modules/syntax-highlight.pl

   1 #!/usr/bin/env perl
   2 # ====================[ syntax-highlight.pl                          ]====================
   3
   4 =head1 NAME
   5
   6 syntax-highlight - An Oddmuse module for syntax highlighting of source
   7 snippets.
   8
   9 =head1 INSTALLATION
  10
  11 syntax-highlight is easily installable; move this file into the
  12 B<wiki/modules/> directory for your Oddmuse Wiki.
  13
  14 =head1 DEPENDENCIES
  15
  16 This module relies on L<Syntax::Highlight::Engine::Kate> to do the
  17 actual coloring.  On a Debian, the Kate syntax highlighter is
  18 available as package C<libsyntax-highlight-engine-kate-perl>.
  19
  20 =cut
  21 package OddMuse;
  22
  23 $ModulesDescription .= '<p><a href="http://git.savannah.gnu.org/cgit/oddmuse.git/tree/modules/syntax-highlight.pl">syntax-highlight.pl</a>, see <a href="http://www.oddmuse.org/wiki/Syntax_Highlighting">Syntax Highlight Extension</a></p>';
  24
  25 use Syntax::Highlight::Engine::Kate;
  26
  27 =head1 VARIABLES
  28
  29 =head2 %NativeLanguageToKateLanguage
  30
  31 Languages supported by the Kate syntax highlighter.  Use the value of
  32 this variable as a reference to setup language mapper for your markup
  33 module.  See C<HighlightSyntax>.
  34
  35 =cut
  36
  37 my %NativeLanguageToKateLanguage =
  38   (
  39    ".desktop"            => ".desktop",
  40    "4GL"                 => "4GL",
  41    "4GL-PER"             => "4GL-PER",
  42    "ABC"                 => "ABC",
  43    "AHDL"                => "AHDL",
  44    "ANSI C89"            => "ANSI C89",
  45    "ASP"                 => "ASP",
  46    "AVR Assembler"       => "AVR Assembler",
  47    "AWK"                 => "AWK",
  48    "Ada"                 => "Ada",
  49    "Alerts"              => "Alerts",
  50    "Ansys"               => "Ansys",
  51    "Apache Configuration"=> "Apache Configuration",
  52    "Asm6502"             => "Asm6502",
  53    "Bash"                => "Bash",
  54    "BibTeX"              => "BibTeX",
  55    "C"                   => "C",
  56    "C#"                  => "C#",
  57    "C++"                 => "C++",
  58    "CGiS"                => "CGiS",
  59    "CMake"               => "CMake",
  60    "CSS"                 => "CSS",
  61    "CSS/PHP"             => "CSS/PHP",
  62    "CUE Sheet"           => "CUE Sheet",
  63    "Cg"                  => "Cg",
  64    "ChangeLog"           => "ChangeLog",
  65    "Cisco"               => "Cisco",
  66    "Clipper"             => "Clipper",
  67    "ColdFusion"          => "ColdFusion",
  68    "Common Lisp"         => "Common Lisp",
  69    "Component-Pascal"    => "Component-Pascal",
  70    "D"                   => "D",
  71    "Debian Changelog"    => "Debian Changelog",
  72    "Debian Control"      => "Debian Control",
  73    "Diff"                => "Diff",
  74    "Doxygen"             => "Doxygen",
  75    "E Language"          => "E Language",
  76    "Eiffel"              => "Eiffel",
  77    "Email"               => "Email",
  78    "Euphoria"            => "Euphoria",
  79    "Fortran"             => "Fortran",
  80    "FreeBASIC"           => "FreeBASIC",
  81    "GDL"                 => "GDL",
  82    "GLSL"                => "GLSL",
  83    "GNU Assembler"       => "GNU Assembler",
  84    "GNU Gettext"         => "GNU Gettext",
  85    "HTML"                => "HTML",
  86    "Haskell"             => "Haskell",
  87    "IDL"                 => "IDL",
  88    "ILERPG"              => "ILERPG",
  89    "INI Files"           => "INI Files",
  90    "Inform"              => "Inform",
  91    "Intel x86 (NASM)"    => "Intel x86 (NASM)",
