Mercurial > hg > soundsoftware-site
comparison vendor/gems/coderay-0.9.7/lib/coderay/scanners/.svn/text-base/php.rb.svn-base @ 210:0579821a129a
Update to Redmine trunk rev 4802
author | Chris Cannam |
---|---|
date | Tue, 08 Feb 2011 13:51:46 +0000 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
128:07fa8a8b56a8 | 210:0579821a129a |
---|---|
1 module CodeRay | |
2 module Scanners | |
3 | |
4 load :html | |
5 | |
6 # Original by Stefan Walk. | |
7 class PHP < Scanner | |
8 | |
9 register_for :php | |
10 file_extension 'php' | |
11 | |
12 KINDS_NOT_LOC = HTML::KINDS_NOT_LOC | |
13 | |
14 def setup | |
15 @html_scanner = CodeRay.scanner :html, :tokens => @tokens, :keep_tokens => true, :keep_state => true | |
16 end | |
17 | |
18 def reset_instance | |
19 super | |
20 @html_scanner.reset | |
21 end | |
22 | |
23 module Words | |
24 | |
25 # according to http://www.php.net/manual/en/reserved.keywords.php | |
26 KEYWORDS = %w[ | |
27 abstract and array as break case catch class clone const continue declare default do else elseif | |
28 enddeclare endfor endforeach endif endswitch endwhile extends final for foreach function global | |
29 goto if implements interface instanceof namespace new or private protected public static switch | |
30 throw try use var while xor | |
31 cfunction old_function | |
32 ] | |
33 | |
34 TYPES = %w[ int integer float double bool boolean string array object resource ] | |
35 | |
36 LANGUAGE_CONSTRUCTS = %w[ | |
37 die echo empty exit eval include include_once isset list | |
38 require require_once return print unset | |
39 ] | |
40 | |
41 CLASSES = %w[ Directory stdClass __PHP_Incomplete_Class exception php_user_filter Closure ] | |
42 | |
43 # according to http://php.net/quickref.php on 2009-04-21; | |
44 # all functions with _ excluded (module functions) and selected additional functions | |
45 BUILTIN_FUNCTIONS = %w[ | |
46 abs acos acosh addcslashes addslashes aggregate array arsort ascii2ebcdic asin asinh asort assert atan atan2 | |
47 atanh basename bcadd bccomp bcdiv bcmod bcmul bcpow bcpowmod bcscale bcsqrt bcsub bin2hex bindec | |
48 bindtextdomain bzclose bzcompress bzdecompress bzerrno bzerror bzerrstr bzflush bzopen bzread bzwrite | |
49 calculhmac ceil chdir checkdate checkdnsrr chgrp chmod chop chown chr chroot clearstatcache closedir closelog | |
50 compact constant copy cos cosh count crc32 crypt current date dcgettext dcngettext deaggregate decbin dechex | |
51 decoct define defined deg2rad delete dgettext die dirname diskfreespace dl dngettext doubleval each | |
52 ebcdic2ascii echo empty end ereg eregi escapeshellarg escapeshellcmd eval exec exit exp explode expm1 extract | |
53 fclose feof fflush fgetc fgetcsv fgets fgetss file fileatime filectime filegroup fileinode filemtime fileowner | |
54 fileperms filepro filesize filetype floatval flock floor flush fmod fnmatch fopen fpassthru fprintf fputcsv | |
55 fputs fread frenchtojd fscanf fseek fsockopen fstat ftell ftok ftruncate fwrite getallheaders getcwd getdate | |
56 getenv gethostbyaddr gethostbyname gethostbynamel getimagesize getlastmod getmxrr getmygid getmyinode getmypid | |
57 getmyuid getopt getprotobyname getprotobynumber getrandmax getrusage getservbyname getservbyport gettext | |
