTING);
1924 goto restart;
1925 }
1926
1927 <ST_IN_SCRIPTING>{BNUM} {
1928 /* The +/- 2 skips "0b" */
1929 size_t len = yyleng - 2;
1930 char *end, *bin = yytext + 2;
1931 bool contains_underscores;
1932
1933 /* Skip any leading 0s */
1934 while (len > 0 && (*bin == '0' || *bin == '_')) {
1935 ++bin;
1936 --len;
1937 }
1938
1939 contains_underscores = (memchr(bin, '_', len) != NULL);
1940
1941 if (contains_underscores) {
1942 bin = estrndup(bin, len);
1943 strip_underscores(bin, &len);
1944 }
1945
1946 if (len < SIZEOF_ZEND_LONG * 8) {
1947 if (len == 0) {
1948 ZVAL_LONG(zendlval, 0);
1949 } else {
1950 errno = 0;
1951 ZVAL_LONG(zendlval, ZEND_STRTOL(bin, &end, 2));
1952 ZEND_ASSERT(!errno && end == bin + len);
1953 }
1954 if (contains_underscores) {
1955 efree(bin);
1956 }
1957 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
1958 } else {
1959 ZVAL_DOUBLE(zendlval, zend_bin_strtod(bin, (const char **)&end));
1960 /* errno isn't checked since we allow HUGE_VAL/INF overflow */
1961 ZEND_ASSERT(end == bin + len);
1962 if (contains_underscores) {
1963 efree(bin);
1964 }
1965 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
1966 }
1967 }
1968
1969 <ST_IN_SCRIPTING>{ONUM} {
1970 /* The +/- 2 skips "0o" */
1971 size_t len = yyleng - 2;
1972 char *end, *octal = yytext + 2;
1973 bool contains_underscores = (memchr(octal, '_', len) != NULL);
1974
1975 /* Skip any leading 0s */
1976 while (len > 0 && (*octal == '0' || *octal == '_')) {
1977 ++octal;
1978 --len;
1979 }
1980
1981 if (len == 0) {
1982 ZVAL_LONG(zendlval, 0);
1983 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
1984 }
1985
1986 if (contains_underscores) {
1987 octal = estrndup(octal, len);
1988 strip_underscores(octal, &len);
1989 }
1990
1991 errno = 0;
1992
1993 ZVAL_LONG(zendlval, ZEND_STRTOL(octal, &end, 8));
1994
1995 ZEND_ASSERT(end == octal + len);
1996
1997 if (!errno) {
1998 if (contains_underscores) {
1999 efree(octal);
2000 }
2001 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
2002 }
2003
2004 /* Overflow */
2005 ZEND_ASSERT(errno == ERANGE);
2006 /* Reset errno */
2007 errno = 0;
2008
2009 /* zend_oct_strtod skips leading '0' */
2010 ZVAL_DOUBLE(zendlval, zend_oct_strtod(octal, (const char **)&end));
2011 ZEND_ASSERT(!errno);
2012 ZEND_ASSERT(end == octal + len);
2013 if (contains_underscores) {
2014 efree(octal);
2015 }
2016 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
2017 }
2018
2019 <ST_IN_SCRIPTING>{LNUM} {
2020 size_t len = yyleng;
2021 char *end, *lnum = yytext;
2022 bool is_octal = lnum[0] == '0';
2023 bool contains_underscores = (memchr(lnum, '_', len) != NULL);
2024
2025 if (contains_underscores) {
2026 lnum = estrndup(lnum, len);
2027 strip_underscores(lnum, &len);
2028 }
2029
2030 /* Digits 8 and 9 are illegal in octal literals. */
2031 if (is_octal) {
2032 size_t i;
2033 for (i = 0; i < len; i++) {
2034 if (lnum[i] == '8' || lnum[i] == '9') {
2035 zend_throw_exception(zend_ce_parse_error, "Invalid numeric literal", 0);
2036 if (PARSER_MODE()) {
2037 if (contains_underscores) {
2038 efree(lnum);
2039 }
2040 ZVAL_UNDEF(zendlval);
2041 RETURN_TOKEN(T_ERROR);
2042 }
2043
2044 /* Continue in order to determine if this is T_LNUMBER or T_DNUMBER. */
2045 len = i;
2046 break;
2047 }
2048 }
2049 }
2050
2051
2052 if (len < MAX_LENGTH_OF_LONG - 1) { /* Won't overflow */
2053 errno = 0;
2054 /* base must be passed explicitly for correct parse error on Windows */
2055 ZVAL_LONG(zendlval, ZEND_STRTOL(lnum, &end, is_octal ? 8 : 10));
2056 ZEND_ASSERT(end == lnum + len);
2057 } else {
2058 errno = 0;
2059 ZVAL_LONG(zendlval, ZEND_STRTOL(lnum, &end, 0));
2060 if (errno == ERANGE) { /* Overflow */
2061 errno = 0;
2062 if (is_octal) { /* octal overflow */
2063 ZVAL_DOUBLE(zendlval, zend_oct_strtod(lnum, (const char **)&end));
2064 } else {
2065 ZVAL_DOUBLE(zendlval, zend_strtod(lnum, (const char **)&end));
2066 }
2067 ZEND_ASSERT(end == lnum + len);
2068 if (contains_underscores) {
2069 efree(lnum);
2070 }
