ext/scintilla/src/LexTeX.cxx

   1 // Scintilla source code edit control\r
   2 \r
   3 // File: LexTeX.cxx - general context conformant tex coloring scheme\r
   4 // Author: Hans Hagen - PRAGMA ADE - Hasselt NL - www.pragma-ade.com\r
   5 // Version: September 28, 2003\r
   6 \r
   7 // Copyright: 1998-2003 by Neil Hodgson <neilh@scintilla.org>\r
   8 // The License.txt file describes the conditions under which this software may be distributed.\r
   9 \r
  10 // This lexer is derived from the one written for the texwork environment (1999++) which in\r
  11 // turn is inspired on texedit (1991++) which finds its roots in wdt (1986).\r
  12 \r
  13 // If you run into strange boundary cases, just tell me and I'll look into it.\r
  14 \r
  15 \r
  16 // TeX Folding code added by instanton (soft_share@126.com) with borrowed code from VisualTeX source by Alex Romanenko.\r
  17 // Version: June 22, 2007\r
  18 \r
  19 #include <stdlib.h>\r
  20 #include <string.h>\r
  21 #include <ctype.h>\r
  22 #include <stdio.h>\r
  23 #include <stdarg.h>\r
  24 \r
  25 #include "Platform.h"\r
  26 \r
  27 #include "PropSet.h"\r
  28 #include "Accessor.h"\r
  29 #include "KeyWords.h"\r
  30 #include "Scintilla.h"\r
  31 #include "SciLexer.h"\r
  32 #include "StyleContext.h"\r
  33 \r
  34 #ifdef SCI_NAMESPACE\r
  35 using namespace Scintilla;\r
  36 #endif\r
  37 \r
  38 // val SCE_TEX_DEFAULT = 0\r
  39 // val SCE_TEX_SPECIAL = 1\r
  40 // val SCE_TEX_GROUP   = 2\r
  41 // val SCE_TEX_SYMBOL  = 3\r
  42 // val SCE_TEX_COMMAND = 4\r
  43 // val SCE_TEX_TEXT    = 5\r
  44 \r
  45 // Definitions in SciTEGlobal.properties:\r
  46 //\r
  47 // TeX Highlighting\r
  48 //\r
  49 // # Default\r
  50 // style.tex.0=fore:#7F7F00\r
  51 // # Special\r
  52 // style.tex.1=fore:#007F7F\r
  53 // # Group\r
  54 // style.tex.2=fore:#880000\r
  55 // # Symbol\r
  56 // style.tex.3=fore:#7F7F00\r
  57 // # Command\r
  58 // style.tex.4=fore:#008800\r
  59 // # Text\r
  60 // style.tex.5=fore:#000000\r
  61 \r
  62 // lexer.tex.interface.default=0\r
  63 // lexer.tex.comment.process=0\r
  64 \r
  65 // todo: lexer.tex.auto.if\r
  66 \r
  67 // Auxiliary functions:\r
  68 \r
  69 static inline bool endOfLine(Accessor &styler, unsigned int i) {\r
  70         return\r
  71       (styler[i] == '\n') || ((styler[i] == '\r') && (styler.SafeGetCharAt(i + 1) != '\n')) ;\r
  72 }\r
  73 \r
  74 static inline bool isTeXzero(int ch) {\r
  75         return\r
  76       (ch == '%') ;\r
  77 }\r
  78 \r
  79 static inline bool isTeXone(int ch) {\r
  80         return\r
  81       (ch == '[') || (ch == ']') || (ch == '=') || (ch == '#') ||\r
  82       (ch == '(') || (ch == ')') || (ch == '<') || (ch == '>') ||\r
  83       (ch == '"') ;\r
  84 }\r
  85 \r
  86 static inline bool isTeXtwo(int ch) {\r
  87         return\r
  88       (ch == '{') || (ch == '}') || (ch == '$') ;\r
  89 }\r
  90 \r
  91 static inline bool isTeXthree(int ch) {\r
  92         return\r
  93       (ch == '~') || (ch == '^') || (ch == '_') || (ch == '&') ||\r
  94       (ch == '-') || (ch == '+') || (ch == '\"') || (ch == '`') ||\r
  95       (ch == '/') || (ch == '|') || (ch == '%') ;\r
  96 }\r
  97 \r
  98 static inline bool isTeXfour(int ch) {\r
  99         return\r
 100       (ch == '\\') ;\r
 101 }\r
 102 \r
 103 static inline bool isTeXfive(int ch) {\r
 104         return\r
 105       ((ch >= 'a') && (ch <= 'z')) || ((ch >= 'A') && (ch <= 'Z')) ||\r
 106       (ch == '@') || (ch == '!') || (ch == '?') ;\r
 107 }\r
 108 \r
 109 static inline bool isTeXsix(int ch) {\r
 110         return\r
 111       (ch == ' ') ;\r
 112 }\r
