ext/scintilla/src/LexCPP.cxx

   1 // Scintilla source code edit control
   2 /** @file LexCPP.cxx
   3  ** Lexer for C++, C, Java, and JavaScript.
   4  **/
   5 // Copyright 1998-2005 by Neil Hodgson <neilh@scintilla.org>
   6 // The License.txt file describes the conditions under which this software may be distributed.
   7
   8 #include <stdlib.h>
   9 #include <string.h>
  10 #include <ctype.h>
  11 #include <stdio.h>
  12 #include <stdarg.h>
  13
  14 #include "Platform.h"
  15
  16 #include "PropSet.h"
  17 #include "Accessor.h"
  18 #include "StyleContext.h"
  19 #include "KeyWords.h"
  20 #include "Scintilla.h"
  21 #include "SciLexer.h"
  22 #include "CharacterSet.h"
  23
  24 #ifdef SCI_NAMESPACE
  25 using namespace Scintilla;
  26 #endif
  27
  28 static bool IsSpaceEquiv(int state) {
  29         return (state <= SCE_C_COMMENTDOC) ||
  30                 // including SCE_C_DEFAULT, SCE_C_COMMENT, SCE_C_COMMENTLINE
  31                 (state == SCE_C_COMMENTLINEDOC) || (state == SCE_C_COMMENTDOCKEYWORD) ||
  32                 (state == SCE_C_COMMENTDOCKEYWORDERROR);
  33 }
  34
  35 // Preconditions: sc.currentPos points to a character after '+' or '-'.
  36 // The test for pos reaching 0 should be redundant,
  37 // and is in only for safety measures.
  38 // Limitation: this code will give the incorrect answer for code like
  39 // a = b+++/ptn/...
  40 // Putting a space between the '++' post-inc operator and the '+' binary op
  41 // fixes this, and is highly recommended for readability anyway.
  42 static bool FollowsPostfixOperator(StyleContext &sc, Accessor &styler) {
  43         int pos = (int) sc.currentPos;
  44         while (--pos > 0) {
  45                 char ch = styler[pos];
  46                 if (ch == '+' || ch == '-') {
  47                         return styler[pos - 1] == ch;
  48                 }
  49         }
  50         return false;
  51 }
  52
  53 static void ColouriseCppDoc(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],
  54                             Accessor &styler, bool caseSensitive) {
  55
  56         WordList &keywords = *keywordlists[0];
  57         WordList &keywords2 = *keywordlists[1];
  58         WordList &keywords3 = *keywordlists[2];
  59         WordList &keywords4 = *keywordlists[3];
  60
  61         bool stylingWithinPreprocessor = styler.GetPropertyInt("styling.within.preprocessor") != 0;
  62
  63         CharacterSet setOKBeforeRE(CharacterSet::setNone, "([{=,:;!%^&*|?~+-");
  64         CharacterSet setCouldBePostOp(CharacterSet::setNone, "+-");
  65
  66         CharacterSet setDoxygen(CharacterSet::setAlpha, "$@\\&<>#{}[]");
  67
  68         CharacterSet setWordStart(CharacterSet::setAlpha, "_", 0x80, true);
  69         CharacterSet setWord(CharacterSet::setAlphaNum, "._", 0x80, true);
  70         if (styler.GetPropertyInt("lexer.cpp.allow.dollars", 1) != 0) {
  71                 setWordStart.Add('$');
  72                 setWord.Add('$');
  73         }
  74
  75         int chPrevNonWhite = ' ';
  76         int visibleChars = 0;
  77         bool lastWordWasUUID = false;
  78         int styleBeforeDCKeyword = SCE_C_DEFAULT;
  79         bool continuationLine = false;
  80
  81         if (initStyle == SCE_C_PREPROCESSOR) {
  82                 // Set continuationLine if last character of previous line is '\'
  83                 int lineCurrent = styler.GetLine(startPos);
  84                 if (lineCurrent > 0) {
  85                         int chBack = styler.SafeGetCharAt(startPos-1, 0);
  86                         int chBack2 = styler.SafeGetCharAt(startPos-2, 0);
  87                         int lineEndChar = '!';
  88                         if (chBack2 == '\r' && chBack == '\n') {
  89                                 lineEndChar = styler.SafeGetCharAt(startPos-3, 0);
  90                         } else if (chBack == '\n' || chBack == '\r') {
  91                                 lineEndChar = chBack2;
  92                         }
  93                         continuationLine = lineEndChar == '\\';
  94                 }
  95         }
  96
  97         // look back to set chPrevNonWhite properly for better regex colouring
  98         if (startPos > 0) {
  99                 int back = startPos;
 100                 while (--back && IsSpaceEquiv(styler.StyleAt(back)))
 101                         ;
 102                 if (styler.StyleAt(back) == SCE_C_OPERATOR) {
 103                         chPrevNonWhite = styler.SafeGetCharAt(back);
 104                 }
 105         }
 106
 107         StyleContext sc(startPos, length, initStyle, styler);
 108
 109         for (; sc.More(); sc.Forward()) {
 110
 111                 if (sc.atLineStart) {
 112                         if (sc.state == SCE_C_STRING) {
 113                                 // Prevent SCE_C_STRINGEOL from leaking back to previous line which
 114                                 // ends with a line continuation by locking in the state upto this position.