2071 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
2072 }
2073 ZEND_ASSERT(end == lnum + len);
2074 }
2075 ZEND_ASSERT(!errno);
2076 if (contains_underscores) {
2077 efree(lnum);
2078 }
2079 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
2080 }
2081
2082 <ST_IN_SCRIPTING>{HNUM} {
2083 /* The +/- 2 skips "0x" */
2084 size_t len = yyleng - 2;
2085 char *end, *hex = yytext + 2;
2086 bool contains_underscores;
2087
2088 /* Skip any leading 0s */
2089 while (len > 0 && (*hex == '0' || *hex == '_')) {
2090 ++hex;
2091 --len;
2092 }
2093
2094 contains_underscores = (memchr(hex, '_', len) != NULL);
2095
2096 if (contains_underscores) {
2097 hex = estrndup(hex, len);
2098 strip_underscores(hex, &len);
2099 }
2100
2101 if (len < SIZEOF_ZEND_LONG * 2 || (len == SIZEOF_ZEND_LONG * 2 && *hex <= '7')) {
2102 if (len == 0) {
2103 ZVAL_LONG(zendlval, 0);
2104 } else {
2105 errno = 0;
2106 ZVAL_LONG(zendlval, ZEND_STRTOL(hex, &end, 16));
2107 ZEND_ASSERT(!errno && end == hex + len);
2108 }
2109 if (contains_underscores) {
2110 efree(hex);
2111 }
2112 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
2113 } else {
2114 ZVAL_DOUBLE(zendlval, zend_hex_strtod(hex, (const char **)&end));
2115 /* errno isn't checked since we allow HUGE_VAL/INF overflow */
2116 ZEND_ASSERT(end == hex + len);
2117 if (contains_underscores) {
2118 efree(hex);
2119 }
2120 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
2121 }
2122 }
2123
2124 <ST_VAR_OFFSET>[0]|([1-9][0-9]*) { /* Offset could be treated as a long */
2125 if (yyleng < MAX_LENGTH_OF_LONG - 1 || (yyleng == MAX_LENGTH_OF_LONG - 1 && strcmp(yytext, long_min_digits) < 0)) {
2126 char *end;
2127 errno = 0;
2128 ZVAL_LONG(zendlval, ZEND_STRTOL(yytext, &end, 10));
2129 if (errno == ERANGE) {
2130 goto string;
2131 }
2132 ZEND_ASSERT(end == yytext + yyleng);
2133 } else {
2134 string:
2135 ZVAL_STRINGL(zendlval, yytext, yyleng);
2136 }
2137 RETURN_TOKEN_WITH_VAL(T_NUM_STRING);
2138 }
2139
2140 <ST_VAR_OFFSET>{LNUM}|{HNUM}|{BNUM}|{ONUM} { /* Offset must be treated as a string */
2141 if (yyleng == 1) {
2142 ZVAL_INTERNED_STR(zendlval, ZSTR_CHAR((zend_uchar)*(yytext)));
2143 } else {
2144 ZVAL_STRINGL(zendlval, yytext, yyleng);
2145 }
2146 RETURN_TOKEN_WITH_VAL(T_NUM_STRING);
2147 }
2148
2149 <ST_IN_SCRIPTING>{DNUM}|{EXPONENT_DNUM} {
2150 const char *end;
2151 size_t len = yyleng;
2152 char *dnum = yytext;
2153 bool contains_underscores = (memchr(dnum, '_', len) != NULL);
2154
2155 if (contains_underscores) {
2156 dnum = estrndup(dnum, len);
2157 strip_underscores(dnum, &len);
2158 }
2159
2160 ZVAL_DOUBLE(zendlval, zend_strtod(dnum, &end));
2161 /* errno isn't checked since we allow HUGE_VAL/INF overflow */
2162 ZEND_ASSERT(end == dnum + len);
2163 if (contains_underscores) {
2164 efree(dnum);
2165 }
2166 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
2167 }
2168
2169 <ST_IN_SCRIPTING>"__CLASS__" {
2170 RETURN_TOKEN_WITH_IDENT(T_CLASS_C);
2171 }
2172
2173 <ST_IN_SCRIPTING>"__TRAIT__" {
2174 RETURN_TOKEN_WITH_IDENT(T_TRAIT_C);
2175 }
2176
2177 <ST_IN_SCRIPTING>"__FUNCTION__" {
2178 RETURN_TOKEN_WITH_IDENT(T_FUNC_C);
2179 }
2180
2181 <ST_IN_SCRIPTING>"__METHOD__" {
2182 RETURN_TOKEN_WITH_IDENT(T_METHOD_C);
2183 }
2184
2185 <ST_IN_SCRIPTING>"__LINE__" {
2186 RETURN_TOKEN_WITH_IDENT(T_LINE);
2187 }
2188
2189 <ST_IN_SCRIPTING>"__FILE__" {
2190 RETURN_TOKEN_WITH_IDENT(T_FILE);
2191 }
2192
2193 <ST_IN_SCRIPTING>"__DIR__" {
2194 RETURN_TOKEN_WITH_IDENT(T_DIR);
2195 }
2196
2197 <ST_IN_SCRIPTING>"__NAMESPACE__" {
2198 RETURN_TOKEN_WITH_IDENT(T_NS_C);
2199 }
2200
2201 <SHEBANG>"#!" .* {NEWLINE} {
2202 CG(zend_lineno)++;
2203 BEGIN(INITIAL);
2204 goto restart;
2205 }
2206
2207 <SHEBANG>{ANY_CHAR} {
2208 yyless(0);
2209 BEGIN(INITIAL);
2210 goto restart;
2211 }
2212
2213 <INITIAL>"<?=" {
2214 BEGIN(ST_IN_SCRIPTING);
2215 if (PARSER_MODE()) {
2216 /* We'll reject this as an identifier in zend_lex_tstring. */
2217 RETURN_TOKEN_WITH_IDENT(T_ECHO);
2218 }
