contrib/src/stc/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 1998-2001 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 "KeyWords.h"
  19 #include "Scintilla.h"
  20 #include "SciLexer.h"
  21
  22 static bool IsOKBeforeRE(int ch) {
  23         return (ch == '(') || (ch == '=') || (ch == ',');
  24 }
  25
  26 static void getRange(unsigned int start,
  27                 unsigned int end,
  28                 Accessor &styler,
  29                 char *s,
  30                 unsigned int len) {
  31         unsigned int i = 0;
  32         while ((i < end - start + 1) && (i < len-1)) {
  33                 s[i] = styler[start + i];
  34                 i++;
  35         }
  36         s[i] = '\0';
  37 }
  38
  39 inline bool IsASpace(int ch) {
  40     return (ch == ' ') || ((ch >= 0x09) && (ch <= 0x0d));
  41 }
  42
  43 inline bool IsAWordChar(int  ch) {
  44         return (ch < 0x80) && (isalnum(ch) || ch == '.' || ch == '_');
  45 }
  46
  47 inline bool IsAWordStart(int ch) {
  48         return (ch < 0x80) && (isalnum(ch) || ch == '_');
  49 }
  50
  51 inline bool IsADigit(int ch) {
  52         return (ch >= '0') && (ch <= '9');
  53 }
  54
  55 // All languages handled so far can treat all characters >= 0x80 as one class
  56 // which just continues the current token or starts an identifier if in default.
  57 // DBCS treated specially as the second character can be < 0x80 and hence
  58 // syntactically significant. UTF-8 avoids this as all trail bytes are >= 0x80
  59 class ColouriseContext {
  60         Accessor &styler;
  61         int lengthDoc;
  62         int currentPos;
  63         ColouriseContext& operator=(const ColouriseContext&) {
  64                 return *this;
  65         }
  66 public:
  67         bool atEOL;
  68         int state;
  69         int chPrev;
  70         int ch;
  71         int chNext;
  72
  73         ColouriseContext(unsigned int startPos, int length,
  74                         int initStyle, Accessor &styler_) :
  75                 styler(styler_),
  76                 lengthDoc(startPos + length),
  77                 currentPos(startPos),
  78                 atEOL(false),
  79                 state(initStyle),
  80                 chPrev(0),
  81                 ch(0),
  82                 chNext(0) {
  83                 styler.StartAt(startPos);
  84                 styler.StartSegment(startPos);
  85                 int pos = currentPos;
  86                 ch = static_cast<unsigned char>(styler.SafeGetCharAt(pos));
  87                 if (styler.IsLeadByte(static_cast<char>(ch))) {
  88                         pos++;
  89                         ch = ch << 8;
  90                         ch |= static_cast<unsigned char>(styler.SafeGetCharAt(pos));
  91                 }
  92                 chNext = static_cast<unsigned char>(styler.SafeGetCharAt(pos+1));
  93                 if (styler.IsLeadByte(static_cast<char>(chNext))) {
  94                         chNext = chNext << 8;
  95                         chNext |= static_cast<unsigned char>(styler.SafeGetCharAt(pos+2));
  96                 }
  97                 atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
  98         }
  99         void Complete() {
 100                 styler.ColourTo(currentPos - 1, state);
 101         }
 102         bool More() {
 103                 return currentPos <= lengthDoc;
 104         }
 105         void Forward() {
 106                 // A lot of this is repeated from the constructor - TODO: merge code
 107                 chPrev = ch;
 108                 currentPos++;
 109                 if (ch >= 0x100)
 110                         currentPos++;
 111                 ch = chNext;
 112                 chNext = static_cast<unsigned char>(styler.SafeGetCharAt(currentPos+1));
 113                 if (styler.IsLeadByte(static_cast<char>(chNext))) {
 114                         chNext = chNext << 8;
 115                         chNext |= static_cast<unsigned char>(styler.SafeGetCharAt(currentPos + 2));
 116                 }
 117                 // Trigger on CR only (Mac style) or either on LF from CR+LF (Dos/Win) or on LF alone (Unix)
 118                 // Avoid triggering two times on Dos/Win
 119                 // End of line
 120                 atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
 121         }
 122         void ChangeState(int state_) {
 123                 state = state_;
 124         }
 125         void SetState(int state_) {
 126                 styler.ColourTo(currentPos - 1, state);
 127                 state = state_;
 128         }
 129         void ForwardSetState(int state_) {
 130                 Forward();
 131                 styler.ColourTo(currentPos - 1, state);
 132                 state = state_;
 133         }
 134         void GetCurrent(char *s, int len) {
 135                 getRange(styler.GetStartSegment(), currentPos - 1, styler, s, len);
 136         }
 137         int LengthCurrent() {
