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: Zeev Suraski <zeev@php.net> |
16 | Jani Taskinen <jani@php.net> |
17 | Marcus Boerger <helly@php.net> |
18 | Nuno Lopes <nlopess@php.net> |
19 | Scott MacVicar <scottmac@php.net> |
20 +----------------------------------------------------------------------+
21 */
22
23 #include <errno.h>
24 #include "zend.h"
25 #include "zend_API.h"
26 #include "zend_globals.h"
27 #include <zend_ini_parser.h>
28 #include "zend_ini_scanner.h"
29
30 #ifdef YYDEBUG
31 #undef YYDEBUG
32 #endif
33
34 #if 0
35 # define YYDEBUG(s, c) printf("state: %d char: %c\n", s, c)
36 #else
37 # define YYDEBUG(s, c)
38 #endif
39
40 #include "zend_ini_scanner_defs.h"
41
42 #define YYCTYPE unsigned char
43 /* allow the scanner to read one null byte after the end of the string (from ZEND_MMAP_AHEAD)
44 * so that if will be able to terminate to match the current token (e.g. non-enclosed string) */
45 #define YYFILL(n) { if (YYCURSOR > YYLIMIT) return 0; }
46 #define YYCURSOR SCNG(yy_cursor)
47 #define YYLIMIT SCNG(yy_limit)
48 #define YYMARKER SCNG(yy_marker)
49
50 #define YYGETCONDITION() SCNG(yy_state)
51 #define YYSETCONDITION(s) SCNG(yy_state) = s
52
53 #define STATE(name) yyc##name
54
55 /* emulate flex constructs */
56 #define BEGIN(state) YYSETCONDITION(STATE(state))
57 #define YYSTATE YYGETCONDITION()
58 #define yytext ((char*)SCNG(yy_text))
59 #define yyleng SCNG(yy_leng)
60 #define yyless(x) do { YYCURSOR = (unsigned char*)yytext + x; \
61 yyleng = (unsigned int)x; } while(0)
62
63 /* #define yymore() goto yymore_restart */
64
65 /* perform sanity check. If this message is triggered you should
66 increase the ZEND_MMAP_AHEAD value in the zend_streams.h file */
67 /*!max:re2c */
68 #if ZEND_MMAP_AHEAD < (YYMAXFILL + 1)
69 # error ZEND_MMAP_AHEAD should be greater than YYMAXFILL
70 #endif
71
72
73 /* How it works (for the core ini directives):
74 * ===========================================
75 *
76 * 1. Scanner scans file for tokens and passes them to parser.
77 * 2. Parser parses the tokens and passes the name/value pairs to the callback
78 * function which stores them in the configuration hash table.
79 * 3. Later REGISTER_INI_ENTRIES() is called which triggers the actual
80 * registering of ini entries and uses zend_get_configuration_directive()
81 * to fetch the previously stored name/value pair from configuration hash table
82 * and registers the static ini entries which match the name to the value
83 * into EG(ini_directives) hash table.
84 * 4. PATH section entries are used per-request from down to top, each overriding
85 * previous if one exists. zend_alter_ini_entry() is called for each entry.
86 * Settings in PATH section are ZEND_INI_SYSTEM accessible and thus mimics the
87 * php_admin_* directives used within Apache httpd.conf when PHP is compiled as
88 * module for Apache.
89 * 5. User defined ini files (like .htaccess for apache) are parsed for each request and
90 * stored in separate hash defined by SAPI.