2219 RETURN_TOKEN(T_OPEN_TAG_WITH_ECHO);
2220 }
2221
2222
2223 <INITIAL>"<?php"([ \t]|{NEWLINE}) {
2224 HANDLE_NEWLINE(yytext[yyleng-1]);
2225 BEGIN(ST_IN_SCRIPTING);
2226 RETURN_OR_SKIP_TOKEN(T_OPEN_TAG);
2227 }
2228
2229 <INITIAL>"<?php" {
2230 /* Allow <?php followed by end of file. */
2231 if (YYCURSOR == YYLIMIT) {
2232 BEGIN(ST_IN_SCRIPTING);
2233 RETURN_OR_SKIP_TOKEN(T_OPEN_TAG);
2234 }
2235 /* Degenerate case: <?phpX is interpreted as <? phpX with short tags. */
2236 if (CG(short_tags)) {
2237 yyless(2);
2238 BEGIN(ST_IN_SCRIPTING);
2239 RETURN_OR_SKIP_TOKEN(T_OPEN_TAG);
2240 }
2241 goto inline_char_handler;
2242 }
2243
2244 <INITIAL>"<?" {
2245 if (CG(short_tags)) {
2246 BEGIN(ST_IN_SCRIPTING);
2247 RETURN_OR_SKIP_TOKEN(T_OPEN_TAG);
2248 } else {
2249 goto inline_char_handler;
2250 }
2251 }
2252
2253 <INITIAL>{ANY_CHAR} {
2254 if (YYCURSOR > YYLIMIT) {
2255 RETURN_END_TOKEN;
2256 }
2257
2258 inline_char_handler:
2259
2260 while (1) {
2261 YYCTYPE *ptr = memchr(YYCURSOR, '<', YYLIMIT - YYCURSOR);
2262
2263 YYCURSOR = ptr ? ptr + 1 : YYLIMIT;
2264
2265 if (YYCURSOR >= YYLIMIT) {
2266 break;
2267 }
2268
2269 if (*YYCURSOR == '?') {
2270 if (CG(short_tags) /* <? */
2271 || (*(YYCURSOR + 1) == '=') /* <?= */
2272 || (!strncasecmp((char*)YYCURSOR + 1, "php", 3) && /* <?php[ \t\r\n] */
2273 (YYCURSOR + 4 == YYLIMIT ||
2274 YYCURSOR[4] == ' ' || YYCURSOR[4] == '\t' ||
2275 YYCURSOR[4] == '\n' || YYCURSOR[4] == '\r'))
2276 ) {
2277 YYCURSOR--;
2278 break;
2279 }
2280 }
2281 }
2282
2283 yyleng = YYCURSOR - SCNG(yy_text);
2284
2285 if (SCNG(output_filter)) {
2286 size_t readsize;
2287 char *s = NULL;
2288 size_t sz = 0;
2289 // TODO: avoid reallocation ???
2290 readsize = SCNG(output_filter)((unsigned char **)&s, &sz, (unsigned char *)yytext, (size_t)yyleng);
2291 ZVAL_STRINGL(zendlval, s, sz);
2292 efree(s);
2293 if (readsize < yyleng) {
2294 yyless(readsize);
2295 }
2296 } else if (yyleng == 1) {
2297 ZVAL_INTERNED_STR(zendlval, ZSTR_CHAR((zend_uchar)*yytext));
2298 } else {
2299 ZVAL_STRINGL(zendlval, yytext, yyleng);
2300 }
2301 HANDLE_NEWLINES(yytext, yyleng);
2302 RETURN_TOKEN_WITH_VAL(T_INLINE_HTML);
2303 }
2304
2305
2306 /* Make sure a label character follows "->" or "?->", otherwise there is no property
2307 * and "->"/"?->" will be taken literally
2308 */
2309 <ST_DOUBLE_QUOTES,ST_HEREDOC,ST_BACKQUOTE>"$"{LABEL}"->"[a-zA-Z_\x80-\xff] {
2310 yyless(yyleng - 3);
2311 yy_push_state(ST_LOOKING_FOR_PROPERTY);
2312 RETURN_TOKEN_WITH_STR(T_VARIABLE, 1);
2313 }
2314
2315 <ST_DOUBLE_QUOTES,ST_HEREDOC,ST_BACKQUOTE>"$"{LABEL}"?->"[a-zA-Z_\x80-\xff] {
2316 yyless(yyleng - 4);
2317 yy_push_state(ST_LOOKING_FOR_PROPERTY);
2318 RETURN_TOKEN_WITH_STR(T_VARIABLE, 1);
2319 }
2320
2321 /* A [ always designates a variable offset, regardless of what follows
2322 */
2323 <ST_DOUBLE_QUOTES,ST_HEREDOC,ST_BACKQUOTE>"$"{LABEL}"[" {
2324 yyless(yyleng - 1);
2325 yy_push_state(ST_VAR_OFFSET);
2326 RETURN_TOKEN_WITH_STR(T_VARIABLE, 1);
2327 }
2328
2329 <ST_IN_SCRIPTING,ST_DOUBLE_QUOTES,ST_HEREDOC,ST_BACKQUOTE,ST_VAR_OFFSET>"$"{LABEL} {
2330 RETURN_TOKEN_WITH_STR(T_VARIABLE, 1);
2331 }
2332
2333 <ST_VAR_OFFSET>"]" {
2334 yy_pop_state();
2335 RETURN_TOKEN(']');
2336 }
2337
2338 <ST_VAR_OFFSET>{TOKENS}|[[(){}"`] {
2339 /* Only '[' or '-' can be valid, but returning other tokens will allow a more explicit parse error */
2340 RETURN_TOKEN(yytext[0]);
2341 }
2342
2343 <ST_VAR_OFFSET>[ \n\r\t\\'#] {
2344 /* Invalid rule to return a more explicit parse error with proper line number */
2345 yyless(0);
2346 yy_pop_state();
2347 ZVAL_NULL(zendlval);
2348 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2349 }
2350
2351 <ST_IN_SCRIPTING>"namespace"("\\"{LABEL})+ {
2352 RETURN_TOKEN_WITH_STR(T_NAME_RELATIVE, sizeof("namespace\\") - 1);
2353 }
2354
2355 <ST_IN_SCRIPTING>{LABEL}("\\"{LABEL})+ {
2356 RETURN_TOKEN_WITH_STR(T_NAME_QUALIFIED, 0);
2357 }
2358
2359 <ST_IN_SCRIPTING>"\\"{LABEL}("\\"{LABEL})* {
2360 RETURN_TOKEN_WITH_STR(T_NAME_FULLY_QUALIFIED, 1);