 115                                 sc.SetState(SCE_C_STRING);
 116                         }
 117                         // Reset states to begining of colourise so no surprises
 118                         // if different sets of lines lexed.
 119                         visibleChars = 0;
 120                         lastWordWasUUID = false;
 121                 }
 122
 123                 // Handle line continuation generically.
 124                 if (sc.ch == '\\') {
 125                         if (sc.chNext == '\n' || sc.chNext == '\r') {
 126                                 sc.Forward();
 127                                 if (sc.ch == '\r' && sc.chNext == '\n') {
 128                                         sc.Forward();
 129                                 }
 130                                 continuationLine = true;
 131                                 continue;
 132                         }
 133                 }
 134
 135                 // Determine if the current state should terminate.
 136                 switch (sc.state) {
 137                         case SCE_C_OPERATOR:
 138                                 sc.SetState(SCE_C_DEFAULT);
 139                                 break;
 140                         case SCE_C_NUMBER:
 141                                 // We accept almost anything because of hex. and number suffixes
 142                                 if (!setWord.Contains(sc.ch)) {
 143                                         sc.SetState(SCE_C_DEFAULT);
 144                                 }
 145                                 break;
 146                         case SCE_C_IDENTIFIER:
 147                                 if (!setWord.Contains(sc.ch) || (sc.ch == '.')) {
 148                                         char s[1000];
 149                                         if (caseSensitive) {
 150                                                 sc.GetCurrent(s, sizeof(s));
 151                                         } else {
 152                                                 sc.GetCurrentLowered(s, sizeof(s));
 153                                         }
 154                                         if (keywords.InList(s)) {
 155                                                 lastWordWasUUID = strcmp(s, "uuid") == 0;
 156                                                 sc.ChangeState(SCE_C_WORD);
 157                                         } else if (keywords2.InList(s)) {
 158                                                 sc.ChangeState(SCE_C_WORD2);
 159                                         } else if (keywords4.InList(s)) {
 160                                                 sc.ChangeState(SCE_C_GLOBALCLASS);
 161                                         }
 162                                         sc.SetState(SCE_C_DEFAULT);
 163                                 }
 164                                 break;
 165                         case SCE_C_PREPROCESSOR:
 166                                 if (sc.atLineStart && !continuationLine) {
 167                                         sc.SetState(SCE_C_DEFAULT);
 168                                 } else if (stylingWithinPreprocessor) {
 169                                         if (IsASpace(sc.ch)) {
 170                                                 sc.SetState(SCE_C_DEFAULT);
 171                                         }
 172                                 } else {
 173                                         if (sc.Match('/', '*') || sc.Match('/', '/')) {
 174                                                 sc.SetState(SCE_C_DEFAULT);
 175                                         }
 176                                 }
 177                                 break;
 178                         case SCE_C_COMMENT:
 179                                 if (sc.Match('*', '/')) {
 180                                         sc.Forward();
 181                                         sc.ForwardSetState(SCE_C_DEFAULT);
 182                                 }
 183                                 break;
 184                         case SCE_C_COMMENTDOC:
 185                                 if (sc.Match('*', '/')) {
 186                                         sc.Forward();
 187                                         sc.ForwardSetState(SCE_C_DEFAULT);
 188                                 } else if (sc.ch == '@' || sc.ch == '\\') { // JavaDoc and Doxygen support
 189                                         // Verify that we have the conditions to mark a comment-doc-keyword
 190                                         if ((IsASpace(sc.chPrev) || sc.chPrev == '*') && (!IsASpace(sc.chNext))) {
 191                                                 styleBeforeDCKeyword = SCE_C_COMMENTDOC;
 192                                                 sc.SetState(SCE_C_COMMENTDOCKEYWORD);
 193                                         }
 194                                 }
 195                                 break;
 196                         case SCE_C_COMMENTLINE:
 197                                 if (sc.atLineStart) {
 198                                         sc.SetState(SCE_C_DEFAULT);
 199                                 }
 200                                 break;
 201                         case SCE_C_COMMENTLINEDOC:
 202                                 if (sc.atLineStart) {
 203                                         sc.SetState(SCE_C_DEFAULT);
