Chris@909
|
1 module CodeRay
|
Chris@909
|
2 module Scanners
|
Chris@909
|
3
|
Chris@909
|
4 load :html
|
Chris@909
|
5
|
Chris@909
|
6 # Scanner for PHP.
|
Chris@909
|
7 #
|
Chris@909
|
8 # Original by Stefan Walk.
|
Chris@909
|
9 class PHP < Scanner
|
Chris@909
|
10
|
Chris@909
|
11 register_for :php
|
Chris@909
|
12 file_extension 'php'
|
Chris@909
|
13 encoding 'BINARY'
|
Chris@909
|
14
|
Chris@909
|
15 KINDS_NOT_LOC = HTML::KINDS_NOT_LOC
|
Chris@909
|
16
|
Chris@909
|
17 protected
|
Chris@909
|
18
|
Chris@909
|
19 def setup
|
Chris@909
|
20 @html_scanner = CodeRay.scanner :html, :tokens => @tokens, :keep_tokens => true, :keep_state => true
|
Chris@909
|
21 end
|
Chris@909
|
22
|
Chris@909
|
23 def reset_instance
|
Chris@909
|
24 super
|
Chris@909
|
25 @html_scanner.reset
|
Chris@909
|
26 end
|
Chris@909
|
27
|
Chris@909
|
28 module Words # :nodoc:
|
Chris@909
|
29
|
Chris@909
|
30 # according to http://www.php.net/manual/en/reserved.keywords.php
|
Chris@909
|
31 KEYWORDS = %w[
|
Chris@909
|
32 abstract and array as break case catch class clone const continue declare default do else elseif
|
Chris@909
|
33 enddeclare endfor endforeach endif endswitch endwhile extends final for foreach function global
|
Chris@909
|
34 goto if implements interface instanceof namespace new or private protected public static switch
|
Chris@909
|
35 throw try use var while xor
|
Chris@909
|
36 cfunction old_function
|
Chris@909
|
37 ]
|
Chris@909
|
38
|
Chris@909
|
39 TYPES = %w[ int integer float double bool boolean string array object resource ]
|
Chris@909
|
40
|
Chris@909
|
41 LANGUAGE_CONSTRUCTS = %w[
|
Chris@909
|
42 die echo empty exit eval include include_once isset list
|
Chris@909
|
43 require require_once return print unset
|
Chris@909
|
44 ]
|
Chris@909
|
45
|
Chris@909
|
46 CLASSES = %w[ Directory stdClass __PHP_Incomplete_Class exception php_user_filter Closure ]
|
Chris@909
|
47
|
Chris@909
|
48 # according to http://php.net/quickref.php on 2009-04-21;
|
Chris@909
|
49 # all functions with _ excluded (module functions) and selected additional functions
|
Chris@909
|
50 BUILTIN_FUNCTIONS = %w[
|
Chris@909
|
51 abs acos acosh addcslashes addslashes aggregate array arsort ascii2ebcdic asin asinh asort assert atan atan2
|
Chris@909
|
52 atanh basename bcadd bccomp bcdiv bcmod bcmul bcpow bcpowmod bcscale bcsqrt bcsub bin2hex bindec
|
Chris@909
|
53 bindtextdomain bzclose bzcompress bzdecompress bzerrno bzerror bzerrstr bzflush bzopen bzread bzwrite
|
Chris@909
|
54 calculhmac ceil chdir checkdate checkdnsrr chgrp chmod chop chown chr chroot clearstatcache closedir closelog
|
Chris@909
|
55 compact constant copy cos cosh count crc32 crypt current date dcgettext dcngettext deaggregate decbin dechex
|
Chris@909
|
56 decoct define defined deg2rad delete dgettext die dirname diskfreespace dl dngettext doubleval each
|
Chris@909
|
57 ebcdic2ascii echo empty end ereg eregi escapeshellarg escapeshellcmd eval exec exit exp explode expm1 extract
|
Chris@909
|
58 fclose feof fflush fgetc fgetcsv fgets fgetss file fileatime filectime filegroup fileinode filemtime fileowner
|
Chris@909
|
59 fileperms filepro filesize filetype floatval flock floor flush fmod fnmatch fopen fpassthru fprintf fputcsv
|
Chris@909
|
60 fputs fread frenchtojd fscanf fseek fsockopen fstat ftell ftok ftruncate fwrite getallheaders getcwd getdate
|
Chris@909
|
61 getenv gethostbyaddr gethostbyname gethostbynamel getimagesize getlastmod getmxrr getmygid getmyinode getmypid
|
Chris@909
|
62 getmyuid getopt getprotobyname getprotobynumber getrandmax getrusage getservbyname getservbyport gettext