2361 }
2362
2363 <ST_IN_SCRIPTING>"\\" {
2364 RETURN_TOKEN(T_NS_SEPARATOR);
2365 }
2366
2367 <ST_IN_SCRIPTING,ST_VAR_OFFSET>{LABEL} {
2368 RETURN_TOKEN_WITH_STR(T_STRING, 0);
2369 }
2370
2371
2372 <ST_IN_SCRIPTING,ST_LOOKING_FOR_PROPERTY>"#"|"//" {
2373 while (YYCURSOR < YYLIMIT) {
2374 switch (*YYCURSOR++) {
2375 case '\r':
2376 case '\n':
2377 YYCURSOR--;
2378 break;
2379 case '?':
2380 if (*YYCURSOR == '>') {
2381 YYCURSOR--;
2382 break;
2383 }
2384 ZEND_FALLTHROUGH;
2385 default:
2386 continue;
2387 }
2388
2389 break;
2390 }
2391
2392 yyleng = YYCURSOR - SCNG(yy_text);
2393 RETURN_OR_SKIP_TOKEN(T_COMMENT);
2394 }
2395
2396 <ST_IN_SCRIPTING,ST_LOOKING_FOR_PROPERTY>"/*"|"/**"{WHITESPACE} {
2397 int doc_com;
2398
2399 if (yyleng > 2) {
2400 doc_com = 1;
2401 RESET_DOC_COMMENT();
2402 } else {
2403 doc_com = 0;
2404 }
2405
2406 while (YYCURSOR < YYLIMIT) {
2407 if (*YYCURSOR++ == '*' && *YYCURSOR == '/') {
2408 break;
2409 }
2410 }
2411
2412 if (YYCURSOR < YYLIMIT) {
2413 YYCURSOR++;
2414 } else {
2415 zend_throw_exception_ex(zend_ce_parse_error, 0, "Unterminated comment starting line %d", CG(zend_lineno));
2416 if (PARSER_MODE()) {
2417 RETURN_TOKEN(T_ERROR);
2418 }
2419 }
2420
2421 yyleng = YYCURSOR - SCNG(yy_text);
2422 HANDLE_NEWLINES(yytext, yyleng);
2423
2424 if (doc_com) {
2425 CG(doc_comment) = zend_string_init(yytext, yyleng, 0);
2426 RETURN_OR_SKIP_TOKEN(T_DOC_COMMENT);
2427 }
2428
2429 RETURN_OR_SKIP_TOKEN(T_COMMENT);
2430 }
2431
2432 <ST_LOOKING_FOR_PROPERTY>{ANY_CHAR} {
2433 yyless(0);
2434 yy_pop_state();
2435 goto restart;
2436 }
2437
2438 <ST_IN_SCRIPTING>"?>"{NEWLINE}? {
2439 BEGIN(INITIAL);
2440 if (yytext[yyleng-1] != '>') {
2441 CG(increment_lineno) = 1;
2442 }
2443 if (PARSER_MODE()) {
2444 RETURN_TOKEN(';'); /* implicit ';' at php-end tag */
2445 }
2446 RETURN_TOKEN(T_CLOSE_TAG);
2447 }
2448
2449
2450 <ST_IN_SCRIPTING>b?['] {
2451 char *s, *t;
2452 char *end;
2453 int bprefix = (yytext[0] != '\'') ? 1 : 0;
2454
2455 while (1) {
2456 if (YYCURSOR < YYLIMIT) {
2457 if (*YYCURSOR == '\'') {
2458 YYCURSOR++;
2459 yyleng = YYCURSOR - SCNG(yy_text);
2460
2461 break;
2462 } else if (*YYCURSOR++ == '\\' && YYCURSOR < YYLIMIT) {
2463 YYCURSOR++;
2464 }
2465 } else {
2466 yyleng = YYLIMIT - SCNG(yy_text);
2467
2468 /* Unclosed single quotes; treat similar to double quotes, but without a separate token
2469 * for ' (unrecognized by parser), instead of old flex fallback to "Unexpected character..."
2470 * rule, which continued in ST_IN_SCRIPTING state after the quote */
2471 ZVAL_NULL(zendlval);
2472 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2473 }
2474 }
2475
2476 if (yyleng-bprefix-2 <= 1) {
2477 if (yyleng-bprefix-2 < 1) {
2478 ZVAL_EMPTY_STRING(zendlval);
2479 } else {
2480 zend_uchar c = (zend_uchar)*(yytext+bprefix+1);
2481 if (c == '\n' || c == '\r') {
2482 CG(zend_lineno)++;
2483 }
2484 ZVAL_INTERNED_STR(zendlval, ZSTR_CHAR(c));
2485 }
2486 goto skip_escape_conversion;
2487 }
2488 ZVAL_STRINGL(zendlval, yytext+bprefix+1, yyleng-bprefix-2);
2489
2490 /* convert escape sequences */
2491 s = Z_STRVAL_P(zendlval);
2492 end = s+Z_STRLEN_P(zendlval);
2493 while (1) {
2494 if (UNEXPECTED(*s=='\\')) {
2495 break;
2496 }
2497 if (*s == '\n' || (*s == '\r' && (*(s+1) != '\n'))) {
2498 CG(zend_lineno)++;
2499 }
2500 s++;
2501 if (s == end) {
2502 goto skip_escape_conversion;
2503 }
2504 }
2505
2506 t = s;
2507 while (s<end) {
2508 if (*s=='\\') {
2509 s++;
2510 if (*s == '\\' || *s == '\'') {
2511 *t++ = *s;
2512 } else {
2513 *t++ = '\\';
2514 *t++ = *s;
2515 }
2516 } else {
2517 *t++ = *s;
2518 }
2519 if (*s == '\n' || (*s == '\r' && (*(s+1) != '\n'))) {
2520 CG(zend_lineno)++;
2521 }
2522 s++;
2523 }
2524 *t = 0;
2525 Z_STRLEN_P(zendlval) = t - Z_STRVAL_P(zendlval);
2526
2527 skip_escape_conversion:
2528 if (SCNG(output_filter)) {
2529 size_t sz = 0;
2530 char *str = NULL;
2531 zend_string *new_str;
2532 s = Z_STRVAL_P(zendlval);
2533 // TODO: avoid reallocation ???
