ext/scintilla/src/LexTeX.cxx

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