58 gettimeofday gettype glob gmdate gmmktime gmstrftime gregoriantojd gzclose gzcompress gzdecode gzdeflate | |
59 gzencode gzeof gzfile gzgetc gzgets gzgetss gzinflate gzopen gzpassthru gzputs gzread gzrewind gzseek gztell | |
60 gzuncompress gzwrite hash header hebrev hebrevc hexdec htmlentities htmlspecialchars hypot iconv idate | |
61 implode include intval ip2long iptcembed iptcparse isset | |
62 jddayofweek jdmonthname jdtofrench jdtogregorian jdtojewish jdtojulian jdtounix jewishtojd join jpeg2wbmp | |
63 juliantojd key krsort ksort lcfirst lchgrp lchown levenshtein link linkinfo list localeconv localtime log | |
64 log10 log1p long2ip lstat ltrim mail main max md5 metaphone mhash microtime min mkdir mktime msql natcasesort | |
65 natsort next ngettext nl2br nthmac octdec opendir openlog | |
66 ord overload pack passthru pathinfo pclose pfsockopen phpcredits phpinfo phpversion pi png2wbmp popen pos pow | |
67 prev print printf putenv quotemeta rad2deg rand range rawurldecode rawurlencode readdir readfile readgzfile | |
68 readline readlink realpath recode rename require reset rewind rewinddir rmdir round rsort rtrim scandir | |
69 serialize setcookie setlocale setrawcookie settype sha1 shuffle signeurlpaiement sin sinh sizeof sleep snmpget | |
70 snmpgetnext snmprealwalk snmpset snmpwalk snmpwalkoid sort soundex split spliti sprintf sqrt srand sscanf stat | |
71 strcasecmp strchr strcmp strcoll strcspn strftime stripcslashes stripos stripslashes stristr strlen | |
72 strnatcasecmp strnatcmp strncasecmp strncmp strpbrk strpos strptime strrchr strrev strripos strrpos strspn | |
73 strstr strtok strtolower strtotime strtoupper strtr strval substr symlink syslog system tan tanh tempnam | |
74 textdomain time tmpfile touch trim uasort ucfirst ucwords uksort umask uniqid unixtojd unlink unpack | |
75 unserialize unset urldecode urlencode usleep usort vfprintf virtual vprintf vsprintf wordwrap | |
76 array_change_key_case array_chunk array_combine array_count_values array_diff array_diff_assoc | |
77 array_diff_key array_diff_uassoc array_diff_ukey array_fill array_fill_keys array_filter array_flip | |
78 array_intersect array_intersect_assoc array_intersect_key array_intersect_uassoc array_intersect_ukey | |
79 array_key_exists array_keys array_map array_merge array_merge_recursive array_multisort array_pad | |
80 array_pop array_product array_push array_rand array_reduce array_reverse array_search array_shift | |
81 array_slice array_splice array_sum array_udiff array_udiff_assoc array_udiff_uassoc array_uintersect | |
82 array_uintersect_assoc array_uintersect_uassoc array_unique array_unshift array_values array_walk | |
83 array_walk_recursive | |
84 assert_options base_convert base64_decode base64_encode | |
85 chunk_split class_exists class_implements class_parents | |
86 count_chars debug_backtrace debug_print_backtrace debug_zval_dump | |
87 error_get_last error_log error_reporting extension_loaded | |
88 file_exists file_get_contents file_put_contents load_file | |
89 func_get_arg func_get_args func_num_args function_exists | |
90 get_browser get_called_class get_cfg_var get_class get_class_methods get_class_vars | |
91 get_current_user get_declared_classes get_declared_interfaces get_defined_constants | |
92 get_defined_functions get_defined_vars get_extension_funcs get_headers get_html_translation_table | |