|
Chris@909
|
63 gettimeofday gettype glob gmdate gmmktime gmstrftime gregoriantojd gzclose gzcompress gzdecode gzdeflate
|
Chris@909
|
64 gzencode gzeof gzfile gzgetc gzgets gzgetss gzinflate gzopen gzpassthru gzputs gzread gzrewind gzseek gztell
|
Chris@909
|
65 gzuncompress gzwrite hash header hebrev hebrevc hexdec htmlentities htmlspecialchars hypot iconv idate
|
Chris@909
|
66 implode include intval ip2long iptcembed iptcparse isset
|
Chris@909
|
67 jddayofweek jdmonthname jdtofrench jdtogregorian jdtojewish jdtojulian jdtounix jewishtojd join jpeg2wbmp
|
Chris@909
|
68 juliantojd key krsort ksort lcfirst lchgrp lchown levenshtein link linkinfo list localeconv localtime log
|
Chris@909
|
69 log10 log1p long2ip lstat ltrim mail main max md5 metaphone mhash microtime min mkdir mktime msql natcasesort
|
Chris@909
|
70 natsort next ngettext nl2br nthmac octdec opendir openlog
|
Chris@909
|
71 ord overload pack passthru pathinfo pclose pfsockopen phpcredits phpinfo phpversion pi png2wbmp popen pos pow
|
Chris@909
|
72 prev print printf putenv quotemeta rad2deg rand range rawurldecode rawurlencode readdir readfile readgzfile
|
Chris@909
|
73 readline readlink realpath recode rename require reset rewind rewinddir rmdir round rsort rtrim scandir
|
Chris@909
|
74 serialize setcookie setlocale setrawcookie settype sha1 shuffle signeurlpaiement sin sinh sizeof sleep snmpget
|
Chris@909
|
75 snmpgetnext snmprealwalk snmpset snmpwalk snmpwalkoid sort soundex split spliti sprintf sqrt srand sscanf stat
|
Chris@909
|
76 strcasecmp strchr strcmp strcoll strcspn strftime stripcslashes stripos stripslashes stristr strlen
|
Chris@909
|
77 strnatcasecmp strnatcmp strncasecmp strncmp strpbrk strpos strptime strrchr strrev strripos strrpos strspn
|
Chris@909
|
78 strstr strtok strtolower strtotime strtoupper strtr strval substr symlink syslog system tan tanh tempnam
|
Chris@909
|
79 textdomain time tmpfile touch trim uasort ucfirst ucwords uksort umask uniqid unixtojd unlink unpack
|
Chris@909
|
80 unserialize unset urldecode urlencode usleep usort vfprintf virtual vprintf vsprintf wordwrap
|
Chris@909
|
81 array_change_key_case array_chunk array_combine array_count_values array_diff array_diff_assoc
|
Chris@909
|
82 array_diff_key array_diff_uassoc array_diff_ukey array_fill array_fill_keys array_filter array_flip
|
Chris@909
|
83 array_intersect array_intersect_assoc array_intersect_key array_intersect_uassoc array_intersect_ukey
|
Chris@909
|
84 array_key_exists array_keys array_map array_merge array_merge_recursive array_multisort array_pad
|
Chris@909
|
85 array_pop array_product array_push array_rand array_reduce array_reverse array_search array_shift
|
Chris@909
|
86 array_slice array_splice array_sum array_udiff array_udiff_assoc array_udiff_uassoc array_uintersect
|
Chris@909
|
87 array_uintersect_assoc array_uintersect_uassoc array_unique array_unshift array_values array_walk
|
Chris@909
|
88 array_walk_recursive
|
Chris@909
|
89 assert_options base_convert base64_decode base64_encode
|
Chris@909
|
90 chunk_split class_exists class_implements class_parents
|
Chris@909
|
91 count_chars debug_backtrace debug_print_backtrace debug_zval_dump
|
Chris@909
|
92 error_get_last error_log error_reporting extension_loaded
|
Chris@909
|
93 file_exists file_get_contents file_put_contents load_file
|
Chris@909
|
94 func_get_arg func_get_args func_num_args function_exists
|
Chris@909
|
95 get_browser get_called_class get_cfg_var get_class get_class_methods get_class_vars
|
Chris@909
|
96 get_current_user get_declared_classes get_declared_interfaces get_defined_constants