 113 \r
 114 static inline bool isTeXseven(int ch) {\r
 115         return\r
 116       (ch == '^') ;\r
 117 }\r
 118 \r
 119 // Interface determination\r
 120 \r
 121 static int CheckTeXInterface(\r
 122     unsigned int startPos,\r
 123     int length,\r
 124     Accessor &styler,\r
 125         int defaultInterface) {\r
 126 \r
 127     char lineBuffer[1024] ;\r
 128         unsigned int linePos = 0 ;\r
 129 \r
 130     // some day we can make something lexer.tex.mapping=(all,0)(nl,1)(en,2)...\r
 131 \r
 132     if (styler.SafeGetCharAt(0) == '%') {\r
 133         for (unsigned int i = 0; i < startPos + length; i++) {\r
 134             lineBuffer[linePos++] = styler.SafeGetCharAt(i) ;\r
 135             if (endOfLine(styler, i) || (linePos >= sizeof(lineBuffer) - 1)) {\r
 136                 lineBuffer[linePos] = '\0';\r
 137                 if (strstr(lineBuffer, "interface=all")) {\r
 138                     return 0 ;\r
 139                                 } else if (strstr(lineBuffer, "interface=tex")) {\r
 140                     return 1 ;\r
 141                 } else if (strstr(lineBuffer, "interface=nl")) {\r
 142                     return 2 ;\r
 143                 } else if (strstr(lineBuffer, "interface=en")) {\r
 144                     return 3 ;\r
 145                 } else if (strstr(lineBuffer, "interface=de")) {\r
 146                     return 4 ;\r
 147                 } else if (strstr(lineBuffer, "interface=cz")) {\r
 148                     return 5 ;\r
 149                 } else if (strstr(lineBuffer, "interface=it")) {\r
 150                     return 6 ;\r
 151                 } else if (strstr(lineBuffer, "interface=ro")) {\r
 152                     return 7 ;\r
 153                 } else if (strstr(lineBuffer, "interface=latex")) {\r
 154                                         // we will move latex cum suis up to 91+ when more keyword lists are supported\r
 155                     return 8 ;\r
 156                                 } else if (styler.SafeGetCharAt(1) == 'D' && strstr(lineBuffer, "%D \\module")) {\r
 157                                         // better would be to limit the search to just one line\r
 158                                         return 3 ;\r
 159                 } else {\r
 160                     return defaultInterface ;\r
 161                 }\r
 162             }\r
 163                 }\r
 164     }\r
 165 \r
 166     return defaultInterface ;\r
 167 }\r
 168 \r
 169 static void ColouriseTeXDoc(\r
 170     unsigned int startPos,\r
 171     int length,\r
 172     int,\r
 173     WordList *keywordlists[],\r
 174     Accessor &styler) {\r
 175 \r
 176         styler.StartAt(startPos) ;\r
 177         styler.StartSegment(startPos) ;\r
 178 \r
 179         bool processComment   = styler.GetPropertyInt("lexer.tex.comment.process",   0) == 1 ;\r
 180         bool useKeywords      = styler.GetPropertyInt("lexer.tex.use.keywords",      1) == 1 ;\r
 181         bool autoIf           = styler.GetPropertyInt("lexer.tex.auto.if",           1) == 1 ;\r
 182         int  defaultInterface = styler.GetPropertyInt("lexer.tex.interface.default", 1) ;\r
 183 \r
 184         char key[100] ;\r
 185         int  k ;\r
 186         bool newifDone = false ;\r
 187         bool inComment = false ;\r
 188 \r
 189         int currentInterface = CheckTeXInterface(startPos,length,styler,defaultInterface) ;\r
 190 \r
 191     if (currentInterface == 0) {\r
 192         useKeywords = false ;\r
 193         currentInterface = 1 ;\r
 194     }\r
 195 \r
 196     WordList &keywords = *keywordlists[currentInterface-1] ;\r
