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) && SCNG(script_filtered) != lex_state->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, op_array);
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) != SUCCESS && 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() != SUCCESS && 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 OPTIONAL_WHITESPACE [ \n\r\t]*
1372 /* We don't use re2c with bounds checking, we just return 0 bytes if we read past the input.
1373 * If we use wildcard matching for comments, we can read past the input, which crashes
1374 * once we try to report a syntax error because the 0 bytes are not actually part of
1375 * the token. We prevent this by not allowing 0 bytes, which already aren't valid anyway. */
1376 MULTI_LINE_COMMENT "/*"([^*\x00]*"*"+)([^*/\x00][^*\x00]*"*"+)*"/"
1377 SINGLE_LINE_COMMENT "//"[^\x00\n\r]*[\n\r]
1378 HASH_COMMENT "#"(([^[\x00][^\x00\n\r]*[\n\r])|[\n\r])
1379 WHITESPACE_OR_COMMENTS ({WHITESPACE}|{MULTI_LINE_COMMENT}|{SINGLE_LINE_COMMENT}|{HASH_COMMENT})+
1380 OPTIONAL_WHITESPACE_OR_COMMENTS ({WHITESPACE}|{MULTI_LINE_COMMENT}|{SINGLE_LINE_COMMENT}|{HASH_COMMENT})*
1381
1382 /* compute yyleng before each rule */
1383 <!*> := yyleng = YYCURSOR - SCNG(yy_text);
1384
1385 <ST_IN_SCRIPTING>"exit" {
1386 RETURN_TOKEN_WITH_IDENT(T_EXIT);
1387 }
1388
1389 <ST_IN_SCRIPTING>"die" {
1390 RETURN_TOKEN_WITH_IDENT(T_EXIT);
1391 }
1392
1393 <ST_IN_SCRIPTING>"fn" {
1394 RETURN_TOKEN_WITH_IDENT(T_FN);
1395 }
1396
1397 <ST_IN_SCRIPTING>"function" {
1398 RETURN_TOKEN_WITH_IDENT(T_FUNCTION);
1399 }
1400
1401 <ST_IN_SCRIPTING>"const" {
1402 RETURN_TOKEN_WITH_IDENT(T_CONST);
1403 }
1404
1405 <ST_IN_SCRIPTING>"return" {
1406 RETURN_TOKEN_WITH_IDENT(T_RETURN);
1407 }
1408
1409 <ST_IN_SCRIPTING>"#[" {
1410 enter_nesting('[');
1411 RETURN_TOKEN(T_ATTRIBUTE);
1412 }
1413
1414 <ST_IN_SCRIPTING>"yield"{WHITESPACE_OR_COMMENTS}"from"[^a-zA-Z0-9_\x80-\xff] {
1415 yyless(yyleng - 1);
1416 HANDLE_NEWLINES(yytext, yyleng);
1417 RETURN_TOKEN_WITH_IDENT(T_YIELD_FROM);
1418 }
1419
1420 <ST_IN_SCRIPTING>"yield" {
1421 RETURN_TOKEN_WITH_IDENT(T_YIELD);
1422 }
1423
1424 <ST_IN_SCRIPTING>"try" {
1425 RETURN_TOKEN_WITH_IDENT(T_TRY);
1426 }
1427
1428 <ST_IN_SCRIPTING>"catch" {
1429 RETURN_TOKEN_WITH_IDENT(T_CATCH);
1430 }
1431
1432 <ST_IN_SCRIPTING>"finally" {
1433 RETURN_TOKEN_WITH_IDENT(T_FINALLY);
1434 }
1435
1436 <ST_IN_SCRIPTING>"throw" {
1437 RETURN_TOKEN_WITH_IDENT(T_THROW);
1438 }
1439
1440 <ST_IN_SCRIPTING>"if" {
1441 RETURN_TOKEN_WITH_IDENT(T_IF);
1442 }
1443
1444 <ST_IN_SCRIPTING>"elseif" {
1445 RETURN_TOKEN_WITH_IDENT(T_ELSEIF);
1446 }
1447
1448 <ST_IN_SCRIPTING>"endif" {
1449 RETURN_TOKEN_WITH_IDENT(T_ENDIF);
1450 }
1451
1452 <ST_IN_SCRIPTING>"else" {
1453 RETURN_TOKEN_WITH_IDENT(T_ELSE);
1454 }
1455
1456 <ST_IN_SCRIPTING>"while" {
1457 RETURN_TOKEN_WITH_IDENT(T_WHILE);
1458 }
1459
1460 <ST_IN_SCRIPTING>"endwhile" {
1461 RETURN_TOKEN_WITH_IDENT(T_ENDWHILE);
1462 }
1463
1464 <ST_IN_SCRIPTING>"do" {
1465 RETURN_TOKEN_WITH_IDENT(T_DO);
1466 }
1467
1468 <ST_IN_SCRIPTING>"for" {
1469 RETURN_TOKEN_WITH_IDENT(T_FOR);
1470 }
1471
1472 <ST_IN_SCRIPTING>"endfor" {
1473 RETURN_TOKEN_WITH_IDENT(T_ENDFOR);
1474 }
1475
1476 <ST_IN_SCRIPTING>"foreach" {
1477 RETURN_TOKEN_WITH_IDENT(T_FOREACH);
1478 }
1479
1480 <ST_IN_SCRIPTING>"endforeach" {
1481 RETURN_TOKEN_WITH_IDENT(T_ENDFOREACH);
1482 }
1483
1484 <ST_IN_SCRIPTING>"declare" {
1485 RETURN_TOKEN_WITH_IDENT(T_DECLARE);
1486 }
1487
1488 <ST_IN_SCRIPTING>"enddeclare" {
1489 RETURN_TOKEN_WITH_IDENT(T_ENDDECLARE);
1490 }
1491
1492 <ST_IN_SCRIPTING>"instanceof" {
1493 RETURN_TOKEN_WITH_IDENT(T_INSTANCEOF);
1494 }
1495
1496 <ST_IN_SCRIPTING>"as" {
1497 RETURN_TOKEN_WITH_IDENT(T_AS);
1498 }
1499
1500 <ST_IN_SCRIPTING>"switch" {
1501 RETURN_TOKEN_WITH_IDENT(T_SWITCH);
1502 }
1503
1504 <ST_IN_SCRIPTING>"match" {
1505 RETURN_TOKEN_WITH_IDENT(T_MATCH);
1506 }
1507
1508 <ST_IN_SCRIPTING>"endswitch" {
1509 RETURN_TOKEN_WITH_IDENT(T_ENDSWITCH);
1510 }
1511
1512 <ST_IN_SCRIPTING>"case" {
1513 RETURN_TOKEN_WITH_IDENT(T_CASE);
1514 }
1515
1516 <ST_IN_SCRIPTING>"default" {
1517 RETURN_TOKEN_WITH_IDENT(T_DEFAULT);
1518 }
1519
1520 <ST_IN_SCRIPTING>"break" {
1521 RETURN_TOKEN_WITH_IDENT(T_BREAK);
1522 }
1523
1524 <ST_IN_SCRIPTING>"continue" {
1525 RETURN_TOKEN_WITH_IDENT(T_CONTINUE);
1526 }
1527
1528 <ST_IN_SCRIPTING>"goto" {
1529 RETURN_TOKEN_WITH_IDENT(T_GOTO);
1530 }
1531
1532 <ST_IN_SCRIPTING>"echo" {
1533 RETURN_TOKEN_WITH_IDENT(T_ECHO);
1534 }
1535
1536 <ST_IN_SCRIPTING>"print" {
1537 RETURN_TOKEN_WITH_IDENT(T_PRINT);
1538 }
1539
1540 <ST_IN_SCRIPTING>"class" {
1541 RETURN_TOKEN_WITH_IDENT(T_CLASS);
1542 }
1543
1544 <ST_IN_SCRIPTING>"interface" {
1545 RETURN_TOKEN_WITH_IDENT(T_INTERFACE);
1546 }
1547
1548 <ST_IN_SCRIPTING>"trait" {
1549 RETURN_TOKEN_WITH_IDENT(T_TRAIT);
1550 }
1551
1552 /*
1553 * The enum keyword must be followed by whitespace and another identifier.
1554 * This avoids the BC break of using enum in classes, namespaces, functions and constants.
