scintilla/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 #include <stdlib.h>
  16 #include <string.h>
  17 #include <ctype.h>
  18 #include <stdio.h>
  19 #include <stdarg.h>
  20
  21 #include "Platform.h"
  22
  23 #include "PropSet.h"
  24 #include "Accessor.h"
  25 #include "KeyWords.h"
  26 #include "Scintilla.h"
  27 #include "SciLexer.h"
  28 #include "StyleContext.h"
  29
  30 // val SCE_TEX_DEFAULT = 0
  31 // val SCE_TEX_SPECIAL = 1
  32 // val SCE_TEX_GROUP   = 2
  33 // val SCE_TEX_SYMBOL  = 3
  34 // val SCE_TEX_COMMAND = 4
  35 // val SCE_TEX_TEXT    = 5
  36
  37 // Definitions in SciTEGlobal.properties:
  38 //
  39 // TeX Highlighting
  40 //
  41 // # Default
  42 // style.tex.0=fore:#7F7F00
  43 // # Special
  44 // style.tex.1=fore:#007F7F
  45 // # Group
  46 // style.tex.2=fore:#880000
  47 // # Symbol
  48 // style.tex.3=fore:#7F7F00
  49 // # Command
  50 // style.tex.4=fore:#008800
  51 // # Text
  52 // style.tex.5=fore:#000000
  53
  54 // lexer.tex.interface.default=0
  55 // lexer.tex.comment.process=0
  56
  57 // todo: lexer.tex.auto.if
  58
  59 // Auxiliary functions:
  60
  61 static inline bool endOfLine(Accessor &styler, unsigned int i) {
  62         return
  63       (styler[i] == '\n') || ((styler[i] == '\r') && (styler.SafeGetCharAt(i + 1) != '\n')) ;
  64 }
  65
  66 static inline bool isTeXzero(int ch) {
  67         return
  68       (ch == '%') ;
  69 }
  70
  71 static inline bool isTeXone(int ch) {
  72         return
  73       (ch == '[') || (ch == ']') || (ch == '=') || (ch == '#') ||
  74       (ch == '(') || (ch == ')') || (ch == '<') || (ch == '>') ||
  75       (ch == '"') ;
  76 }
  77
  78 static inline bool isTeXtwo(int ch) {
  79         return
  80       (ch == '{') || (ch == '}') || (ch == '$') ;
  81 }
  82
  83 static inline bool isTeXthree(int ch) {
  84         return
  85       (ch == '~') || (ch == '^') || (ch == '_') || (ch == '&') ||
  86       (ch == '-') || (ch == '+') || (ch == '\"') || (ch == '`') ||
  87       (ch == '/') || (ch == '|') || (ch == '%') ;
  88 }
  89
  90 static inline bool isTeXfour(int ch) {
  91         return
  92       (ch == '\\') ;
  93 }
  94
  95 static inline bool isTeXfive(int ch) {
  96         return
  97       ((ch >= 'a') && (ch <= 'z')) || ((ch >= 'A') && (ch <= 'Z')) ||
  98       (ch == '@') || (ch == '!') || (ch == '?') ;
  99 }
 100
 101 static inline bool isTeXsix(int ch) {
 102         return
 103       (ch == ' ') ;
 104 }
 105
 106 static inline bool isTeXseven(int ch) {
 107         return
 108       (ch == '^') ;
 109 }
 110
 111 // Interface determination
 112
 113 static int CheckTeXInterface(
 114     unsigned int startPos,
 115     int length,
 116     Accessor &styler,
 117         int defaultInterface) {
 118
 119     char lineBuffer[1024] ;
 120         unsigned int linePos = 0 ;
 121
 122     // some day we can make something lexer.tex.mapping=(all,0)(nl,1)(en,2)...
 123
 124     if (styler.SafeGetCharAt(0) == '%') {
 125         for (unsigned int i = 0; i < startPos + length; i++) {
 126             lineBuffer[linePos++] = styler.SafeGetCharAt(i) ;
 127             if (endOfLine(styler, i) || (linePos >= sizeof(lineBuffer) - 1)) {
 128                 lineBuffer[linePos] = '\0';
 129                 if (strstr(lineBuffer, "interface=all")) {
 130                     return 0 ;
 131                                 } else if (strstr(lineBuffer, "interface=tex")) {
 132                     return 1 ;
 133                 } else if (strstr(lineBuffer, "interface=nl")) {
 134                     return 2 ;
 135                 } else if (strstr(lineBuffer, "interface=en")) {
 136                     return 3 ;
 137                 } else if (strstr(lineBuffer, "interface=de")) {
 138                     return 4 ;
 139                 } else if (strstr(lineBuffer, "interface=cz")) {
 140                     return 5 ;
 141                 } else if (strstr(lineBuffer, "interface=it")) {
 142                     return 6 ;
 143                 } else if (strstr(lineBuffer, "interface=ro")) {
 144                     return 7 ;
 145                 } else if (strstr(lineBuffer, "interface=latex")) {
 146                                         // we will move latex cum suis up to 91+ when more keyword lists are supported
 147                     return 8 ;
 148                                 } else if (styler.SafeGetCharAt(1) == 'D' && strstr(lineBuffer, "%D \\module")) {
 149                                         // better would be to limit the search to just one line
 150                                         return 3 ;
 151                 } else {
 152                     return defaultInterface ;
 153                 }
 154             }
 155                 }
 156     }
 157
 158     return defaultInterface ;
 159 }
 160
 161 static void ColouriseTeXDoc(
 162     unsigned int startPos,
 163     int length,
 164     int,
 165     WordList *keywordlists[],
 166     Accessor &styler) {
 167
 168         styler.StartAt(startPos) ;
 169         styler.StartSegment(startPos) ;
 170
 171         bool processComment   = styler.GetPropertyInt("lexer.tex.comment.process",   0) == 1 ;
 172     bool useKeywords      = styler.GetPropertyInt("lexer.tex.use.keywords",      1) == 1 ;
 173         bool autoIf           = styler.GetPropertyInt("lexer.tex.auto.if",           1) == 1 ;
 174     int  defaultInterface = styler.GetPropertyInt("lexer.tex.interface.default", 1) ;
 175
 176         char key[100] ;
 177         int  k ;
 178         bool newifDone = false ;
 179         bool inComment = false ;
 180
 181         int currentInterface = CheckTeXInterface(startPos,length,styler,defaultInterface) ;
 182
 183     if (currentInterface == 0) {
 184         useKeywords = false ;
 185         currentInterface = 1 ;
 186     }
 187