91 */
92
93 /* TODO: (ordered by importance :-)
94 * ===============================================================================
95 *
96 * - Separate constant lookup totally from plain strings (using CONSTANT pattern)
97 * - Add #if .. #else .. #endif and ==, !=, <, > , <=, >= operators
98 * - Add #include "some.ini"
99 * - Allow variables to refer to options also when using parse_ini_file()
100 *
101 */
102
103 /* Globals Macros */
104 #define SCNG INI_SCNG
105 #ifdef ZTS
106 ZEND_API ts_rsrc_id ini_scanner_globals_id;
107 ZEND_API size_t ini_scanner_globals_offset;
108 #else
109 ZEND_API zend_ini_scanner_globals ini_scanner_globals;
110 #endif
111
112 #define ZEND_SYSTEM_INI CG(ini_parser_unbuffered_errors)
113
114 /* Eat leading whitespace */
115 #define EAT_LEADING_WHITESPACE() \
116 while (yyleng) { \
117 if (yytext[0] == ' ' || yytext[0] == '\t') { \
118 SCNG(yy_text)++; \
119 yyleng--; \
120 } else { \
121 break; \
122 } \
123 }
124
125 /* Eat trailing whitespace + extra char */
126 #define EAT_TRAILING_WHITESPACE_EX(ch) \
127 while (yyleng && ( \
128 (ch != 'X' && yytext[yyleng - 1] == ch) || \
129 yytext[yyleng - 1] == '\n' || \
130 yytext[yyleng - 1] == '\r' || \
131 yytext[yyleng - 1] == '\t' || \
132 yytext[yyleng - 1] == ' ') \
133 ) { \
134 yyleng--; \
135 }
136
137 /* Eat trailing whitespace */
138 #define EAT_TRAILING_WHITESPACE() EAT_TRAILING_WHITESPACE_EX('X')
139
140 #define zend_ini_copy_value(retval, str, len) \
141 ZVAL_NEW_STR(retval, zend_string_init(str, len, ZEND_SYSTEM_INI))
142
143
144 #define RETURN_TOKEN(type, str, len) { \
145 if (SCNG(scanner_mode) == ZEND_INI_SCANNER_TYPED && \
146 (YYSTATE == STATE(ST_VALUE) || YYSTATE == STATE(ST_RAW))) {\
147 zend_ini_copy_typed_value(ini_lval, type, str, len); \
148 } else { \
149 zend_ini_copy_value(ini_lval, str, len); \
150 } \
151 return type; \
152 }
153
convert_to_number(zval * retval,const char * str,const int str_len)154 static inline zend_result convert_to_number(zval *retval, const char *str, const int str_len)
155 {
156 zend_uchar type;
157 int overflow;
158 zend_long lval;
159 double dval;
160
161 if ((type = is_numeric_string_ex(str, str_len, &lval, &dval, 0, &overflow, NULL)) != 0) {
162 if (type == IS_LONG) {
163 ZVAL_LONG(retval, lval);
164 return SUCCESS;
165 } else if (type == IS_DOUBLE && !overflow) {
166 ZVAL_DOUBLE(retval, dval);
167 return SUCCESS;
168 }
169 }
170
171 return FAILURE;
172 }
173
zend_ini_copy_typed_value(zval * retval,const int type,const char * str,int len)174 static void zend_ini_copy_typed_value(zval *retval, const int type, const char *str, int len)
175 {
176 switch (type) {
177 case BOOL_FALSE:
178 case BOOL_TRUE:
179 ZVAL_BOOL(retval, type == BOOL_TRUE);
180 break;
181
182 case NULL_NULL:
183 ZVAL_NULL(retval);
184 break;
185
186 case TC_NUMBER:
187 if (convert_to_number(retval, str, len) == SUCCESS) {
188 break;
189 }
190 /* intentional fall-through */
191 default:
192 zend_ini_copy_value(retval, str, len);
193 }
194 }
195
_yy_push_state(int new_state)196 static void _yy_push_state(int new_state)
197 {
198 zend_stack_push(&SCNG(state_stack), (void *) &YYGETCONDITION());
199 YYSETCONDITION(new_state);
200 }
201
202 #define yy_push_state(state_and_tsrm) _yy_push_state(yyc##state_and_tsrm)
203
yy_pop_state(void)204 static void yy_pop_state(void)
205 {
206 int *stack_state = zend_stack_top(&SCNG(state_stack));
207 YYSETCONDITION(*stack_state);
208 zend_stack_del_top(&SCNG(state_stack));
209 }
210
yy_scan_buffer(char * str,unsigned int len)211 static void yy_scan_buffer(char *str, unsigned int len)
212 {
213 YYCURSOR = (YYCTYPE*)str;
214 SCNG(yy_start) = YYCURSOR;
215 YYLIMIT = YYCURSOR + len;
216 }
217
218 #define ini_filename SCNG(filename)
219
220 /* {{{ init_ini_scanner() */
init_ini_scanner(int scanner_mode,zend_file_handle * fh)221 static zend_result init_ini_scanner(int scanner_mode, zend_file_handle *fh)
222 {
223 /* Sanity check */
224 if (scanner_mode != ZEND_INI_SCANNER_NORMAL && scanner_mode != ZEND_INI_SCANNER_RAW && scanner_mode != ZEND_INI_SCANNER_TYPED) {
225 zend_error(E_WARNING, "Invalid scanner mode");
226 return FAILURE;
227 }
228
229 SCNG(lineno) = 1;
230 SCNG(scanner_mode) = scanner_mode;
231 SCNG(yy_in) = fh;
232
233 if (fh != NULL) {
234 ini_filename = zend_strndup(fh->filename, strlen(fh->filename));
235 } else {
236 ini_filename = NULL;
237 }
238
239 zend_stack_init(&SCNG(state_stack), sizeof(int));
240 BEGIN(INITIAL);
241
242 return SUCCESS;
243 }
244 /* }}} */
245
246 /* {{{ shutdown_ini_scanner() */
shutdown_ini_scanner(void)247 void shutdown_ini_scanner(void)
248 {
249 zend_stack_destroy(&SCNG(state_stack));
250 if (ini_filename) {
251 free(ini_filename);
252 }
253 }
254 /* }}} */
255
256 /* {{{ zend_ini_scanner_get_lineno() */
zend_ini_scanner_get_lineno(void)257 ZEND_COLD int zend_ini_scanner_get_lineno(void)
258 {
259 return SCNG(lineno);
260 }
261 /* }}} */
262
263 /* {{{ zend_ini_scanner_get_filename() */
zend_ini_scanner_get_filename(void)264 ZEND_COLD char *zend_ini_scanner_get_filename(void)
265 {
266 return ini_filename ? ini_filename : "Unknown";
267 }
268 /* }}} */
269
270 /* {{{ zend_ini_open_file_for_scanning() */
zend_ini_open_file_for_scanning(zend_file_handle * fh,int scanner_mode)271 zend_result zend_ini_open_file_for_scanning(zend_file_handle *fh, int scanner_mode)
272 {
273 char *buf;
274 size_t size;
275
276 if (zend_stream_fixup(fh, &buf, &size) == FAILURE) {
277 return FAILURE;
278 }
279
280 if (init_ini_scanner(scanner_mode, fh) == FAILURE) {
281 zend_file_handle_dtor(fh);
282 return FAILURE;
283 }
284
285 yy_scan_buffer(buf, (unsigned int)size);
286
287 return SUCCESS;
288 }
289 /* }}} */
290
291 /* {{{ zend_ini_prepare_string_for_scanning() */
zend_ini_prepare_string_for_scanning(char * str,int scanner_mode)292 zend_result zend_ini_prepare_string_for_scanning(char *str, int scanner_mode)
293 {
294 int len = (int)strlen(str);