 204                                 } else if (sc.ch == '@' || sc.ch == '\\') { // JavaDoc and Doxygen support
 205                                         // Verify that we have the conditions to mark a comment-doc-keyword
 206                                         if ((IsASpace(sc.chPrev) || sc.chPrev == '/' || sc.chPrev == '!') && (!IsASpace(sc.chNext))) {
 207                                                 styleBeforeDCKeyword = SCE_C_COMMENTLINEDOC;
 208                                                 sc.SetState(SCE_C_COMMENTDOCKEYWORD);
 209                                         }
 210                                 }
 211                                 break;
 212                         case SCE_C_COMMENTDOCKEYWORD:
 213                                 if ((styleBeforeDCKeyword == SCE_C_COMMENTDOC) && sc.Match('*', '/')) {
 214                                         sc.ChangeState(SCE_C_COMMENTDOCKEYWORDERROR);
 215                                         sc.Forward();
 216                                         sc.ForwardSetState(SCE_C_DEFAULT);
 217                                 } else if (!setDoxygen.Contains(sc.ch)) {
 218                                         char s[100];
 219                                         if (caseSensitive) {
 220                                                 sc.GetCurrent(s, sizeof(s));
 221                                         } else {
 222                                                 sc.GetCurrentLowered(s, sizeof(s));
 223                                         }
 224                                         if (!IsASpace(sc.ch) || !keywords3.InList(s + 1)) {
 225                                                 sc.ChangeState(SCE_C_COMMENTDOCKEYWORDERROR);
 226                                         }
 227                                         sc.SetState(styleBeforeDCKeyword);
 228                                 }
 229                                 break;
 230                         case SCE_C_STRING:
 231                                 if (sc.atLineEnd) {
 232                                         sc.ChangeState(SCE_C_STRINGEOL);
 233                                 } else if (sc.ch == '\\') {
 234                                         if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\') {
 235                                                 sc.Forward();
 236                                         }
 237                                 } else if (sc.ch == '\"') {
 238                                         sc.ForwardSetState(SCE_C_DEFAULT);
 239                                 }
 240                                 break;
 241                         case SCE_C_CHARACTER:
 242                                 if (sc.atLineEnd) {
 243                                         sc.ChangeState(SCE_C_STRINGEOL);
 244                                 } else if (sc.ch == '\\') {
 245                                         if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\') {
 246                                                 sc.Forward();
 247                                         }
 248                                 } else if (sc.ch == '\'') {
 249                                         sc.ForwardSetState(SCE_C_DEFAULT);
 250                                 }
 251                                 break;
 252                         case SCE_C_REGEX:
 253                                 if (sc.atLineStart) {
 254                                         sc.SetState(SCE_C_DEFAULT);
 255                                 } else if (sc.ch == '/') {
 256                                         sc.Forward();
 257                                         while ((sc.ch < 0x80) && islower(sc.ch))
 258                                                 sc.Forward();    // gobble regex flags
 259                                         sc.SetState(SCE_C_DEFAULT);
 260                                 } else if (sc.ch == '\\') {
 261                                         // Gobble up the quoted character
 262                                         if (sc.chNext == '\\' || sc.chNext == '/') {
 263                                                 sc.Forward();
 264                                         }
 265                                 }
 266                                 break;
 267                         case SCE_C_STRINGEOL:
 268                                 if (sc.atLineStart) {
 269                                         sc.SetState(SCE_C_DEFAULT);
 270                                 }
 271                                 break;
 272                         case SCE_C_VERBATIM:
 273                                 if (sc.ch == '\"') {
 274                                         if (sc.chNext == '\"') {
 275                                                 sc.Forward();
 276                                         } else {
 277                                                 sc.ForwardSetState(SCE_C_DEFAULT);
 278                                         }
 279                                 }
 280                                 break;
 281                         case SCE_C_UUID:
 282                                 if (sc.ch == '\r' || sc.ch == '\n' || sc.ch == ')') {
 283                                         sc.SetState(SCE_C_DEFAULT);
 284                                 }
 285                 }
 286
 287                 // Determine if a new state should be entered.