1555 */
1556 <ST_IN_SCRIPTING>"enum"{WHITESPACE_OR_COMMENTS}("extends"|"implements") {
1557 yyless(4);
1558 RETURN_TOKEN_WITH_STR(T_STRING, 0);
1559 }
1560 <ST_IN_SCRIPTING>"enum"{WHITESPACE_OR_COMMENTS}[a-zA-Z_\x80-\xff] {
1561 yyless(4);
1562 RETURN_TOKEN_WITH_IDENT(T_ENUM);
1563 }
1564
1565 <ST_IN_SCRIPTING>"extends" {
1566 RETURN_TOKEN_WITH_IDENT(T_EXTENDS);
1567 }
1568
1569 <ST_IN_SCRIPTING>"implements" {
1570 RETURN_TOKEN_WITH_IDENT(T_IMPLEMENTS);
1571 }
1572
1573 <ST_IN_SCRIPTING>"->" {
1574 yy_push_state(ST_LOOKING_FOR_PROPERTY);
1575 RETURN_TOKEN(T_OBJECT_OPERATOR);
1576 }
1577
1578 <ST_IN_SCRIPTING>"?->" {
1579 yy_push_state(ST_LOOKING_FOR_PROPERTY);
1580 RETURN_TOKEN(T_NULLSAFE_OBJECT_OPERATOR);
1581 }
1582
1583 <ST_IN_SCRIPTING,ST_LOOKING_FOR_PROPERTY>{WHITESPACE}+ {
1584 goto return_whitespace;
1585 }
1586
1587 <ST_LOOKING_FOR_PROPERTY>"->" {
1588 RETURN_TOKEN(T_OBJECT_OPERATOR);
1589 }
1590
1591 <ST_LOOKING_FOR_PROPERTY>"?->" {
1592 RETURN_TOKEN(T_NULLSAFE_OBJECT_OPERATOR);
1593 }
1594
1595 <ST_LOOKING_FOR_PROPERTY>{LABEL} {
1596 yy_pop_state();
1597 RETURN_TOKEN_WITH_STR(T_STRING, 0);
1598 }
1599
1600 <ST_IN_SCRIPTING>"::" {
1601 RETURN_TOKEN(T_PAAMAYIM_NEKUDOTAYIM);
1602 }
1603
1604 <ST_IN_SCRIPTING>"..." {
1605 RETURN_TOKEN(T_ELLIPSIS);
1606 }
1607
1608 <ST_IN_SCRIPTING>"??" {
1609 RETURN_TOKEN(T_COALESCE);
1610 }
1611
1612 <ST_IN_SCRIPTING>"new" {
1613 RETURN_TOKEN_WITH_IDENT(T_NEW);
1614 }
1615
1616 <ST_IN_SCRIPTING>"clone" {
1617 RETURN_TOKEN_WITH_IDENT(T_CLONE);
1618 }
1619
1620 <ST_IN_SCRIPTING>"var" {
1621 RETURN_TOKEN_WITH_IDENT(T_VAR);
1622 }
1623
1624 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("int"|"integer"){TABS_AND_SPACES}")" {
1625 RETURN_TOKEN(T_INT_CAST);
1626 }
1627
1628 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("double"|"float"){TABS_AND_SPACES}")" {
1629 RETURN_TOKEN(T_DOUBLE_CAST);
1630 }
1631
1632 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}"real"{TABS_AND_SPACES}")" {
1633 if (PARSER_MODE()) {
1634 zend_throw_exception(zend_ce_parse_error, "The (real) cast has been removed, use (float) instead", 0);
1635 RETURN_TOKEN(T_ERROR);
1636 }
1637 RETURN_TOKEN(T_DOUBLE_CAST);
1638 }
1639
1640 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("string"|"binary"){TABS_AND_SPACES}")" {
1641 RETURN_TOKEN(T_STRING_CAST);
1642 }
1643
1644 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}"array"{TABS_AND_SPACES}")" {
1645 RETURN_TOKEN(T_ARRAY_CAST);
1646 }
1647
1648 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}"object"{TABS_AND_SPACES}")" {
1649 RETURN_TOKEN(T_OBJECT_CAST);
1650 }
1651
1652 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("bool"|"boolean"){TABS_AND_SPACES}")" {
1653 RETURN_TOKEN(T_BOOL_CAST);
1654 }
1655
1656 <ST_IN_SCRIPTING>"("{TABS_AND_SPACES}("unset"){TABS_AND_SPACES}")" {
1657 RETURN_TOKEN(T_UNSET_CAST);
1658 }
1659
1660 <ST_IN_SCRIPTING>"eval" {
1661 RETURN_TOKEN_WITH_IDENT(T_EVAL);
1662 }
1663
1664 <ST_IN_SCRIPTING>"include" {
1665 RETURN_TOKEN_WITH_IDENT(T_INCLUDE);
1666 }
1667
1668 <ST_IN_SCRIPTING>"include_once" {
1669 RETURN_TOKEN_WITH_IDENT(T_INCLUDE_ONCE);
1670 }
1671
1672 <ST_IN_SCRIPTING>"require" {
1673 RETURN_TOKEN_WITH_IDENT(T_REQUIRE);
1674 }
1675
1676 <ST_IN_SCRIPTING>"require_once" {
1677 RETURN_TOKEN_WITH_IDENT(T_REQUIRE_ONCE);
1678 }
1679
1680 <ST_IN_SCRIPTING>"namespace" {
1681 RETURN_TOKEN_WITH_IDENT(T_NAMESPACE);
1682 }
1683
1684 <ST_IN_SCRIPTING>"use" {
1685 RETURN_TOKEN_WITH_IDENT(T_USE);
1686 }
1687
1688 <ST_IN_SCRIPTING>"insteadof" {
1689 RETURN_TOKEN_WITH_IDENT(T_INSTEADOF);
1690 }
1691
1692 <ST_IN_SCRIPTING>"global" {
1693 RETURN_TOKEN_WITH_IDENT(T_GLOBAL);
1694 }
1695
1696 <ST_IN_SCRIPTING>"isset" {
1697 RETURN_TOKEN_WITH_IDENT(T_ISSET);
1698 }
1699
1700 <ST_IN_SCRIPTING>"empty" {
1701 RETURN_TOKEN_WITH_IDENT(T_EMPTY);
1702 }
1703
1704 <ST_IN_SCRIPTING>"__halt_compiler" {
1705 RETURN_TOKEN_WITH_IDENT(T_HALT_COMPILER);
1706 }
1707
1708 <ST_IN_SCRIPTING>"static" {
1709 RETURN_TOKEN_WITH_IDENT(T_STATIC);
1710 }
1711
1712 <ST_IN_SCRIPTING>"abstract" {
1713 RETURN_TOKEN_WITH_IDENT(T_ABSTRACT);
1714 }
1715
1716 <ST_IN_SCRIPTING>"final" {
1717 RETURN_TOKEN_WITH_IDENT(T_FINAL);
1718 }
1719
1720 <ST_IN_SCRIPTING>"private" {
1721 RETURN_TOKEN_WITH_IDENT(T_PRIVATE);
1722 }
1723
1724 <ST_IN_SCRIPTING>"protected" {
1725 RETURN_TOKEN_WITH_IDENT(T_PROTECTED);
1726 }
1727
1728 <ST_IN_SCRIPTING>"public(set)" {
1729 RETURN_TOKEN_WITH_IDENT(T_PUBLIC_SET);
1730 }
1731
1732 <ST_IN_SCRIPTING>"protected(set)" {
1733 RETURN_TOKEN_WITH_IDENT(T_PROTECTED_SET);
1734 }
1735
1736 <ST_IN_SCRIPTING>"private(set)" {
1737 RETURN_TOKEN_WITH_IDENT(T_PRIVATE_SET);
1738 }
1739
1740 <ST_IN_SCRIPTING>"public" {
1741 RETURN_TOKEN_WITH_IDENT(T_PUBLIC);
1742 }
1743
1744 <ST_IN_SCRIPTING>"readonly" {
1745 RETURN_TOKEN_WITH_IDENT(T_READONLY);
1746 }
1747
1748 <ST_IN_SCRIPTING>"unset" {
1749 RETURN_TOKEN_WITH_IDENT(T_UNSET);
1750 }
1751
1752 <ST_IN_SCRIPTING>"=>" {
1753 RETURN_TOKEN(T_DOUBLE_ARROW);
1754 }
1755
1756 <ST_IN_SCRIPTING>"list" {
1757 RETURN_TOKEN_WITH_IDENT(T_LIST);
1758 }
1759
1760 <ST_IN_SCRIPTING>"array" {
1761 RETURN_TOKEN_WITH_IDENT(T_ARRAY);
1762 }
1763
1764 <ST_IN_SCRIPTING>"callable" {
1765 RETURN_TOKEN_WITH_IDENT(T_CALLABLE);
1766 }
1767
1768 <ST_IN_SCRIPTING>"++" {
1769 RETURN_TOKEN(T_INC);
1770 }
1771
1772 <ST_IN_SCRIPTING>"--" {
1773 RETURN_TOKEN(T_DEC);
1774 }
1775
1776 <ST_IN_SCRIPTING>"===" {
1777 RETURN_TOKEN(T_IS_IDENTICAL);
1778 }
1779
1780 <ST_IN_SCRIPTING>"!==" {
1781 RETURN_TOKEN(T_IS_NOT_IDENTICAL);
1782 }
1783
1784 <ST_IN_SCRIPTING>"==" {
1785 RETURN_TOKEN(T_IS_EQUAL);
1786 }
1787
1788 <ST_IN_SCRIPTING>"!="|"<>" {
1789 RETURN_TOKEN(T_IS_NOT_EQUAL);
1790 }
1791
1792 <ST_IN_SCRIPTING>"<=>" {
1793 RETURN_TOKEN(T_SPACESHIP);
1794 }
1795
1796 <ST_IN_SCRIPTING>"<=" {
1797 RETURN_TOKEN(T_IS_SMALLER_OR_EQUAL);
1798 }
1799
1800 <ST_IN_SCRIPTING>">=" {
1801 RETURN_TOKEN(T_IS_GREATER_OR_EQUAL);
1802 }
1803
1804 <ST_IN_SCRIPTING>"+=" {
1805 RETURN_TOKEN(T_PLUS_EQUAL);
1806 }
1807
1808 <ST_IN_SCRIPTING>"-=" {
1809 RETURN_TOKEN(T_MINUS_EQUAL);
1810 }
1811
1812 <ST_IN_SCRIPTING>"*=" {
1813 RETURN_TOKEN(T_MUL_EQUAL);
1814 }
1815
1816 <ST_IN_SCRIPTING>"*\*" {