 197 \r
 198         StyleContext sc(startPos, length, SCE_TEX_TEXT, styler);\r
 199 \r
 200         bool going = sc.More() ; // needed because of a fuzzy end of file state\r
 201 \r
 202         for (; going; sc.Forward()) {\r
 203 \r
 204                 if (! sc.More()) { going = false ; } // we need to go one behind the end of text\r
 205 \r
 206                 if (inComment) {\r
 207                         if (sc.atLineEnd) {\r
 208                                 sc.SetState(SCE_TEX_TEXT) ;\r
 209                                 newifDone = false ;\r
 210                                 inComment = false ;\r
 211                         }\r
 212                 } else {\r
 213                         if (! isTeXfive(sc.ch)) {\r
 214                                 if (sc.state == SCE_TEX_COMMAND) {\r
 215                                         if (sc.LengthCurrent() == 1) { // \<noncstoken>\r
 216                                                 if (isTeXseven(sc.ch) && isTeXseven(sc.chNext)) {\r
 217                                                         sc.Forward(2) ; // \^^ and \^^<token>\r
 218                                                 }\r
 219                                                 sc.ForwardSetState(SCE_TEX_TEXT) ;\r
 220                                         } else {\r
 221                                                 sc.GetCurrent(key, sizeof(key)-1) ;\r
 222                                                 k = strlen(key) ;\r
 223                                                 memmove(key,key+1,k) ; // shift left over escape token\r
 224                                                 key[k] = '\0' ;\r
 225                                                 k-- ;\r
 226                                                 if (! keywords || ! useKeywords) {\r
 227                                                         sc.SetState(SCE_TEX_COMMAND) ;\r
 228                                                         newifDone = false ;\r
 229                                                 } else if (k == 1) { //\<cstoken>\r
 230                                                         sc.SetState(SCE_TEX_COMMAND) ;\r
 231                                                         newifDone = false ;\r
 232                                                 } else if (keywords.InList(key)) {\r
 233                                                 sc.SetState(SCE_TEX_COMMAND) ;\r
 234                                                         newifDone = autoIf && (strcmp(key,"newif") == 0) ;\r
 235                                                 } else if (autoIf && ! newifDone && (key[0] == 'i') && (key[1] == 'f') && keywords.InList("if")) {\r
 236                                                 sc.SetState(SCE_TEX_COMMAND) ;\r
 237                                                 } else {\r
 238                                                         sc.ChangeState(SCE_TEX_TEXT) ;\r
 239                                                         sc.SetState(SCE_TEX_TEXT) ;\r
 240                                                         newifDone = false ;\r
 241                                                 }\r
 242                                         }\r
 243                                 }\r
 244                                 if (isTeXzero(sc.ch)) {\r
 245                                         sc.SetState(SCE_TEX_SYMBOL) ;\r
 246                                         sc.ForwardSetState(SCE_TEX_DEFAULT) ;\r
 247                                         inComment = ! processComment ;\r
 248                                         newifDone = false ;\r