295
296 if (init_ini_scanner(scanner_mode, NULL) == FAILURE) {
297 return FAILURE;
298 }
299
300 yy_scan_buffer(str, len);
301
302 return SUCCESS;
303 }
304 /* }}} */
305
306 /* {{{ zend_ini_escape_string() */
zend_ini_escape_string(zval * lval,char * str,int len,char quote_type)307 static void zend_ini_escape_string(zval *lval, char *str, int len, char quote_type)
308 {
309 register char *s, *t;
310 char *end;
311
312 zend_ini_copy_value(lval, str, len);
313
314 /* convert escape sequences */
315 s = t = Z_STRVAL_P(lval);
316 end = s + Z_STRLEN_P(lval);
317
318 while (s < end) {
319 if (*s == '\\') {
320 s++;
321 if (s >= end) {
322 *t++ = '\\';
323 continue;
324 }
325 switch (*s) {
326 case '"':
327 if (*s != quote_type) {
328 *t++ = '\\';
329 *t++ = *s;
330 break;
331 }
332 case '\\':
333 case '$':
334 *t++ = *s;
335 Z_STRLEN_P(lval)--;
336 break;
337 default:
338 *t++ = '\\';
339 *t++ = *s;
340 break;
341 }
342 } else {
343 *t++ = *s;
344 }
345 if (*s == '\n' || (*s == '\r' && (*(s+1) != '\n'))) {
346 SCNG(lineno)++;
347 }
348 s++;
349 }
350 *t = 0;
351 }
352 /* }}} */
353
ini_lex(zval * ini_lval)354 int ini_lex(zval *ini_lval)
355 {
356 restart:
357 SCNG(yy_text) = YYCURSOR;
358
359 /* yymore_restart: */
360 /* detect EOF */
361 if (YYCURSOR >= YYLIMIT) {
362 if (YYSTATE == STATE(ST_VALUE) || YYSTATE == STATE(ST_RAW)) {
363 BEGIN(INITIAL);
364 return END_OF_LINE;
365 }
366 return 0;
367 }
368
369 /* Eat any UTF-8 BOM we find in the first 3 bytes */
370 if (YYCURSOR == SCNG(yy_start) && YYCURSOR + 3 < YYLIMIT) {
371 if (memcmp(YYCURSOR, "\xef\xbb\xbf", 3) == 0) {
372 YYCURSOR += 3;
373 goto restart;
374 }
375 }
376 /*!re2c
377 re2c:yyfill:check = 0;
378 LNUM [0-9]+
379 DNUM ([0-9]*[\.][0-9]+)|([0-9]+[\.][0-9]*)
380 NUMBER [-]?{LNUM}|{DNUM}
381 ANY_CHAR (.|[\n\t])
382 NEWLINE ("\r"|"\n"|"\r\n")
383 TABS_AND_SPACES [ \t]
384 WHITESPACE [ \t]+
385 CONSTANT [a-zA-Z_][a-zA-Z0-9_]*
386 LABEL [^=\n\r\t;&|^$~(){}!"\[]+
387 TOKENS [:,.\[\]"'()&|^+-/*=%$!~<>?@{}]
388 OPERATORS [&|^~()!]
389 DOLLAR_CURLY "${"
390
391 SECTION_RAW_CHARS [^\]\n\r]
392 SINGLE_QUOTED_CHARS [^']
393 RAW_VALUE_CHARS [^\n\r;\000]
394
395 LITERAL_DOLLAR ("$"([^{\000]|("\\"{ANY_CHAR})))
396 VALUE_CHARS ([^$= \t\n\r;&|^~()!"'\000]|{LITERAL_DOLLAR})
397 SECTION_VALUE_CHARS ([^$\n\r;"'\]\\]|("\\"{ANY_CHAR})|{LITERAL_DOLLAR})
398
399 <!*> := yyleng = YYCURSOR - SCNG(yy_text);
400
401 <INITIAL>"[" { /* Section start */
402 /* Enter section data lookup state */
403 if (SCNG(scanner_mode) == ZEND_INI_SCANNER_RAW) {
404 BEGIN(ST_SECTION_RAW);
405 } else {
406 BEGIN(ST_SECTION_VALUE);
407 }
408 return TC_SECTION;
409 }
410
411 <ST_VALUE,ST_SECTION_VALUE,ST_OFFSET>"'"{SINGLE_QUOTED_CHARS}+"'" { /* Raw string */
412 /* Eat leading and trailing single quotes */
413 if (yytext[0] == '\'' && yytext[yyleng - 1] == '\'') {
414 SCNG(yy_text)++;
415 yyleng = yyleng - 2;
416 }
417 RETURN_TOKEN(TC_RAW, yytext, yyleng);
418 }
419