 188     WordList &keywords = *keywordlists[currentInterface-1] ;
 189
 190         StyleContext sc(startPos, length, SCE_TEX_TEXT, styler);
 191
 192         bool going = sc.More() ; // needed because of a fuzzy end of file state
 193
 194         for (; going; sc.Forward()) {
 195
 196                 if (! sc.More()) { going = false ; } // we need to go one behind the end of text
 197
 198                 if (inComment) {
 199                         if (sc.atLineEnd) {
 200                                 sc.SetState(SCE_TEX_TEXT) ;
 201                                 newifDone = false ;
 202                                 inComment = false ;
 203                         }
 204                 } else {
 205                         if (! isTeXfive(sc.ch)) {
 206                                 if (sc.state == SCE_TEX_COMMAND) {
 207                                         if (sc.LengthCurrent() == 1) { // \<noncstoken>
 208                                                 if (isTeXseven(sc.ch) && isTeXseven(sc.chNext)) {
 209                                                         sc.Forward(2) ; // \^^ and \^^<token>
 210                                                 }
 211                                                 sc.ForwardSetState(SCE_TEX_TEXT) ;
 212                                         } else {
 213                                                 sc.GetCurrent(key, sizeof(key)-1) ;
 214                                                 k = strlen(key) ;
 215                                                 memmove(key,key+1,k) ; // shift left over escape token
 216                                                 key[k] = '\0' ;
 217                                                 k-- ;
 218                                                 if (! keywords || ! useKeywords) {
 219                                                         sc.SetState(SCE_TEX_COMMAND) ;
 220                                                         newifDone = false ;
 221                                                 } else if (k == 1) { //\<cstoken>
 222                                                         sc.SetState(SCE_TEX_COMMAND) ;
 223                                                         newifDone = false ;
 224                                                 } else if (keywords.InList(key)) {
 225                                                 sc.SetState(SCE_TEX_COMMAND) ;
 226                                                         newifDone = autoIf && (strcmp(key,"newif") == 0) ;
 227                                                 } else if (autoIf && ! newifDone && (key[0] == 'i') && (key[1] == 'f') && keywords.InList("if")) {
 228                                                 sc.SetState(SCE_TEX_COMMAND) ;
 229                                                 } else {
 230                                                         sc.ChangeState(SCE_TEX_TEXT) ;
 231                                                         sc.SetState(SCE_TEX_TEXT) ;
 232                                                         newifDone = false ;
 233                                                 }
 234                                         }
 235                                 }
 236                                 if (isTeXzero(sc.ch)) {
 237                                         sc.SetState(SCE_TEX_SYMBOL) ;
 238                                         sc.ForwardSetState(SCE_TEX_DEFAULT) ;
 239                                         inComment = ! processComment ;
 240                                         newifDone = false ;
 241                                 } else if (isTeXseven(sc.ch) && isTeXseven(sc.chNext)) {
 242                                         sc.SetState(SCE_TEX_TEXT) ;
 243                                         sc.ForwardSetState(SCE_TEX_TEXT) ;
 244                                 } else if (isTeXone(sc.ch)) {
 245                                         sc.SetState(SCE_TEX_SPECIAL) ;
 246                                         newifDone = false ;
 247                                 } else if (isTeXtwo(sc.ch)) {
 248                                         sc.SetState(SCE_TEX_GROUP) ;
 249                                         newifDone = false ;
 250                                 } else if (isTeXthree(sc.ch)) {
 251                                         sc.SetState(SCE_TEX_SYMBOL) ;
 252                                         newifDone = false ;
 253                                 } else if (isTeXfour(sc.ch)) {
 254                                         sc.SetState(SCE_TEX_COMMAND) ;
 255                                 } else if (isTeXsix(sc.ch)) {
 256                                         sc.SetState(SCE_TEX_TEXT) ;
 257                                 } else if (sc.atLineEnd) {
 258                                         sc.SetState(SCE_TEX_TEXT) ;
 259                                         newifDone = false ;
 260                                         inComment = false ;
 261                                 } else {
 262                                         sc.SetState(SCE_TEX_TEXT) ;
 263                                 }
 264                         } else if (sc.state != SCE_TEX_COMMAND) {
 265                                 sc.SetState(SCE_TEX_TEXT) ;
 266                         }
 267                 }
 268         }
 269         sc.ChangeState(SCE_TEX_TEXT) ;
 270         sc.Complete();
 271
 272 }
 273
 274
 275 // Hooks into the system:
 276
 277 static const char * const texWordListDesc[] = {
 278     "TeX, eTeX, pdfTeX, Omega"
 279     "ConTeXt Dutch",
 280     "ConTeXt English",
 281     "ConTeXt German",
 282     "ConTeXt Czech",
 283     "ConTeXt Italian",
 284     "ConTeXt Romanian",
 285         0,
 286 } ;
 287
 288 LexerModule lmTeX(SCLEX_TEX, ColouriseTeXDoc, "tex", 0, texWordListDesc);