 249                                 } else if (isTeXseven(sc.ch) && isTeXseven(sc.chNext)) {\r
 250                                         sc.SetState(SCE_TEX_TEXT) ;\r
 251                                         sc.ForwardSetState(SCE_TEX_TEXT) ;\r
 252                                 } else if (isTeXone(sc.ch)) {\r
 253                                         sc.SetState(SCE_TEX_SPECIAL) ;\r
 254                                         newifDone = false ;\r
 255                                 } else if (isTeXtwo(sc.ch)) {\r
 256                                         sc.SetState(SCE_TEX_GROUP) ;\r
 257                                         newifDone = false ;\r
 258                                 } else if (isTeXthree(sc.ch)) {\r
 259                                         sc.SetState(SCE_TEX_SYMBOL) ;\r
 260                                         newifDone = false ;\r
 261                                 } else if (isTeXfour(sc.ch)) {\r
 262                                         sc.SetState(SCE_TEX_COMMAND) ;\r
 263                                 } else if (isTeXsix(sc.ch)) {\r
 264                                         sc.SetState(SCE_TEX_TEXT) ;\r
 265                                 } else if (sc.atLineEnd) {\r
 266                                         sc.SetState(SCE_TEX_TEXT) ;\r
 267                                         newifDone = false ;\r
 268                                         inComment = false ;\r
 269                                 } else {\r
 270                                         sc.SetState(SCE_TEX_TEXT) ;\r
 271                                 }\r
 272                         } else if (sc.state != SCE_TEX_COMMAND) {\r
 273                                 sc.SetState(SCE_TEX_TEXT) ;\r
 274                         }\r
 275                 }\r
 276         }\r
 277         sc.ChangeState(SCE_TEX_TEXT) ;\r
 278         sc.Complete();\r
 279 \r
 280 }\r
 281 \r
 282 \r
 283 static inline bool isNumber(int ch) {\r
 284         return\r
 285       (ch == '0') || (ch == '1') || (ch == '2') || \r
 286       (ch == '3') || (ch == '4') || (ch == '5') || \r
 287       (ch == '6') || (ch == '7') || (ch == '8') || (ch == '9');\r
 288 }\r
 289 \r
 290 static inline bool isWordChar(int ch) {\r
 291         return ((ch >= 'a') && (ch <= 'z')) || ((ch >= 'A') && (ch <= 'Z'));\r
 292 }\r
 293 \r
 294 static int ParseTeXCommand(unsigned int pos, Accessor &styler, char *command)\r
 295 {\r
 296   int length=0;\r
 297   char ch=styler.SafeGetCharAt(pos+1);\r
 298   \r
 299   if(ch==',' || ch==':' || ch==';' || ch=='%'){\r
 300       command[0]=ch;\r
 301       command[1]=0;\r
 302           return 1;\r
 303   }\r
 304 \r
 305   // find end\r
 306      while(isWordChar(ch) && !isNumber(ch) && ch!='_' && ch!='.' && length<100){\r
 307           command[length]=ch;\r
 308           length++;\r
 309           ch=styler.SafeGetCharAt(pos+length+1);\r
 310      }\r
 311      \r
 312   command[length]='\0';   \r
 313   if(!length) return 0;\r
 314   return length+1;\r
 315 }\r
 316 \r
 317 static int classifyFoldPointTeXPaired(const char* s) {\r
 318         int lev=0; \r
 319         if (!(isdigit(s[0]) || (s[0] == '.'))){\r
 320                 if (strcmp(s, "begin")==0||strcmp(s,"FoldStart")==0||\r
 321                         strcmp(s,"abstract")==0||strcmp(s,"unprotect")==0||\r
 322                         strcmp(s,"title")==0||strncmp(s,"start",5)==0||strncmp(s,"Start",5)==0||\r
 323                         strcmp(s,"documentclass")==0||strncmp(s,"if",2)==0\r
 324                         )\r
 325                         lev=1;\r