1817 RETURN_TOKEN(T_POW);
1818 }
1819
1820 <ST_IN_SCRIPTING>"*\*=" {
1821 RETURN_TOKEN(T_POW_EQUAL);
1822 }
1823
1824 <ST_IN_SCRIPTING>"/=" {
1825 RETURN_TOKEN(T_DIV_EQUAL);
1826 }
1827
1828 <ST_IN_SCRIPTING>".=" {
1829 RETURN_TOKEN(T_CONCAT_EQUAL);
1830 }
1831
1832 <ST_IN_SCRIPTING>"%=" {
1833 RETURN_TOKEN(T_MOD_EQUAL);
1834 }
1835
1836 <ST_IN_SCRIPTING>"<<=" {
1837 RETURN_TOKEN(T_SL_EQUAL);
1838 }
1839
1840 <ST_IN_SCRIPTING>">>=" {
1841 RETURN_TOKEN(T_SR_EQUAL);
1842 }
1843
1844 <ST_IN_SCRIPTING>"&=" {
1845 RETURN_TOKEN(T_AND_EQUAL);
1846 }
1847
1848 <ST_IN_SCRIPTING>"|=" {
1849 RETURN_TOKEN(T_OR_EQUAL);
1850 }
1851
1852 <ST_IN_SCRIPTING>"^=" {
1853 RETURN_TOKEN(T_XOR_EQUAL);
1854 }
1855
1856 <ST_IN_SCRIPTING>"??=" {
1857 RETURN_TOKEN(T_COALESCE_EQUAL);
1858 }
1859
1860 <ST_IN_SCRIPTING>"||" {
1861 RETURN_TOKEN(T_BOOLEAN_OR);
1862 }
1863
1864 <ST_IN_SCRIPTING>"&&" {
1865 RETURN_TOKEN(T_BOOLEAN_AND);
1866 }
1867
1868 <ST_IN_SCRIPTING>"OR" {
1869 RETURN_TOKEN_WITH_IDENT(T_LOGICAL_OR);
1870 }
1871
1872 <ST_IN_SCRIPTING>"AND" {
1873 RETURN_TOKEN_WITH_IDENT(T_LOGICAL_AND);
1874 }
1875
1876 <ST_IN_SCRIPTING>"XOR" {
1877 RETURN_TOKEN_WITH_IDENT(T_LOGICAL_XOR);
1878 }
1879
1880 <ST_IN_SCRIPTING>"<<" {
1881 RETURN_TOKEN(T_SL);
1882 }
1883
1884 <ST_IN_SCRIPTING>">>" {
1885 RETURN_TOKEN(T_SR);
1886 }
1887
1888 <ST_IN_SCRIPTING>"&"{OPTIONAL_WHITESPACE_OR_COMMENTS}("$"|"...") {
1889 yyless(1);
1890 RETURN_TOKEN(T_AMPERSAND_FOLLOWED_BY_VAR_OR_VARARG);
1891 }
1892
1893 <ST_IN_SCRIPTING>"&" {
1894 RETURN_TOKEN(T_AMPERSAND_NOT_FOLLOWED_BY_VAR_OR_VARARG);
1895 }
1896
1897 <ST_IN_SCRIPTING>"]"|")" {
1898 /* Check that ] and ) match up properly with a preceding [ or ( */
1899 RETURN_EXIT_NESTING_TOKEN(yytext[0]);
1900 }
1901
1902 <ST_IN_SCRIPTING>"["|"(" {
1903 enter_nesting(yytext[0]);
1904 RETURN_TOKEN(yytext[0]);
1905 }
1906
1907 <ST_IN_SCRIPTING>{TOKENS} {
1908 RETURN_TOKEN(yytext[0]);
1909 }
1910
1911
1912 <ST_IN_SCRIPTING>"{" {
1913 yy_push_state(ST_IN_SCRIPTING);
1914 enter_nesting('{');
1915 RETURN_TOKEN('{');
1916 }
1917
1918
1919 <ST_DOUBLE_QUOTES,ST_BACKQUOTE,ST_HEREDOC>"${" {
1920 yy_push_state(ST_LOOKING_FOR_VARNAME);
1921 enter_nesting('{');
1922 RETURN_TOKEN(T_DOLLAR_OPEN_CURLY_BRACES);
1923 }
1924
1925 <ST_IN_SCRIPTING>"}" {
1926 RESET_DOC_COMMENT();
1927 if (!zend_stack_is_empty(&SCNG(state_stack))) {
1928 yy_pop_state();
1929 }
1930 RETURN_EXIT_NESTING_TOKEN('}');
1931 }
1932
1933
1934 <ST_LOOKING_FOR_VARNAME>{LABEL}[[}] {
1935 yyless(yyleng - 1);
1936 yy_pop_state();
1937 yy_push_state(ST_IN_SCRIPTING);
1938 RETURN_TOKEN_WITH_STR(T_STRING_VARNAME, 0);
1939 }
1940
1941
1942 <ST_LOOKING_FOR_VARNAME>{ANY_CHAR} {
1943 yyless(0);
1944 yy_pop_state();
1945 yy_push_state(ST_IN_SCRIPTING);
1946 goto restart;
1947 }
1948
1949 <ST_IN_SCRIPTING>{BNUM} {
1950 /* The +/- 2 skips "0b" */
1951 size_t len = yyleng - 2;
1952 char *end, *bin = yytext + 2;
1953 bool contains_underscores;
1954
1955 /* Skip any leading 0s */
1956 while (len > 0 && (*bin == '0' || *bin == '_')) {
1957 ++bin;
1958 --len;
1959 }
1960
1961 contains_underscores = (memchr(bin, '_', len) != NULL);
1962
1963 if (contains_underscores) {
1964 bin = estrndup(bin, len);
1965 strip_underscores(bin, &len);
1966 }
1967
1968 if (len < SIZEOF_ZEND_LONG * 8) {
1969 if (len == 0) {
1970 ZVAL_LONG(zendlval, 0);
1971 } else {
1972 errno = 0;
1973 ZVAL_LONG(zendlval, ZEND_STRTOL(bin, &end, 2));
1974 ZEND_ASSERT(!errno && end == bin + len);
1975 }
1976 if (contains_underscores) {
1977 efree(bin);
1978 }
1979 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
1980 } else {
1981 ZVAL_DOUBLE(zendlval, zend_bin_strtod(bin, (const char **)&end));
1982 /* errno isn't checked since we allow HUGE_VAL/INF overflow */
1983 ZEND_ASSERT(end == bin + len);
1984 if (contains_underscores) {
1985 efree(bin);
1986 }
1987 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
1988 }
1989 }
1990
1991 <ST_IN_SCRIPTING>{ONUM} {
1992 /* The +/- 2 skips "0o" */
1993 size_t len = yyleng - 2;
1994 char *end, *octal = yytext + 2;
1995 bool contains_underscores = (memchr(octal, '_', len) != NULL);
1996
1997 /* Skip any leading 0s */
1998 while (len > 0 && (*octal == '0' || *octal == '_')) {
1999 ++octal;
2000 --len;
2001 }
2002
2003 if (len == 0) {
2004 ZVAL_LONG(zendlval, 0);
2005 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
2006 }
2007
2008 if (contains_underscores) {
2009 octal = estrndup(octal, len);
2010 strip_underscores(octal, &len);
2011 }
2012
2013 errno = 0;
2014
2015 ZVAL_LONG(zendlval, ZEND_STRTOL(octal, &end, 8));
2016
2017 ZEND_ASSERT(end == octal + len);
2018
2019 if (!errno) {
2020 if (contains_underscores) {
2021 efree(octal);
2022 }
2023 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
2024 }
2025
2026 /* Overflow */
2027 ZEND_ASSERT(errno == ERANGE);
2028 /* Reset errno */
2029 errno = 0;
2030
2031 /* zend_oct_strtod skips leading '0' */
2032 ZVAL_DOUBLE(zendlval, zend_oct_strtod(octal, (const char **)&end));
2033 ZEND_ASSERT(!errno);
2034 ZEND_ASSERT(end == octal + len);
2035 if (contains_underscores) {
2036 efree(octal);
2037 }
2038 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
2039 }
2040
2041 <ST_IN_SCRIPTING>{LNUM} {
2042 size_t len = yyleng;
2043 char *end, *lnum = yytext;
2044 bool is_octal = lnum[0] == '0';
2045 bool contains_underscores = (memchr(lnum, '_', len) != NULL);
2046
2047 if (contains_underscores) {
2048 lnum = estrndup(lnum, len);
2049 strip_underscores(lnum, &len);
2050 }
2051
2052 /* Digits 8 and 9 are illegal in octal literals. */
2053 if (is_octal) {
2054 size_t i;
2055 for (i = 0; i < len; i++) {
2056 if (lnum[i] == '8' || lnum[i] == '9') {
2057 zend_throw_exception(zend_ce_parse_error, "Invalid numeric literal", 0);
2058 if (PARSER_MODE()) {
2059 if (contains_underscores) {
2060 efree(lnum);
2061 }
2062 ZVAL_UNDEF(zendlval);
2063 RETURN_TOKEN(T_ERROR);
2064 }
2065
2066 /* Continue in order to determine if this is T_LNUMBER or T_DNUMBER. */
2067 len = i;
2068 break;
2069 }
2070 }
2071 }
2072
2073
2074 if (len < MAX_LENGTH_OF_LONG - 1) { /* Won't overflow */
2075 errno = 0;
2076 /* base must be passed explicitly for correct parse error on Windows */
2077 ZVAL_LONG(zendlval, ZEND_STRTOL(lnum, &end, is_octal ? 8 : 10));
