ext/scintilla/lexers/LexTAL.cxx

   1 // Scintilla source code edit control
   2 /** @file LexTAL.cxx
   3  ** Lexer for TAL
   4  ** Based on LexPascal.cxx
   5  ** Written by Laurent le Tynevez
   6  ** Updated by Simon Steele <s.steele@pnotepad.org> September 2002
   7  ** Updated by Mathias Rauen <scite@madshi.net> May 2003 (Delphi adjustments)
   8  ** Updated by Rod Falck, Aug 2006 Converted to TAL
   9  **/
  10
  11 #include <stdlib.h>
  12 #include <string.h>
  13 #include <stdio.h>
  14 #include <stdarg.h>
  15 #include <assert.h>
  16 #include <ctype.h>
  17
  18 #include "ILexer.h"
  19 #include "Scintilla.h"
  20 #include "SciLexer.h"
  21
  22 #include "WordList.h"
  23 #include "LexAccessor.h"
  24 #include "Accessor.h"
  25 #include "StyleContext.h"
  26 #include "CharacterSet.h"
  27 #include "LexerModule.h"
  28
  29 #ifdef SCI_NAMESPACE
  30 using namespace Scintilla;
  31 #endif
  32
  33 inline bool isTALoperator(char ch)
  34         {
  35         return ch == '\'' || ch == '@' || ch == '#' || isoperator(ch);
  36         }
  37
  38 inline bool isTALwordchar(char ch)
  39         {
  40         return ch == '$' || ch == '^' || iswordchar(ch);
  41         }
  42
  43 inline bool isTALwordstart(char ch)
  44         {
  45         return ch == '$' || ch == '^' || iswordstart(ch);
  46         }
  47
  48 static void getRange(unsigned int start,
  49                 unsigned int end,
  50                 Accessor &styler,
  51                 char *s,
  52                 unsigned int len) {
  53         unsigned int i = 0;
  54         while ((i < end - start + 1) && (i < len-1)) {
  55                 s[i] = static_cast<char>(tolower(styler[start + i]));
  56                 i++;
  57         }
  58         s[i] = '\0';
  59 }
  60
  61 static bool IsStreamCommentStyle(int style) {
  62         return style == SCE_C_COMMENT ||
  63                 style == SCE_C_COMMENTDOC ||
  64                 style == SCE_C_COMMENTDOCKEYWORD ||
  65                 style == SCE_C_COMMENTDOCKEYWORDERROR;
  66 }
  67
  68 static void ColourTo(Accessor &styler, unsigned int end, unsigned int attr, bool bInAsm) {
  69         if ((bInAsm) && (attr == SCE_C_OPERATOR || attr == SCE_C_NUMBER || attr == SCE_C_DEFAULT || attr == SCE_C_WORD || attr == SCE_C_IDENTIFIER)) {
  70                 styler.ColourTo(end, SCE_C_REGEX);
  71         } else
  72                 styler.ColourTo(end, attr);
  73 }
  74
  75 // returns 1 if the item starts a class definition, and -1 if the word is "end", and 2 if the word is "asm"
  76 static int classifyWordTAL(unsigned int start, unsigned int end, /*WordList &keywords*/WordList *keywordlists[], Accessor &styler, bool bInAsm) {
  77         int ret = 0;
  78
  79         WordList& keywords = *keywordlists[0];
  80         WordList& builtins = *keywordlists[1];
  81         WordList& nonreserved_keywords = *keywordlists[2];
  82
  83         char s[100];
  84         getRange(start, end, styler, s, sizeof(s));
  85
  86         char chAttr = SCE_C_IDENTIFIER;
  87         if (isdigit(s[0]) || (s[0] == '.')) {
  88                 chAttr = SCE_C_NUMBER;
  89         }
  90         else {
  91                 if (keywords.InList(s)) {
  92                         chAttr = SCE_C_WORD;
  93
  94                         if (strcmp(s, "asm") == 0) {
  95                                 ret = 2;
  96                         }
  97                         else if (strcmp(s, "end") == 0) {
  98                                 ret = -1;
  99                         }
 100                 }
 101                 else if (s[0] == '$' || builtins.InList(s)) {
 102                         chAttr = SCE_C_WORD2;
 103                 }
 104                 else if (nonreserved_keywords.InList(s)) {
 105                         chAttr = SCE_C_UUID;
 106                 }
 107         }
 108         ColourTo(styler, end, chAttr, (bInAsm && ret != -1));
 109         return ret;
 110 }
 111
 112 static int classifyFoldPointTAL(const char* s) {
 113         int lev = 0;
 114         if (!(isdigit(s[0]) || (s[0] == '.'))) {
 115                 if (strcmp(s, "begin") == 0 ||
 116                         strcmp(s, "block") == 0) {
 117                         lev=1;
 118                 } else if (strcmp(s, "end") == 0) {
 119                         lev=-1;
 120                 }
 121         }
 122         return lev;
 123 }
 124
 125 static void ColouriseTALDoc(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],
 126         Accessor &styler) {
 127
 128         styler.StartAt(startPos);
 129
 130         int state = initStyle;
 131         if (state == SCE_C_CHARACTER)   // Does not leak onto next line