 138                 return currentPos - styler.GetStartSegment();
 139         }
 140         bool Match(char ch0) {
 141                 return ch == ch0;
 142         }
 143         bool Match(char ch0, char ch1) {
 144                 return (ch == ch0) && (chNext == ch1);
 145         }
 146         bool Match(const char *s) {
 147                 if (ch != *s)
 148                         return false;
 149                 s++;
 150                 if (chNext != *s)
 151                         return false;
 152                 s++;
 153                 for (int n=2; *s; n++) {
 154                         if (*s != styler.SafeGetCharAt(currentPos+n))
 155                                 return false;
 156                         s++;
 157                 }
 158                 return true;
 159         }
 160 };
 161
 162 static void ColouriseCppDoc(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],
 163                             Accessor &styler) {
 164
 165         WordList &keywords = *keywordlists[0];
 166         WordList &keywords2 = *keywordlists[1];
 167
 168         bool stylingWithinPreprocessor = styler.GetPropertyInt("styling.within.preprocessor");
 169
 170         if (initStyle == SCE_C_STRINGEOL)       // Does not leak onto next line
 171                 initStyle = SCE_C_DEFAULT;
 172
 173         int chPrevNonWhite = ' ';
 174         int visibleChars = 0;
 175         bool lastWordWasUUID = false;
 176
 177         ColouriseContext cc(startPos, length, initStyle, styler);
 178
 179         for (; cc.More(); cc.Forward()) {
 180
 181                 if (cc.state == SCE_C_STRINGEOL) {
 182                         if (cc.atEOL) {
 183                                 cc.SetState(SCE_C_DEFAULT);
 184                         }
 185                 } else if (cc.state == SCE_C_OPERATOR) {
 186                         cc.SetState(SCE_C_DEFAULT);
 187                 } else if (cc.state == SCE_C_NUMBER) {
 188                         if (!IsAWordChar(cc.ch)) {
 189                                 cc.SetState(SCE_C_DEFAULT);
 190                         }
 191                 } else if (cc.state == SCE_C_IDENTIFIER) {
 192                         if (!IsAWordChar(cc.ch) || (cc.ch == '.')) {
 193                                 char s[100];
 194                                 cc.GetCurrent(s, sizeof(s));
 195                                 if (keywords.InList(s)) {
 196                                         lastWordWasUUID = strcmp(s, "uuid") == 0;
 197                                         cc.ChangeState(SCE_C_WORD);
 198                                 } else if (keywords2.InList(s)) {
 199                                         cc.ChangeState(SCE_C_WORD2);
 200                                 }
 201                                 cc.SetState(SCE_C_DEFAULT);
 202                         }
 203                 } if (cc.state == SCE_C_PREPROCESSOR) {
 204                         if (stylingWithinPreprocessor) {
 205                                 if (IsASpace(cc.ch)) {
 206                                         cc.SetState(SCE_C_DEFAULT);
 207                                 }
 208                         } else {
 209                                 if (cc.atEOL && (cc.chPrev != '\\')) {
 210                                         cc.SetState(SCE_C_DEFAULT);
 211                                 }
 212                         }
 213                 } else if (cc.state == SCE_C_COMMENT) {
 214                         if (cc.Match('*', '/')) {
 215                                 cc.Forward();
 216                                 cc.ForwardSetState(SCE_C_DEFAULT);
 217                         }
 218                 } else if (cc.state == SCE_C_COMMENTDOC) {
 219                         if (cc.Match('*', '/')) {
 220                                 cc.Forward();
 221                                 cc.ForwardSetState(SCE_C_DEFAULT);
 222                         }
 223                 } else if (cc.state == SCE_C_COMMENTLINE || cc.state == SCE_C_COMMENTLINEDOC) {
 224                         if (cc.ch == '\r' || cc.ch == '\n') {
 225                                 cc.SetState(SCE_C_DEFAULT);
 226                         }
 227                 } else if (cc.state == SCE_C_STRING) {
 228                         if (cc.ch == '\\') {
 229                                 if (cc.chNext == '\"' || cc.chNext == '\'' || cc.chNext == '\\') {
 230                                         cc.Forward();
 231                                 }
 232                         } else if (cc.ch == '\"') {
 233                                 cc.ForwardSetState(SCE_C_DEFAULT);
 234                         } else if ((cc.atEOL) && (cc.chPrev != '\\')) {
 235                                 cc.ChangeState(SCE_C_STRINGEOL);
 236                         }
 237                 } else if (cc.state == SCE_C_CHARACTER) {
 238                         if ((cc.ch == '\r' || cc.ch == '\n') && (cc.chPrev != '\\')) {