93 get_include_path get_included_files get_loaded_extensions get_magic_quotes_gpc get_magic_quotes_runtime | |
94 get_meta_tags get_object_vars get_parent_class get_required_filesget_resource_type | |
95 gc_collect_cycles gc_disable gc_enable gc_enabled | |
96 halt_compiler headers_list headers_sent highlight_file highlight_string | |
97 html_entity_decode htmlspecialchars_decode | |
98 in_array include_once inclued_get_data | |
99 is_a is_array is_binary is_bool is_buffer is_callable is_dir is_double is_executable is_file is_finite | |
100 is_float is_infinite is_int is_integer is_link is_long is_nan is_null is_numeric is_object is_readable | |
101 is_real is_resource is_scalar is_soap_fault is_string is_subclass_of is_unicode is_uploaded_file | |
102 is_writable is_writeable | |
103 locale_get_default locale_set_default | |
104 number_format override_function parse_str parse_url | |
105 php_check_syntax php_ini_loaded_file php_ini_scanned_files php_logo_guid php_sapi_name | |
106 php_strip_whitespace php_uname | |
107 preg_filter preg_grep preg_last_error preg_match preg_match_all preg_quote preg_replace | |
108 preg_replace_callback preg_split print_r | |
109 require_once register_shutdown_function register_tick_function | |
110 set_error_handler set_exception_handler set_file_buffer set_include_path | |
111 set_magic_quotes_runtime set_time_limit shell_exec | |
112 str_getcsv str_ireplace str_pad str_repeat str_replace str_rot13 str_shuffle str_split str_word_count | |
113 strip_tags substr_compare substr_count substr_replace | |
114 time_nanosleep time_sleep_until | |
115 token_get_all token_name trigger_error | |
116 unregister_tick_function use_soap_error_handler user_error | |
117 utf8_decode utf8_encode var_dump var_export | |
118 version_compare | |
119 zend_logo_guid zend_thread_id zend_version | |
120 create_function call_user_func_array | |
121 posix_access posix_ctermid posix_get_last_error posix_getcwd posix_getegid | |
122 posix_geteuid posix_getgid posix_getgrgid posix_getgrnam posix_getgroups | |
123 posix_getlogin posix_getpgid posix_getpgrp posix_getpid posix_getppid | |
124 posix_getpwnam posix_getpwuid posix_getrlimit posix_getsid posix_getuid | |
125 posix_initgroups posix_isatty posix_kill posix_mkfifo posix_mknod | |
126 posix_setegid posix_seteuid posix_setgid posix_setpgid posix_setsid | |
127 posix_setuid posix_strerror posix_times posix_ttyname posix_uname | |
128 pcntl_alarm pcntl_exec pcntl_fork pcntl_getpriority pcntl_setpriority | |
129 pcntl_signal pcntl_signal_dispatch pcntl_sigprocmask pcntl_sigtimedwait | |
130 pcntl_sigwaitinfo pcntl_wait pcntl_waitpid pcntl_wexitstatus pcntl_wifexited | |
131 pcntl_wifsignaled pcntl_wifstopped pcntl_wstopsig pcntl_wtermsig | |
132 ] | |
133 # TODO: more built-in PHP functions? | |
134 | |
135 EXCEPTIONS = %w[ | |
136 E_ERROR E_WARNING E_PARSE E_NOTICE E_CORE_ERROR E_CORE_WARNING E_COMPILE_ERROR E_COMPILE_WARNING | |
137 E_USER_ERROR E_USER_WARNING E_USER_NOTICE E_DEPRECATED E_USER_DEPRECATED E_ALL E_STRICT | |
138 ] | |
139 | |
140 CONSTANTS = %w[ | |
141 null true false self parent | |
142 __LINE__ __DIR__ __FILE__ __LINE__ | |
143 __CLASS__ __NAMESPACE__ __METHOD__ __FUNCTION__ | |
144 PHP_VERSION PHP_MAJOR_VERSION PHP_MINOR_VERSION PHP_RELEASE_VERSION PHP_VERSION_ID PHP_EXTRA_VERSION PHP_ZTS | |