 288                 if (sc.state == SCE_C_DEFAULT) {
 289                         if (sc.Match('@', '\"')) {
 290                                 sc.SetState(SCE_C_VERBATIM);
 291                                 sc.Forward();
 292                         } else if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
 293                                 if (lastWordWasUUID) {
 294                                         sc.SetState(SCE_C_UUID);
 295                                         lastWordWasUUID = false;
 296                                 } else {
 297                                         sc.SetState(SCE_C_NUMBER);
 298                                 }
 299                         } else if (setWordStart.Contains(sc.ch) || (sc.ch == '@')) {
 300                                 if (lastWordWasUUID) {
 301                                         sc.SetState(SCE_C_UUID);
 302                                         lastWordWasUUID = false;
 303                                 } else {
 304                                         sc.SetState(SCE_C_IDENTIFIER);
 305                                 }
 306                         } else if (sc.Match('/', '*')) {
 307                                 if (sc.Match("/**") || sc.Match("/*!")) {       // Support of Qt/Doxygen doc. style
 308                                         sc.SetState(SCE_C_COMMENTDOC);
 309                                 } else {
 310                                         sc.SetState(SCE_C_COMMENT);
 311                                 }
 312                                 sc.Forward();   // Eat the * so it isn't used for the end of the comment
 313                         } else if (sc.Match('/', '/')) {
 314                                 if ((sc.Match("///") && !sc.Match("////")) || sc.Match("//!"))
 315                                         // Support of Qt/Doxygen doc. style
 316                                         sc.SetState(SCE_C_COMMENTLINEDOC);
 317                                 else
 318                                         sc.SetState(SCE_C_COMMENTLINE);
 319                         } else if (sc.ch == '/' && setOKBeforeRE.Contains(chPrevNonWhite) &&
 320                                 (!setCouldBePostOp.Contains(chPrevNonWhite) || !FollowsPostfixOperator(sc, styler))) {
 321                                 sc.SetState(SCE_C_REGEX);       // JavaScript's RegEx
 322                         } else if (sc.ch == '\"') {
 323                                 sc.SetState(SCE_C_STRING);
 324                         } else if (sc.ch == '\'') {
 325                                 sc.SetState(SCE_C_CHARACTER);
 326                         } else if (sc.ch == '#' && visibleChars == 0) {
 327                                 // Preprocessor commands are alone on their line
 328                                 sc.SetState(SCE_C_PREPROCESSOR);
 329                                 // Skip whitespace between # and preprocessor word
 330                                 do {
 331                                         sc.Forward();
 332                                 } while ((sc.ch == ' ' || sc.ch == '\t') && sc.More());
 333                                 if (sc.atLineEnd) {
 334                                         sc.SetState(SCE_C_DEFAULT);
 335                                 }
 336                         } else if (isoperator(static_cast<char>(sc.ch))) {
 337                                 sc.SetState(SCE_C_OPERATOR);
 338                         }
 339                 }
 340
 341                 if (!IsASpace(sc.ch) && !IsSpaceEquiv(sc.state)) {
 342                         chPrevNonWhite = sc.ch;
 343                         visibleChars++;
 344                 }
 345                 continuationLine = false;
 346         }
 347         sc.Complete();
 348 }
 349
 350 static bool IsStreamCommentStyle(int style) {
 351         return style == SCE_C_COMMENT ||
 352                 style == SCE_C_COMMENTDOC ||
 353                 style == SCE_C_COMMENTDOCKEYWORD ||
 354                 style == SCE_C_COMMENTDOCKEYWORDERROR;
 355 }
 356
 357 // Store both the current line's fold level and the next lines in the
 358 // level store to make it easy to pick up with each increment
 359 // and to make it possible to fiddle the current level for "} else {".
