lexers/texinfo.lua

   1 -- Copyright 2014-2015 stef@ailleurs.land. See LICENSE.
   2 -- Plain Texinfo version 5.2 LPeg lexer
   3 -- Freely inspired from Mitchell work and valuable help from him too !
   4
   5 -- Directives are processed (more or less) in the Reference Card Texinfo order
   6 -- Reference Card page for each directive group is in comment for reference
   7
   8 --[[
   9 Note: Improving Fold Points use with Texinfo
  10
  11 At the very beginning of your Texinfo file, it could be wised to insert theses
  12 alias :
  13
  14 @alias startchapter = comment
  15 @alias endchapter = comment
  16
  17 Then use this to begin each chapter :
  18
  19 @endchapter --------------------------------------------------------------------
  20 @chapter CHAPTER TITLE
  21 @startchapter ------------------------------------------------------------------
  22
  23 With the use of Scintilla's `SCI_FOLDALL(SC_FOLDACTION_TOGGLE)` or Textadept's
  24 `buffer:fold_all(buffer.FOLDACTION_TOGGLE)`, you have then a nice chapter
  25 folding, useful with large documents.
  26 ]]
  27
  28 local l = require('lexer')
  29 local token, word_match = l.token, l.word_match
  30 local P, R, S = lpeg.P, lpeg.R, lpeg.S
  31
  32 local M = {_NAME = 'texinfo'}
  33
  34 -------------------------------------------------------------------------------
  35 -- Common processing
  36 -------------------------------------------------------------------------------
  37
  38 -- Whitespace.
  39 local ws = token(l.WHITESPACE, l.space^1)
  40
  41 -- Comments.
  42 local line_comment = '@c' * l.nonnewline_esc^0
  43 --local line_comment_long = '@comment' * l.nonnewline_esc^0
  44 local block_comment = '@ignore' * (l.any - '@end ignore')^0 *
  45                       P('@end ignore')^-1
  46 local comment = token(l.COMMENT, line_comment + block_comment)
  47
  48 -- Strings.
  49 local string = token(l.STRING, l.delimited_range('{}', false, true, true))
  50
  51 -- Numbers.
  52 local number = token(l.NUMBER, l.float + l.integer)
  53
  54 -- Identifiers
  55 local identifier = token(l.IDENTIFIER, l.word)
  56
  57 -------------------------------------------------------------------------------
  58 -- Common keywords
  59 -------------------------------------------------------------------------------
  60
  61 local keyword_base = word_match({
  62   'end',
  63   -- Beginning a Texinfo document (page 1, column 1)
  64   'setfilename', 'settitle', 'insertcopying',
  65   -- Beginning a Texinfo document > Internationlization (page 1, column 1)
  66   'documentencoding', 'documentlanguage', 'frenchspacing',
  67   -- Beginning a Texinfo document > Info directory specification and HTML
  68   -- document description (page 1, column 1)
  69   'dircategory', 'direntry', 'documentdescription',
  70   -- Beginning a Texinfo document > Titre pages (page 1, column 1)
  71   'shorttitlepage', 'center', 'titlefont', 'title', 'subtitle', 'author',
  72   -- Beginning a Texinfo document > Tables of contents (page 1, column 2)
  73   'shortcontents', 'summarycontents', 'contents', 'setcontentsaftertitlepage',
  74   'setshortcontentsaftertitlepage',
  75   -- Nodes (page 1, column 2)
  76   'node', 'top', 'anchor', 'novalidate',
  77   -- Menus (page 1, column 2)
  78   'menu', 'detailmenu',
  79   -- Cross references > Within the Info system (page 1, column 3)
  80   'xref', 'pxref', 'ref', 'inforef', 'xrefautomaticsectiontitle',
  81   -- Cross references > Outside of info (page 1, column 3)
  82   'url', 'cite',
  83   -- Marking text > Markup for regular text (page 1, column 3)
  84   'var', 'dfn', 'acronym', 'abbr',
  85   -- Marking text > Markup for litteral text (page 1, column 3)
  86   'code', 'file', 'command', 'env', 'option', 'kbd', 'key', 'email',
  87   'indicateurl', 'samp', 'verb',
  88   -- Marking text > GUI sequences (page 2, column 1)
  89   'clicksequence', 'click', 'clickstyle', 'arrow',
  90   -- Marking text > Math (page 2, column 1)
  91   'math', 'minus', 'geq', 'leq',
  92   -- Marking text > Explicit font selection (page 2, column 1)