|
Chris@909
|
97 get_defined_functions get_defined_vars get_extension_funcs get_headers get_html_translation_table
|
Chris@909
|
98 get_include_path get_included_files get_loaded_extensions get_magic_quotes_gpc get_magic_quotes_runtime
|
Chris@909
|
99 get_meta_tags get_object_vars get_parent_class get_required_filesget_resource_type
|
Chris@909
|
100 gc_collect_cycles gc_disable gc_enable gc_enabled
|
Chris@909
|
101 halt_compiler headers_list headers_sent highlight_file highlight_string
|
Chris@909
|
102 html_entity_decode htmlspecialchars_decode
|
Chris@909
|
103 in_array include_once inclued_get_data
|
Chris@909
|
104 is_a is_array is_binary is_bool is_buffer is_callable is_dir is_double is_executable is_file is_finite
|
Chris@909
|
105 is_float is_infinite is_int is_integer is_link is_long is_nan is_null is_numeric is_object is_readable
|
Chris@909
|
106 is_real is_resource is_scalar is_soap_fault is_string is_subclass_of is_unicode is_uploaded_file
|
Chris@909
|
107 is_writable is_writeable
|
Chris@909
|
108 locale_get_default locale_set_default
|
Chris@909
|
109 number_format override_function parse_str parse_url
|
Chris@909
|
110 php_check_syntax php_ini_loaded_file php_ini_scanned_files php_logo_guid php_sapi_name
|
Chris@909
|
111 php_strip_whitespace php_uname
|
Chris@909
|
112 preg_filter preg_grep preg_last_error preg_match preg_match_all preg_quote preg_replace
|
Chris@909
|
113 preg_replace_callback preg_split print_r
|
Chris@909
|
114 require_once register_shutdown_function register_tick_function
|
Chris@909
|
115 set_error_handler set_exception_handler set_file_buffer set_include_path
|
Chris@909
|
116 set_magic_quotes_runtime set_time_limit shell_exec
|
Chris@909
|
117 str_getcsv str_ireplace str_pad str_repeat str_replace str_rot13 str_shuffle str_split str_word_count
|
Chris@909
|
118 strip_tags substr_compare substr_count substr_replace
|
Chris@909
|
119 time_nanosleep time_sleep_until
|
Chris@909
|
120 token_get_all token_name trigger_error
|
Chris@909
|
121 unregister_tick_function use_soap_error_handler user_error
|
Chris@909
|
122 utf8_decode utf8_encode var_dump var_export
|
Chris@909
|
123 version_compare
|
Chris@909
|
124 zend_logo_guid zend_thread_id zend_version
|
Chris@909
|
125 create_function call_user_func_array
|
Chris@909
|
126 posix_access posix_ctermid posix_get_last_error posix_getcwd posix_getegid
|
Chris@909
|
127 posix_geteuid posix_getgid posix_getgrgid posix_getgrnam posix_getgroups
|
Chris@909
|
128 posix_getlogin posix_getpgid posix_getpgrp posix_getpid posix_getppid
|
Chris@909
|
129 posix_getpwnam posix_getpwuid posix_getrlimit posix_getsid posix_getuid
|
Chris@909
|
130 posix_initgroups posix_isatty posix_kill posix_mkfifo posix_mknod
|
Chris@909
|
131 posix_setegid posix_seteuid posix_setgid posix_setpgid posix_setsid
|
Chris@909
|
132 posix_setuid posix_strerror posix_times posix_ttyname posix_uname
|
Chris@909
|
133 pcntl_alarm pcntl_exec pcntl_fork pcntl_getpriority pcntl_setpriority
|
Chris@909
|
134 pcntl_signal pcntl_signal_dispatch pcntl_sigprocmask pcntl_sigtimedwait
|
Chris@909
|
135 pcntl_sigwaitinfo pcntl_wait pcntl_waitpid pcntl_wexitstatus pcntl_wifexited
|
Chris@909
|
136 pcntl_wifsignaled pcntl_wifstopped pcntl_wstopsig pcntl_wtermsig
|
Chris@909
|
137 ]
|
Chris@909
|
138 # TODO: more built-in PHP functions?
|
Chris@909
|
139
|
Chris@909
|
140 EXCEPTIONS = %w[
|
Chris@909
|
141 E_ERROR E_WARNING E_PARSE E_NOTICE E_CORE_ERROR E_CORE_WARNING E_COMPILE_ERROR E_COMPILE_WARNING
|
Chris@909
|
142 E_USER_ERROR E_USER_WARNING E_USER_NOTICE E_DEPRECATED E_USER_DEPRECATED E_ALL E_STRICT