 239                                 cc.ChangeState(SCE_C_STRINGEOL);
 240                         } else if (cc.ch == '\\') {
 241                                 if (cc.chNext == '\"' || cc.chNext == '\'' || cc.chNext == '\\') {
 242                                         cc.Forward();
 243                                 }
 244                         } else if (cc.ch == '\'') {
 245                                 cc.ForwardSetState(SCE_C_DEFAULT);
 246                         }
 247                 } else if (cc.state == SCE_C_REGEX) {
 248                         if (cc.ch == '\r' || cc.ch == '\n' || cc.ch == '/') {
 249                                 cc.ForwardSetState(SCE_C_DEFAULT);
 250                         } else if (cc.ch == '\\') {
 251                                 // Gobble up the quoted character
 252                                 if (cc.chNext == '\\' || cc.chNext == '/') {
 253                                         cc.Forward();
 254                                 }
 255                         }
 256                 } else if (cc.state == SCE_C_VERBATIM) {
 257                         if (cc.ch == '\"') {
 258                                 if (cc.chNext == '\"') {
 259                                         cc.Forward();
 260                                 } else {
 261                                         cc.ForwardSetState(SCE_C_DEFAULT);
 262                                 }
 263                         }
 264                 } else if (cc.state == SCE_C_UUID) {
 265                         if (cc.ch == '\r' || cc.ch == '\n' || cc.ch == ')') {
 266                                 cc.SetState(SCE_C_DEFAULT);
 267                         }
 268                 }
 269
 270                 if (cc.state == SCE_C_DEFAULT) {
 271                         if (cc.Match('@', '\"')) {
 272                                 cc.SetState(SCE_C_VERBATIM);
 273                                 cc.Forward();
 274                         } else if (IsADigit(cc.ch) || (cc.ch == '.' && IsADigit(cc.chNext))) {
 275                                 if (lastWordWasUUID) {
 276                                         cc.SetState(SCE_C_UUID);
 277                                         lastWordWasUUID = false;
 278                                 } else {
 279                                         cc.SetState(SCE_C_NUMBER);
 280                                 }
 281                         } else if (IsAWordStart(cc.ch) || (cc.ch == '@')) {
 282                                 if (lastWordWasUUID) {
 283                                         cc.SetState(SCE_C_UUID);
 284                                         lastWordWasUUID = false;
 285                                 } else {
 286                                         cc.SetState(SCE_C_IDENTIFIER);
 287                                 }
 288                         } else if (cc.Match('/', '*')) {
 289                                 if (cc.Match("/**") || cc.Match("/*!")) // Support of Qt/Doxygen doc. style
 290                                         cc.SetState(SCE_C_COMMENTDOC);
 291                                 else
 292                                         cc.SetState(SCE_C_COMMENT);
 293                                 cc.Forward();   // Eat the * so it isn't used for the end of the comment
 294                         } else if (cc.Match('/', '/')) {
 295                                 if (cc.Match("///") || cc.Match("//!")) // Support of Qt/Doxygen doc. style
 296                                         cc.SetState(SCE_C_COMMENTLINEDOC);
 297                                 else
 298                                         cc.SetState(SCE_C_COMMENTLINE);
 299                         } else if (cc.ch == '/' && IsOKBeforeRE(chPrevNonWhite)) {
 300                                 cc.SetState(SCE_C_REGEX);
 301                         } else if (cc.ch == '\"') {
 302                                 cc.SetState(SCE_C_STRING);
 303                         } else if (cc.ch == '\'') {
 304                                 cc.SetState(SCE_C_CHARACTER);
 305                         } else if (cc.ch == '#' && visibleChars == 0) {
 306                                 // Preprocessor commands are alone on their line
 307                                 cc.SetState(SCE_C_PREPROCESSOR);
 308                                 // Skip whitespace between # and preprocessor word
 309                                 do {
 310                                         cc.Forward();
 311                                 } while (IsASpace(cc.ch) && cc.More());
 312                         } else if (isoperator(static_cast<char>(cc.ch))) {
 313                                 cc.SetState(SCE_C_OPERATOR);
 314                         }
 315                 }
 316                 if (cc.atEOL) {
 317                         // Reset states to begining of colourise so no surprises
 318                         // if different sets of lines lexed.
