ext/scintilla/lexers/LexVisualProlog.cxx

   1 // Scintilla source code edit control
   2 /** @file LexVisualProlog.cxx
   3 ** Lexer for Visual Prolog.
   4 **/
   5 // Author Thomas Linder Puls, Prolog Development Denter A/S, http://www.visual-prolog.com
   6 // Based on Lexer for C++, C, Java, and JavaScript.
   7 // Copyright 1998-2005 by Neil Hodgson <neilh@scintilla.org>
   8 // The License.txt file describes the conditions under which this software may be distributed.
   9
  10 #include <stdlib.h>
  11 #include <string.h>
  12 #include <ctype.h>
  13 #include <stdio.h>
  14 #include <stdarg.h>
  15 #include <assert.h>
  16
  17 #ifdef _MSC_VER
  18 #pragma warning(disable: 4786)
  19 #endif
  20
  21 #include <string>
  22 #include <vector>
  23 #include <map>
  24 #include <algorithm>
  25
  26 #include "ILexer.h"
  27 #include "Scintilla.h"
  28 #include "SciLexer.h"
  29
  30 #include "WordList.h"
  31 #include "LexAccessor.h"
  32 #include "Accessor.h"
  33 #include "StyleContext.h"
  34 #include "CharacterSet.h"
  35 #include "LexerModule.h"
  36 #include "OptionSet.h"
  37
  38 #ifdef SCI_NAMESPACE
  39 using namespace Scintilla;
  40 #endif
  41
  42 // Options used for LexerVisualProlog
  43 struct OptionsVisualProlog {
  44     OptionsVisualProlog() {
  45     }
  46 };
  47
  48 static const char *const visualPrologWordLists[] = {
  49     "Major keywords (class, predicates, ...)",
  50     "Minor keywords (if, then, try, ...)",
  51     "Directive keywords without the '#' (include, requires, ...)",
  52     "Documentation keywords without the '@' (short, detail, ...)",
  53     0,
  54 };
  55
  56 struct OptionSetVisualProlog : public OptionSet<OptionsVisualProlog> {
  57     OptionSetVisualProlog() {
  58         DefineWordListSets(visualPrologWordLists);
  59     }
  60 };
  61
  62 class LexerVisualProlog : public ILexer {
  63     WordList majorKeywords;
  64     WordList minorKeywords;
  65     WordList directiveKeywords;
  66     WordList docKeywords;
  67     OptionsVisualProlog options;
  68     OptionSetVisualProlog osVisualProlog;
  69 public:
  70     LexerVisualProlog() {
  71     }
  72     virtual ~LexerVisualProlog() {
  73     }
  74     void SCI_METHOD Release() {
  75         delete this;
  76     }
  77     int SCI_METHOD Version() const {
  78         return lvOriginal;
  79     }
  80     const char * SCI_METHOD PropertyNames() {
  81         return osVisualProlog.PropertyNames();
  82     }
  83     int SCI_METHOD PropertyType(const char *name) {
  84         return osVisualProlog.PropertyType(name);
  85     }
  86     const char * SCI_METHOD DescribeProperty(const char *name) {
  87         return osVisualProlog.DescribeProperty(name);
  88     }
  89     int SCI_METHOD PropertySet(const char *key, const char *val);
  90     const char * SCI_METHOD DescribeWordListSets() {
  91         return osVisualProlog.DescribeWordListSets();
  92     }
  93     int SCI_METHOD WordListSet(int n, const char *wl);
  94     void SCI_METHOD Lex(unsigned int startPos, int length, int initStyle, IDocument *pAccess);
  95     void SCI_METHOD Fold(unsigned int startPos, int length, int initStyle, IDocument *pAccess);
  96
  97     void * SCI_METHOD PrivateCall(int, void *) {
  98         return 0;
  99     }
 100
 101     static ILexer *LexerFactoryVisualProlog() {
 102         return new LexerVisualProlog();
 103     }
 104 };
 105
 106 int SCI_METHOD LexerVisualProlog::PropertySet(const char *key, const char *val) {
 107     if (osVisualProlog.PropertySet(&options, key, val)) {
 108         return 0;
 109     }
 110     return -1;
 111 }
 112
 113 int SCI_METHOD LexerVisualProlog::WordListSet(int n, const char *wl) {
 114     WordList *wordListN = 0;
 115     switch (n) {
 116     case 0:
 117         wordListN = &majorKeywords;