2078 ZEND_ASSERT(end == lnum + len);
2079 } else {
2080 errno = 0;
2081 ZVAL_LONG(zendlval, ZEND_STRTOL(lnum, &end, 0));
2082 if (errno == ERANGE) { /* Overflow */
2083 errno = 0;
2084 if (is_octal) { /* octal overflow */
2085 ZVAL_DOUBLE(zendlval, zend_oct_strtod(lnum, (const char **)&end));
2086 } else {
2087 ZVAL_DOUBLE(zendlval, zend_strtod(lnum, (const char **)&end));
2088 }
2089 ZEND_ASSERT(end == lnum + len);
2090 if (contains_underscores) {
2091 efree(lnum);
2092 }
2093 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
2094 }
2095 ZEND_ASSERT(end == lnum + len);
2096 }
2097 ZEND_ASSERT(!errno);
2098 if (contains_underscores) {
2099 efree(lnum);
2100 }
2101 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
2102 }
2103
2104 <ST_IN_SCRIPTING>{HNUM} {
2105 /* The +/- 2 skips "0x" */
2106 size_t len = yyleng - 2;
2107 char *end, *hex = yytext + 2;
2108 bool contains_underscores;
2109
2110 /* Skip any leading 0s */
2111 while (len > 0 && (*hex == '0' || *hex == '_')) {
2112 ++hex;
2113 --len;
2114 }
2115
2116 contains_underscores = (memchr(hex, '_', len) != NULL);
2117
2118 if (contains_underscores) {
2119 hex = estrndup(hex, len);
2120 strip_underscores(hex, &len);
2121 }
2122
2123 if (len < SIZEOF_ZEND_LONG * 2 || (len == SIZEOF_ZEND_LONG * 2 && *hex <= '7')) {
2124 if (len == 0) {
2125 ZVAL_LONG(zendlval, 0);
2126 } else {
2127 errno = 0;
2128 ZVAL_LONG(zendlval, ZEND_STRTOL(hex, &end, 16));
2129 ZEND_ASSERT(!errno && end == hex + len);
2130 }
2131 if (contains_underscores) {
2132 efree(hex);
2133 }
2134 RETURN_TOKEN_WITH_VAL(T_LNUMBER);
2135 } else {
2136 ZVAL_DOUBLE(zendlval, zend_hex_strtod(hex, (const char **)&end));
2137 /* errno isn't checked since we allow HUGE_VAL/INF overflow */
2138 ZEND_ASSERT(end == hex + len);
2139 if (contains_underscores) {
2140 efree(hex);
2141 }
2142 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
2143 }
2144 }
2145
2146 <ST_VAR_OFFSET>[0]|([1-9][0-9]*) { /* Offset could be treated as a long */
2147 if (yyleng < MAX_LENGTH_OF_LONG - 1 || (yyleng == MAX_LENGTH_OF_LONG - 1 && strcmp(yytext, long_min_digits) < 0)) {
2148 char *end;
2149 errno = 0;
2150 ZVAL_LONG(zendlval, ZEND_STRTOL(yytext, &end, 10));
2151 if (errno == ERANGE) {
2152 goto string;
2153 }
2154 ZEND_ASSERT(end == yytext + yyleng);
2155 } else {
2156 string:
2157 ZVAL_STRINGL(zendlval, yytext, yyleng);
2158 }
2159 RETURN_TOKEN_WITH_VAL(T_NUM_STRING);
2160 }
2161
2162 <ST_VAR_OFFSET>{LNUM}|{HNUM}|{BNUM}|{ONUM} { /* Offset must be treated as a string */
2163 if (yyleng == 1) {
2164 ZVAL_INTERNED_STR(zendlval, ZSTR_CHAR((zend_uchar)*(yytext)));
2165 } else {
2166 ZVAL_STRINGL(zendlval, yytext, yyleng);
2167 }
2168 RETURN_TOKEN_WITH_VAL(T_NUM_STRING);
2169 }
2170
2171 <ST_IN_SCRIPTING>{DNUM}|{EXPONENT_DNUM} {
2172 const char *end;
2173 size_t len = yyleng;
2174 char *dnum = yytext;
2175 bool contains_underscores = (memchr(dnum, '_', len) != NULL);
2176
2177 if (contains_underscores) {
2178 dnum = estrndup(dnum, len);
2179 strip_underscores(dnum, &len);
2180 }
2181
2182 ZVAL_DOUBLE(zendlval, zend_strtod(dnum, &end));
2183 /* errno isn't checked since we allow HUGE_VAL/INF overflow */
2184 ZEND_ASSERT(end == dnum + len);
2185 if (contains_underscores) {
2186 efree(dnum);
2187 }
2188 RETURN_TOKEN_WITH_VAL(T_DNUMBER);
2189 }
2190
2191 <ST_IN_SCRIPTING>"__CLASS__" {
2192 RETURN_TOKEN_WITH_IDENT(T_CLASS_C);
2193 }
2194
2195 <ST_IN_SCRIPTING>"__TRAIT__" {
2196 RETURN_TOKEN_WITH_IDENT(T_TRAIT_C);
2197 }
2198
2199 <ST_IN_SCRIPTING>"__FUNCTION__" {
2200 RETURN_TOKEN_WITH_IDENT(T_FUNC_C);
2201 }
2202
2203 <ST_IN_SCRIPTING>"__PROPERTY__" {
2204 RETURN_TOKEN_WITH_IDENT(T_PROPERTY_C);
2205 }
2206
2207 <ST_IN_SCRIPTING>"__METHOD__" {
2208 RETURN_TOKEN_WITH_IDENT(T_METHOD_C);
2209 }
2210
2211 <ST_IN_SCRIPTING>"__LINE__" {
2212 RETURN_TOKEN_WITH_IDENT(T_LINE);
2213 }
2214
2215 <ST_IN_SCRIPTING>"__FILE__" {
2216 RETURN_TOKEN_WITH_IDENT(T_FILE);
2217 }
2218
2219 <ST_IN_SCRIPTING>"__DIR__" {
2220 RETURN_TOKEN_WITH_IDENT(T_DIR);
2221 }
2222
2223 <ST_IN_SCRIPTING>"__NAMESPACE__" {
2224 RETURN_TOKEN_WITH_IDENT(T_NS_C);
2225 }
2226
2227 <SHEBANG>"#!" .* {NEWLINE} {
2228 CG(zend_lineno)++;
2229 BEGIN(INITIAL);
2230 goto restart;
2231 }
2232
2233 <SHEBANG>{ANY_CHAR} {
2234 yyless(0);
2235 BEGIN(INITIAL);
2236 goto restart;
2237 }
2238
2239 <INITIAL>"<?=" {
2240 BEGIN(ST_IN_SCRIPTING);
2241 if (PARSER_MODE()) {
2242 /* We'll reject this as an identifier in zend_lex_tstring. */
2243 RETURN_TOKEN_WITH_IDENT(T_ECHO);
2244 }
2245 RETURN_TOKEN(T_OPEN_TAG_WITH_ECHO);
2246 }
2247
2248
2249 <INITIAL>"<?php"([ \t]|{NEWLINE}) {
2250 HANDLE_NEWLINE(yytext[yyleng-1]);
2251 BEGIN(ST_IN_SCRIPTING);
2252 RETURN_OR_SKIP_TOKEN(T_OPEN_TAG);
2253 }
2254
2255 <INITIAL>"<?php" {
2256 /* Allow <?php followed by end of file. */
2257 if (YYCURSOR == YYLIMIT) {
2258 BEGIN(ST_IN_SCRIPTING);
2259 RETURN_OR_SKIP_TOKEN(T_OPEN_TAG);
2260 }
2261 /* Degenerate case: <?phpX is interpreted as <? phpX with short tags. */
2262 if (CG(short_tags)) {
2263 yyless(2);
2264 BEGIN(ST_IN_SCRIPTING);
2265 RETURN_OR_SKIP_TOKEN(T_OPEN_TAG);
2266 }
2267 goto inline_char_handler;
2268 }
2269
2270 <INITIAL>"<?" {
2271 if (CG(short_tags)) {
2272 BEGIN(ST_IN_SCRIPTING);
2273 RETURN_OR_SKIP_TOKEN(T_OPEN_TAG);
2274 } else {
2275 goto inline_char_handler;
2276 }
2277 }
2278
2279 <INITIAL>{ANY_CHAR} {
2280 if (YYCURSOR > YYLIMIT) {
2281 RETURN_END_TOKEN;
2282 }
2283
2284 inline_char_handler:
2285
2286 while (1) {
2287 YYCTYPE *ptr = memchr(YYCURSOR, '<', YYLIMIT - YYCURSOR);
2288
2289 YYCURSOR = ptr ? ptr + 1 : YYLIMIT;
2290
2291 if (YYCURSOR >= YYLIMIT) {
2292 break;
2293 }
2294
2295 if (*YYCURSOR == '?') {
2296 if (CG(short_tags) /* <? */
2297 || (*(YYCURSOR + 1) == '=') /* <?= */
2298 || (!strncasecmp((char*)YYCURSOR + 1, "php", 3) && /* <?php[ \t\r\n] */
2299 (YYCURSOR + 4 == YYLIMIT ||
2300 YYCURSOR[4] == ' ' || YYCURSOR[4] == '\t' ||
2301 YYCURSOR[4] == '\n' || YYCURSOR[4] == '\r'))
2302 ) {
2303 YYCURSOR--;
2304 break;
2305 }
2306 }
2307 }
2308
2309 yyleng = YYCURSOR - SCNG(yy_text);
2310
2311 if (SCNG(output_filter)) {
2312 size_t readsize;
2313 char *s = NULL;
2314 size_t sz = 0;
2315 // TODO: avoid reallocation ???
