ext/scintilla/src/LexCSS.cxx

   1 // Scintilla source code edit control\r
   2 /** @file LexCSS.cxx\r
   3  ** Lexer for Cascading Style Sheets\r
   4  ** Written by Jakub Vrána\r
   5  ** Improved by Philippe Lhoste (CSS2)\r
   6  **/\r
   7 // Copyright 1998-2002 by Neil Hodgson <neilh@scintilla.org>\r
   8 // The License.txt file describes the conditions under which this software may be distributed.\r
   9 \r
  10 #include <stdlib.h>\r
  11 #include <string.h>\r
  12 #include <ctype.h>\r
  13 #include <stdio.h>\r
  14 #include <stdarg.h>\r
  15 \r
  16 #include "Platform.h"\r
  17 \r
  18 #include "PropSet.h"\r
  19 #include "Accessor.h"\r
  20 #include "StyleContext.h"\r
  21 #include "KeyWords.h"\r
  22 #include "Scintilla.h"\r
  23 #include "SciLexer.h"\r
  24 \r
  25 #ifdef SCI_NAMESPACE\r
  26 using namespace Scintilla;\r
  27 #endif\r
  28 \r
  29 \r
  30 static inline bool IsAWordChar(const unsigned int ch) {\r
  31         /* FIXME:\r
  32          * The CSS spec allows "ISO 10646 characters U+00A1 and higher" to be treated as word chars.\r
  33          * Unfortunately, we are only getting string bytes here, and not full unicode characters. We cannot guarantee\r
  34          * that our byte is between U+0080 - U+00A0 (to return false), so we have to allow all characters U+0080 and higher\r
  35          */\r
  36         return ch >= 0x80 || isalnum(ch) || ch == '-' || ch == '_';\r
  37 }\r
  38 \r
  39 inline bool IsCssOperator(const int ch) {\r
  40         if (!((ch < 0x80) && isalnum(ch)) &&\r
  41                 (ch == '{' || ch == '}' || ch == ':' || ch == ',' || ch == ';' ||\r
  42                  ch == '.' || ch == '#' || ch == '!' || ch == '@' ||\r
  43                  /* CSS2 */\r
  44                  ch == '*' || ch == '>' || ch == '+' || ch == '=' || ch == '~' || ch == '|' ||\r
  45                  ch == '[' || ch == ']' || ch == '(' || ch == ')')) {\r
  46                 return true;\r
  47         }\r
  48         return false;\r
  49 }\r
  50 \r
  51 static void ColouriseCssDoc(unsigned int startPos, int length, int initStyle, WordList *keywordlists[], Accessor &styler) {\r
  52         WordList &css1Props = *keywordlists[0];\r
  53         WordList &pseudoClasses = *keywordlists[1];\r
  54         WordList &css2Props = *keywordlists[2];\r
  55         WordList &css3Props = *keywordlists[3];\r
  56         WordList &pseudoElements = *keywordlists[4];\r
  57         WordList &exProps = *keywordlists[5];\r
  58         WordList &exPseudoClasses = *keywordlists[6];\r
  59         WordList &exPseudoElements = *keywordlists[7];\r
  60 \r
  61         StyleContext sc(startPos, length, initStyle, styler);\r
  62 \r
  63         int lastState = -1; // before operator\r
  64         int lastStateC = -1; // before comment\r
  65         int op = ' '; // last operator\r
  66         int opPrev = ' '; // last operator\r
  67 \r
  68         for (; sc.More(); sc.Forward()) {\r
  69                 if (sc.state == SCE_CSS_COMMENT && sc.Match('*', '/')) {\r
  70                         if (lastStateC == -1) {\r
  71                                 // backtrack to get last state:\r
  72                                 // comments are like whitespace, so we must return to the previous state\r
  73                                 unsigned int i = startPos;\r
  74                                 for (; i > 0; i--) {\r
  75                                         if ((lastStateC = styler.StyleAt(i-1)) != SCE_CSS_COMMENT) {\r
  76                                                 if (lastStateC == SCE_CSS_OPERATOR) {\r
  77                                                         op = styler.SafeGetCharAt(i-1);\r
  78                                                         opPrev = styler.SafeGetCharAt(i-2);\r
  79                                                         while (--i) {\r
  80                                                                 lastState = styler.StyleAt(i-1);\r
  81                                                                 if (lastState != SCE_CSS_OPERATOR && lastState != SCE_CSS_COMMENT)\r
  82                                                                         break;\r
  83                                                         }\r
  84                                                         if (i == 0)\r
  85                                                                 lastState = SCE_CSS_DEFAULT;\r
  86                                                 }\r
  87                                                 break;\r
  88                                         }\r