|
Chris@909
|
143 ]
|
Chris@909
|
144
|
Chris@909
|
145 CONSTANTS = %w[
|
Chris@909
|
146 null true false self parent
|
Chris@909
|
147 __LINE__ __DIR__ __FILE__ __LINE__
|
Chris@909
|
148 __CLASS__ __NAMESPACE__ __METHOD__ __FUNCTION__
|
Chris@909
|
149 PHP_VERSION PHP_MAJOR_VERSION PHP_MINOR_VERSION PHP_RELEASE_VERSION PHP_VERSION_ID PHP_EXTRA_VERSION PHP_ZTS
|
Chris@909
|
150 PHP_DEBUG PHP_MAXPATHLEN PHP_OS PHP_SAPI PHP_EOL PHP_INT_MAX PHP_INT_SIZE DEFAULT_INCLUDE_PATH
|
Chris@909
|
151 PEAR_INSTALL_DIR PEAR_EXTENSION_DIR PHP_EXTENSION_DIR PHP_PREFIX PHP_BINDIR PHP_LIBDIR PHP_DATADIR
|
Chris@909
|
152 PHP_SYSCONFDIR PHP_LOCALSTATEDIR PHP_CONFIG_FILE_PATH PHP_CONFIG_FILE_SCAN_DIR PHP_SHLIB_SUFFIX
|
Chris@909
|
153 PHP_OUTPUT_HANDLER_START PHP_OUTPUT_HANDLER_CONT PHP_OUTPUT_HANDLER_END
|
Chris@909
|
154 __COMPILER_HALT_OFFSET__
|
Chris@909
|
155 EXTR_OVERWRITE EXTR_SKIP EXTR_PREFIX_SAME EXTR_PREFIX_ALL EXTR_PREFIX_INVALID EXTR_PREFIX_IF_EXISTS
|
Chris@909
|
156 EXTR_IF_EXISTS SORT_ASC SORT_DESC SORT_REGULAR SORT_NUMERIC SORT_STRING CASE_LOWER CASE_UPPER COUNT_NORMAL
|
Chris@909
|
157 COUNT_RECURSIVE ASSERT_ACTIVE ASSERT_CALLBACK ASSERT_BAIL ASSERT_WARNING ASSERT_QUIET_EVAL CONNECTION_ABORTED
|
Chris@909
|
158 CONNECTION_NORMAL CONNECTION_TIMEOUT INI_USER INI_PERDIR INI_SYSTEM INI_ALL M_E M_LOG2E M_LOG10E M_LN2 M_LN10
|
Chris@909
|
159 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
|
Chris@909
|
160 CRYPT_MD5 CRYPT_BLOWFISH DIRECTORY_SEPARATOR SEEK_SET SEEK_CUR SEEK_END LOCK_SH LOCK_EX LOCK_UN LOCK_NB
|
Chris@909
|
161 HTML_SPECIALCHARS HTML_ENTITIES ENT_COMPAT ENT_QUOTES ENT_NOQUOTES INFO_GENERAL INFO_CREDITS
|
Chris@909
|
162 INFO_CONFIGURATION INFO_MODULES INFO_ENVIRONMENT INFO_VARIABLES INFO_LICENSE INFO_ALL CREDITS_GROUP
|
Chris@909
|
163 CREDITS_GENERAL CREDITS_SAPI CREDITS_MODULES CREDITS_DOCS CREDITS_FULLPAGE CREDITS_QA CREDITS_ALL STR_PAD_LEFT
|
Chris@909
|
164 STR_PAD_RIGHT STR_PAD_BOTH PATHINFO_DIRNAME PATHINFO_BASENAME PATHINFO_EXTENSION PATH_SEPARATOR CHAR_MAX
|
Chris@909
|
165 LC_CTYPE LC_NUMERIC LC_TIME LC_COLLATE LC_MONETARY LC_ALL LC_MESSAGES ABDAY_1 ABDAY_2 ABDAY_3 ABDAY_4 ABDAY_5
|
Chris@909
|
166 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
|
Chris@909
|
167 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
|
Chris@909
|
168 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
|
Chris@909
|
169 ALT_DIGITS INT_CURR_SYMBOL CURRENCY_SYMBOL CRNCYSTR MON_DECIMAL_POINT MON_THOUSANDS_SEP MON_GROUPING
|
Chris@909
|
170 POSITIVE_SIGN NEGATIVE_SIGN INT_FRAC_DIGITS FRAC_DIGITS P_CS_PRECEDES P_SEP_BY_SPACE N_CS_PRECEDES
|
Chris@909
|
171 N_SEP_BY_SPACE P_SIGN_POSN N_SIGN_POSN DECIMAL_POINT RADIXCHAR THOUSANDS_SEP THOUSEP GROUPING YESEXPR NOEXPR
|
Chris@909
|
172 YESSTR NOSTR CODESET LOG_EMERG LOG_ALERT LOG_CRIT LOG_ERR LOG_WARNING LOG_NOTICE LOG_INFO LOG_DEBUG LOG_KERN
|
Chris@909
|
173 LOG_USER LOG_MAIL LOG_DAEMON LOG_AUTH LOG_SYSLOG LOG_LPR LOG_NEWS LOG_UUCP LOG_CRON LOG_AUTHPRIV LOG_LOCAL0
|
Chris@909
|
174 LOG_LOCAL1 LOG_LOCAL2 LOG_LOCAL3 LOG_LOCAL4 LOG_LOCAL5 LOG_LOCAL6 LOG_LOCAL7 LOG_PID LOG_CONS LOG_ODELAY
|
Chris@909
|
175 LOG_NDELAY LOG_NOWAIT LOG_PERROR
|
Chris@909
|
176 ]
|
Chris@909
|
177
|
Chris@909
|
178 PREDEFINED = %w[
|
Chris@909
|
179 $GLOBALS $_SERVER $_GET $_POST $_FILES $_REQUEST $_SESSION $_ENV
|
Chris@909
|
180 $_COOKIE $php_errormsg $HTTP_RAW_POST_DATA $http_response_header
|
Chris@909
|
181 $argc $argv
|
Chris@909
|
182 ]
|
Chris@909
|
183
|
Chris@909
|
184 IDENT_KIND = WordList::CaseIgnoring.new(:ident).
|
Chris@909
|
185 add(KEYWORDS, :keyword).