145 PHP_DEBUG PHP_MAXPATHLEN PHP_OS PHP_SAPI PHP_EOL PHP_INT_MAX PHP_INT_SIZE DEFAULT_INCLUDE_PATH | |
146 PEAR_INSTALL_DIR PEAR_EXTENSION_DIR PHP_EXTENSION_DIR PHP_PREFIX PHP_BINDIR PHP_LIBDIR PHP_DATADIR | |
147 PHP_SYSCONFDIR PHP_LOCALSTATEDIR PHP_CONFIG_FILE_PATH PHP_CONFIG_FILE_SCAN_DIR PHP_SHLIB_SUFFIX | |
148 PHP_OUTPUT_HANDLER_START PHP_OUTPUT_HANDLER_CONT PHP_OUTPUT_HANDLER_END | |
149 __COMPILER_HALT_OFFSET__ | |
150 EXTR_OVERWRITE EXTR_SKIP EXTR_PREFIX_SAME EXTR_PREFIX_ALL EXTR_PREFIX_INVALID EXTR_PREFIX_IF_EXISTS | |
151 EXTR_IF_EXISTS SORT_ASC SORT_DESC SORT_REGULAR SORT_NUMERIC SORT_STRING CASE_LOWER CASE_UPPER COUNT_NORMAL | |
152 COUNT_RECURSIVE ASSERT_ACTIVE ASSERT_CALLBACK ASSERT_BAIL ASSERT_WARNING ASSERT_QUIET_EVAL CONNECTION_ABORTED | |
153 CONNECTION_NORMAL CONNECTION_TIMEOUT INI_USER INI_PERDIR INI_SYSTEM INI_ALL M_E M_LOG2E M_LOG10E M_LN2 M_LN10 | |
154 M_PI M_PI_2 M_PI_4 M_1_PI M_2_PI M_2_SQRTPI M_SQRT2 M_SQRT1_2 CRYPT_SALT_LENGTH CRYPT_STD_DES CRYPT_EXT_DES | |
155 CRYPT_MD5 CRYPT_BLOWFISH DIRECTORY_SEPARATOR SEEK_SET SEEK_CUR SEEK_END LOCK_SH LOCK_EX LOCK_UN LOCK_NB | |
156 HTML_SPECIALCHARS HTML_ENTITIES ENT_COMPAT ENT_QUOTES ENT_NOQUOTES INFO_GENERAL INFO_CREDITS | |
157 INFO_CONFIGURATION INFO_MODULES INFO_ENVIRONMENT INFO_VARIABLES INFO_LICENSE INFO_ALL CREDITS_GROUP | |
158 CREDITS_GENERAL CREDITS_SAPI CREDITS_MODULES CREDITS_DOCS CREDITS_FULLPAGE CREDITS_QA CREDITS_ALL STR_PAD_LEFT | |
159 STR_PAD_RIGHT STR_PAD_BOTH PATHINFO_DIRNAME PATHINFO_BASENAME PATHINFO_EXTENSION PATH_SEPARATOR CHAR_MAX | |
160 LC_CTYPE LC_NUMERIC LC_TIME LC_COLLATE LC_MONETARY LC_ALL LC_MESSAGES ABDAY_1 ABDAY_2 ABDAY_3 ABDAY_4 ABDAY_5 | |
161 ABDAY_6 ABDAY_7 DAY_1 DAY_2 DAY_3 DAY_4 DAY_5 DAY_6 DAY_7 ABMON_1 ABMON_2 ABMON_3 ABMON_4 ABMON_5 ABMON_6 | |
162 ABMON_7 ABMON_8 ABMON_9 ABMON_10 ABMON_11 ABMON_12 MON_1 MON_2 MON_3 MON_4 MON_5 MON_6 MON_7 MON_8 MON_9 | |
163 MON_10 MON_11 MON_12 AM_STR PM_STR D_T_FMT D_FMT T_FMT T_FMT_AMPM ERA ERA_YEAR ERA_D_T_FMT ERA_D_FMT ERA_T_FMT | |
164 ALT_DIGITS INT_CURR_SYMBOL CURRENCY_SYMBOL CRNCYSTR MON_DECIMAL_POINT MON_THOUSANDS_SEP MON_GROUPING | |
165 POSITIVE_SIGN NEGATIVE_SIGN INT_FRAC_DIGITS FRAC_DIGITS P_CS_PRECEDES P_SEP_BY_SPACE N_CS_PRECEDES | |
166 N_SEP_BY_SPACE P_SIGN_POSN N_SIGN_POSN DECIMAL_POINT RADIXCHAR THOUSANDS_SEP THOUSEP GROUPING YESEXPR NOEXPR | |
167 YESSTR NOSTR CODESET LOG_EMERG LOG_ALERT LOG_CRIT LOG_ERR LOG_WARNING LOG_NOTICE LOG_INFO LOG_DEBUG LOG_KERN | |
168 LOG_USER LOG_MAIL LOG_DAEMON LOG_AUTH LOG_SYSLOG LOG_LPR LOG_NEWS LOG_UUCP LOG_CRON LOG_AUTHPRIV LOG_LOCAL0 | |
169 LOG_LOCAL1 LOG_LOCAL2 LOG_LOCAL3 LOG_LOCAL4 LOG_LOCAL5 LOG_LOCAL6 LOG_LOCAL7 LOG_PID LOG_CONS LOG_ODELAY | |
170 LOG_NDELAY LOG_NOWAIT LOG_PERROR | |
171 ] | |
172 | |
173 PREDEFINED = %w[ | |
174 $GLOBALS $_SERVER $_GET $_POST $_FILES $_REQUEST $_SESSION $_ENV | |
175 $_COOKIE $php_errormsg $HTTP_RAW_POST_DATA $http_response_header | |
176 $argc $argv | |
177 ] | |
178 | |
179 IDENT_KIND = CaseIgnoringWordList.new(:ident). | |
180 add(KEYWORDS, :reserved). | |
181 add(TYPES, :pre_type). | |
182 add(LANGUAGE_CONSTRUCTS, :reserved). | |
183 add(BUILTIN_FUNCTIONS, :predefined). | |
184 add(CLASSES, :pre_constant). | |
185 add(EXCEPTIONS, :exception). | |