  89                                 }\r
  90                                 if (i == 0)\r
  91                                         lastStateC = SCE_CSS_DEFAULT;\r
  92                         }\r
  93                         sc.Forward();\r
  94                         sc.ForwardSetState(lastStateC);\r
  95                 }\r
  96 \r
  97                 if (sc.state == SCE_CSS_COMMENT)\r
  98                         continue;\r
  99 \r
 100                 if (sc.state == SCE_CSS_DOUBLESTRING || sc.state == SCE_CSS_SINGLESTRING) {\r
 101                         if (sc.ch != (sc.state == SCE_CSS_DOUBLESTRING ? '\"' : '\''))\r
 102                                 continue;\r
 103                         unsigned int i = sc.currentPos;\r
 104                         while (i && styler[i-1] == '\\')\r
 105                                 i--;\r
 106                         if ((sc.currentPos - i) % 2 == 1)\r
 107                                 continue;\r
 108                         sc.ForwardSetState(SCE_CSS_VALUE);\r
 109                 }\r
 110 \r
 111                 if (sc.state == SCE_CSS_OPERATOR) {\r
 112                         if (op == ' ') {\r
 113                                 unsigned int i = startPos;\r
 114                                 op = styler.SafeGetCharAt(i-1);\r
 115                                 opPrev = styler.SafeGetCharAt(i-2);\r
 116                                 while (--i) {\r
 117                                         lastState = styler.StyleAt(i-1);\r
 118                                         if (lastState != SCE_CSS_OPERATOR && lastState != SCE_CSS_COMMENT)\r
 119                                                 break;\r
 120                                 }\r
 121                         }\r
 122                         switch (op) {\r
 123                         case '@':\r
 124                                 if (lastState == SCE_CSS_DEFAULT)\r
 125                                         sc.SetState(SCE_CSS_DIRECTIVE);\r
 126                                 break;\r
 127                         case '>':\r
 128                         case '+':\r
 129                                 if (lastState == SCE_CSS_TAG || lastState == SCE_CSS_CLASS || lastState == SCE_CSS_ID ||\r
 130                                         lastState == SCE_CSS_PSEUDOCLASS || lastState == SCE_CSS_EXTENDED_PSEUDOCLASS || lastState == SCE_CSS_UNKNOWN_PSEUDOCLASS)\r
 131                                         sc.SetState(SCE_CSS_DEFAULT);\r
 132                                 break;\r
 133                         case '[':\r
 134                                 if (lastState == SCE_CSS_TAG || lastState == SCE_CSS_DEFAULT || lastState == SCE_CSS_CLASS || lastState == SCE_CSS_ID ||\r
 135                                         lastState == SCE_CSS_PSEUDOCLASS || lastState == SCE_CSS_EXTENDED_PSEUDOCLASS || lastState == SCE_CSS_UNKNOWN_PSEUDOCLASS)\r
 136                                         sc.SetState(SCE_CSS_ATTRIBUTE);\r
 137                                 break;\r
 138                         case ']':\r
 139                                 if (lastState == SCE_CSS_ATTRIBUTE)\r
 140                                         sc.SetState(SCE_CSS_TAG);\r
 141                                 break;\r
 142                         case '{':\r
 143                                 if (lastState == SCE_CSS_DIRECTIVE)\r
 144                                         sc.SetState(SCE_CSS_DEFAULT);\r
 145                                 else if (lastState == SCE_CSS_TAG)\r
 146                                         sc.SetState(SCE_CSS_IDENTIFIER);\r
 147                                 break;\r
 148                         case '}':\r
 149                                 if (lastState == SCE_CSS_DEFAULT || lastState == SCE_CSS_VALUE || lastState == SCE_CSS_IMPORTANT ||\r
 150                                         lastState == SCE_CSS_IDENTIFIER || lastState == SCE_CSS_IDENTIFIER2 || lastState == SCE_CSS_IDENTIFIER3)\r
 151                                         sc.SetState(SCE_CSS_DEFAULT);\r
 152                                 break;\r
 153                         case '(':\r
 154                                 if (lastState == SCE_CSS_PSEUDOCLASS)\r
 155                                         sc.SetState(SCE_CSS_TAG);\r
 156                                 else if (lastState == SCE_CSS_EXTENDED_PSEUDOCLASS)\r
 157                                         sc.SetState(SCE_CSS_EXTENDED_PSEUDOCLASS);\r
 158                                 break;\r
 159                         case ')':\r