 319                         chPrevNonWhite = ' ';
 320                         visibleChars = 0;
 321                         lastWordWasUUID = false;
 322                 }
 323                 if (!IsASpace(cc.ch)) {
 324                         chPrevNonWhite = cc.ch;
 325                         visibleChars++;
 326                 }
 327         }
 328         cc.Complete();
 329 }
 330
 331 static void FoldCppDoc(unsigned int startPos, int length, int initStyle, WordList *[],
 332                             Accessor &styler) {
 333         bool foldComment = styler.GetPropertyInt("fold.comment");
 334         bool foldCompact = styler.GetPropertyInt("fold.compact", 1);
 335         unsigned int endPos = startPos + length;
 336         int visibleChars = 0;
 337         int lineCurrent = styler.GetLine(startPos);
 338         int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
 339         int levelCurrent = levelPrev;
 340         char chNext = styler[startPos];
 341         int styleNext = styler.StyleAt(startPos);
 342         int style = initStyle;
 343         for (unsigned int i = startPos; i < endPos; i++) {
 344                 char ch = chNext;
 345                 chNext = styler.SafeGetCharAt(i + 1);
 346                 int stylePrev = style;
 347                 style = styleNext;
 348                 styleNext = styler.StyleAt(i + 1);
 349                 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
 350                 if (foldComment &&
 351                         (style == SCE_C_COMMENT || style == SCE_C_COMMENTDOC)) {
 352                         if (style != stylePrev) {
 353                                 levelCurrent++;
 354                         } else if ((style != styleNext) && !atEOL) {
 355                                 // Comments don't end at end of line and the next character may be unstyled.
 356                                 levelCurrent--;
 357                         }
 358                 }
 359                 if (style == SCE_C_OPERATOR) {
 360                         if (ch == '{') {
 361                                 levelCurrent++;
 362                         } else if (ch == '}') {
 363                                 levelCurrent--;
 364                         }
 365                 }
 366                 if (atEOL) {
 367                         int lev = levelPrev;
 368                         if (visibleChars == 0 && foldCompact)
 369                                 lev |= SC_FOLDLEVELWHITEFLAG;
 370                         if ((levelCurrent > levelPrev) && (visibleChars > 0))
 371                                 lev |= SC_FOLDLEVELHEADERFLAG;
 372                         if (lev != styler.LevelAt(lineCurrent)) {
 373                                 styler.SetLevel(lineCurrent, lev);
 374                         }
 375                         lineCurrent++;
 376                         levelPrev = levelCurrent;
 377                         visibleChars = 0;
 378                 }
 379                 if (!isspacechar(ch))
 380                         visibleChars++;
 381         }
 382         // Fill in the real level of the next line, keeping the current flags as they will be filled in later
 383         int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;
 384         styler.SetLevel(lineCurrent, levelPrev | flagsNext);
 385 }
 386
 387 LexerModule lmCPP(SCLEX_CPP, ColouriseCppDoc, "cpp", FoldCppDoc);
 388 LexerModule lmTCL(SCLEX_TCL, ColouriseCppDoc, "tcl", FoldCppDoc);