186 add(CONSTANTS, :pre_constant) | |
187 | |
188 VARIABLE_KIND = WordList.new(:local_variable). | |
189 add(PREDEFINED, :predefined) | |
190 end | |
191 | |
192 module RE | |
193 | |
194 PHP_START = / | |
195 <script\s+[^>]*?language\s*=\s*"php"[^>]*?> | | |
196 <script\s+[^>]*?language\s*=\s*'php'[^>]*?> | | |
197 <\?php\d? | | |
198 <\?(?!xml) | |
199 /xi | |
200 | |
201 PHP_END = %r! | |
202 </script> | | |
203 \?> | |
204 !xi | |
205 | |
206 HTML_INDICATOR = /<!DOCTYPE html|<(?:html|body|div|p)[> ]/i | |
207 | |
208 IDENTIFIER = /[a-z_\x7f-\xFF][a-z0-9_\x7f-\xFF]*/i | |
209 VARIABLE = /\$#{IDENTIFIER}/ | |
210 | |
211 OPERATOR = / | |
212 \.(?!\d)=? | # dot that is not decimal point, string concatenation | |
213 && | \|\| | # logic | |
214 :: | -> | => | # scope, member, dictionary | |
215 \\(?!\n) | # namespace | |
216 \+\+ | -- | # increment, decrement | |
217 [,;?:()\[\]{}] | # simple delimiters | |
218 [-+*\/%&|^]=? | # ordinary math, binary logic, assignment shortcuts | |
219 [~$] | # whatever | |
220 =& | # reference assignment | |
221 [=!]=?=? | <> | # comparison and assignment | |
222 <<=? | >>=? | [<>]=? # comparison and shift | |
223 /x | |
224 | |
225 end | |
226 | |
227 def scan_tokens tokens, options | |
228 if string.respond_to?(:encoding) | |
229 unless string.encoding == Encoding::ASCII_8BIT | |
230 self.string = string.encode Encoding::ASCII_8BIT, | |
231 :invalid => :replace, :undef => :replace, :replace => '?' | |
232 end | |
233 end | |
234 | |
235 if check(RE::PHP_START) || # starts with <? | |
236 (match?(/\s*<\S/) && exist?(RE::PHP_START)) || # starts with tag and contains <? | |
237 exist?(RE::HTML_INDICATOR) || | |
238 check(/.{1,100}#{RE::PHP_START}/om) # PHP start after max 100 chars | |
239 # is HTML with embedded PHP, so start with HTML | |
240 states = [:initial] | |
241 else | |
242 # is just PHP, so start with PHP surrounded by HTML | |
243 states = [:initial, :php] | |
244 end | |
245 | |
246 label_expected = true | |
247 case_expected = false | |
248 | |
249 heredoc_delimiter = nil | |
250 delimiter = nil | |
251 modifier = nil | |
252 | |
253 until eos? | |
254 | |
255 match = nil | |
256 kind = nil | |
257 | |
258 case states.last | |
259 | |
260 when :initial # HTML | |
261 if scan RE::PHP_START | |
262 kind = :inline_delimiter | |
263 label_expected = true | |
264 states << :php | |
265 else | |
266 match = scan_until(/(?=#{RE::PHP_START})/o) || scan_until(/\z/) | |
267 @html_scanner.tokenize match unless match.empty? | |
268 next | |
269 end | |
270 | |
271 when :php | |
272 if match = scan(/\s+/) | |
273 tokens << [match, :space] | |
274 next | |
275 | |
276 elsif scan(%r! (?m: \/\* (?: .*? \*\/ | .* ) ) | (?://|\#) .*? (?=#{RE::PHP_END}|$) !xo) | |
277 kind = :comment | |
278 | |
279 elsif match = scan(RE::IDENTIFIER) | |
280 kind = Words::IDENT_KIND[match] | |
281 if kind == :ident && label_expected && check(/:(?!:)/) | |
282 kind = :label | |
283 label_expected = true | |
284 else | |
285 label_expected = false | |
286 if kind == :ident && match =~ /^[A-Z]/ | |
287 kind = :constant | |
288 elsif kind == :reserved | |
289 case match | |
290 when 'class' | |
291 states << :class_expected | |
292 when 'function' | |
293 states << :function_expected | |
294 when 'case', 'default' | |
295 case_expected = true | |