2534 SCNG(output_filter)((unsigned char **)&str, &sz, (unsigned char *)s, (size_t)Z_STRLEN_P(zendlval));
2535 new_str = zend_string_init(str, sz, 0);
2536 if (str != s) {
2537 efree(str);
2538 }
2539 zend_string_release_ex(Z_STR_P(zendlval), 0);
2540 ZVAL_STR(zendlval, new_str);
2541 }
2542 RETURN_TOKEN_WITH_VAL(T_CONSTANT_ENCAPSED_STRING);
2543 }
2544
2545
2546 <ST_IN_SCRIPTING>b?["] {
2547 int bprefix = (yytext[0] != '"') ? 1 : 0;
2548
2549 while (YYCURSOR < YYLIMIT) {
2550 switch (*YYCURSOR++) {
2551 case '"':
2552 yyleng = YYCURSOR - SCNG(yy_text);
2553 if (EXPECTED(zend_scan_escape_string(zendlval, yytext+bprefix+1, yyleng-bprefix-2, '"') == SUCCESS)
2554 || !PARSER_MODE()) {
2555 RETURN_TOKEN_WITH_VAL(T_CONSTANT_ENCAPSED_STRING);
2556 } else {
2557 RETURN_TOKEN(T_ERROR);
2558 }
2559 case '$':
2560 if (IS_LABEL_START(*YYCURSOR) || *YYCURSOR == '{') {
2561 break;
2562 }
2563 continue;
2564 case '{':
2565 if (*YYCURSOR == '$') {
2566 break;
2567 }
2568 continue;
2569 case '\\':
2570 if (YYCURSOR < YYLIMIT) {
2571 YYCURSOR++;
2572 }
2573 ZEND_FALLTHROUGH;
2574 default:
2575 continue;
2576 }
2577
2578 YYCURSOR--;
2579 break;
2580 }
2581
2582 /* Remember how much was scanned to save rescanning */
2583 SET_DOUBLE_QUOTES_SCANNED_LENGTH(YYCURSOR - SCNG(yy_text) - yyleng);
2584
2585 YYCURSOR = SCNG(yy_text) + yyleng;
2586
2587 BEGIN(ST_DOUBLE_QUOTES);
2588 RETURN_TOKEN('"');
2589 }
2590
2591
2592 <ST_IN_SCRIPTING>b?"<<<"{TABS_AND_SPACES}({LABEL}|([']{LABEL}['])|(["]{LABEL}["])){NEWLINE} {
2593 char *s;
2594 unsigned char *saved_cursor;
2595 int bprefix = (yytext[0] != '<') ? 1 : 0, spacing = 0, indentation = 0;
2596 zend_heredoc_label *heredoc_label = emalloc(sizeof(zend_heredoc_label));
2597 bool is_heredoc = 1;
2598
2599 CG(zend_lineno)++;
2600 heredoc_label->length = yyleng-bprefix-3-1-(yytext[yyleng-2]=='\r'?1:0);
2601 s = yytext+bprefix+3;
2602 while ((*s == ' ') || (*s == '\t')) {
2603 s++;
2604 heredoc_label->length--;
2605 }
2606
2607 if (*s == '\'') {
2608 s++;
2609 heredoc_label->length -= 2;
2610 is_heredoc = 0;
2611
2612 BEGIN(ST_NOWDOC);
2613 } else {
2614 if (*s == '"') {
2615 s++;
2616 heredoc_label->length -= 2;
2617 }
2618
2619 BEGIN(ST_HEREDOC);
2620 }
2621
2622 heredoc_label->label = estrndup(s, heredoc_label->length);
2623 heredoc_label->indentation_uses_spaces = 0;
2624 heredoc_label->indentation = 0;
2625 saved_cursor = YYCURSOR;
2626
2627 zend_ptr_stack_push(&SCNG(heredoc_label_stack), (void *) heredoc_label);
2628
2629 while (YYCURSOR < YYLIMIT && (*YYCURSOR == ' ' || *YYCURSOR == '\t')) {
2630 if (*YYCURSOR == '\t') {
2631 spacing |= HEREDOC_USING_TABS;
2632 } else {
2633 spacing |= HEREDOC_USING_SPACES;
2634 }
2635 ++YYCURSOR;
2636 ++indentation;
2637 }
2638
2639 if (YYCURSOR == YYLIMIT) {
2640 YYCURSOR = saved_cursor;
2641 RETURN_TOKEN(T_START_HEREDOC);
2642 }
2643
2644 /* Check for ending label on the next line */
2645 if (heredoc_label->length < YYLIMIT - YYCURSOR && !memcmp(YYCURSOR, s, heredoc_label->length)) {
2646 if (!IS_LABEL_SUCCESSOR(YYCURSOR[heredoc_label->length])) {
2647 if (spacing == (HEREDOC_USING_SPACES | HEREDOC_USING_TABS)) {
2648 zend_throw_exception(zend_ce_parse_error, "Invalid indentation - tabs and spaces cannot be mixed", 0);
2649 if (PARSER_MODE()) {
2650 RETURN_TOKEN(T_ERROR);
2651 }
2652 }
2653
2654 YYCURSOR = saved_cursor;
2655 heredoc_label->indentation = indentation;
2656
2657 BEGIN(ST_END_HEREDOC);
2658 RETURN_TOKEN(T_START_HEREDOC);
2659 }
2660 }
2661
2662 YYCURSOR = saved_cursor;
2663
2664 if (is_heredoc && !SCNG(heredoc_scan_ahead)) {
2665 zend_lex_state current_state;
2666 zend_string *saved_doc_comment = CG(doc_comment);
2667 int heredoc_nesting_level = 1;
2668 int first_token = 0;
2669 int error = 0;
2670
2671 zend_save_lexical_state(¤t_state);
2672
2673 SCNG(heredoc_scan_ahead) = 1;
2674 SCNG(heredoc_indentation) = 0;
2675 SCNG(heredoc_indentation_uses_spaces) = 0;
2676 LANG_SCNG(on_event) = NULL;
2677 CG(doc_comment) = NULL;
2678
