mirror of
https://github.com/ralsina/tartrazine.git
synced 2024-11-10 05:22:23 +00:00
400 lines
14 KiB
XML
400 lines
14 KiB
XML
|
<lexer>
|
||
|
|
<config>
|
||
|
|
<name>Perl</name>
|
||
|
|
<alias>perl</alias>
|
||
|
|
<alias>pl</alias>
|
||
|
|
<filename>*.pl</filename>
|
||
|
|
<filename>*.pm</filename>
|
||
|
|
<filename>*.t</filename>
|
||
|
|
<mime_type>text/x-perl</mime_type>
|
||
|
|
<mime_type>application/x-perl</mime_type>
|
||
|
|
<dot_all>true</dot_all>
|
||
|
|
</config>
|
||
|
|
<rules>
|
||
|
|
<state name="root">
|
||
|
|
<rule pattern="\A\#!.+?$">
|
||
|
|
<token type="CommentHashbang"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\#.*?$">
|
||
|
|
<token type="CommentSingle"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="^=[a-zA-Z0-9]+\s+.*?\n=cut">
|
||
|
|
<token type="CommentMultiline"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(continue|foreach|unless|return|elsif|CHECK|while|BEGIN|reset|print|until|next|else|INIT|then|last|redo|case|our|new|for|END|if|do|my)\b">
|
||
|
|
<token type="Keyword"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(format)(\s+)(\w+)(\s*)(=)(\s*\n)">
|
||
|
|
<bygroups>
|
||
|
|
<token type="Keyword"/>
|
||
|
|
<token type="Text"/>
|
||
|
|
<token type="Name"/>
|
||
|
|
<token type="Text"/>
|
||
|
|
<token type="Punctuation"/>
|
||
|
|
<token type="Text"/>
|
||
|
|
</bygroups>
|
||
|
|
<push state="format"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(eq|lt|gt|le|ge|ne|not|and|or|cmp)\b">
|
||
|
|
<token type="OperatorWord"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="s/(\\\\|\\[^\\]|[^\\/])*/(\\\\|\\[^\\]|[^\\/])*/[egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="s!(\\\\|\\!|[^!])*!(\\\\|\\!|[^!])*![egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="s\\(\\\\|[^\\])*\\(\\\\|[^\\])*\\[egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="s@(\\\\|\\[^\\]|[^\\@])*@(\\\\|\\[^\\]|[^\\@])*@[egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="s%(\\\\|\\[^\\]|[^\\%])*%(\\\\|\\[^\\]|[^\\%])*%[egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="s\{(\\\\|\\[^\\]|[^\\}])*\}\s*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<push state="balanced-regex"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="s<(\\\\|\\[^\\]|[^\\>])*>\s*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<push state="balanced-regex"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="s\[(\\\\|\\[^\\]|[^\\\]])*\]\s*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<push state="balanced-regex"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="s\((\\\\|\\[^\\]|[^\\)])*\)\s*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<push state="balanced-regex"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="m?/(\\\\|\\[^\\]|[^\\/\n])*/[gcimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="m(?=[/!\\{<\[(@%$])">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<push state="balanced-regex"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="((?<==~)|(?<=\())\s*/(\\\\|\\[^\\]|[^\\/])*/[gcimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\s+">
|
||
|
|
<token type="Text"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(getprotobynumber|getprotobyname|getservbyport|getservbyname|gethostbyname|gethostbyaddr|getnetbyaddr|getnetbyname|setprotoent|setpriority|getsockname|getprotoent|getpriority|getpeername|endprotoent|setsockopt|endhostent|setservent|socketpair|getsockopt|endservent|sethostent|getservent|gethostent|getnetent|wantarray|localtime|quotemeta|setnetent|prototype|endnetent|rewinddir|endpwent|endgrent|getpwent|getpwnam|getgrgid|dbmclose|continue|closedir|shutdown|setpwent|shmwrite|syswrite|truncate|formline|setgrent|getgrent|readpipe|getgrnam|readlink|readline|getpwuid|getlogin|telldir|opendir|shmread|readdir|getpgrp|getppid|waitpid|binmode|reverse|sprintf|unshift|symlink|seekdir|sysopen|defined|dbmopen|sysread|sysseek|connect|lcfirst|ucfirst|setpgrp|syscall|import|chroot|system|gmtime|unlink|unpack|semget|semctl|select|msgrcv|length|printf|listen|scalar|caller|rindex|substr|splice|rename|msgctl|msgget|delete|msgsnd|exists|values|socket|fileno|format|accept|shmget|shmctl|ioctl|alarm|atan2|utime|untie|bless|chdir|chmod|undef|umask|chomp|times|close|crypt|srand|split|mkdir|sleep|lstat|study|rmdir|local|fcntl|flock|write|shift|index|semop|chown|tell|join|send|warn|kill|wait|grep|glob|bind|last|pack|getc|fork|link|seek|push|rand|read|exit|sort|exec|open|eval|sqrt|redo|stat|chop|next|each|dump|time|recv|tied|goto|keys|pipe|map|tie|cos|die|eof|pop|sin|pos|oct|ref|exp|int|ord|log|our|hex|abs|vec|chr|uc|tr|my|lc)\b">
|
||
|
|