|
Chris@909
|
186 add(TYPES, :predefined_type).
|
Chris@909
|
187 add(LANGUAGE_CONSTRUCTS, :keyword).
|
Chris@909
|
188 add(BUILTIN_FUNCTIONS, :predefined).
|
Chris@909
|
189 add(CLASSES, :predefined_constant).
|
Chris@909
|
190 add(EXCEPTIONS, :exception).
|
Chris@909
|
191 add(CONSTANTS, :predefined_constant)
|
Chris@909
|
192
|
Chris@909
|
193 VARIABLE_KIND = WordList.new(:local_variable).
|
Chris@909
|
194 add(PREDEFINED, :predefined)
|
Chris@909
|
195 end
|
Chris@909
|
196
|
Chris@909
|
197 module RE # :nodoc:
|
Chris@909
|
198
|
Chris@909
|
199 PHP_START = /
|
Chris@909
|
200 <script\s+[^>]*?language\s*=\s*"php"[^>]*?> |
|
Chris@909
|
201 <script\s+[^>]*?language\s*=\s*'php'[^>]*?> |
|
Chris@909
|
202 <\?php\d? |
|
Chris@909
|
203 <\?(?!xml)
|
Chris@909
|
204 /xi
|
Chris@909
|
205
|
Chris@909
|
206 PHP_END = %r!
|
Chris@909
|
207 </script> |
|
Chris@909
|
208 \?>
|
Chris@909
|
209 !xi
|
Chris@909
|
210
|
Chris@909
|
211 HTML_INDICATOR = /<!DOCTYPE html|<(?:html|body|div|p)[> ]/i
|
Chris@909
|
212
|
Chris@909
|
213 IDENTIFIER = /[a-z_\x7f-\xFF][a-z0-9_\x7f-\xFF]*/i
|
Chris@909
|
214 VARIABLE = /\$#{IDENTIFIER}/
|
Chris@909
|
215
|
Chris@909
|
216 OPERATOR = /
|
Chris@909
|
217 \.(?!\d)=? | # dot that is not decimal point, string concatenation
|
Chris@909
|
218 && | \|\| | # logic
|
Chris@909
|
219 :: | -> | => | # scope, member, dictionary
|
Chris@909
|
220 \\(?!\n) | # namespace
|
Chris@909
|
221 \+\+ | -- | # increment, decrement
|
Chris@909
|
222 [,;?:()\[\]{}] | # simple delimiters
|
Chris@909
|
223 [-+*\/%&|^]=? | # ordinary math, binary logic, assignment shortcuts
|
Chris@909
|
224 [~$] | # whatever
|
Chris@909
|
225 =& | # reference assignment
|
Chris@909
|
226 [=!]=?=? | <> | # comparison and assignment
|
Chris@909
|
227 <<=? | >>=? | [<>]=? # comparison and shift
|
Chris@909
|
228 /x
|
Chris@909
|
229
|
Chris@909
|
230 end
|
Chris@909
|
231
|
Chris@909
|
232 protected
|
Chris@909
|
233
|
Chris@909
|
234 def scan_tokens encoder, options
|
Chris@909
|
235
|
Chris@909
|
236 if check(RE::PHP_START) || # starts with <?
|
Chris@909
|
237 (match?(/\s*<\S/) && check(/.{1,1000}#{RE::PHP_START}/om)) || # starts with tag and contains <?
|
Chris@909
|
238 check(/.{0,1000}#{RE::HTML_INDICATOR}/om) ||
|
Chris@909
|
239 check(/.{1,100}#{RE::PHP_START}/om) # PHP start after max 100 chars
|
Chris@909
|
240 # is HTML with embedded PHP, so start with HTML
|
Chris@909
|
241 states = [:initial]
|
Chris@909
|
242 else
|
Chris@909
|
243 # is just PHP, so start with PHP surrounded by HTML
|
Chris@909
|
244 states = [:initial, :php]
|
Chris@909
|
245 end
|
Chris@909
|
246
|
Chris@909
|
247 label_expected = true
|
Chris@909
|
248 case_expected = false
|
Chris@909
|
249
|
Chris@909
|
250 heredoc_delimiter = nil
|
Chris@909
|
251 delimiter = nil
|
Chris@909
|
252 modifier = nil
|
Chris@909
|
253
|
Chris@909
|
254 until eos?
|
Chris@909
|
255
|
Chris@909
|
256 case states.last
|
Chris@909
|
257
|
Chris@909
|
258 when :initial # HTML
|
Chris@909
|
259 if match = scan(RE::PHP_START)
|
Chris@909
|
260 encoder.text_token match, :inline_delimiter
|
Chris@909
|
261 label_expected = true
|
Chris@909
|
262 states << :php
|
Chris@909
|
263 else
|
Chris@909
|
264 match = scan_until(/(?=#{RE::PHP_START})/o) || scan_rest
|
Chris@909
|
265 @html_scanner.tokenize match unless match.empty?