 326                 if (strcmp(s, "end")==0||strcmp(s,"FoldStop")==0||\r
 327                         strcmp(s,"maketitle")==0||strcmp(s,"protect")==0||\r
 328                         strncmp(s,"stop",4)==0||strncmp(s,"Stop",4)==0||\r
 329                         strcmp(s,"fi")==0\r
 330                         ) \r
 331                 lev=-1;\r
 332         }\r
 333         return lev;\r
 334 }\r
 335 \r
 336 static int classifyFoldPointTeXUnpaired(const char* s) {\r
 337         int lev=0; \r
 338         if (!(isdigit(s[0]) || (s[0] == '.'))){\r
 339                 if (strcmp(s,"part")==0||\r
 340                         strcmp(s,"chapter")==0||\r
 341                         strcmp(s,"section")==0||\r
 342                         strcmp(s,"subsection")==0||\r
 343                         strcmp(s,"subsubsection")==0||\r
 344                         strcmp(s,"CJKfamily")==0||\r
 345                         strcmp(s,"appendix")==0||\r
 346                         strcmp(s,"Topic")==0||strcmp(s,"topic")==0||\r
 347                         strcmp(s,"subject")==0||strcmp(s,"subsubject")==0||\r
 348                         strcmp(s,"def")==0||strcmp(s,"gdef")==0||strcmp(s,"edef")==0||\r
 349                         strcmp(s,"xdef")==0||strcmp(s,"framed")==0||\r
 350                         strcmp(s,"frame")==0||\r
 351                         strcmp(s,"foilhead")==0||strcmp(s,"overlays")==0||strcmp(s,"slide")==0\r
 352                         ){\r
 353                             lev=1;\r
 354                         }\r
 355         }\r
 356         return lev;\r
 357 }\r
 358 \r
 359 static bool IsTeXCommentLine(int line, Accessor &styler) {\r
 360         int pos = styler.LineStart(line);\r
 361         int eol_pos = styler.LineStart(line + 1) - 1;\r
 362         \r
 363         int startpos = pos;\r
 364 \r
 365         while (startpos<eol_pos){\r
 366                 char ch = styler[startpos];\r
 367                 if (ch!='%' && ch!=' ') return false;\r
 368                 else if (ch=='%') return true;\r
 369                 startpos++;\r
 370         }               \r
 371 \r
 372         return false;\r
 373 }\r
 374 \r
 375 // FoldTeXDoc: borrowed from VisualTeX with modifications\r
 376 \r
 377 static void FoldTexDoc(unsigned int startPos, int length, int, WordList *[], Accessor &styler) \r
 378 {\r
 379         bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;\r
 380         unsigned int endPos = startPos+length;\r
 381         int visibleChars=0;\r
 382         int lineCurrent=styler.GetLine(startPos);\r
 383         int levelPrev=styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;\r
 384         int levelCurrent=levelPrev;\r
 385         char chNext=styler[startPos];\r
 386         char buffer[100]="";\r
 387         \r
 388         for (unsigned int i=startPos; i < endPos; i++) {\r
 389                 char ch=chNext;\r
 390                 chNext=styler.SafeGetCharAt(i+1);\r
 391                 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');\r
 392 \r
 393         if(ch=='\\') {\r
 394             ParseTeXCommand(i, styler, buffer);\r
 395                         levelCurrent += classifyFoldPointTeXPaired(buffer)+classifyFoldPointTeXUnpaired(buffer);\r
 396                 }\r
 397 \r
 398                 if (levelCurrent > SC_FOLDLEVELBASE && ((ch == '\r' || ch=='\n') && (chNext == '\\'))) {\r
 399             ParseTeXCommand(i+1, styler, buffer);\r
 400                         levelCurrent -= classifyFoldPointTeXUnpaired(buffer);\r
 401                 }\r
 402 \r
 403         char chNext2;\r