420 <ST_SECTION_RAW,ST_SECTION_VALUE>"]"{TABS_AND_SPACES}*{NEWLINE}? { /* End of section */
421 BEGIN(INITIAL);
422 SCNG(lineno)++;
423 return ']';
424 }
425
426 <INITIAL>{LABEL}"["{TABS_AND_SPACES}* { /* Start of option with offset */
427 /* Eat leading whitespace */
428 EAT_LEADING_WHITESPACE();
429
430 /* Eat trailing whitespace and [ */
431 EAT_TRAILING_WHITESPACE_EX('[');
432
433 /* Enter offset lookup state */
434 BEGIN(ST_OFFSET);
435
436 RETURN_TOKEN(TC_OFFSET, yytext, yyleng);
437 }
438
439 <ST_OFFSET>{TABS_AND_SPACES}*"]" { /* End of section or an option offset */
440 BEGIN(INITIAL);
441 return ']';
442 }
443
444 <ST_DOUBLE_QUOTES,ST_SECTION_VALUE,ST_VALUE,ST_OFFSET>{DOLLAR_CURLY} { /* Variable start */
445 yy_push_state(ST_VARNAME);
446 return TC_DOLLAR_CURLY;
447 }
448
449 <ST_VARNAME>{LABEL} { /* Variable name */
450 /* Eat leading whitespace */
451 EAT_LEADING_WHITESPACE();
452
453 /* Eat trailing whitespace */
454 EAT_TRAILING_WHITESPACE();
455
456 RETURN_TOKEN(TC_VARNAME, yytext, yyleng);
457 }
458
459 <ST_VARNAME>"}" { /* Variable end */
460 yy_pop_state();
461 return '}';
462 }
463
464 <INITIAL,ST_VALUE>("true"|"on"|"yes"){TABS_AND_SPACES}* { /* TRUE value (when used outside option value/offset this causes parse error!) */
465 RETURN_TOKEN(BOOL_TRUE, "1", 1);
466 }
467
468 <INITIAL,ST_VALUE>("false"|"off"|"no"|"none"){TABS_AND_SPACES}* { /* FALSE value (when used outside option value/offset this causes parse error!)*/
469 RETURN_TOKEN(BOOL_FALSE, "", 0);
470 }
471
472 <INITIAL,ST_VALUE>("null"){TABS_AND_SPACES}* {
473 RETURN_TOKEN(NULL_NULL, "", 0);
474 }
475
476 <INITIAL>{LABEL} { /* Get option name */
477 /* Eat leading whitespace */
478 EAT_LEADING_WHITESPACE();
479
480 /* Eat trailing whitespace */
481 EAT_TRAILING_WHITESPACE();
482
483 RETURN_TOKEN(TC_LABEL, yytext, yyleng);
484 }
485
486 <INITIAL>{TABS_AND_SPACES}*[=]{TABS_AND_SPACES}* { /* Start option value */
487 if (SCNG(scanner_mode) == ZEND_INI_SCANNER_RAW) {
488 BEGIN(ST_RAW);
489 } else {
490 BEGIN(ST_VALUE);
491 }
492 return '=';
493 }
494
495 <ST_RAW>{RAW_VALUE_CHARS} { /* Raw value, only used when SCNG(scanner_mode) == ZEND_INI_SCANNER_RAW. */
496 unsigned char *sc = NULL;
497 EAT_LEADING_WHITESPACE();
498 while (YYCURSOR < YYLIMIT) {
499 switch (*YYCURSOR) {
500 case '\n':
501 case '\r':
502 goto end_raw_value_chars;
503 break;
504 case ';':
505 if (sc == NULL) {
506 sc = YYCURSOR;
507 }
508 YYCURSOR++;
509 break;
510 case '"':
511 if (yytext[0] == '"') {
512 sc = NULL;
513 }
514 YYCURSOR++;
515 break;
516 default:
517 YYCURSOR++;
518 break;
519 }
520 }
521 end_raw_value_chars:
522 if (sc) {
523 yyleng = sc - SCNG(yy_text);
524 } else {
525 yyleng = YYCURSOR - SCNG(yy_text);
526 }
527
528 EAT_TRAILING_WHITESPACE();
529
530 /* Eat leading and trailing double quotes */
531 if (yyleng > 1 && yytext[0] == '"' && yytext[yyleng - 1] == '"') {
532 SCNG(yy_text)++;
533 yyleng = yyleng - 2;
534 }
535
536 RETURN_TOKEN(TC_RAW, yytext, yyleng);