|
Chris@909
|
266 end
|
Chris@909
|
267
|
Chris@909
|
268 when :php
|
Chris@909
|
269 if match = scan(/\s+/)
|
Chris@909
|
270 encoder.text_token match, :space
|
Chris@909
|
271
|
Chris@909
|
272 elsif match = scan(%r! (?m: \/\* (?: .*? \*\/ | .* ) ) | (?://|\#) .*? (?=#{RE::PHP_END}|$) !xo)
|
Chris@909
|
273 encoder.text_token match, :comment
|
Chris@909
|
274
|
Chris@909
|
275 elsif match = scan(RE::IDENTIFIER)
|
Chris@909
|
276 kind = Words::IDENT_KIND[match]
|
Chris@909
|
277 if kind == :ident && label_expected && check(/:(?!:)/)
|
Chris@909
|
278 kind = :label
|
Chris@909
|
279 label_expected = true
|
Chris@909
|
280 else
|
Chris@909
|
281 label_expected = false
|
Chris@909
|
282 if kind == :ident && match =~ /^[A-Z]/
|
Chris@909
|
283 kind = :constant
|
Chris@909
|
284 elsif kind == :keyword
|
Chris@909
|
285 case match
|
Chris@909
|
286 when 'class'
|
Chris@909
|
287 states << :class_expected
|
Chris@909
|
288 when 'function'
|
Chris@909
|
289 states << :function_expected
|
Chris@909
|
290 when 'case', 'default'
|
Chris@909
|
291 case_expected = true
|
Chris@909
|
292 end
|
Chris@909
|
293 elsif match == 'b' && check(/['"]/) # binary string literal
|
Chris@909
|
294 modifier = match
|
Chris@909
|
295 next
|
Chris@909
|
296 end
|
Chris@909
|
297 end
|
Chris@909
|
298 encoder.text_token match, kind
|
Chris@909
|
299
|
Chris@909
|
300 elsif match = scan(/(?:\d+\.\d*|\d*\.\d+)(?:e[-+]?\d+)?|\d+e[-+]?\d+/i)
|
Chris@909
|
301 label_expected = false
|
Chris@909
|
302 encoder.text_token match, :float
|
Chris@909
|
303
|
Chris@909
|
304 elsif match = scan(/0x[0-9a-fA-F]+/)
|
Chris@909
|
305 label_expected = false
|
Chris@909
|
306 encoder.text_token match, :hex
|
Chris@909
|
307
|
Chris@909
|
308 elsif match = scan(/\d+/)
|
Chris@909
|
309 label_expected = false
|
Chris@909
|
310 encoder.text_token match, :integer
|
Chris@909
|
311
|
Chris@909
|
312 elsif match = scan(/['"`]/)
|
Chris@909
|
313 encoder.begin_group :string
|
Chris@909
|
314 if modifier
|
Chris@909
|
315 encoder.text_token modifier, :modifier
|
Chris@909
|
316 modifier = nil
|
Chris@909
|
317 end
|
Chris@909
|
318 delimiter = match
|
Chris@909
|
319 encoder.text_token match, :delimiter
|
Chris@909
|
320 states.push match == "'" ? :sqstring : :dqstring
|
Chris@909
|
321
|
Chris@909
|
322 elsif match = scan(RE::VARIABLE)
|
Chris@909
|
323 label_expected = false
|
Chris@909
|
324 encoder.text_token match, Words::VARIABLE_KIND[match]
|
Chris@909
|
325
|
Chris@909
|
326 elsif match = scan(/\{/)
|
Chris@909
|
327 encoder.text_token match, :operator
|
Chris@909
|
328 label_expected = true
|
Chris@909
|
329 states.push :php
|
Chris@909
|
330
|
Chris@909
|
331 elsif match = scan(/\}/)
|
Chris@909
|
332 if states.size == 1
|
Chris@909
|
333 encoder.text_token match, :error
|
Chris@909
|
334 else
|
Chris@909
|
335 states.pop
|
Chris@909
|
336 if states.last.is_a?(::Array)
|
Chris@909
|
337 delimiter = states.last[1]
|
Chris@909
|
338 states[-1] = states.last[0]
|
Chris@909
|
339 encoder.text_token match, :delimiter
|
Chris@909
|
340 encoder.end_group :inline
|
Chris@909
|
341 else
|
Chris@909
|
342 encoder.text_token match, :operator
|
Chris@909
|
343 label_expected = true
|
Chris@909
|
344 end
|
Chris@909
|
345 end
|
Chris@909
|
346
|
Chris@909
|
347 elsif match = scan(/@/)
|
Chris@909
|
348 label_expected = false
|
Chris@909
|
349 encoder.text_token match, :exception
|
Chris@909
|
350
|
Chris@909
|
351 elsif match = scan(RE::PHP_END)