 118         break;
 119     case 1:
 120         wordListN = &minorKeywords;
 121         break;
 122     case 2:
 123         wordListN = &directiveKeywords;
 124         break;
 125     case 3:
 126         wordListN = &docKeywords;
 127         break;
 128     }
 129     int firstModification = -1;
 130     if (wordListN) {
 131         WordList wlNew;
 132         wlNew.Set(wl);
 133         if (*wordListN != wlNew) {
 134             wordListN->Set(wl);
 135             firstModification = 0;
 136         }
 137     }
 138     return firstModification;
 139 }
 140
 141 // Functor used to truncate history
 142 struct After {
 143     int line;
 144     After(int line_) : line(line_) {}
 145 };
 146
 147 // Look ahead to see which colour "end" should have (takes colour after the following keyword)
 148 static void endLookAhead(char s[], LexAccessor &styler, int start, CharacterSet &setIdentifier) {
 149     char ch = styler.SafeGetCharAt(start, '\n');
 150     while (' ' == ch) {
 151         start++;
 152         ch = styler.SafeGetCharAt(start, '\n');
 153     }
 154     int i = 0;
 155     while (i < 100 && setIdentifier.Contains(ch)){
 156         s[i] = ch;
 157         i++;
 158         ch = styler.SafeGetCharAt(start + i, '\n');
 159     }
 160     s[i] = '\0';
 161 }
 162
 163 static void forwardEscapeLiteral(StyleContext &sc, int OwnChar, int EscapeState) {
 164     sc.Forward();
 165     if (sc.ch == OwnChar || sc.ch == '\\' || sc.ch == 'n' || sc.ch == 'l' || sc.ch == 'r' || sc.ch == 't') {
 166         sc.ChangeState(EscapeState);
 167     } else if (sc.ch == 'u') {
 168         if (IsADigit(sc.chNext, 16)) {
 169             sc.Forward();
 170             if (IsADigit(sc.chNext, 16)) {
 171                 sc.Forward();
 172                 if (IsADigit(sc.chNext, 16)) {
 173                     sc.Forward();
 174                     if (IsADigit(sc.chNext, 16)) {
 175                         sc.Forward();
 176                         sc.ChangeState(EscapeState);
 177                     }
 178                 }
 179             }
 180         }
 181     }
 182 }
 183
 184 void SCI_METHOD LexerVisualProlog::Lex(unsigned int startPos, int length, int initStyle, IDocument *pAccess) {
 185     LexAccessor styler(pAccess);
 186
 187     CharacterSet setDoxygen(CharacterSet::setAlpha, "$@\\&<>#{}[]");
 188
 189     CharacterSet setLowerStart(CharacterSet::setLower);
 190     CharacterSet setVariableStart(CharacterSet::setUpper);
 191     CharacterSet setIdentifier(CharacterSet::setAlphaNum, "_", 0x80, true);
 192
 193     int styleBeforeDocKeyword = SCE_VISUALPROLOG_DEFAULT;
 194
 195     int currentLine = styler.GetLine(startPos);
 196
 197     int nestLevel = 0;
 198     if (currentLine >= 1)
 199     {
 200         nestLevel = styler.GetLineState(currentLine - 1);
 201     }
 202
 203     StyleContext sc(startPos, length, initStyle, styler, 0x7f);
 204
 205     // Truncate ppDefineHistory before current line
 206
 207     for (; sc.More(); sc.Forward()) {
 208
 209         if (sc.atLineEnd) {
 210             // Update the line state, so it can be seen by next line
 211             styler.SetLineState(currentLine, nestLevel);
 212             currentLine++;
 213         }
 214
 215         if (sc.atLineStart) {
 216             if ((sc.state == SCE_VISUALPROLOG_STRING) || (sc.state == SCE_VISUALPROLOG_CHARACTER)) {
 217                 // Prevent SCE_VISUALPROLOG_STRING_EOL from leaking back to previous line which
 218                 // ends with a line continuation by locking in the state upto this position.
 219                 sc.SetState(sc.state);
 220             }
 221         }
 222
 223         const bool atLineEndBeforeSwitch = sc.atLineEnd;
 224
 225         // Determine if the current state should terminate.