 360 static void FoldCppDoc(unsigned int startPos, int length, int initStyle,
 361                                            WordList *[], Accessor &styler) {
 362         bool foldComment = styler.GetPropertyInt("fold.comment") != 0;
 363         bool foldPreprocessor = styler.GetPropertyInt("fold.preprocessor") != 0;
 364         bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;
 365         bool foldAtElse = styler.GetPropertyInt("fold.at.else", 0) != 0;
 366         unsigned int endPos = startPos + length;
 367         int visibleChars = 0;
 368         int lineCurrent = styler.GetLine(startPos);
 369         int levelCurrent = SC_FOLDLEVELBASE;
 370         if (lineCurrent > 0)
 371                 levelCurrent = styler.LevelAt(lineCurrent-1) >> 16;
 372         int levelMinCurrent = levelCurrent;
 373         int levelNext = levelCurrent;
 374         char chNext = styler[startPos];
 375         int styleNext = styler.StyleAt(startPos);
 376         int style = initStyle;
 377         for (unsigned int i = startPos; i < endPos; i++) {
 378                 char ch = chNext;
 379                 chNext = styler.SafeGetCharAt(i + 1);
 380                 int stylePrev = style;
 381                 style = styleNext;
 382                 styleNext = styler.StyleAt(i + 1);
 383                 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
 384                 if (foldComment && IsStreamCommentStyle(style)) {
 385                         if (!IsStreamCommentStyle(stylePrev) && (stylePrev != SCE_C_COMMENTLINEDOC)) {
 386                                 levelNext++;
 387                         } else if (!IsStreamCommentStyle(styleNext) && (styleNext != SCE_C_COMMENTLINEDOC) && !atEOL) {
 388                                 // Comments don't end at end of line and the next character may be unstyled.
 389                                 levelNext--;
 390                         }
 391                 }
 392                 if (foldComment && (style == SCE_C_COMMENTLINE)) {
 393                         if ((ch == '/') && (chNext == '/')) {
 394                                 char chNext2 = styler.SafeGetCharAt(i + 2);
 395                                 if (chNext2 == '{') {
 396                                         levelNext++;
 397                                 } else if (chNext2 == '}') {
 398                                         levelNext--;
 399                                 }
 400                         }
 401                 }
 402                 if (foldPreprocessor && (style == SCE_C_PREPROCESSOR)) {
 403                         if (ch == '#') {
 404                                 unsigned int j = i + 1;
 405                                 while ((j < endPos) && IsASpaceOrTab(styler.SafeGetCharAt(j))) {
 406                                         j++;
 407                                 }
 408                                 if (styler.Match(j, "region") || styler.Match(j, "if")) {
 409                                         levelNext++;
 410                                 } else if (styler.Match(j, "end")) {
 411                                         levelNext--;
 412                                 }
 413                         }
 414                 }
 415                 if (style == SCE_C_OPERATOR) {
 416                         if (ch == '{') {
 417                                 // Measure the minimum before a '{' to allow
 418                                 // folding on "} else {"
 419                                 if (levelMinCurrent > levelNext) {
 420                                         levelMinCurrent = levelNext;
 421                                 }
 422                                 levelNext++;
 423                         } else if (ch == '}') {
 424                                 levelNext--;
 425                         }
 426                 }
 427                 if (!IsASpace(ch))
 428                         visibleChars++;
 429                 if (atEOL || (i == endPos-1)) {
 430                         int levelUse = levelCurrent;
 431                         if (foldAtElse) {
 432                                 levelUse = levelMinCurrent;
 433                         }
 434                         int lev = levelUse | levelNext << 16;
 435                         if (visibleChars == 0 && foldCompact)
 436                                 lev |= SC_FOLDLEVELWHITEFLAG;
 437                         if (levelUse < levelNext)
 438                                 lev |= SC_FOLDLEVELHEADERFLAG;
 439                         if (lev != styler.LevelAt(lineCurrent)) {
 440                                 styler.SetLevel(lineCurrent, lev);
 441                         }
 442                         lineCurrent++;
 443                         levelCurrent = levelNext;
 444                         levelMinCurrent = levelCurrent;
 445                         visibleChars = 0;
 446                 }
 447         }
 448 }
 449
 450 static const char * const cppWordLists[] = {
 451             "Primary keywords and identifiers",
 452             "Secondary keywords and identifiers",
 453             "Documentation comment keywords",
 454             "Unused",
 455             "Global classes and typedefs",
 456             0,
 457         };
 458
 459 static void ColouriseCppDocSensitive(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],
 460                                      Accessor &styler) {
 461         ColouriseCppDoc(startPos, length, initStyle, keywordlists, styler, true);
 462 }
 463
 464 static void ColouriseCppDocInsensitive(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],
 465                                        Accessor &styler) {
 466         ColouriseCppDoc(startPos, length, initStyle, keywordlists, styler, false);
 467 }
 468
 469 LexerModule lmCPP(SCLEX_CPP, ColouriseCppDocSensitive, "cpp", FoldCppDoc, cppWordLists);
 470 LexerModule lmCPPNoCase(SCLEX_CPPNOCASE, ColouriseCppDocInsensitive, "cppnocase", FoldCppDoc, cppWordLists);