<token type="NameBuiltin"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="((__(DATA|DIE|WARN)__)|(STD(IN|OUT|ERR)))\b">
|
||
|
|
<token type="NameBuiltinPseudo"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(<<)([\'"]?)([a-zA-Z_]\w*)(\2;?\n.*?\n)(\3)(\n)">
|
||
|
|
<bygroups>
|
||
|
|
<token type="LiteralString"/>
|
||
|
|
<token type="LiteralString"/>
|
||
|
|
<token type="LiteralStringDelimiter"/>
|
||
|
|
<token type="LiteralString"/>
|
||
|
|
<token type="LiteralStringDelimiter"/>
|
||
|
|
<token type="Text"/>
|
||
|
|
</bygroups>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="__END__">
|
||
|
|
<token type="CommentPreproc"/>
|
||
|
|
<push state="end-part"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\$\^[ADEFHILMOPSTWX]">
|
||
|
|
<token type="NameVariableGlobal"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\$[\\\"\[\]'&`+*.,;=%~?@$!<>(^|/-](?!\w)">
|
||
|
|
<token type="NameVariableGlobal"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="[$@%#]+">
|
||
|
|
<token type="NameVariable"/>
|
||
|
|
<push state="varname"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="0_?[0-7]+(_[0-7]+)*">
|
||
|
|
<token type="LiteralNumberOct"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="0x[0-9A-Fa-f]+(_[0-9A-Fa-f]+)*">
|
||
|
|
<token type="LiteralNumberHex"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="0b[01]+(_[01]+)*">
|
||
|
|
<token type="LiteralNumberBin"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(?i)(\d*(_\d*)*\.\d+(_\d*)*|\d+(_\d*)*\.\d+(_\d*)*)(e[+-]?\d+)?">
|
||
|
|
<token type="LiteralNumberFloat"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(?i)\d+(_\d*)*e[+-]?\d+(_\d*)*">
|
||
|
|
<token type="LiteralNumberFloat"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\d+(_\d+)*">
|
||
|
|
<token type="LiteralNumberInteger"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="'(\\\\|\\[^\\]|[^'\\])*'">
|
||
|
|
<token type="LiteralString"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern=""(\\\\|\\[^\\]|[^"\\])*"">
|
||
|
|
<token type="LiteralString"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="`(\\\\|\\[^\\]|[^`\\])*`">
|
||
|
|
<token type="LiteralStringBacktick"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="<([^\s>]+)>">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(q|qq|qw|qr|qx)\{">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
<push state="cb-string"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(q|qq|qw|qr|qx)\(">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
<push state="rb-string"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(q|qq|qw|qr|qx)\[">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
<push state="sb-string"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(q|qq|qw|qr|qx)\<">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
<push state="lt-string"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(q|qq|qw|qr|qx)([\W_])(.|\n)*?\2">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(package)(\s+)([a-zA-Z_]\w*(?:::[a-zA-Z_]\w*)*)">
|
||
|
|
<bygroups>
|
||
|
|
<token type="Keyword"/>
|
||
|
|
<token type="Text"/>
|
||
|
|
<token type="NameNamespace"/>
|
||
|
|
</bygroups>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(use|require|no)(\s+)([a-zA-Z_]\w*(?:::[a-zA-Z_]\w*)*)">
|
||
|
|
<bygroups>
|
||
|
|
<token type="Keyword"/>
|
||
|
|
<token type="Text"/>
|
||
|
|
<token type="NameNamespace"/>
|
||
|
|
</bygroups>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(sub)(\s+)">
|
||
|
|
<bygroups>
|
||
|
|
<token type="Keyword"/>
|
||
|
|
<token type="Text"/>
|
||
|
|
</bygroups>
|
||
|
|
<push state="funcname"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(package|require|use|no)\b">
|
||
|
|
<token type="Keyword"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(\[\]|\*\*|::|<<|>>|>=|<=>|<=|={3}|!=|=~|!~|&&?|\|\||\.{1,3})">
|
||
|
|
<token type="Operator"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="[-+/*%=<>&^|!\\~]=?">
|
||
|
|
<token type="Operator"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="[()\[\]:;,<>/?{}]">
|
||
|
|
<token type="Punctuation"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(?=\w)">
|
||
|
|
<token type="Name"/>
|
||
|
|
<push state="name"/>
|
||
|
|
</rule>
|
||
|
|
</state>
|
||
|
|
<state name="cb-string">
|
||
|
|
<rule pattern="\\[{}\\]">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\\">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\{">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
<push state="cb-string"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\}">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="[^{}\\]+">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