2679 zend_ptr_stack_reverse_apply(¤t_state.heredoc_label_stack, copy_heredoc_label_stack);
2680
2681 zend_exception_save();
2682 while (heredoc_nesting_level) {
2683 zval zv;
2684 int retval;
2685
2686 ZVAL_UNDEF(&zv);
2687 retval = lex_scan(&zv, NULL);
2688 zval_ptr_dtor_nogc(&zv);
2689
2690 if (EG(exception)) {
2691 zend_clear_exception();
2692 break;
2693 }
2694
2695 if (!first_token) {
2696 first_token = retval;
2697 }
2698
2699 switch (retval) {
2700 case T_START_HEREDOC:
2701 ++heredoc_nesting_level;
2702 break;
2703 case T_END_HEREDOC:
2704 --heredoc_nesting_level;
2705 break;
2706 case END:
2707 heredoc_nesting_level = 0;
2708 }
2709 }
2710 zend_exception_restore();
2711
2712 if (
2713 (first_token == T_VARIABLE
2714 || first_token == T_DOLLAR_OPEN_CURLY_BRACES
2715 || first_token == T_CURLY_OPEN
2716 ) && SCNG(heredoc_indentation)) {
2717 zend_throw_exception_ex(zend_ce_parse_error, 0, "Invalid body indentation level (expecting an indentation level of at least %d)", SCNG(heredoc_indentation));
2718 error = 1;
2719 }
2720
2721 heredoc_label->indentation = SCNG(heredoc_indentation);
2722 heredoc_label->indentation_uses_spaces = SCNG(heredoc_indentation_uses_spaces);
2723
2724 zend_restore_lexical_state(¤t_state);
2725 SCNG(heredoc_scan_ahead) = 0;
2726 CG(increment_lineno) = 0;
2727 CG(doc_comment) = saved_doc_comment;
2728
2729 if (PARSER_MODE() && error) {
2730 RETURN_TOKEN(T_ERROR);
2731 }
2732 }
2733
2734 RETURN_TOKEN(T_START_HEREDOC);
2735 }
2736
2737
2738 <ST_IN_SCRIPTING>[`] {
2739 BEGIN(ST_BACKQUOTE);
2740 RETURN_TOKEN('`');
2741 }
2742
2743
2744 <ST_END_HEREDOC>{ANY_CHAR} {
2745 zend_heredoc_label *heredoc_label = zend_ptr_stack_pop(&SCNG(heredoc_label_stack));
2746
2747 yyleng = heredoc_label->indentation + heredoc_label->length;
2748 YYCURSOR += yyleng - 1;
2749
2750 heredoc_label_dtor(heredoc_label);
2751 efree(heredoc_label);
2752
2753 BEGIN(ST_IN_SCRIPTING);
2754 RETURN_TOKEN(T_END_HEREDOC);
2755 }
2756
2757
2758 <ST_DOUBLE_QUOTES,ST_BACKQUOTE,ST_HEREDOC>"{$" {
2759 yy_push_state(ST_IN_SCRIPTING);
2760 yyless(1);
2761 enter_nesting('{');
2762 RETURN_TOKEN(T_CURLY_OPEN);
2763 }
2764
2765
2766 <ST_DOUBLE_QUOTES>["] {
2767 BEGIN(ST_IN_SCRIPTING);
2768 RETURN_TOKEN('"');
2769 }
2770
2771 <ST_BACKQUOTE>[`] {
2772 BEGIN(ST_IN_SCRIPTING);
2773 RETURN_TOKEN('`');
2774 }
2775
2776
2777 <ST_DOUBLE_QUOTES>{ANY_CHAR} {
2778 if (GET_DOUBLE_QUOTES_SCANNED_LENGTH()) {
2779 YYCURSOR += GET_DOUBLE_QUOTES_SCANNED_LENGTH() - 1;
2780 SET_DOUBLE_QUOTES_SCANNED_LENGTH(0);
2781
2782 goto double_quotes_scan_done;
2783 }
2784
2785 if (YYCURSOR > YYLIMIT) {
2786 RETURN_END_TOKEN;
2787 }
2788 if (yytext[0] == '\\' && YYCURSOR < YYLIMIT) {
2789 YYCURSOR++;
2790 }
2791
2792 while (YYCURSOR < YYLIMIT) {
2793 switch (*YYCURSOR++) {
2794 case '"':
2795 break;
2796 case '$':
2797 if (IS_LABEL_START(*YYCURSOR) || *YYCURSOR == '{') {
2798 break;
2799 }
2800 continue;
2801 case '{':
2802 if (*YYCURSOR == '$') {
2803 break;
2804 }
2805 continue;
2806 case '\\':
2807 if (YYCURSOR < YYLIMIT) {
2808 YYCURSOR++;
2809 }
2810 ZEND_FALLTHROUGH;
2811 default:
2812 continue;
2813 }
2814
2815 YYCURSOR--;
2816 break;
2817 }
2818
2819 double_quotes_scan_done:
2820 yyleng = YYCURSOR - SCNG(yy_text);
2821
2822 if (EXPECTED(zend_scan_escape_string(zendlval, yytext, yyleng, '"') == SUCCESS)
2823 || !PARSER_MODE()) {
2824 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2825 } else {
2826 RETURN_TOKEN(T_ERROR);
2827 }
2828 }
2829
2830
2831 <ST_BACKQUOTE>{ANY_CHAR} {
2832 if (YYCURSOR > YYLIMIT) {
2833 RETURN_END_TOKEN;
2834 }
2835 if (yytext[0] == '\\' && YYCURSOR < YYLIMIT) {
2836 YYCURSOR++;
2837 }
2838
2839 while (YYCURSOR < YYLIMIT) {
2840 switch (*YYCURSOR++) {
2841 case '`':
2842 break;
2843 case '$':
2844 if (IS_LABEL_START(*YYCURSOR) || *YYCURSOR == '{') {
2845 break;
2846 }
2847 continue;
2848 case '{':
2849 if (*YYCURSOR == '$') {
2850 break;
2851 }
2852 continue;
2853 case '\\':
2854 if (YYCURSOR < YYLIMIT) {