 404         char chNext3;\r
 405         char chNext4;\r
 406         char chNext5;\r
 407         chNext2=styler.SafeGetCharAt(i+2);\r
 408         chNext3=styler.SafeGetCharAt(i+3);\r
 409         chNext4=styler.SafeGetCharAt(i+4);\r
 410         chNext5=styler.SafeGetCharAt(i+5);\r
 411 \r
 412         bool atEOfold = (ch == '%') && \r
 413                         (chNext == '%') && (chNext2=='}') && \r
 414                                 (chNext3=='}')&& (chNext4=='-')&& (chNext5=='-');\r
 415 \r
 416         bool atBOfold = (ch == '%') && \r
 417                         (chNext == '%') && (chNext2=='-') && \r
 418                                 (chNext3=='-')&& (chNext4=='{')&& (chNext5=='{');\r
 419 \r
 420         if(atBOfold){\r
 421                 levelCurrent+=1;\r
 422         }\r
 423 \r
 424         if(atEOfold){\r
 425                 levelCurrent-=1;\r
 426         }\r
 427         \r
 428         if(ch=='\\' && chNext=='['){\r
 429                 levelCurrent+=1;\r
 430         }\r
 431         \r
 432         if(ch=='\\' && chNext==']'){\r
 433                 levelCurrent-=1;\r
 434         }\r
 435 \r
 436         bool foldComment = styler.GetPropertyInt("fold.comment") != 0;\r
 437 \r
 438         if (foldComment && atEOL && IsTeXCommentLine(lineCurrent, styler))\r
 439         {\r
 440             if (lineCurrent==0 && IsTeXCommentLine(lineCurrent + 1, styler)\r
 441                                 )\r
 442                 levelCurrent++;\r
 443             else if (lineCurrent!=0 && !IsTeXCommentLine(lineCurrent - 1, styler)\r
 444                && IsTeXCommentLine(lineCurrent + 1, styler)\r
 445                                 )\r
 446                 levelCurrent++;\r
 447             else if (lineCurrent!=0 && IsTeXCommentLine(lineCurrent - 1, styler) &&\r
 448                      !IsTeXCommentLine(lineCurrent+1, styler))\r
 449                 levelCurrent--;\r
 450         }\r
 451 \r
 452 //--------------------------------------------------------------------------------------------- \r
 453                 \r
 454                 if (atEOL) {\r
 455                         int lev = levelPrev;\r
 456                         if (visibleChars == 0 && foldCompact)\r
 457                                 lev |= SC_FOLDLEVELWHITEFLAG;\r
 458                         if ((levelCurrent > levelPrev) && (visibleChars > 0))\r
 459                                 lev |= SC_FOLDLEVELHEADERFLAG;\r
 460                         if (lev != styler.LevelAt(lineCurrent)) {\r
 461                                 styler.SetLevel(lineCurrent, lev);\r
 462                         }\r
 463                         lineCurrent++;\r
 464                         levelPrev = levelCurrent;\r
 465                         visibleChars = 0;\r
 466                 }\r
 467 \r
 468                 if (!isspacechar(ch))\r
 469                         visibleChars++;\r
 470         }\r
 471 \r
 472         // Fill in the real level of the next line, keeping the current flags as they will be filled in later\r
 473         int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;\r
 474         styler.SetLevel(lineCurrent, levelPrev | flagsNext);\r
 475 }\r
 476 \r
 477 \r
 478 \r
 479 \r
 480 static const char * const texWordListDesc[] = {\r
 481     "TeX, eTeX, pdfTeX, Omega",\r
 482     "ConTeXt Dutch",\r
 483     "ConTeXt English",\r
 484     "ConTeXt German",\r
 485     "ConTeXt Czech",\r
 486     "ConTeXt Italian",\r
 487     "ConTeXt Romanian",\r
 488         0,\r
 489 } ;\r
 490 \r
 491 LexerModule lmTeX(SCLEX_TEX,   ColouriseTeXDoc, "tex", FoldTexDoc, texWordListDesc);\r