537 }
538
539 <ST_SECTION_RAW>{SECTION_RAW_CHARS}+ { /* Raw value, only used when SCNG(scanner_mode) == ZEND_INI_SCANNER_RAW. */
540 RETURN_TOKEN(TC_RAW, yytext, yyleng);
541 }
542
543 <ST_VALUE,ST_RAW>{TABS_AND_SPACES}*{NEWLINE} { /* End of option value */
544 BEGIN(INITIAL);
545 SCNG(lineno)++;
546 return END_OF_LINE;
547 }
548
549 <ST_SECTION_VALUE,ST_VALUE,ST_OFFSET>{CONSTANT} { /* Get constant option value */
550 RETURN_TOKEN(TC_CONSTANT, yytext, yyleng);
551 }
552
553 <ST_SECTION_VALUE,ST_VALUE,ST_OFFSET>{NUMBER} { /* Get number option value as string */
554 RETURN_TOKEN(TC_NUMBER, yytext, yyleng);
555 }
556
557 <INITIAL>{TOKENS} { /* Disallow these chars outside option values */
558 return yytext[0];
559 }
560
561 <ST_VALUE>{OPERATORS}{TABS_AND_SPACES}* { /* Boolean operators */
562 return yytext[0];
563 }
564
565 <ST_VALUE>[=] { /* Make = used in option value to trigger error */
566 yyless(0);
567 BEGIN(INITIAL);
568 return END_OF_LINE;
569 }
570
571 <ST_VALUE>{VALUE_CHARS}+ { /* Get everything else as option/offset value */
572 RETURN_TOKEN(TC_STRING, yytext, yyleng);
573 }
574
575 <ST_SECTION_VALUE,ST_OFFSET>{SECTION_VALUE_CHARS}+ { /* Get rest as section/offset value */
576 RETURN_TOKEN(TC_STRING, yytext, yyleng);
577 }
578
579 <ST_SECTION_VALUE,ST_VALUE,ST_OFFSET>{TABS_AND_SPACES}*["] { /* Double quoted '"' string start */
580 yy_push_state(ST_DOUBLE_QUOTES);
581 return '"';
582 }
583
584 <ST_DOUBLE_QUOTES>["]{TABS_AND_SPACES}* { /* Double quoted '"' string ends */
585 yy_pop_state();
586 return '"';
587 }
588
589 <ST_DOUBLE_QUOTES>[^] { /* Escape double quoted string contents */
590 if (YYCURSOR > YYLIMIT) {
591 return 0;
592 }
593
594 while (YYCURSOR < YYLIMIT) {
595 switch (*YYCURSOR++) {
596 case '"':
597 if (YYCURSOR < YYLIMIT && YYCURSOR[-2] == '\\' && *YYCURSOR != '\r' && *YYCURSOR != '\n') {
598 continue;
599 }
600 break;
601 case '$':
602 if (*YYCURSOR == '{') {
603 break;
604 }
605 continue;
606 case '\\':
607 if (YYCURSOR < YYLIMIT && *YYCURSOR != '"') {
608 YYCURSOR++;
609 }
610 /* fall through */
611 default:
612 continue;
613 }
614
615 YYCURSOR--;
616 break;
617 }
618
619 yyleng = YYCURSOR - SCNG(yy_text);
620
621 zend_ini_escape_string(ini_lval, yytext, yyleng, '"');
622 return TC_QUOTED_STRING;
623 }
624
625 <ST_SECTION_VALUE,ST_VALUE,ST_OFFSET>{WHITESPACE} {
626 RETURN_TOKEN(TC_WHITESPACE, yytext, yyleng);
627 }
628
629 <INITIAL,ST_RAW>{TABS_AND_SPACES}+ {
630 /* eat whitespace */
631 goto restart;
632 }
633
634 <INITIAL>{TABS_AND_SPACES}*{NEWLINE} {
635 SCNG(lineno)++;
636 return END_OF_LINE;
637 }
638
639 <INITIAL,ST_VALUE,ST_RAW>{TABS_AND_SPACES}*[;][^\r\n]*{NEWLINE} { /* Comment */
640 BEGIN(INITIAL);
641 SCNG(lineno)++;
642 return END_OF_LINE;
643 }
644
645 <ST_VALUE,ST_RAW>[^] { /* End of option value (if EOF is reached before EOL */
646 BEGIN(INITIAL);
647 return END_OF_LINE;
648 }
649
650 <*>[^] {
651 return 0;
652 }
653
654 */
655 }
656