  92    "JSP"                 => "JSP",
  93    "Java"                => "Java",
  94    "JavaScript"          => "JavaScript",
  95    "JavaScript/PHP"      => "JavaScript/PHP",
  96    "Javadoc"             => "Javadoc",
  97    "KBasic"              => "KBasic",
  98    "Kate File Template"  => "Kate File Template",
  99    "LDIF"                => "LDIF",
 100    "LPC"                 => "LPC",
 101    "LaTeX"               => "LaTeX",
 102    "Lex/Flex"            => "Lex/Flex",
 103    "LilyPond"            => "LilyPond",
 104    "Literate Haskell"    => "Literate Haskell",
 105    "Logtalk"             => "Logtalk",
 106    "Lua"                 => "Lua",
 107    "M3U"                 => "M3U",
 108    "MAB-DB"              => "MAB-DB",
 109    "MIPS Assembler"      => "MIPS Assembler",
 110    "Makefile"            => "Makefile",
 111    "Mason"               => "Mason",
 112    "Matlab"              => "Matlab",
 113    "Modula-2"            => "Modula-2",
 114    "Music Publisher"     => "Music Publisher",
 115    "Objective Caml"      => "Objective Caml",
 116    "Objective-C"         => "Objective-C",
 117    "Octave"              => "Octave",
 118    "PHP (HTML)"          => "PHP (HTML)",
 119    "PHP/PHP"             => "PHP/PHP",
 120    "POV-Ray"             => "POV-Ray",
 121    "Pascal"              => "Pascal",
 122    "Perl"                => "Perl",
 123    "PicAsm"              => "PicAsm",
 124    "Pike"                => "Pike",
 125    "PostScript"          => "PostScript",
 126    "Prolog"              => "Prolog",
 127    "PureBasic"           => "PureBasic",
 128    "Python"              => "Python",
 129    "Quake Script"        => "Quake Script",
 130    "R Script"            => "R Script",
 131    "REXX"                => "REXX",
 132    "RPM Spec"            => "RPM Spec",
 133    "RSI IDL"             => "RSI IDL",
 134    "RenderMan RIB"       => "RenderMan RIB",
 135    "Ruby"                => "Ruby",
 136    "SGML"                => "SGML",
 137    "SML"                 => "SML",
 138    "SQL (MySQL)"         => "SQL (MySQL)",
 139    "SQL (PostgreSQL)"    => "SQL (PostgreSQL)",
 140    "SQL"                 => "SQL",
 141    "Sather"              => "Sather",
 142    "Scheme"              => "Scheme",
 143    "Sieve"               => "Sieve",
 144    "Spice"               => "Spice",
 145    "Stata"               => "Stata",
 146    "TI Basic"            => "TI Basic",
 147    "TaskJuggler"         => "TaskJuggler",
 148    "Tcl/Tk"              => "Tcl/Tk",
 149    "UnrealScript"        => "UnrealScript",
 150    "VHDL"                => "VHDL",
 151    "VRML"                => "VRML",
 152    "Velocity"            => "Velocity",
 153    "Verilog"             => "Verilog",
 154    "WINE Config"         => "WINE Config",
 155    "Wikimedia"           => "Wikimedia",
 156    "XML"                 => "XML",
 157    "Yacc/Bison"          => "Yacc/Bison",
 158    "de_DE"               => "de_DE",
 159    "en_US"               => "en_US",
 160    "ferite"              => "ferite",
 161    "nl"                  => "nl",
 162    "progress"            => "progress",
 163    "scilab"              => "scilab",
 164    "txt2tags"            => "txt2tags",
 165    "x.org Configuration" => "x.org Configuration",
 166    "xHarbour"            => "xHarbour",
 167    "xslt"                => "xslt",
 168    "yacas"               => "yacas"
 169   );
 170
 171 my $syntax_hl =
 172   new Syntax::Highlight::Engine::Kate
 173   (language => 'Perl',
 174    substitutions =>
 175    {
 176     "<" => "&lt;",
 177     ">" => "&gt;",
 178     "&" => "&amp;",
 179     # Oddmuse encloses src blocks within <pre>...</pre>.