 226         switch (sc.state) {
 227         case SCE_VISUALPROLOG_OPERATOR:
 228             sc.SetState(SCE_VISUALPROLOG_DEFAULT);
 229             break;
 230         case SCE_VISUALPROLOG_NUMBER:
 231             // We accept almost anything because of hex. and number suffixes
 232             if (!(setIdentifier.Contains(sc.ch) || (sc.ch == '.') || ((sc.ch == '+' || sc.ch == '-') && (sc.chPrev == 'e' || sc.chPrev == 'E')))) {
 233                 sc.SetState(SCE_VISUALPROLOG_DEFAULT);
 234             }
 235             break;
 236         case SCE_VISUALPROLOG_IDENTIFIER:
 237             if (!setIdentifier.Contains(sc.ch)) {
 238                 char s[1000];
 239                 sc.GetCurrent(s, sizeof(s));
 240                 if (0 == strcmp(s, "end")) {
 241                     endLookAhead(s, styler, sc.currentPos, setIdentifier);
 242                 }
 243                 if (majorKeywords.InList(s)) {
 244                     sc.ChangeState(SCE_VISUALPROLOG_KEY_MAJOR);
 245                 } else if (minorKeywords.InList(s)) {
 246                     sc.ChangeState(SCE_VISUALPROLOG_KEY_MINOR);
 247                 }
 248                 sc.SetState(SCE_VISUALPROLOG_DEFAULT);
 249             }
 250             break;
 251         case SCE_VISUALPROLOG_VARIABLE:
 252         case SCE_VISUALPROLOG_ANONYMOUS:
 253             if (!setIdentifier.Contains(sc.ch)) {
 254                 sc.SetState(SCE_VISUALPROLOG_DEFAULT);
 255             }
 256             break;
 257         case SCE_VISUALPROLOG_KEY_DIRECTIVE:
 258             if (!setLowerStart.Contains(sc.ch)) {
 259                 char s[1000];
 260                 sc.GetCurrent(s, sizeof(s));
 261                 if (!directiveKeywords.InList(s+1)) {
 262                     sc.ChangeState(SCE_VISUALPROLOG_IDENTIFIER);
 263                 }
 264                 sc.SetState(SCE_VISUALPROLOG_DEFAULT);
 265             }
 266             break;
 267         case SCE_VISUALPROLOG_COMMENT_BLOCK:
 268             if (sc.Match('*', '/')) {
 269                 sc.Forward();
 270                 nestLevel--;
 271                 int nextState = (nestLevel == 0) ? SCE_VISUALPROLOG_DEFAULT : SCE_VISUALPROLOG_COMMENT_BLOCK;
 272                 sc.ForwardSetState(nextState);
 273             } else if (sc.Match('/', '*')) {
 274                 sc.Forward();
 275                 nestLevel++;
 276             } else if (sc.ch == '%') {
 277                 sc.SetState(SCE_VISUALPROLOG_COMMENT_LINE);
 278             } else if (sc.ch == '@') {
 279                 styleBeforeDocKeyword = sc.state;
 280                 sc.SetState(SCE_VISUALPROLOG_COMMENT_KEY_ERROR);
 281             }
 282             break;
 283         case SCE_VISUALPROLOG_COMMENT_LINE:
 284             if (sc.atLineEnd) {
 285                 int nextState = (nestLevel == 0) ? SCE_VISUALPROLOG_DEFAULT : SCE_VISUALPROLOG_COMMENT_BLOCK;
 286                 sc.SetState(nextState);
 287             } else if (sc.ch == '@') {
 288                 styleBeforeDocKeyword = sc.state;
 289                 sc.SetState(SCE_VISUALPROLOG_COMMENT_KEY_ERROR);
 290             }
 291             break;
 292         case SCE_VISUALPROLOG_COMMENT_KEY_ERROR:
 293             if (!setDoxygen.Contains(sc.ch)) {
 294                 char s[1000];
 295                 sc.GetCurrent(s, sizeof(s));
 296                 if (docKeywords.InList(s+1)) {
 297                     sc.ChangeState(SCE_VISUALPROLOG_COMMENT_KEY);
 298                 }
 299                 sc.SetState(styleBeforeDocKeyword);
 300             }
 301             if (SCE_VISUALPROLOG_COMMENT_LINE == styleBeforeDocKeyword && sc.atLineStart) {