|
Chris@909
|
352 encoder.text_token match, :inline_delimiter
|
Chris@909
|
353 states = [:initial]
|
Chris@909
|
354
|
Chris@909
|
355 elsif match = scan(/<<<(?:(#{RE::IDENTIFIER})|"(#{RE::IDENTIFIER})"|'(#{RE::IDENTIFIER})')/o)
|
Chris@909
|
356 encoder.begin_group :string
|
Chris@909
|
357 # warn 'heredoc in heredoc?' if heredoc_delimiter
|
Chris@909
|
358 heredoc_delimiter = Regexp.escape(self[1] || self[2] || self[3])
|
Chris@909
|
359 encoder.text_token match, :delimiter
|
Chris@909
|
360 states.push self[3] ? :sqstring : :dqstring
|
Chris@909
|
361 heredoc_delimiter = /#{heredoc_delimiter}(?=;?$)/
|
Chris@909
|
362
|
Chris@909
|
363 elsif match = scan(/#{RE::OPERATOR}/o)
|
Chris@909
|
364 label_expected = match == ';'
|
Chris@909
|
365 if case_expected
|
Chris@909
|
366 label_expected = true if match == ':'
|
Chris@909
|
367 case_expected = false
|
Chris@909
|
368 end
|
Chris@909
|
369 encoder.text_token match, :operator
|
Chris@909
|
370
|
Chris@909
|
371 else
|
Chris@909
|
372 encoder.text_token getch, :error
|
Chris@909
|
373
|
Chris@909
|
374 end
|
Chris@909
|
375
|
Chris@909
|
376 when :sqstring
|
Chris@909
|
377 if match = scan(heredoc_delimiter ? /[^\\\n]+/ : /[^'\\]+/)
|
Chris@909
|
378 encoder.text_token match, :content
|
Chris@909
|
379 elsif !heredoc_delimiter && match = scan(/'/)
|
Chris@909
|
380 encoder.text_token match, :delimiter
|
Chris@909
|
381 encoder.end_group :string
|
Chris@909
|
382 delimiter = nil
|
Chris@909
|
383 label_expected = false
|
Chris@909
|
384 states.pop
|
Chris@909
|
385 elsif heredoc_delimiter && match = scan(/\n/)
|
Chris@909
|
386 if scan heredoc_delimiter
|
Chris@909
|
387 encoder.text_token "\n", :content
|
Chris@909
|
388 encoder.text_token matched, :delimiter
|
Chris@909
|
389 encoder.end_group :string
|
Chris@909
|
390 heredoc_delimiter = nil
|
Chris@909
|
391 label_expected = false
|
Chris@909
|
392 states.pop
|
Chris@909
|
393 else
|
Chris@909
|
394 encoder.text_token match, :content
|
Chris@909
|
395 end
|
Chris@909
|
396 elsif match = scan(heredoc_delimiter ? /\\\\/ : /\\[\\'\n]/)
|
Chris@909
|
397 encoder.text_token match, :char
|
Chris@909
|
398 elsif match = scan(/\\./m)
|
Chris@909
|
399 encoder.text_token match, :content
|
Chris@909
|
400 elsif match = scan(/\\/)
|
Chris@909
|
401 encoder.text_token match, :error
|
Chris@909
|
402 else
|
Chris@909
|
403 states.pop
|
Chris@909
|
404 end
|
Chris@909
|
405
|
Chris@909
|
406 when :dqstring
|
Chris@909
|
407 if match = scan(heredoc_delimiter ? /[^${\\\n]+/ : (delimiter == '"' ? /[^"${\\]+/ : /[^`${\\]+/))
|
Chris@909
|
408 encoder.text_token match, :content
|
Chris@909
|
409 elsif !heredoc_delimiter && match = scan(delimiter == '"' ? /"/ : /`/)
|
Chris@909
|
410 encoder.text_token match, :delimiter
|
Chris@909
|
411 encoder.end_group :string
|
Chris@909
|
412 delimiter = nil
|
Chris@909
|
413 label_expected = false
|
Chris@909
|
414 states.pop
|
Chris@909
|
415 elsif heredoc_delimiter && match = scan(/\n/)
|
Chris@909
|
416 if scan heredoc_delimiter
|
Chris@909
|
417 encoder.text_token "\n", :content
|
Chris@909
|
418 encoder.text_token matched, :delimiter
|
Chris@909
|
419 encoder.end_group :string
|
Chris@909
|
420 heredoc_delimiter = nil
|
Chris@909
|
421 label_expected = false
|
Chris@909
|
422 states.pop
|
Chris@909
|
423 else
|
Chris@909
|
424 encoder.text_token match, :content
|
Chris@909
|
425 end
|
Chris@909
|
426 elsif match = scan(/\\(?:x[0-9A-Fa-f]{1,2}|[0-7]{1,3})/)
|