  93   'sc', 'r', 'i', 'slanted', 'b', 'sansserif', 't',
  94   -- Block environments (page 2, column 1)
  95   'noindent', 'indent', 'exdent',
  96   -- Block environments > Normally filled displays using regular text fonts
  97   -- (page 2, column 1)
  98   'quotation', 'smallquotation', 'indentedblock', 'smallindentedblock',
  99   'raggedright',
 100   -- Block environments > Line-for-line displays using regular test fonts (page
 101   -- 2, column 2)
 102   'format', 'smallformat', 'display', 'smalldisplay', 'flushleft', 'flushright',
 103   -- Block environments > Displays using fixed-width fonts (page 2, column 2)
 104   'lisp', 'smalllisp', 'verbatim',
 105   -- List and tables (page 2, column 2)
 106   'table', 'ftable', 'vtable', 'tab', 'item', 'itemx', 'headitem',
 107   'headitemfont', 'asis',
 108   -- Indices (page 2, column 3)
 109   'cindex', 'findex', 'vindex', 'kindex', 'pindex', 'tindex', 'defcodeindex',
 110   'syncodeindex', 'synindex', 'printindex',
 111   -- Insertions within a paragraph > Characters special to Texinfo (page 2,
 112   -- column 3)
 113   '@', '{', '}', 'backslashcar', 'comma', 'hashcar', ':', '.', '?', '!', 'dmn',
 114   -- Insertions within a paragraph > Accents (page 3, column 1)
 115   -- not implemented
 116   -- Insertions within a paragraph > Non-English characters (page 3, column 1)
 117   -- not implemented
 118   -- Insertions within a paragraph > Other text characters an logos (page 3,
 119   -- column 1)
 120   'bullet', 'dots', 'enddots', 'euro', 'pounds', 'textdegree', 'copyright',
 121   'registeredsymbol', 'TeX', 'LaTeX', 'today',
 122   'guillemetleft', 'guillementright', 'guillemotleft', 'guillemotright',
 123   -- Insertions within a paragraph > Glyphs for code examples (page 3, column 2)
 124   'equiv', 'error', 'expansion', 'point', 'print', 'result',
 125   -- Making and preventing breaks (page 3, column 2)
 126   '*', '/', '-', 'hyphenation', 'tie', 'w', 'refill',
 127   -- Vertical space (page 3, column 2)
 128   'sp', 'page', 'need', 'group', 'vskip'
 129   -- Definition commands (page 3, column 2)
 130   -- not implemented
 131 }, nil, true)
 132
 133 local keyword = token(l.KEYWORD, ('@end' * l.space^1 + '@') * keyword_base)
 134
 135 -------------------------------------------------------------------------------
 136 -- Chapter structuring Keywords
 137 -------------------------------------------------------------------------------
 138
 139 local chapters_base = word_match({
 140   -- Chapter structuring (page 1, column 2)
 141   'lowersections', 'raisesections', 'part',
 142   -- Chapter structuring > Numbered, included in contents (page 1, column 2)
 143   'chapter', 'centerchap',
 144   -- Chapter structuring > Context-dependent, included in contents (page 1,
 145   -- column 2)
 146   'section', 'subsection', 'subsubsection',
 147   -- Chapter structuring > Unumbered, included in contents (page 1, column 2)
 148   'unnumbered', 'unnumberedsec', 'unnumberedsubsec', 'unnumberedsubsection',
 149   'unnumberedsubsubsec', 'unnumberedsubsubsection',
 150   -- Chapter structuring > Letter and numbered, included in contents (page 1,
 151   -- column 2)
 152   'appendix', 'appendixsec', 'appendixsection', 'appendixsubsec',
 153   'appendixsubsection', 'appendixsubsubsec', 'appendixsubsubsection',
 154   -- Chapter structuring > Unumbered, not included in contents, no new page
 155   -- (page 1, column 3)
 156   'chapheading', 'majorheading', 'heading', 'subheading', 'subsubheading'
 157 }, nil, true)
 158
 159 local chapters = token('chapters', ('@end' * l.space^1 + '@') * chapters_base)
 160
 161 -------------------------------------------------------------------------------
 162 -- Directives Keywords
 163 -------------------------------------------------------------------------------
 164
 165 local directives_base = word_match({
 166   'end',
 167   -- Custom keywords for chapter folding
 168   'startchapter', 'endchapter',
 169   -- List and tables (page 2, column 2)
 170   'itemize', 'enumerate',