2316 readsize = SCNG(output_filter)((unsigned char **)&s, &sz, (unsigned char *)yytext, (size_t)yyleng);
2317 ZVAL_STRINGL(zendlval, s, sz);
2318 efree(s);
2319 if (readsize < yyleng) {
2320 yyless(readsize);
2321 }
2322 } else if (yyleng == 1) {
2323 ZVAL_INTERNED_STR(zendlval, ZSTR_CHAR((zend_uchar)*yytext));
2324 } else {
2325 ZVAL_STRINGL(zendlval, yytext, yyleng);
2326 }
2327 HANDLE_NEWLINES(yytext, yyleng);
2328 RETURN_TOKEN_WITH_VAL(T_INLINE_HTML);
2329 }
2330
2331
2332 /* Make sure a label character follows "->" or "?->", otherwise there is no property
2333 * and "->"/"?->" will be taken literally
2334 */
2335 <ST_DOUBLE_QUOTES,ST_HEREDOC,ST_BACKQUOTE>"$"{LABEL}"->"[a-zA-Z_\x80-\xff] {
2336 yyless(yyleng - 3);
2337 yy_push_state(ST_LOOKING_FOR_PROPERTY);
2338 RETURN_TOKEN_WITH_STR(T_VARIABLE, 1);
2339 }
2340
2341 <ST_DOUBLE_QUOTES,ST_HEREDOC,ST_BACKQUOTE>"$"{LABEL}"?->"[a-zA-Z_\x80-\xff] {
2342 yyless(yyleng - 4);
2343 yy_push_state(ST_LOOKING_FOR_PROPERTY);
2344 RETURN_TOKEN_WITH_STR(T_VARIABLE, 1);
2345 }
2346
2347 /* A [ always designates a variable offset, regardless of what follows
2348 */
2349 <ST_DOUBLE_QUOTES,ST_HEREDOC,ST_BACKQUOTE>"$"{LABEL}"[" {
2350 yyless(yyleng - 1);
2351 yy_push_state(ST_VAR_OFFSET);
2352 RETURN_TOKEN_WITH_STR(T_VARIABLE, 1);
2353 }
2354
2355 <ST_IN_SCRIPTING,ST_DOUBLE_QUOTES,ST_HEREDOC,ST_BACKQUOTE,ST_VAR_OFFSET>"$"{LABEL} {
2356 RETURN_TOKEN_WITH_STR(T_VARIABLE, 1);
2357 }
2358
2359 <ST_VAR_OFFSET>"]" {
2360 yy_pop_state();
2361 RETURN_TOKEN(']');
2362 }
2363
2364 <ST_VAR_OFFSET>{TOKENS}|[[(){}"`] {
2365 /* Only '[' or '-' can be valid, but returning other tokens will allow a more explicit parse error */
2366 RETURN_TOKEN(yytext[0]);
2367 }
2368
2369 <ST_VAR_OFFSET>[ \n\r\t\\'#] {
2370 /* Invalid rule to return a more explicit parse error with proper line number */
2371 yyless(0);
2372 yy_pop_state();
2373 ZVAL_NULL(zendlval);
2374 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2375 }
2376
2377 <ST_IN_SCRIPTING>"namespace"("\\"{LABEL})+ {
2378 RETURN_TOKEN_WITH_STR(T_NAME_RELATIVE, sizeof("namespace\\") - 1);
2379 }
2380
2381 <ST_IN_SCRIPTING>{LABEL}("\\"{LABEL})+ {
2382 RETURN_TOKEN_WITH_STR(T_NAME_QUALIFIED, 0);
2383 }
2384
2385 <ST_IN_SCRIPTING>"\\"{LABEL}("\\"{LABEL})* {
2386 RETURN_TOKEN_WITH_STR(T_NAME_FULLY_QUALIFIED, 1);
2387 }
2388
2389 <ST_IN_SCRIPTING>"\\" {
2390 RETURN_TOKEN(T_NS_SEPARATOR);
2391 }
2392
2393 <ST_IN_SCRIPTING,ST_VAR_OFFSET>{LABEL} {
2394 RETURN_TOKEN_WITH_STR(T_STRING, 0);
2395 }
2396
2397
2398 <ST_IN_SCRIPTING,ST_LOOKING_FOR_PROPERTY>"#"|"//" {
2399 while (YYCURSOR < YYLIMIT) {
2400 switch (*YYCURSOR++) {
2401 case '\r':
2402 case '\n':
2403 YYCURSOR--;
2404 break;
2405 case '?':
2406 if (*YYCURSOR == '>') {
2407 YYCURSOR--;
2408 break;
2409 }
2410 ZEND_FALLTHROUGH;
2411 default:
2412 continue;
2413 }
2414
2415 break;
2416 }
2417
2418 yyleng = YYCURSOR - SCNG(yy_text);
2419 RETURN_OR_SKIP_TOKEN(T_COMMENT);
2420 }
2421
2422 <ST_IN_SCRIPTING,ST_LOOKING_FOR_PROPERTY>"/*"|"/**"{WHITESPACE} {
2423 int doc_com;
2424
2425 if (yyleng > 2) {
2426 doc_com = 1;
2427 RESET_DOC_COMMENT();
2428 } else {
2429 doc_com = 0;
2430 }
2431
2432 while (YYCURSOR < YYLIMIT) {
2433 if (*YYCURSOR++ == '*' && *YYCURSOR == '/') {
2434 break;
2435 }
2436 }
2437
2438 if (YYCURSOR < YYLIMIT) {
2439 YYCURSOR++;
2440 } else {
2441 zend_throw_exception_ex(zend_ce_parse_error, 0, "Unterminated comment starting line %d", CG(zend_lineno));
2442 if (PARSER_MODE()) {
2443 RETURN_TOKEN(T_ERROR);
2444 }
2445 }
2446
2447 yyleng = YYCURSOR - SCNG(yy_text);
2448 HANDLE_NEWLINES(yytext, yyleng);
2449
2450 if (doc_com) {
2451 CG(doc_comment) = zend_string_init(yytext, yyleng, 0);
2452 RETURN_OR_SKIP_TOKEN(T_DOC_COMMENT);
2453 }
2454
2455 RETURN_OR_SKIP_TOKEN(T_COMMENT);
2456 }
2457
2458 <ST_LOOKING_FOR_PROPERTY>{ANY_CHAR} {
2459 yyless(0);
2460 yy_pop_state();
2461 goto restart;
2462 }
2463
2464 <ST_IN_SCRIPTING>"?>"{NEWLINE}? {
2465 BEGIN(INITIAL);
2466 if (yytext[yyleng-1] != '>') {
2467 CG(increment_lineno) = 1;
2468 }
2469 if (PARSER_MODE()) {
2470 RETURN_TOKEN(';'); /* implicit ';' at php-end tag */
2471 }
2472 RETURN_TOKEN(T_CLOSE_TAG);
2473 }
2474
2475
2476 <ST_IN_SCRIPTING>b?['] {
2477 char *s, *t;
2478 char *end;
2479 int bprefix = (yytext[0] != '\'') ? 1 : 0;
2480
2481 while (1) {
2482 if (YYCURSOR < YYLIMIT) {
2483 if (*YYCURSOR == '\'') {
2484 YYCURSOR++;
2485 yyleng = YYCURSOR - SCNG(yy_text);
2486
2487 break;
2488 } else if (*YYCURSOR++ == '\\' && YYCURSOR < YYLIMIT) {
2489 YYCURSOR++;
2490 }
2491 } else {
2492 yyleng = YYLIMIT - SCNG(yy_text);
2493
2494 /* Unclosed single quotes; treat similar to double quotes, but without a separate token
2495 * for ' (unrecognized by parser), instead of old flex fallback to "Unexpected character..."