 132                 state = SCE_C_DEFAULT;
 133         char chPrev = ' ';
 134         char chNext = styler[startPos];
 135         unsigned int lengthDoc = startPos + length;
 136
 137         bool bInClassDefinition;
 138
 139         int currentLine = styler.GetLine(startPos);
 140         if (currentLine > 0) {
 141                 styler.SetLineState(currentLine, styler.GetLineState(currentLine-1));
 142                 bInClassDefinition = (styler.GetLineState(currentLine) == 1);
 143         } else {
 144                 styler.SetLineState(currentLine, 0);
 145                 bInClassDefinition = false;
 146         }
 147
 148         bool bInAsm = (state == SCE_C_REGEX);
 149         if (bInAsm)
 150                 state = SCE_C_DEFAULT;
 151
 152         styler.StartSegment(startPos);
 153         int visibleChars = 0;
 154         for (unsigned int i = startPos; i < lengthDoc; i++) {
 155                 char ch = chNext;
 156
 157                 chNext = styler.SafeGetCharAt(i + 1);
 158
 159                 if ((ch == '\r' && chNext != '\n') || (ch == '\n')) {
 160                         // Trigger on CR only (Mac style) or either on LF from CR+LF (Dos/Win) or on LF alone (Unix)
 161                         // Avoid triggering two times on Dos/Win
 162                         // End of line
 163                         if (state == SCE_C_CHARACTER) {
 164                                 ColourTo(styler, i, state, bInAsm);
 165                                 state = SCE_C_DEFAULT;
 166                         }
 167                         visibleChars = 0;
 168                         currentLine++;
 169                         styler.SetLineState(currentLine, (bInClassDefinition ? 1 : 0));
 170                 }
 171
 172                 if (styler.IsLeadByte(ch)) {
 173                         chNext = styler.SafeGetCharAt(i + 2);
 174                         chPrev = ' ';
 175                         i += 1;
 176                         continue;
 177                 }
 178
 179                 if (state == SCE_C_DEFAULT) {
 180                         if (isTALwordstart(ch)) {
 181                                 ColourTo(styler, i-1, state, bInAsm);
 182                                 state = SCE_C_IDENTIFIER;
 183                         } else if (ch == '!' && chNext != '*') {
 184                                 ColourTo(styler, i-1, state, bInAsm);
 185                                 state = SCE_C_COMMENT;
 186                         } else if (ch == '!' && chNext == '*') {
 187                                 ColourTo(styler, i-1, state, bInAsm);
 188                                 state = SCE_C_COMMENTDOC;
 189                         } else if (ch == '-' && chNext == '-') {
 190                                 ColourTo(styler, i-1, state, bInAsm);
 191                                 state = SCE_C_COMMENTLINE;
 192                         } else if (ch == '"') {
 193                                 ColourTo(styler, i-1, state, bInAsm);
 194                                 state = SCE_C_STRING;
 195                         } else if (ch == '?' && visibleChars == 0) {
 196                                 ColourTo(styler, i-1, state, bInAsm);
 197                                 state = SCE_C_PREPROCESSOR;
 198                         } else if (isTALoperator(ch)) {
 199                                 ColourTo(styler, i-1, state, bInAsm);
 200                                 ColourTo(styler, i, SCE_C_OPERATOR, bInAsm);
 201                         }
 202                 } else if (state == SCE_C_IDENTIFIER) {
 203                         if (!isTALwordchar(ch)) {
 204                                 int lStateChange = classifyWordTAL(styler.GetStartSegment(), i - 1, keywordlists, styler, bInAsm);
 205
 206                                 if(lStateChange == 1) {
 207                                         styler.SetLineState(currentLine, 1);
 208                                         bInClassDefinition = true;
 209                                 } else if(lStateChange == 2) {
 210                                         bInAsm = true;
 211                                 } else if(lStateChange == -1) {
 212                                         styler.SetLineState(currentLine, 0);
 213                                         bInClassDefinition = false;
 214                                         bInAsm = false;
 215                                 }
 216
 217                                 state = SCE_C_DEFAULT;
 218                                 chNext = styler.SafeGetCharAt(i + 1);