</rule>
|
||
|
|
</state>
|
||
|
|
<state name="balanced-regex">
|
||
|
|
<rule pattern="/(\\\\|\\[^\\]|[^\\/])*/[egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="!(\\\\|\\[^\\]|[^\\!])*![egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\\(\\\\|[^\\])*\\[egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\{(\\\\|\\[^\\]|[^\\}])*\}[egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="<(\\\\|\\[^\\]|[^\\>])*>[egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\[(\\\\|\\[^\\]|[^\\\]])*\][egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\((\\\\|\\[^\\]|[^\\)])*\)[egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="@(\\\\|\\[^\\]|[^\\@])*@[egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="%(\\\\|\\[^\\]|[^\\%])*%[egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\$(\\\\|\\[^\\]|[^\\$])*\$[egimosx]*">
|
||
|
|
<token type="LiteralStringRegex"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
</state>
|
||
|
|
<state name="lt-string">
|
||
|
|
<rule pattern="\\[<>\\]">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\\">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\<">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
<push state="lt-string"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\>">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="[^<>]+">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
</rule>
|
||
|
|
</state>
|
||
|
|
<state name="format">
|
||
|
|
<rule pattern="\.\n">
|
||
|
|
<token type="LiteralStringInterpol"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="[^\n]*\n">
|
||
|
|
<token type="LiteralStringInterpol"/>
|
||
|
|
</rule>
|
||
|
|
</state>
|
||
|
|
<state name="funcname">
|
||
|
|
<rule pattern="[a-zA-Z_]\w*[!?]?">
|
||
|
|
<token type="NameFunction"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\s+">
|
||
|
|
<token type="Text"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(\([$@%]*\))(\s*)">
|
||
|
|
<bygroups>
|
||
|
|
<token type="Punctuation"/>
|
||
|
|
<token type="Text"/>
|
||
|
|
</bygroups>
|
||
|
|
</rule>
|
||
|
|
<rule pattern=";">
|
||
|
|
<token type="Punctuation"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern=".*?\{">
|
||
|
|
<token type="Punctuation"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
</state>
|
||
|
|
<state name="end-part">
|
||
|
|
<rule pattern=".+">
|
||
|
|
<token type="CommentPreproc"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
</state>
|
||
|
|
<state name="varname">
|
||
|
|
<rule pattern="\s+">
|
||
|
|
<token type="Text"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\{">
|
||
|
|
<token type="Punctuation"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\)|,">
|
||
|
|
<token type="Punctuation"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\w+::">
|
||
|
|
<token type="NameNamespace"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="[\w:]+">
|
||
|
|
<token type="NameVariable"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
</state>
|
||
|
|
<state name="name">
|
||
|
|
<rule pattern="[a-zA-Z_]\w*(::[a-zA-Z_]\w*)*(::)?(?=\s*->)">
|
||
|
|
<token type="NameNamespace"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="[a-zA-Z_]\w*(::[a-zA-Z_]\w*)*::">
|
||
|
|
<token type="NameNamespace"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="[\w:]+">
|
||
|
|
<token type="Name"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="[A-Z_]+(?=\W)">
|
||
|
|
<token type="NameConstant"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="(?=\W)">
|
||
|
|
<token type="Text"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
</state>
|
||
|
|
<state name="rb-string">
|
||
|
|
<rule pattern="\\[()\\]">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\\">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\(">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
<push state="rb-string"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\)">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="[^()]+">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
</rule>
|
||
|
|
</state>
|
||
|
|
<state name="sb-string">
|
||
|
|
<rule pattern="\\[\[\]\\]">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\\">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\[">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
<push state="sb-string"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="\]">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
<pop depth="1"/>
|
||
|
|
</rule>
|
||
|
|
<rule pattern="[^\[\]]+">
|
||
|
|
<token type="LiteralStringOther"/>
|
||
|
|
</rule>
|
||
|
|
</state>
|
||
|
|
</rules>
|
||
|
|
</lexer>
|