2496 * rule, which continued in ST_IN_SCRIPTING state after the quote */
2497 ZVAL_NULL(zendlval);
2498 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2499 }
2500 }
2501
2502 if (yyleng-bprefix-2 <= 1) {
2503 if (yyleng-bprefix-2 < 1) {
2504 ZVAL_EMPTY_STRING(zendlval);
2505 } else {
2506 zend_uchar c = (zend_uchar)*(yytext+bprefix+1);
2507 if (c == '\n' || c == '\r') {
2508 CG(zend_lineno)++;
2509 }
2510 ZVAL_INTERNED_STR(zendlval, ZSTR_CHAR(c));
2511 }
2512 goto skip_escape_conversion;
2513 }
2514 ZVAL_STRINGL(zendlval, yytext+bprefix+1, yyleng-bprefix-2);
2515
2516 /* convert escape sequences */
2517 s = Z_STRVAL_P(zendlval);
2518 end = s+Z_STRLEN_P(zendlval);
2519 while (1) {
2520 if (UNEXPECTED(*s=='\\')) {
2521 break;
2522 }
2523 if (*s == '\n' || (*s == '\r' && (*(s+1) != '\n'))) {
2524 CG(zend_lineno)++;
2525 }
2526 s++;
2527 if (s == end) {
2528 goto skip_escape_conversion;
2529 }
2530 }
2531
2532 t = s;
2533 while (s<end) {
2534 if (*s=='\\') {
2535 s++;
2536 if (*s == '\\' || *s == '\'') {
2537 *t++ = *s;
2538 } else {
2539 *t++ = '\\';
2540 *t++ = *s;
2541 }
2542 } else {
2543 *t++ = *s;
2544 }
2545 if (*s == '\n' || (*s == '\r' && (*(s+1) != '\n'))) {
2546 CG(zend_lineno)++;
2547 }
2548 s++;
2549 }
2550 *t = 0;
2551 Z_STRLEN_P(zendlval) = t - Z_STRVAL_P(zendlval);
2552
2553 skip_escape_conversion:
2554 if (SCNG(output_filter)) {
2555 size_t sz = 0;
2556 char *str = NULL;
2557 zend_string *new_str;
2558 s = Z_STRVAL_P(zendlval);
2559 // TODO: avoid reallocation ???
2560 SCNG(output_filter)((unsigned char **)&str, &sz, (unsigned char *)s, (size_t)Z_STRLEN_P(zendlval));
2561 new_str = zend_string_init(str, sz, 0);
2562 if (str != s) {
2563 efree(str);
2564 }
2565 zend_string_release_ex(Z_STR_P(zendlval), 0);
2566 ZVAL_STR(zendlval, new_str);
2567 }
2568 RETURN_TOKEN_WITH_VAL(T_CONSTANT_ENCAPSED_STRING);
2569 }
2570
2571
2572 <ST_IN_SCRIPTING>b?["] {
2573 int bprefix = (yytext[0] != '"') ? 1 : 0;
2574
2575 while (YYCURSOR < YYLIMIT) {
2576 switch (*YYCURSOR++) {
2577 case '"':
2578 yyleng = YYCURSOR - SCNG(yy_text);
2579 if (EXPECTED(zend_scan_escape_string(zendlval, yytext+bprefix+1, yyleng-bprefix-2, '"') == SUCCESS)
2580 || !PARSER_MODE()) {
2581 RETURN_TOKEN_WITH_VAL(T_CONSTANT_ENCAPSED_STRING);
2582 } else {
2583 RETURN_TOKEN(T_ERROR);
2584 }
2585 case '$':
2586 if (IS_LABEL_START(*YYCURSOR) || *YYCURSOR == '{') {
2587 break;
2588 }
2589 continue;
2590 case '{':
2591 if (*YYCURSOR == '$') {
2592 break;
2593 }
2594 continue;
2595 case '\\':
2596 if (YYCURSOR < YYLIMIT) {
2597 YYCURSOR++;
2598 }
2599 ZEND_FALLTHROUGH;
2600 default:
2601 continue;
2602 }
2603
2604 YYCURSOR--;
2605 break;
2606 }
2607
2608 /* Remember how much was scanned to save rescanning */
2609 SET_DOUBLE_QUOTES_SCANNED_LENGTH(YYCURSOR - SCNG(yy_text) - yyleng);
2610
2611 YYCURSOR = SCNG(yy_text) + yyleng;
2612
2613 BEGIN(ST_DOUBLE_QUOTES);
2614 RETURN_TOKEN('"');
2615 }
2616
2617
2618 <ST_IN_SCRIPTING>b?"<<<"{TABS_AND_SPACES}({LABEL}|([']{LABEL}['])|(["]{LABEL}["])){NEWLINE} {
2619 char *s;
2620 unsigned char *saved_cursor;
2621 int bprefix = (yytext[0] != '<') ? 1 : 0, spacing = 0, indentation = 0;
2622 zend_heredoc_label *heredoc_label = emalloc(sizeof(zend_heredoc_label));
2623 bool is_heredoc = 1;
2624
2625 CG(zend_lineno)++;
2626 heredoc_label->length = yyleng-bprefix-3-1-(yytext[yyleng-2]=='\r'?1:0);
2627 s = yytext+bprefix+3;
2628 while ((*s == ' ') || (*s == '\t')) {
2629 s++;
2630 heredoc_label->length--;
2631 }
2632
2633 if (*s == '\'') {
2634 s++;
2635 heredoc_label->length -= 2;
2636 is_heredoc = 0;
2637
2638 BEGIN(ST_NOWDOC);
2639 } else {
2640 if (*s == '"') {
2641 s++;
2642 heredoc_label->length -= 2;
2643 }
2644
2645 BEGIN(ST_HEREDOC);
2646 }
2647
2648 heredoc_label->label = estrndup(s, heredoc_label->length);
2649 heredoc_label->indentation_uses_spaces = 0;
2650 heredoc_label->indentation = 0;
2651 saved_cursor = YYCURSOR;
2652
2653 zend_ptr_stack_push(&SCNG(heredoc_label_stack), (void *) heredoc_label);
2654
2655 while (YYCURSOR < YYLIMIT && (*YYCURSOR == ' ' || *YYCURSOR == '\t')) {
2656 if (*YYCURSOR == '\t') {
2657 spacing |= HEREDOC_USING_TABS;
2658 } else {
2659 spacing |= HEREDOC_USING_SPACES;
2660 }
2661 ++YYCURSOR;
2662 ++indentation;
2663 }
2664
2665 if (YYCURSOR == YYLIMIT) {
2666 YYCURSOR = saved_cursor;
2667 RETURN_TOKEN(T_START_HEREDOC);
2668 }
2669
2670 /* Check for ending label on the next line */
2671 if (heredoc_label->length < YYLIMIT - YYCURSOR && !memcmp(YYCURSOR, s, heredoc_label->length)) {
2672 if (!IS_LABEL_SUCCESSOR(YYCURSOR[heredoc_label->length])) {
2673 if (spacing == (HEREDOC_USING_SPACES | HEREDOC_USING_TABS)) {
2674 zend_throw_exception(zend_ce_parse_error, "Invalid indentation - tabs and spaces cannot be mixed", 0);
2675 if (PARSER_MODE()) {
2676 RETURN_TOKEN(T_ERROR);
2677 }
2678 }
2679
2680 YYCURSOR = saved_cursor;
2681 heredoc_label->indentation = indentation;
2682
2683 BEGIN(ST_END_HEREDOC);
2684 RETURN_TOKEN(T_START_HEREDOC);
2685 }
2686 }
2687
2688 YYCURSOR = saved_cursor;
2689
2690 if (is_heredoc && !SCNG(heredoc_scan_ahead)) {
2691 zend_lex_state current_state;
2692 zend_string *saved_doc_comment = CG(doc_comment);
2693 int heredoc_nesting_level = 1;
2694 int first_token = 0;
2695 int error = 0;
2696
2697 zend_save_lexical_state(¤t_state);
2698
2699 SCNG(heredoc_scan_ahead) = 1;
2700 SCNG(heredoc_indentation) = 0;
2701 SCNG(heredoc_indentation_uses_spaces) = 0;
2702 LANG_SCNG(on_event) = NULL;
2703 CG(doc_comment) = NULL;
2704
2705 zend_ptr_stack_reverse_apply(¤t_state.heredoc_label_stack, copy_heredoc_label_stack);
2706
2707 zend_exception_save();
2708 while (heredoc_nesting_level) {
2709 zval zv;
2710 int retval;
2711
2712 ZVAL_UNDEF(&zv);
2713 retval = lex_scan(&zv, NULL);
2714 zval_ptr_dtor_nogc(&zv);
2715
2716 if (EG(exception)) {