 219                                 if (ch == '!' && chNext != '*') {
 220                                         state = SCE_C_COMMENT;
 221                                 } else if (ch == '!' && chNext == '*') {
 222                                         ColourTo(styler, i-1, state, bInAsm);
 223                                         state = SCE_C_COMMENTDOC;
 224                                 } else if (ch == '-' && chNext == '-') {
 225                                         state = SCE_C_COMMENTLINE;
 226                                 } else if (ch == '"') {
 227                                         state = SCE_C_STRING;
 228                                 } else if (isTALoperator(ch)) {
 229                                         ColourTo(styler, i, SCE_C_OPERATOR, bInAsm);
 230                                 }
 231                         }
 232                 } else {
 233                         if (state == SCE_C_PREPROCESSOR) {
 234                                 if ((ch == '\r' || ch == '\n') && !(chPrev == '\\' || chPrev == '\r')) {
 235                                         ColourTo(styler, i-1, state, bInAsm);
 236                                         state = SCE_C_DEFAULT;
 237                                 }
 238                         } else if (state == SCE_C_COMMENT) {
 239                                 if (ch == '!' || (ch == '\r' || ch == '\n') ) {
 240                                         ColourTo(styler, i, state, bInAsm);
 241                                         state = SCE_C_DEFAULT;
 242                                 }
 243                         } else if (state == SCE_C_COMMENTDOC) {
 244                                 if (ch == '!' || (ch == '\r' || ch == '\n')) {
 245                                         if (((i > styler.GetStartSegment() + 2) || (
 246                                                 (initStyle == SCE_C_COMMENTDOC) &&
 247                                                 (styler.GetStartSegment() == static_cast<unsigned int>(startPos))))) {
 248                                                         ColourTo(styler, i, state, bInAsm);
 249                                                         state = SCE_C_DEFAULT;
 250                                         }
 251                                 }
 252                         } else if (state == SCE_C_COMMENTLINE) {
 253                                 if (ch == '\r' || ch == '\n') {
 254                                         ColourTo(styler, i-1, state, bInAsm);
 255                                         state = SCE_C_DEFAULT;
 256                                 }
 257                         } else if (state == SCE_C_STRING) {
 258                                 if (ch == '"') {
 259                                         ColourTo(styler, i, state, bInAsm);
 260                                         state = SCE_C_DEFAULT;
 261                                 }
 262                         }
 263                 }
 264         if (!isspacechar(ch))
 265             visibleChars++;
 266                 chPrev = ch;
 267         }
 268         ColourTo(styler, lengthDoc - 1, state, bInAsm);
 269 }
 270
 271 static void FoldTALDoc(unsigned int startPos, int length, int initStyle, WordList *[],
 272                             Accessor &styler) {
 273         bool foldComment = styler.GetPropertyInt("fold.comment") != 0;
 274         bool foldPreprocessor = styler.GetPropertyInt("fold.preprocessor") != 0;
 275         bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;
 276         unsigned int endPos = startPos + length;
 277         int visibleChars = 0;
 278         int lineCurrent = styler.GetLine(startPos);
 279         int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
 280         int levelCurrent = levelPrev;
 281         char chNext = styler[startPos];
 282         int styleNext = styler.StyleAt(startPos);
 283         int style = initStyle;
 284         bool was_end = false;
 285         bool section = false;
 286
 287         int lastStart = 0;
 288
 289         for (unsigned int i = startPos; i < endPos; i++) {
 290                 char ch = chNext;
 291                 chNext = styler.SafeGetCharAt(i + 1);
 292                 int stylePrev = style;
 293                 style = styleNext;
 294                 styleNext = styler.StyleAt(i + 1);
 295                 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
 296
 297                 if (stylePrev == SCE_C_DEFAULT && (style == SCE_C_WORD || style == SCE_C_UUID || style == SCE_C_PREPROCESSOR))
 298                 {
 299                         // Store last word start point.