 302                 sc.SetState(SCE_VISUALPROLOG_DEFAULT);
 303             } else if (SCE_VISUALPROLOG_COMMENT_BLOCK == styleBeforeDocKeyword && sc.atLineStart) {
 304                 sc.SetState(SCE_VISUALPROLOG_COMMENT_BLOCK);
 305             }
 306             break;
 307         case SCE_VISUALPROLOG_STRING_ESCAPE:
 308         case SCE_VISUALPROLOG_STRING_ESCAPE_ERROR:
 309             // return to SCE_VISUALPROLOG_STRING and treat as such (fall-through)
 310             sc.SetState(SCE_VISUALPROLOG_STRING);
 311         case SCE_VISUALPROLOG_STRING:
 312             if (sc.atLineEnd) {
 313                 sc.SetState(SCE_VISUALPROLOG_STRING_EOL_OPEN);
 314             } else if (sc.ch == '"') {
 315                 sc.ForwardSetState(SCE_VISUALPROLOG_DEFAULT);
 316             } else if (sc.ch == '\\') {
 317                 sc.SetState(SCE_VISUALPROLOG_STRING_ESCAPE_ERROR);
 318                 forwardEscapeLiteral(sc, '"', SCE_VISUALPROLOG_STRING_ESCAPE);
 319             }
 320             break;
 321         case SCE_VISUALPROLOG_CHARACTER_TOO_MANY:
 322             if (sc.atLineStart) {
 323                 sc.SetState(SCE_VISUALPROLOG_DEFAULT);
 324             } else if (sc.ch == '\'') {
 325                 sc.SetState(SCE_VISUALPROLOG_CHARACTER);
 326                 sc.ForwardSetState(SCE_VISUALPROLOG_DEFAULT);
 327             }
 328             break;
 329         case SCE_VISUALPROLOG_CHARACTER:
 330             if (sc.atLineEnd) {
 331                 sc.SetState(SCE_VISUALPROLOG_STRING_EOL_OPEN);  // reuse STRING_EOL_OPEN for this
 332             } else if (sc.ch == '\'') {
 333                 sc.SetState(SCE_VISUALPROLOG_CHARACTER_ESCAPE_ERROR);
 334                 sc.ForwardSetState(SCE_VISUALPROLOG_DEFAULT);
 335             } else {
 336                 if (sc.ch == '\\') {
 337                     sc.SetState(SCE_VISUALPROLOG_CHARACTER_ESCAPE_ERROR);
 338                     forwardEscapeLiteral(sc, '\'', SCE_VISUALPROLOG_CHARACTER);
 339                 }
 340                 sc.ForwardSetState(SCE_VISUALPROLOG_CHARACTER);
 341                 if (sc.ch == '\'') {
 342                     sc.ForwardSetState(SCE_VISUALPROLOG_DEFAULT);
 343                 } else {
 344                     sc.SetState(SCE_VISUALPROLOG_CHARACTER_TOO_MANY);
 345                 }
 346             }
 347             break;
 348         case SCE_VISUALPROLOG_STRING_EOL_OPEN:
 349             if (sc.atLineStart) {
 350                 sc.SetState(SCE_VISUALPROLOG_DEFAULT);
 351             }
 352             break;
 353         case SCE_VISUALPROLOG_STRING_VERBATIM_SPECIAL:
 354         case SCE_VISUALPROLOG_STRING_VERBATIM_EOL:
 355             // return to SCE_VISUALPROLOG_STRING_VERBATIM and treat as such (fall-through)
 356             sc.SetState(SCE_VISUALPROLOG_STRING_VERBATIM);
 357         case SCE_VISUALPROLOG_STRING_VERBATIM:
 358             if (sc.atLineEnd) {
 359                 sc.SetState(SCE_VISUALPROLOG_STRING_VERBATIM_EOL);
 360             } else if (sc.ch == '\"') {
 361                 if (sc.chNext == '\"') {
 362                     sc.SetState(SCE_VISUALPROLOG_STRING_VERBATIM_SPECIAL);
 363                     sc.Forward();
 364                 } else {
 365                     sc.ForwardSetState(SCE_VISUALPROLOG_DEFAULT);
 366                 }
 367             }
 368             break;
 369         }
 370
 371         if (sc.atLineEnd && !atLineEndBeforeSwitch) {
 372             // State exit processing consumed characters up to end of line.
 373             currentLine++;
 374         }
 375
 376         // Determine if a new state should be entered.