2717 zend_clear_exception();
2718 break;
2719 }
2720
2721 if (!first_token) {
2722 first_token = retval;
2723 }
2724
2725 switch (retval) {
2726 case T_START_HEREDOC:
2727 ++heredoc_nesting_level;
2728 break;
2729 case T_END_HEREDOC:
2730 --heredoc_nesting_level;
2731 break;
2732 case END:
2733 heredoc_nesting_level = 0;
2734 }
2735 }
2736 zend_exception_restore();
2737
2738 if (
2739 (first_token == T_VARIABLE
2740 || first_token == T_DOLLAR_OPEN_CURLY_BRACES
2741 || first_token == T_CURLY_OPEN
2742 ) && SCNG(heredoc_indentation)) {
2743 zend_throw_exception_ex(zend_ce_parse_error, 0, "Invalid body indentation level (expecting an indentation level of at least %d)", SCNG(heredoc_indentation));
2744 error = 1;
2745 }
2746
2747 heredoc_label->indentation = SCNG(heredoc_indentation);
2748 heredoc_label->indentation_uses_spaces = SCNG(heredoc_indentation_uses_spaces);
2749
2750 zend_restore_lexical_state(¤t_state);
2751 SCNG(heredoc_scan_ahead) = 0;
2752 CG(increment_lineno) = 0;
2753 CG(doc_comment) = saved_doc_comment;
2754
2755 if (PARSER_MODE() && error) {
2756 RETURN_TOKEN(T_ERROR);
2757 }
2758 }
2759
2760 RETURN_TOKEN(T_START_HEREDOC);
2761 }
2762
2763
2764 <ST_IN_SCRIPTING>[`] {
2765 BEGIN(ST_BACKQUOTE);
2766 RETURN_TOKEN('`');
2767 }
2768
2769
2770 <ST_END_HEREDOC>{ANY_CHAR} {
2771 zend_heredoc_label *heredoc_label = zend_ptr_stack_pop(&SCNG(heredoc_label_stack));
2772
2773 yyleng = heredoc_label->indentation + heredoc_label->length;
2774 YYCURSOR += yyleng - 1;
2775
2776 heredoc_label_dtor(heredoc_label);
2777 efree(heredoc_label);
2778
2779 BEGIN(ST_IN_SCRIPTING);
2780 RETURN_TOKEN(T_END_HEREDOC);
2781 }
2782
2783
2784 <ST_DOUBLE_QUOTES,ST_BACKQUOTE,ST_HEREDOC>"{$" {
2785 yy_push_state(ST_IN_SCRIPTING);
2786 yyless(1);
2787 enter_nesting('{');
2788 RETURN_TOKEN(T_CURLY_OPEN);
2789 }
2790
2791
2792 <ST_DOUBLE_QUOTES>["] {
2793 BEGIN(ST_IN_SCRIPTING);
2794 RETURN_TOKEN('"');
2795 }
2796
2797 <ST_BACKQUOTE>[`] {
2798 BEGIN(ST_IN_SCRIPTING);
2799 RETURN_TOKEN('`');
2800 }
2801
2802
2803 <ST_DOUBLE_QUOTES>{ANY_CHAR} {
2804 if (GET_DOUBLE_QUOTES_SCANNED_LENGTH()) {
2805 YYCURSOR += GET_DOUBLE_QUOTES_SCANNED_LENGTH() - 1;
2806 SET_DOUBLE_QUOTES_SCANNED_LENGTH(0);
2807
2808 goto double_quotes_scan_done;
2809 }
2810
2811 if (YYCURSOR > YYLIMIT) {
2812 RETURN_END_TOKEN;
2813 }
2814 if (yytext[0] == '\\' && YYCURSOR < YYLIMIT) {
2815 YYCURSOR++;
2816 }
2817
2818 while (YYCURSOR < YYLIMIT) {
2819 switch (*YYCURSOR++) {
2820 case '"':
2821 break;
2822 case '$':
2823 if (IS_LABEL_START(*YYCURSOR) || *YYCURSOR == '{') {
2824 break;
2825 }
2826 continue;
2827 case '{':
2828 if (*YYCURSOR == '$') {
2829 break;
2830 }
2831 continue;
2832 case '\\':
2833 if (YYCURSOR < YYLIMIT) {
2834 YYCURSOR++;
2835 }
2836 ZEND_FALLTHROUGH;
2837 default:
2838 continue;
2839 }
2840
2841 YYCURSOR--;
2842 break;
2843 }
2844
2845 double_quotes_scan_done:
2846 yyleng = YYCURSOR - SCNG(yy_text);
2847
2848 if (EXPECTED(zend_scan_escape_string(zendlval, yytext, yyleng, '"') == SUCCESS)
2849 || !PARSER_MODE()) {
2850 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2851 } else {
2852 RETURN_TOKEN(T_ERROR);
2853 }
2854 }
2855
2856
2857 <ST_BACKQUOTE>{ANY_CHAR} {
2858 if (YYCURSOR > YYLIMIT) {
2859 RETURN_END_TOKEN;
2860 }
2861 if (yytext[0] == '\\' && YYCURSOR < YYLIMIT) {
2862 YYCURSOR++;
2863 }
2864
2865 while (YYCURSOR < YYLIMIT) {
2866 switch (*YYCURSOR++) {
2867 case '`':
2868 break;
2869 case '$':
2870 if (IS_LABEL_START(*YYCURSOR) || *YYCURSOR == '{') {
2871 break;
2872 }
2873 continue;
2874 case '{':
2875 if (*YYCURSOR == '$') {
2876 break;
2877 }
2878 continue;
2879 case '\\':
2880 if (YYCURSOR < YYLIMIT) {
2881 YYCURSOR++;
2882 }
2883 ZEND_FALLTHROUGH;
2884 default:
2885 continue;
2886 }
2887
2888 YYCURSOR--;
2889 break;
2890 }
2891
2892 yyleng = YYCURSOR - SCNG(yy_text);
2893
2894 if (EXPECTED(zend_scan_escape_string(zendlval, yytext, yyleng, '`') == SUCCESS)
2895 || !PARSER_MODE()) {
2896 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2897 } else {
2898 RETURN_TOKEN(T_ERROR);
2899 }
2900 }
2901
2902
2903 <ST_HEREDOC>{ANY_CHAR} {
2904 zend_heredoc_label *heredoc_label = zend_ptr_stack_top(&SCNG(heredoc_label_stack));
2905 int newline = 0, indentation = 0, spacing = 0;
2906
2907 if (YYCURSOR > YYLIMIT) {
2908 RETURN_END_TOKEN;
2909 }
2910
2911 YYCURSOR--;
2912
2913 while (YYCURSOR < YYLIMIT) {
2914 switch (*YYCURSOR++) {
2915 case '\r':
2916 if (*YYCURSOR == '\n') {
2917 YYCURSOR++;
2918 }
2919 ZEND_FALLTHROUGH;
2920 case '\n':
2921 indentation = spacing = 0;
2922
2923 while (YYCURSOR < YYLIMIT && (*YYCURSOR == ' ' || *YYCURSOR == '\t')) {
2924 if (*YYCURSOR == '\t') {
2925 spacing |= HEREDOC_USING_TABS;
2926 } else {
2927 spacing |= HEREDOC_USING_SPACES;
2928 }
2929 ++YYCURSOR;
2930 ++indentation;
2931 }
2932
2933 if (YYCURSOR == YYLIMIT) {
2934 yyleng = YYCURSOR - SCNG(yy_text);
2935 HANDLE_NEWLINES(yytext, yyleng);
2936 ZVAL_NULL(zendlval);
2937 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
2938 }
2939
2940 /* Check for ending label on the next line */
2941 if (IS_LABEL_START(*YYCURSOR) && heredoc_label->length < YYLIMIT - YYCURSOR && !memcmp(YYCURSOR, heredoc_label->label, heredoc_label->length)) {
2942 if (IS_LABEL_SUCCESSOR(YYCURSOR[heredoc_label->length])) {
2943 continue;
2944 }
2945
2946 if (spacing == (HEREDOC_USING_SPACES | HEREDOC_USING_TABS)) {
2947 zend_throw_exception(zend_ce_parse_error, "Invalid indentation - tabs and spaces cannot be mixed", 0);
2948 if (PARSER_MODE()) {
2949 RETURN_TOKEN(T_ERROR);
2950 }
2951 }
2952
2953 /* newline before label will be subtracted from returned text, but
2954 * yyleng/yytext will include it, for zend_highlight/strip, tokenizer, etc. */
2955 if (YYCURSOR[-indentation - 2] == '\r' && YYCURSOR[-indentation - 1] == '\n') {
2956 newline = 2; /* Windows newline */