 300                         lastStart = i;
 301                 }
 302
 303                 if (stylePrev == SCE_C_WORD || style == SCE_C_UUID || stylePrev == SCE_C_PREPROCESSOR) {
 304                         if(isTALwordchar(ch) && !isTALwordchar(chNext)) {
 305                                 char s[100];
 306                                 getRange(lastStart, i, styler, s, sizeof(s));
 307                                 if (stylePrev == SCE_C_PREPROCESSOR && strcmp(s, "?section") == 0)
 308                                         {
 309                                         section = true;
 310                                         levelCurrent = 1;
 311                                         levelPrev = 0;
 312                                         }
 313                                 else if (stylePrev == SCE_C_WORD || stylePrev == SCE_C_UUID)
 314                                         {
 315                                         if (strcmp(s, "block") == 0)
 316                                                 {
 317                                                 // block keyword is ignored immediately after end keyword
 318                                                 if (!was_end)
 319                                                         levelCurrent++;
 320                                                 }
 321                                         else
 322                                                 levelCurrent += classifyFoldPointTAL(s);
 323                                         if (strcmp(s, "end") == 0)
 324                                                 {
 325                                                 was_end = true;
 326                                                 }
 327                                         else
 328                                                 {
 329                                                 was_end = false;
 330                                                 }
 331                                         }
 332                         }
 333                 }
 334
 335                 if (foldComment && (style == SCE_C_COMMENTLINE)) {
 336                         if ((ch == '/') && (chNext == '/')) {
 337                                 char chNext2 = styler.SafeGetCharAt(i + 2);
 338                                 if (chNext2 == '{') {
 339                                         levelCurrent++;
 340                                 } else if (chNext2 == '}') {
 341                                         levelCurrent--;
 342                                 }
 343                         }
 344                 }
 345
 346                 if (foldPreprocessor && (style == SCE_C_PREPROCESSOR)) {
 347                         if (ch == '{' && chNext == '$') {
 348                                 unsigned int j=i+2; // skip {$
 349                                 while ((j<endPos) && IsASpaceOrTab(styler.SafeGetCharAt(j))) {
 350                                         j++;
 351                                 }
 352                                 if (styler.Match(j, "region") || styler.Match(j, "if")) {
 353                                         levelCurrent++;
 354                                 } else if (styler.Match(j, "end")) {
 355                                         levelCurrent--;
 356                                 }
 357                         }
 358                 }
 359
 360                 if (foldComment && IsStreamCommentStyle(style)) {
 361                         if (!IsStreamCommentStyle(stylePrev)) {
 362                                 levelCurrent++;
 363                         } else if (!IsStreamCommentStyle(styleNext) && !atEOL) {
 364                                 // Comments don't end at end of line and the next character may be unstyled.
 365                                 levelCurrent--;
 366                         }
 367                 }
 368
 369                 if (atEOL) {
 370                         int lev = levelPrev | SC_FOLDLEVELBASE;
 371                         if (visibleChars == 0 && foldCompact)
 372                                 lev |= SC_FOLDLEVELWHITEFLAG;
 373                         if ((levelCurrent > levelPrev || section) && (visibleChars > 0))
 374                                 lev |= SC_FOLDLEVELHEADERFLAG;
 375                         if (lev != styler.LevelAt(lineCurrent)) {
 376                                 styler.SetLevel(lineCurrent, lev);
 377                         }
 378                         lineCurrent++;
 379                         levelPrev = levelCurrent;
 380                         visibleChars = 0;
 381                         section = false;
 382                 }
 383
 384                 if (!isspacechar(ch))
 385                         visibleChars++;
 386         }
 387
 388         // Fill in the real level of the next line, keeping the current flags as they will be filled in later
 389         int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;
 390         styler.SetLevel(lineCurrent, levelPrev | flagsNext);
 391 }
 392
 393 static const char * const TALWordListDesc[] = {
 394         "Keywords",
 395         "Builtins",
 396         0
 397 };
 398
 399 LexerModule lmTAL(SCLEX_TAL, ColouriseTALDoc, "TAL", FoldTALDoc, TALWordListDesc);