296 end | |
297 elsif match == 'b' && check(/['"]/) # binary string literal | |
298 modifier = match | |
299 next | |
300 end | |
301 end | |
302 | |
303 elsif scan(/(?:\d+\.\d*|\d*\.\d+)(?:e[-+]?\d+)?|\d+e[-+]?\d+/i) | |
304 label_expected = false | |
305 kind = :float | |
306 | |
307 elsif scan(/0x[0-9a-fA-F]+/) | |
308 label_expected = false | |
309 kind = :hex | |
310 | |
311 elsif scan(/\d+/) | |
312 label_expected = false | |
313 kind = :integer | |
314 | |
315 elsif scan(/'/) | |
316 tokens << [:open, :string] | |
317 if modifier | |
318 tokens << [modifier, :modifier] | |
319 modifier = nil | |
320 end | |
321 kind = :delimiter | |
322 states.push :sqstring | |
323 | |
324 elsif match = scan(/["`]/) | |
325 tokens << [:open, :string] | |
326 if modifier | |
327 tokens << [modifier, :modifier] | |
328 modifier = nil | |
329 end | |
330 delimiter = match | |
331 kind = :delimiter | |
332 states.push :dqstring | |
333 | |
334 elsif match = scan(RE::VARIABLE) | |
335 label_expected = false | |
336 kind = Words::VARIABLE_KIND[match] | |
337 | |
338 elsif scan(/\{/) | |
339 kind = :operator | |
340 label_expected = true | |
341 states.push :php | |
342 | |
343 elsif scan(/\}/) | |
344 if states.size == 1 | |
345 kind = :error | |
346 else | |
347 states.pop | |
348 if states.last.is_a?(::Array) | |
349 delimiter = states.last[1] | |
350 states[-1] = states.last[0] | |
351 tokens << [matched, :delimiter] | |
352 tokens << [:close, :inline] | |
353 next | |
354 else | |
355 kind = :operator | |
356 label_expected = true | |
357 end | |
358 end | |
359 | |
360 elsif scan(/@/) | |
361 label_expected = false | |
362 kind = :exception | |
363 | |
364 elsif scan RE::PHP_END | |
365 kind = :inline_delimiter | |
366 states = [:initial] | |
367 | |
368 elsif match = scan(/<<<(?:(#{RE::IDENTIFIER})|"(#{RE::IDENTIFIER})"|'(#{RE::IDENTIFIER})')/o) | |
369 tokens << [:open, :string] | |
370 warn 'heredoc in heredoc?' if heredoc_delimiter | |
371 heredoc_delimiter = Regexp.escape(self[1] || self[2] || self[3]) | |
372 kind = :delimiter | |
373 states.push self[3] ? :sqstring : :dqstring | |
374 heredoc_delimiter = /#{heredoc_delimiter}(?=;?$)/ | |
375 | |
376 elsif match = scan(/#{RE::OPERATOR}/o) | |
377 label_expected = match == ';' | |
378 if case_expected | |
379 label_expected = true if match == ':' | |
380 case_expected = false | |
381 end | |
382 kind = :operator | |
383 | |
384 else | |
385 getch | |
386 kind = :error | |
387 | |
388 end | |
389 | |
390 when :sqstring | |
391 if scan(heredoc_delimiter ? /[^\\\n]+/ : /[^'\\]+/) | |
392 kind = :content | |
393 elsif !heredoc_delimiter && scan(/'/) | |
394 tokens << [matched, :delimiter] | |
395 tokens << [:close, :string] | |
396 delimiter = nil | |
397 label_expected = false | |
398 states.pop | |
399 next | |
400 elsif heredoc_delimiter && match = scan(/\n/) | |
401 kind = :content | |
402 if scan heredoc_delimiter | |
403 tokens << ["\n", :content] | |
404 tokens << [matched, :delimiter] | |
405 tokens << [:close, :string] | |
406 heredoc_delimiter = nil | |
407 label_expected = false | |
408 states.pop | |
409 next | |
410 end | |
411 elsif scan(heredoc_delimiter ? /\\\\/ : /\\[\\'\n]/) | |
412 kind = :char | |
413 elsif scan(/\\./m) | |
414 kind = :content | |
415 elsif scan(/\\/) | |