2957 } else {
2958 newline = 1;
2959 }
2960
2961 CG(increment_lineno) = 1; /* For newline before label */
2962
2963 if (SCNG(heredoc_scan_ahead)) {
2964 SCNG(heredoc_indentation) = indentation;
2965 SCNG(heredoc_indentation_uses_spaces) = (spacing == HEREDOC_USING_SPACES);
2966 } else {
2967 YYCURSOR -= indentation;
2968 }
2969
2970 BEGIN(ST_END_HEREDOC);
2971
2972 goto heredoc_scan_done;
2973 }
2974 continue;
2975 case '$':
2976 if (IS_LABEL_START(*YYCURSOR) || *YYCURSOR == '{') {
2977 break;
2978 }
2979 continue;
2980 case '{':
2981 if (*YYCURSOR == '$') {
2982 break;
2983 }
2984 continue;
2985 case '\\':
2986 if (YYCURSOR < YYLIMIT && *YYCURSOR != '\n' && *YYCURSOR != '\r') {
2987 YYCURSOR++;
2988 }
2989 ZEND_FALLTHROUGH;
2990 default:
2991 continue;
2992 }
2993
2994 YYCURSOR--;
2995 break;
2996 }
2997
2998 heredoc_scan_done:
2999
3000 yyleng = YYCURSOR - SCNG(yy_text);
3001 ZVAL_STRINGL(zendlval, yytext, yyleng - newline);
3002
3003 if (!SCNG(heredoc_scan_ahead) && !EG(exception) && PARSER_MODE()) {
3004 bool newline_at_start = *(yytext - 1) == '\n' || *(yytext - 1) == '\r';
3005 zend_string *copy = Z_STR_P(zendlval);
3006
3007 if (!strip_multiline_string_indentation(
3008 zendlval, heredoc_label->indentation, heredoc_label->indentation_uses_spaces,
3009 newline_at_start, newline != 0)) {
3010 RETURN_TOKEN(T_ERROR);
3011 }
3012
3013 if (UNEXPECTED(zend_scan_escape_string(zendlval, ZSTR_VAL(copy), ZSTR_LEN(copy), 0) != SUCCESS)) {
3014 zend_string_efree(copy);
3015 RETURN_TOKEN(T_ERROR);
3016 }
3017
3018 zend_string_efree(copy);
3019 } else {
3020 HANDLE_NEWLINES(yytext, yyleng - newline);
3021 }
3022
3023 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
3024 }
3025
3026
3027 <ST_NOWDOC>{ANY_CHAR} {
3028 zend_heredoc_label *heredoc_label = zend_ptr_stack_top(&SCNG(heredoc_label_stack));
3029 int newline = 0, indentation = 0, spacing = -1;
3030
3031 if (YYCURSOR > YYLIMIT) {
3032 RETURN_END_TOKEN;
3033 }
3034
3035 YYCURSOR--;
3036
3037 while (YYCURSOR < YYLIMIT) {
3038 switch (*YYCURSOR++) {
3039 case '\r':
3040 if (*YYCURSOR == '\n') {
3041 YYCURSOR++;
3042 }
3043 ZEND_FALLTHROUGH;
3044 case '\n':
3045 indentation = spacing = 0;
3046
3047 while (YYCURSOR < YYLIMIT && (*YYCURSOR == ' ' || *YYCURSOR == '\t')) {
3048 if (*YYCURSOR == '\t') {
3049 spacing |= HEREDOC_USING_TABS;
3050 } else {
3051 spacing |= HEREDOC_USING_SPACES;
3052 }
3053 ++YYCURSOR;
3054 ++indentation;
3055 }
3056
3057 if (YYCURSOR == YYLIMIT) {
3058 yyleng = YYCURSOR - SCNG(yy_text);
3059 HANDLE_NEWLINES(yytext, yyleng);
3060 ZVAL_NULL(zendlval);
3061 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
3062 }
3063
3064 /* Check for ending label on the next line */
3065 if (IS_LABEL_START(*YYCURSOR) && heredoc_label->length < YYLIMIT - YYCURSOR && !memcmp(YYCURSOR, heredoc_label->label, heredoc_label->length)) {
3066 if (IS_LABEL_SUCCESSOR(YYCURSOR[heredoc_label->length])) {
3067 continue;
3068 }
3069
3070 if (spacing == (HEREDOC_USING_SPACES | HEREDOC_USING_TABS)) {
3071 zend_throw_exception(zend_ce_parse_error, "Invalid indentation - tabs and spaces cannot be mixed", 0);
3072 if (PARSER_MODE()) {
3073 RETURN_TOKEN(T_ERROR);
3074 }
3075 }
3076
3077 /* newline before label will be subtracted from returned text, but
3078 * yyleng/yytext will include it, for zend_highlight/strip, tokenizer, etc. */
3079 if (YYCURSOR[-indentation - 2] == '\r' && YYCURSOR[-indentation - 1] == '\n') {
3080 newline = 2; /* Windows newline */
3081 } else {
3082 newline = 1;
3083 }
3084
3085 CG(increment_lineno) = 1; /* For newline before label */
3086
3087 YYCURSOR -= indentation;
3088 heredoc_label->indentation = indentation;
3089
3090 BEGIN(ST_END_HEREDOC);
3091
3092 goto nowdoc_scan_done;
3093 }
3094 ZEND_FALLTHROUGH;
3095 default:
3096 continue;
3097 }
3098 }
3099
3100 nowdoc_scan_done:
3101 yyleng = YYCURSOR - SCNG(yy_text);
3102 ZVAL_STRINGL(zendlval, yytext, yyleng - newline);
3103
3104 if (!EG(exception) && spacing != -1 && PARSER_MODE()) {
3105 bool newline_at_start = *(yytext - 1) == '\n' || *(yytext - 1) == '\r';
3106 if (!strip_multiline_string_indentation(
3107 zendlval, indentation, spacing == HEREDOC_USING_SPACES,
3108 newline_at_start, newline != 0)) {
3109 RETURN_TOKEN(T_ERROR);
3110 }
3111 }
3112
3113 HANDLE_NEWLINES(yytext, yyleng - newline);
3114 RETURN_TOKEN_WITH_VAL(T_ENCAPSED_AND_WHITESPACE);
3115 }
3116
3117
3118 <ST_IN_SCRIPTING,ST_VAR_OFFSET>{ANY_CHAR} {
3119 if (YYCURSOR > YYLIMIT) {
3120 RETURN_END_TOKEN;
3121 }
3122
3123 RETURN_TOKEN(T_BAD_CHARACTER);
3124 }
3125
3126 */
3127
3128 emit_token_with_str:
3129 zend_copy_value(zendlval, (yytext + offset), (yyleng - offset));
3130
3131 emit_token_with_val:
3132 if (PARSER_MODE()) {
3133 ZEND_ASSERT(Z_TYPE_P(zendlval) != IS_UNDEF);
3134 elem->ast = zend_ast_create_zval_with_lineno(zendlval, start_line);
3135 }
3136
3137 emit_token:
3138 if (SCNG(on_event)) {
3139 SCNG(on_event)(ON_TOKEN, token, start_line, yytext, yyleng, SCNG(on_event_context));
3140 }
3141 return token;
3142
3143 emit_token_with_ident:
3144 if (PARSER_MODE()) {
3145 elem->ident = SCNG(yy_text);
3146 }
3147 if (SCNG(on_event)) {
3148 SCNG(on_event)(ON_TOKEN, token, start_line, yytext, yyleng, SCNG(on_event_context));
3149 }
3150 return token;
3151
3152 return_whitespace:
3153 HANDLE_NEWLINES(yytext, yyleng);
3154 if (SCNG(on_event)) {
3155 SCNG(on_event)(ON_TOKEN, T_WHITESPACE, start_line, yytext, yyleng, SCNG(on_event_context));
3156 }
3157 if (PARSER_MODE()) {
3158 start_line = CG(zend_lineno);
3159 goto restart;
3160 } else {
3161 return T_WHITESPACE;
3162 }
3163
3164 skip_token:
3165 if (SCNG(on_event)) {
3166 SCNG(on_event)(ON_TOKEN, token, start_line, yytext, yyleng, SCNG(on_event_context));
3167 }
3168 start_line = CG(zend_lineno);
3169 goto restart;
3170 }
3171