2855 YYCURSOR++;
2856 }
2857 ZEND_FALLTHROUGH;
2858 default:
2859 continue;
2860 }
2861
2862 YYCURSOR--;
2863 break;
2864 }
2865
2866 yyleng = YYCURSOR - SCNG(yy_text);
2867
2868 if (EXPECTED(zend_scan_escape_string(zendlval, yytext, yyleng, '`') == SUCCESS)
2869 || !PARSER_MODE()) {
2870 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2871 } else {
2872 RETURN_TOKEN(T_ERROR);
2873 }
2874 }
2875
2876
2877 <ST_HEREDOC>{ANY_CHAR} {
2878 zend_heredoc_label *heredoc_label = zend_ptr_stack_top(&SCNG(heredoc_label_stack));
2879 int newline = 0, indentation = 0, spacing = 0;
2880
2881 if (YYCURSOR > YYLIMIT) {
2882 RETURN_END_TOKEN;
2883 }
2884
2885 YYCURSOR--;
2886
2887 while (YYCURSOR < YYLIMIT) {
2888 switch (*YYCURSOR++) {
2889 case '\r':
2890 if (*YYCURSOR == '\n') {
2891 YYCURSOR++;
2892 }
2893 ZEND_FALLTHROUGH;
2894 case '\n':
2895 indentation = spacing = 0;
2896
2897 while (YYCURSOR < YYLIMIT && (*YYCURSOR == ' ' || *YYCURSOR == '\t')) {
2898 if (*YYCURSOR == '\t') {
2899 spacing |= HEREDOC_USING_TABS;
2900 } else {
2901 spacing |= HEREDOC_USING_SPACES;
2902 }
2903 ++YYCURSOR;
2904 ++indentation;
2905 }
2906
2907 if (YYCURSOR == YYLIMIT) {
2908 yyleng = YYCURSOR - SCNG(yy_text);
2909 HANDLE_NEWLINES(yytext, yyleng);
2910 ZVAL_NULL(zendlval);
2911 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2912 }
2913
2914 /* Check for ending label on the next line */
2915 if (IS_LABEL_START(*YYCURSOR) && heredoc_label->length < YYLIMIT - YYCURSOR && !memcmp(YYCURSOR, heredoc_label->label, heredoc_label->length)) {
2916 if (IS_LABEL_SUCCESSOR(YYCURSOR[heredoc_label->length])) {
2917 continue;
2918 }
2919
2920 if (spacing == (HEREDOC_USING_SPACES | HEREDOC_USING_TABS)) {
2921 zend_throw_exception(zend_ce_parse_error, "Invalid indentation - tabs and spaces cannot be mixed", 0);
2922 if (PARSER_MODE()) {
2923 RETURN_TOKEN(T_ERROR);
2924 }
2925 }
2926
2927 /* newline before label will be subtracted from returned text, but
2928 * yyleng/yytext will include it, for zend_highlight/strip, tokenizer, etc. */
2929 if (YYCURSOR[-indentation - 2] == '\r' && YYCURSOR[-indentation - 1] == '\n') {
2930 newline = 2; /* Windows newline */
2931 } else {
2932 newline = 1;
2933 }
2934
2935 CG(increment_lineno) = 1; /* For newline before label */
2936
2937 if (SCNG(heredoc_scan_ahead)) {
2938 SCNG(heredoc_indentation) = indentation;
2939 SCNG(heredoc_indentation_uses_spaces) = (spacing == HEREDOC_USING_SPACES);
2940 } else {
2941 YYCURSOR -= indentation;
2942 }
2943
2944 BEGIN(ST_END_HEREDOC);
2945
2946 goto heredoc_scan_done;
2947 }
2948 continue;
2949 case '$':
2950 if (IS_LABEL_START(*YYCURSOR) || *YYCURSOR == '{') {
2951 break;
2952 }
2953 continue;
2954 case '{':
2955 if (*YYCURSOR == '$') {
2956 break;
2957 }
2958 continue;
2959 case '\\':
2960 if (YYCURSOR < YYLIMIT && *YYCURSOR != '\n' && *YYCURSOR != '\r') {
2961 YYCURSOR++;
2962 }
2963 ZEND_FALLTHROUGH;
2964 default:
2965 continue;
2966 }
2967
2968 YYCURSOR--;
2969 break;
2970 }
2971
2972 heredoc_scan_done:
2973
2974 yyleng = YYCURSOR - SCNG(yy_text);
2975 ZVAL_STRINGL(zendlval, yytext, yyleng - newline);
2976
2977 if (!SCNG(heredoc_scan_ahead) && !EG(exception) && PARSER_MODE()) {
2978 bool newline_at_start = *(yytext - 1) == '\n' || *(yytext - 1) == '\r';
2979 zend_string *copy = Z_STR_P(zendlval);
2980
2981 if (!strip_multiline_string_indentation(
2982 zendlval, heredoc_label->indentation, heredoc_label->indentation_uses_spaces,
2983 newline_at_start, newline != 0)) {
2984 RETURN_TOKEN(T_ERROR);
2985 }
2986
2987 if (UNEXPECTED(zend_scan_escape_string(zendlval, ZSTR_VAL(copy), ZSTR_LEN(copy), 0) != SUCCESS)) {
2988 zend_string_efree(copy);
2989 RETURN_TOKEN(T_ERROR);
2990 }
2991
2992 zend_string_efree(copy);
2993 } else {
2994 HANDLE_NEWLINES(yytext, yyleng - newline);
2995 }
2996
2997 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2998 }
2999
3000
3001 <ST_NOWDOC>{ANY_CHAR} {
3002 zend_heredoc_label *heredoc_label = zend_ptr_stack_top(&SCNG(heredoc_label_stack));