 180     # " " => "&nbsp;",
 181     # "\t" => "&nbsp;&nbsp;&nbsp;",
 182     # "\n" => "<br/>\n",
 183    },
 184    format_table =>
 185    {
 186     Alert       => ["<span style=\"color: #0000ff\">"            , "</span>"],
 187     BaseN       => ["<span style=\"color: #007f00\">"            , "</span>"],
 188     # BString   => ["<span style=\"color: #c9a7ff\">"            , "</span>"],
 189     BString     => [""                                           , ""],
 190     Char        => ["<span style=\"color: #ff00ff\">"            , "</span>"],
 191     Comment     => ["<span style=\"color: #7f7f7f\"><em>"        , "</em></span>"],
 192     DataType    => ["<span style=\"color: #0000ff\">"            , "</span>"],
 193     DecVal      => ["<span style=\"color: #00007f\">"            , "</span>"],
 194     Error       => ["<span style=\"color: #ff0000\"><strong><em>", "</em></strong></span>"],
 195     Float       => ["<span style=\"color: #00007f\">"            , "</span>"],
 196     Function    => ["<span style=\"color: #007f00\">"            , "</span>"],
 197     IString     => ["<span style=\"color: #ff0000\">"            , "<//span>"],
 198     Keyword     => ["<strong>"                                   , "</strong>"],
 199     Normal      => [""                                           , ""],
 200     Operator    => ["<span style=\"color: #ffa500\">"            , "</span>"],
 201     Others      => ["<span style=\"color: #b03060\">"            , "</span>"],
 202     RegionMarker=> ["<span style=\"color: #96b9ff\"><em>"        , "</em></span>"],
 203     Reserved    => ["<span style=\"color: #9b30ff\"><strong>"    , "</strong></span>"],
 204     String      => ["<span style=\"color: #ff0000\">"            , "</span>"],
 205     Variable    => ["<span style=\"color: #0000ff\"><strong>"    , "</strong></span>"],
 206     Warning     => ["<span style=\"color: #0000ff\"><strong><em>", "</strong></em></span>"],
 207    },
 208   );
 209
 210 =head1 FUNCTIONS
 211
 212 =head2 HighlightSyntax($src, $kateLang)
 213
 214 Markup C<$quoted_src> in C<$kateLang> and return the resulting HTML.
 215 If C<$kateLang> is unknown to Kate, return C<$quoted_src>.  Assume
 216 C<$quoted_src> as quoted.  Unquote the source block with
 217 C<UnquoteHtml> before passing it to Kate for highlighting.
 218
 219 Before using this function, you may have to setup up a hash that maps
 220 from the name of the language as it is designated by your markup to
 221 the how the language itself is called by Kate.  Use
 222 C<%NativeLanguageToKateLanguage> as a template to setup this hash.
 223
 224 Invoke C<HighlightSyntax> as below.
 225
 226         my %OrgLanguageToKateLanguage = ("perl"=>"Perl");
 227         # Do Syntax Highlighting
 228         if (defined(&HighlightSyntax)) {
 229           $lines = HighlightSyntax($src, $OrgLanguageToKateLanguage{$lang});
 230         }
 231
 232 Here, the hash C<%OrgLanguageToKateLanguage> maps the language from
 233 Emacs/Org-mode to Kate.  As seen above, C<Perl> blocks in
 234 Emacs/Org-mode markup are designated by C<perl> while in Kate they are
 235 designated by C<Perl>.
 236
 237 =cut
 238
 239 sub HighlightSyntax {
 240   my ($quoted_src, $kateLang) = @_;
 241
 242   # No highlighter for current language
 243   return $quoted_src
 244     unless $syntax_hl->languagePlug($kateLang);
 245
 246   # Use highlighter
 247   my $unquoted_src = UnquoteHtml($quoted_src);
 248   $syntax_hl->language($kateLang);
 249   return $syntax_hl->highlightText($unquoted_src);
 250 }
 251
 252 =head1 COPYRIGHT AND LICENSE
 253
 254 Copyright 2013       by Jambunathan K <kjambunathan@gmail.com>.
 255
 256 This program is free software; you can redistribute it and/or modify
 257 it under the terms of the GNU General Public License as published by
 258 the Free Software Foundation; either version 3 of the License, or
 259 (at your option) any later version.
 260
 261 This program is distributed in the hope that it will be useful,
 262 but WITHOUT ANY WARRANTY; without even the implied warranty of
 263 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 264 GNU General Public License for more details.
 265
 266 You should have received a copy of the GNU General Public License
 267 along with this program. If not, see L<http://www.gnu.org/licenses/>.
 268
 269 =cut