416 kind = :error | |
417 end | |
418 | |
419 when :dqstring | |
420 if scan(heredoc_delimiter ? /[^${\\\n]+/ : (delimiter == '"' ? /[^"${\\]+/ : /[^`${\\]+/)) | |
421 kind = :content | |
422 elsif !heredoc_delimiter && scan(delimiter == '"' ? /"/ : /`/) | |
423 tokens << [matched, :delimiter] | |
424 tokens << [:close, :string] | |
425 delimiter = nil | |
426 label_expected = false | |
427 states.pop | |
428 next | |
429 elsif heredoc_delimiter && match = scan(/\n/) | |
430 kind = :content | |
431 if scan heredoc_delimiter | |
432 tokens << ["\n", :content] | |
433 tokens << [matched, :delimiter] | |
434 tokens << [:close, :string] | |
435 heredoc_delimiter = nil | |
436 label_expected = false | |
437 states.pop | |
438 next | |
439 end | |
440 elsif scan(/\\(?:x[0-9A-Fa-f]{1,2}|[0-7]{1,3})/) | |
441 kind = :char | |
442 elsif scan(heredoc_delimiter ? /\\[nrtvf\\$]/ : (delimiter == '"' ? /\\[nrtvf\\$"]/ : /\\[nrtvf\\$`]/)) | |
443 kind = :char | |
444 elsif scan(/\\./m) | |
445 kind = :content | |
446 elsif scan(/\\/) | |
447 kind = :error | |
448 elsif match = scan(/#{RE::VARIABLE}/o) | |
449 kind = :local_variable | |
450 if check(/\[#{RE::IDENTIFIER}\]/o) | |
451 tokens << [:open, :inline] | |
452 tokens << [match, :local_variable] | |
453 tokens << [scan(/\[/), :operator] | |
454 tokens << [scan(/#{RE::IDENTIFIER}/o), :ident] | |
455 tokens << [scan(/\]/), :operator] | |
456 tokens << [:close, :inline] | |
457 next | |
458 elsif check(/\[/) | |
459 match << scan(/\[['"]?#{RE::IDENTIFIER}?['"]?\]?/o) | |
460 kind = :error | |
461 elsif check(/->#{RE::IDENTIFIER}/o) | |
462 tokens << [:open, :inline] | |
463 tokens << [match, :local_variable] | |
464 tokens << [scan(/->/), :operator] | |
465 tokens << [scan(/#{RE::IDENTIFIER}/o), :ident] | |
466 tokens << [:close, :inline] | |
467 next | |
468 elsif check(/->/) | |
469 match << scan(/->/) | |
470 kind = :error | |
471 end | |
472 elsif match = scan(/\{/) | |
473 if check(/\$/) | |
474 kind = :delimiter | |
475 states[-1] = [states.last, delimiter] | |
476 delimiter = nil | |
477 states.push :php | |
478 tokens << [:open, :inline] | |
479 else | |
480 kind = :string | |
481 end | |
482 elsif scan(/\$\{#{RE::IDENTIFIER}\}/o) | |
483 kind = :local_variable | |
484 elsif scan(/\$/) | |
485 kind = :content | |
486 end | |
487 | |
488 when :class_expected | |
489 if scan(/\s+/) | |
490 kind = :space | |
491 elsif match = scan(/#{RE::IDENTIFIER}/o) | |
492 kind = :class | |
493 states.pop | |
494 else | |
495 states.pop | |
496 next | |
497 end | |
498 | |
499 when :function_expected | |
500 if scan(/\s+/) | |
501 kind = :space | |
502 elsif scan(/&/) | |
503 kind = :operator | |
504 elsif match = scan(/#{RE::IDENTIFIER}/o) | |
505 kind = :function | |
506 states.pop | |
507 else | |
508 states.pop | |
509 next | |
510 end | |
511 | |
512 else | |
513 raise_inspect 'Unknown state!', tokens, states | |
514 end | |
515 | |
516 match ||= matched | |
517 if $CODERAY_DEBUG and not kind | |
518 raise_inspect 'Error token %p in line %d' % | |
519 [[match, kind], line], tokens, states | |
520 end | |
521 raise_inspect 'Empty token', tokens, states unless match | |
522 | |
523 tokens << [match, kind] | |
524 | |
525 end | |
526 | |
527 tokens | |
528 end | |
529 | |
530 end | |
531 | |
532 end | |
533 end |