Chris@909
|
427 encoder.text_token match, :char
|
Chris@909
|
428 elsif match = scan(heredoc_delimiter ? /\\[nrtvf\\$]/ : (delimiter == '"' ? /\\[nrtvf\\$"]/ : /\\[nrtvf\\$`]/))
|
Chris@909
|
429 encoder.text_token match, :char
|
Chris@909
|
430 elsif match = scan(/\\./m)
|
Chris@909
|
431 encoder.text_token match, :content
|
Chris@909
|
432 elsif match = scan(/\\/)
|
Chris@909
|
433 encoder.text_token match, :error
|
Chris@909
|
434 elsif match = scan(/#{RE::VARIABLE}/o)
|
Chris@909
|
435 if check(/\[#{RE::IDENTIFIER}\]/o)
|
Chris@909
|
436 encoder.begin_group :inline
|
Chris@909
|
437 encoder.text_token match, :local_variable
|
Chris@909
|
438 encoder.text_token scan(/\[/), :operator
|
Chris@909
|
439 encoder.text_token scan(/#{RE::IDENTIFIER}/o), :ident
|
Chris@909
|
440 encoder.text_token scan(/\]/), :operator
|
Chris@909
|
441 encoder.end_group :inline
|
Chris@909
|
442 elsif check(/\[/)
|
Chris@909
|
443 match << scan(/\[['"]?#{RE::IDENTIFIER}?['"]?\]?/o)
|
Chris@909
|
444 encoder.text_token match, :error
|
Chris@909
|
445 elsif check(/->#{RE::IDENTIFIER}/o)
|
Chris@909
|
446 encoder.begin_group :inline
|
Chris@909
|
447 encoder.text_token match, :local_variable
|
Chris@909
|
448 encoder.text_token scan(/->/), :operator
|
Chris@909
|
449 encoder.text_token scan(/#{RE::IDENTIFIER}/o), :ident
|
Chris@909
|
450 encoder.end_group :inline
|
Chris@909
|
451 elsif check(/->/)
|
Chris@909
|
452 match << scan(/->/)
|
Chris@909
|
453 encoder.text_token match, :error
|
Chris@909
|
454 else
|
Chris@909
|
455 encoder.text_token match, :local_variable
|
Chris@909
|
456 end
|
Chris@909
|
457 elsif match = scan(/\{/)
|
Chris@909
|
458 if check(/\$/)
|
Chris@909
|
459 encoder.begin_group :inline
|
Chris@909
|
460 states[-1] = [states.last, delimiter]
|
Chris@909
|
461 delimiter = nil
|
Chris@909
|
462 states.push :php
|
Chris@909
|
463 encoder.text_token match, :delimiter
|
Chris@909
|
464 else
|
Chris@909
|
465 encoder.text_token match, :content
|
Chris@909
|
466 end
|
Chris@909
|
467 elsif match = scan(/\$\{#{RE::IDENTIFIER}\}/o)
|
Chris@909
|
468 encoder.text_token match, :local_variable
|
Chris@909
|
469 elsif match = scan(/\$/)
|
Chris@909
|
470 encoder.text_token match, :content
|
Chris@909
|
471 else
|
Chris@909
|
472 states.pop
|
Chris@909
|
473 end
|
Chris@909
|
474
|
Chris@909
|
475 when :class_expected
|
Chris@909
|
476 if match = scan(/\s+/)
|
Chris@909
|
477 encoder.text_token match, :space
|
Chris@909
|
478 elsif match = scan(/#{RE::IDENTIFIER}/o)
|
Chris@909
|
479 encoder.text_token match, :class
|
Chris@909
|
480 states.pop
|
Chris@909
|
481 else
|
Chris@909
|
482 states.pop
|
Chris@909
|
483 end
|
Chris@909
|
484
|
Chris@909
|
485 when :function_expected
|
Chris@909
|
486 if match = scan(/\s+/)
|
Chris@909
|
487 encoder.text_token match, :space
|
Chris@909
|
488 elsif match = scan(/&/)
|
Chris@909
|
489 encoder.text_token match, :operator
|
Chris@909
|
490 elsif match = scan(/#{RE::IDENTIFIER}/o)
|
Chris@909
|
491 encoder.text_token match, :function
|
Chris@909
|
492 states.pop
|
Chris@909
|
493 else
|
Chris@909
|
494 states.pop
|
Chris@909
|
495 end
|
Chris@909
|
496
|
Chris@909
|
497 else
|
Chris@909
|
498 raise_inspect 'Unknown state!', encoder, states
|
Chris@909
|
499 end
|
Chris@909
|
500
|
Chris@909
|
501 end
|
Chris@909
|
502
|
Chris@909
|
503 encoder
|
Chris@909
|
504 end
|
Chris@909
|
505
|
Chris@909
|
506 end
|
Chris@909
|
507
|
Chris@909
|
508 end
|
Chris@909
|
509 end
|