 377         if (sc.state == SCE_VISUALPROLOG_DEFAULT) {
 378             if (sc.Match('@', '\"')) {
 379                 sc.SetState(SCE_VISUALPROLOG_STRING_VERBATIM);
 380                 sc.Forward();
 381             } else if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
 382                 sc.SetState(SCE_VISUALPROLOG_NUMBER);
 383             } else if (setLowerStart.Contains(sc.ch)) {
 384                 sc.SetState(SCE_VISUALPROLOG_IDENTIFIER);
 385             } else if (setVariableStart.Contains(sc.ch)) {
 386                 sc.SetState(SCE_VISUALPROLOG_VARIABLE);
 387             } else if (sc.ch == '_') {
 388                 sc.SetState(SCE_VISUALPROLOG_ANONYMOUS);
 389             } else if (sc.Match('/', '*')) {
 390                 sc.SetState(SCE_VISUALPROLOG_COMMENT_BLOCK);
 391                 nestLevel = 1;
 392                 sc.Forward();   // Eat the * so it isn't used for the end of the comment
 393             } else if (sc.ch == '%') {
 394                 sc.SetState(SCE_VISUALPROLOG_COMMENT_LINE);
 395             } else if (sc.ch == '\"') {
 396                 sc.SetState(SCE_VISUALPROLOG_STRING);
 397             } else if (sc.ch == '\'') {
 398                 sc.SetState(SCE_VISUALPROLOG_CHARACTER);
 399             } else if (sc.ch == '#') {
 400                 sc.SetState(SCE_VISUALPROLOG_KEY_DIRECTIVE);
 401             } else if (isoperator(static_cast<char>(sc.ch)) || sc.ch == '\\') {
 402                 sc.SetState(SCE_VISUALPROLOG_OPERATOR);
 403             }
 404         }
 405
 406     }
 407     sc.Complete();
 408     styler.Flush();
 409 }
 410
 411 // Store both the current line's fold level and the next lines in the
 412 // level store to make it easy to pick up with each increment
 413 // and to make it possible to fiddle the current level for "} else {".
 414
 415 void SCI_METHOD LexerVisualProlog::Fold(unsigned int startPos, int length, int initStyle, IDocument *pAccess) {
 416
 417     LexAccessor styler(pAccess);
 418
 419     unsigned int endPos = startPos + length;
 420     int visibleChars = 0;
 421     int currentLine = styler.GetLine(startPos);
 422     int levelCurrent = SC_FOLDLEVELBASE;
 423     if (currentLine > 0)
 424         levelCurrent = styler.LevelAt(currentLine-1) >> 16;
 425     int levelMinCurrent = levelCurrent;
 426     int levelNext = levelCurrent;
 427     char chNext = styler[startPos];
 428     int styleNext = styler.StyleAt(startPos);
 429     int style = initStyle;
 430     for (unsigned int i = startPos; i < endPos; i++) {
 431         char ch = chNext;
 432         chNext = styler.SafeGetCharAt(i + 1);
 433         style = styleNext;
 434         styleNext = styler.StyleAt(i + 1);
 435         bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
 436         if (style == SCE_VISUALPROLOG_OPERATOR) {
 437             if (ch == '{') {
 438                 // Measure the minimum before a '{' to allow
 439                 // folding on "} else {"
 440                 if (levelMinCurrent > levelNext) {
 441                     levelMinCurrent = levelNext;
 442                 }
 443                 levelNext++;
 444             } else if (ch == '}') {
 445                 levelNext--;
 446             }
 447         }
 448         if (!IsASpace(ch))
 449             visibleChars++;
 450         if (atEOL || (i == endPos-1)) {
 451             int levelUse = levelCurrent;
 452             int lev = levelUse | levelNext << 16;
 453             if (levelUse < levelNext)
 454                 lev |= SC_FOLDLEVELHEADERFLAG;
 455             if (lev != styler.LevelAt(currentLine)) {
 456                 styler.SetLevel(currentLine, lev);
 457             }
 458             currentLine++;
 459             levelCurrent = levelNext;
 460             levelMinCurrent = levelCurrent;
 461             if (atEOL && (i == static_cast<unsigned int>(styler.Length()-1))) {
 462                 // There is an empty line at end of file so give it same level and empty
 463                 styler.SetLevel(currentLine, (levelCurrent | levelCurrent << 16) | SC_FOLDLEVELWHITEFLAG);
 464             }
 465             visibleChars = 0;
 466         }
 467     }
 468 }
 469
 470 LexerModule lmVisualProlog(SCLEX_VISUALPROLOG, LexerVisualProlog::LexerFactoryVisualProlog, "visualprolog", visualPrologWordLists);