 171   -- Beginning a Texinfo document (page 1, column 1)
 172   'titlepage', 'copying',
 173   -- Block environments (page 2, column 1)
 174   'cartouche',
 175   -- Block environments > Displays using fixed-width fonts (page 2, column 2)
 176   'example', 'smallexample',
 177   -- List and tables (page 2, column 2)
 178   'multitable',
 179   -- Floating Displays (page 2, column 3)
 180   'float', 'listoffloats', 'caption', 'shortcaption', 'image',
 181   -- Floating Displays > Footnotes (page 2, column 3)
 182   'footnote', 'footnotestyle',
 183   -- Conditionally (in)visible text > Output formats (page 3, column 3)
 184   'ifdocbook', 'ifhtml', 'ifinfo', 'ifplaintext', 'iftex', 'ifxml',
 185   'ifnotdocbook', 'ifnothtml', 'ifnotplaintext',
 186   'ifnottex', 'ifnotxml', 'ifnotinfo', 'inlinefmt', 'inlinefmtifelse',
 187   -- Conditionally (in)visible text > Raw formatter text (page 4, column 1)
 188   'docbook', 'html', 'tex', 'xml', 'inlineraw',
 189   -- Conditionally (in)visible text > Documents variables (page 4, column 1)
 190   'set', 'clear', 'value', 'ifset', 'ifclear', 'inlineifset', 'inlineifclear',
 191   -- Conditionally (in)visible text > Testing for commands (page 4, column 1)
 192   'ifcommanddefined', 'ifcommandnotdefined', 'end',
 193   -- Defining new Texinfo commands (page 4, column 1)
 194   'alias', 'macro', 'unmacro', 'definfounclose',
 195   -- File inclusion (page 4, column 1)
 196   'include', 'verbatiminclude',
 197   -- Formatting and headers footers for TeX (page 4, column 1)
 198   'allowcodebreaks', 'finalout', 'fonttextsize',
 199   -- Formatting and headers footers for TeX > paper size (page 4, column 2)
 200   'smallbook', 'afourpaper', 'afivepaper', 'afourlatex', 'afourwide',
 201   'pagesizes',
 202   -- Formatting and headers footers for TeX > Page headers and footers (page 4,
 203   -- column 2)
 204   -- not implemented
 205   -- Document preferences (page 4, column 2)
 206   -- not implemented
 207   -- Ending a Texinfo document (page 4, column 2)
 208   'bye'
 209 }, nil, case_insensitive_tags)
 210
 211 local directives = token('directives',
 212                          ('@end' * l.space^1 + '@') * directives_base)
 213
 214 -------------------------------------------------------------------------------
 215 -- Special keywords
 216 -------------------------------------------------------------------------------
 217
 218 -- Italics
 219 local emph = token('emph', '@emph' * l.delimited_range('{}', false, true, true))
 220
 221 -- Bold
 222 local strong = token('strong',
 223                      '@strong' * l.delimited_range('{}', false, true, true))
 224
 225 -------------------------------------------------------------------------------
 226 -- Rules, Tokenstyles and Folds points
 227 -------------------------------------------------------------------------------
 228
 229 M._rules = {
 230   {'whitespace', ws},
 231   {'directives', directives},
 232   {'chapters', chapters},
 233   {'keyword', keyword},
 234   {'emph', emph},
 235   {'strong', strong},
 236   {'identifier', identifier},
 237   {'string', string},
 238   {'number', number},
 239   {'comment', comment},
 240 }
 241
 242 M._tokenstyles = {
 243   directives = l.STYLE_FUNCTION,
 244   chapters = l.STYLE_CLASS,
 245   emph = l.STYLE_STRING..',italics',
 246   strong = l.STYLE_STRING..',bold'
 247 }
 248
 249 M._foldsymbols = {
 250   _patterns = {'@end %l+', '@%l+'},
 251   directives = {
 252     ['@titlepage'] = 1, ['@end titlepage'] = -1,
 253     ['@copying'] = 1, ['@end copying'] = -1,
 254
 255     ['@ifset'] = 1, ['@end ifset'] = -1,
 256     ['@tex'] = 1, ['@end tex'] = -1,
 257
 258     ['@itemize'] = 1, ['@end itemize'] = -1,
 259     ['@enumerate'] = 1, ['@end enumerate'] = -1,
 260     ['@multitable'] = 1, ['@end multitable'] = -1,
 261
 262     ['@example'] = 1, ['@end example'] = -1,
 263     ['@smallexample'] = 1, ['@end smallexample'] = -1,
 264     ['@cartouche'] = 1, ['@end cartouche'] = -1,
 265
 266     ['@startchapter'] = 1, ['@endchapter'] = -1,
 267   }
 268 }
 269
 270 return M