3003 int newline = 0, indentation = 0, spacing = -1;
3004
3005 if (YYCURSOR > YYLIMIT) {
3006 RETURN_END_TOKEN;
3007 }
3008
3009 YYCURSOR--;
3010
3011 while (YYCURSOR < YYLIMIT) {
3012 switch (*YYCURSOR++) {
3013 case '\r':
3014 if (*YYCURSOR == '\n') {
3015 YYCURSOR++;
3016 }
3017 ZEND_FALLTHROUGH;
3018 case '\n':
3019 indentation = spacing = 0;
3020
3021 while (YYCURSOR < YYLIMIT && (*YYCURSOR == ' ' || *YYCURSOR == '\t')) {
3022 if (*YYCURSOR == '\t') {
3023 spacing |= HEREDOC_USING_TABS;
3024 } else {
3025 spacing |= HEREDOC_USING_SPACES;
3026 }
3027 ++YYCURSOR;
3028 ++indentation;
3029 }
3030
3031 if (YYCURSOR == YYLIMIT) {
3032 yyleng = YYCURSOR - SCNG(yy_text);
3033 HANDLE_NEWLINES(yytext, yyleng);
3034 ZVAL_NULL(zendlval);
3035 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
3036 }
3037
3038 /* Check for ending label on the next line */
3039 if (IS_LABEL_START(*YYCURSOR) && heredoc_label->length < YYLIMIT - YYCURSOR && !memcmp(YYCURSOR, heredoc_label->label, heredoc_label->length)) {
3040 if (IS_LABEL_SUCCESSOR(YYCURSOR[heredoc_label->length])) {
3041 continue;
3042 }
3043
3044 if (spacing == (HEREDOC_USING_SPACES | HEREDOC_USING_TABS)) {
3045 zend_throw_exception(zend_ce_parse_error, "Invalid indentation - tabs and spaces cannot be mixed", 0);
3046 if (PARSER_MODE()) {
3047 RETURN_TOKEN(T_ERROR);
3048 }
3049 }
3050
3051 /* newline before label will be subtracted from returned text, but
3052 * yyleng/yytext will include it, for zend_highlight/strip, tokenizer, etc. */
3053 if (YYCURSOR[-indentation - 2] == '\r' && YYCURSOR[-indentation - 1] == '\n') {
3054 newline = 2; /* Windows newline */
3055 } else {
3056 newline = 1;
3057 }
3058
3059 CG(increment_lineno) = 1; /* For newline before label */
3060
3061 YYCURSOR -= indentation;
3062 heredoc_label->indentation = indentation;
3063
3064 BEGIN(ST_END_HEREDOC);
3065
3066 goto nowdoc_scan_done;
3067 }
3068 ZEND_FALLTHROUGH;
3069 default:
3070 continue;
3071 }
3072 }
3073
3074 nowdoc_scan_done:
3075 yyleng = YYCURSOR - SCNG(yy_text);
3076 ZVAL_STRINGL(zendlval, yytext, yyleng - newline);
3077
3078 if (!EG(exception) && spacing != -1 && PARSER_MODE()) {
3079 bool newline_at_start = *(yytext - 1) == '\n' || *(yytext - 1) == '\r';
3080 if (!strip_multiline_string_indentation(
3081 zendlval, indentation, spacing == HEREDOC_USING_SPACES,
3082 newline_at_start, newline != 0)) {
3083 RETURN_TOKEN(T_ERROR);
3084 }
3085 }
3086
3087 HANDLE_NEWLINES(yytext, yyleng - newline);
3088 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
3089 }
3090
3091
3092 <ST_IN_SCRIPTING,ST_VAR_OFFSET>{ANY_CHAR} {
3093 if (YYCURSOR > YYLIMIT) {
3094 RETURN_END_TOKEN;
3095 }
3096
3097 RETURN_TOKEN(T_BAD_CHARACTER);
3098 }
3099
3100 */
3101
3102 emit_token_with_str:
3103 zend_copy_value(zendlval, (yytext + offset), (yyleng - offset));
3104
3105 emit_token_with_val:
3106 if (PARSER_MODE()) {
3107 ZEND_ASSERT(Z_TYPE_P(zendlval) != IS_UNDEF);
3108 elem->ast = zend_ast_create_zval_with_lineno(zendlval, start_line);
3109 }
3110
3111 emit_token:
3112 if (SCNG(on_event)) {
3113 SCNG(on_event)(ON_TOKEN, token, start_line, yytext, yyleng, SCNG(on_event_context));
3114 }
3115 return token;
3116
3117 emit_token_with_ident:
3118 if (PARSER_MODE()) {
3119 elem->ident = SCNG(yy_text);
3120 }
3121 if (SCNG(on_event)) {
3122 SCNG(on_event)(ON_TOKEN, token, start_line, yytext, yyleng, SCNG(on_event_context));
3123 }
3124 return token;
3125
3126 return_whitespace:
3127 HANDLE_NEWLINES(yytext, yyleng);
3128 if (SCNG(on_event)) {
3129 SCNG(on_event)(ON_TOKEN, T_WHITESPACE, start_line, yytext, yyleng, SCNG(on_event_context));
3130 }
3131 if (PARSER_MODE()) {
3132 start_line = CG(zend_lineno);
3133 goto restart;
3134 } else {
3135 return T_WHITESPACE;
3136 }
3137
3138 skip_token:
3139 if (SCNG(on_event)) {
3140 SCNG(on_event)(ON_TOKEN, token, start_line, yytext, yyleng, SCNG(on_event_context));
3141 }
3142 start_line = CG(zend_lineno);
3143 goto restart;
3144 }
3145