 160                                 if (lastState == SCE_CSS_TAG || lastState == SCE_CSS_DEFAULT || lastState == SCE_CSS_CLASS || lastState == SCE_CSS_ID ||\r
 161                                         lastState == SCE_CSS_PSEUDOCLASS || lastState == SCE_CSS_EXTENDED_PSEUDOCLASS || lastState == SCE_CSS_UNKNOWN_PSEUDOCLASS ||\r
 162                                         lastState == SCE_CSS_PSEUDOELEMENT || lastState == SCE_CSS_EXTENDED_PSEUDOELEMENT)\r
 163                                         sc.SetState(SCE_CSS_TAG);\r
 164                                 break;\r
 165                         case ':':\r
 166                                 if (lastState == SCE_CSS_TAG || lastState == SCE_CSS_DEFAULT || lastState == SCE_CSS_CLASS || lastState == SCE_CSS_ID ||\r
 167                                         lastState == SCE_CSS_PSEUDOCLASS || lastState == SCE_CSS_EXTENDED_PSEUDOCLASS || lastState == SCE_CSS_UNKNOWN_PSEUDOCLASS ||\r
 168                                         lastState == SCE_CSS_PSEUDOELEMENT || lastState == SCE_CSS_EXTENDED_PSEUDOELEMENT)\r
 169                                         sc.SetState(SCE_CSS_PSEUDOCLASS);\r
 170                                 else if (lastState == SCE_CSS_IDENTIFIER || lastState == SCE_CSS_IDENTIFIER2 ||\r
 171                                         lastState == SCE_CSS_IDENTIFIER3 || lastState == SCE_CSS_EXTENDED_IDENTIFIER ||\r
 172                                         lastState == SCE_CSS_UNKNOWN_IDENTIFIER)\r
 173                                         sc.SetState(SCE_CSS_VALUE);\r
 174                                 break;\r
 175                         case '.':\r
 176                                 if (lastState == SCE_CSS_TAG || lastState == SCE_CSS_DEFAULT || lastState == SCE_CSS_CLASS || lastState == SCE_CSS_ID ||\r
 177                                         lastState == SCE_CSS_PSEUDOCLASS || lastState == SCE_CSS_EXTENDED_PSEUDOCLASS || lastState == SCE_CSS_UNKNOWN_PSEUDOCLASS)\r
 178                                         sc.SetState(SCE_CSS_CLASS);\r
 179                                 break;\r
 180                         case '#':\r
 181                                 if (lastState == SCE_CSS_TAG || lastState == SCE_CSS_DEFAULT || lastState == SCE_CSS_CLASS || lastState == SCE_CSS_ID ||\r
 182                                         lastState == SCE_CSS_PSEUDOCLASS || lastState == SCE_CSS_EXTENDED_PSEUDOCLASS || lastState == SCE_CSS_UNKNOWN_PSEUDOCLASS)\r
 183                                         sc.SetState(SCE_CSS_ID);\r
 184                                 break;\r
 185                         case ',':\r
 186                         case '|':\r
 187                         case '~':\r
 188                                 if (lastState == SCE_CSS_TAG)\r
 189                                         sc.SetState(SCE_CSS_DEFAULT);\r
 190                                 break;\r
 191                         case ';':\r
 192                                 if (lastState == SCE_CSS_DIRECTIVE)\r
 193                                         sc.SetState(SCE_CSS_DEFAULT);\r
 194                                 else if (lastState == SCE_CSS_VALUE || lastState == SCE_CSS_IMPORTANT)\r
 195                                         sc.SetState(SCE_CSS_IDENTIFIER);\r
 196                                 break;\r
 197                         case '!':\r
 198                                 if (lastState == SCE_CSS_VALUE)\r
 199                                         sc.SetState(SCE_CSS_IMPORTANT);\r
 200                                 break;\r
 201                         }\r
 202                 }\r
 203 \r
 204                 if (IsAWordChar(sc.ch)) {\r
 205                         if (sc.state == SCE_CSS_DEFAULT)\r
 206                                 sc.SetState(SCE_CSS_TAG);\r
 207                         continue;\r
 208                 }\r
 209 \r
 210                 if (sc.ch == '*' && sc.state == SCE_CSS_DEFAULT) {\r
 211                         sc.SetState(SCE_CSS_TAG);\r
 212                         continue;\r
 213                 }\r
 214 \r
 215                 if (IsAWordChar(sc.chPrev) && (\r
 216                         sc.state == SCE_CSS_IDENTIFIER || sc.state == SCE_CSS_IDENTIFIER2 ||\r
 217                         sc.state == SCE_CSS_IDENTIFIER3 || sc.state == SCE_CSS_EXTENDED_IDENTIFIER ||\r
 218                         sc.state == SCE_CSS_UNKNOWN_IDENTIFIER ||\r
 219                         sc.state == SCE_CSS_PSEUDOCLASS || sc.state == SCE_CSS_PSEUDOELEMENT ||\r
 220                         sc.state == SCE_CSS_EXTENDED_PSEUDOCLASS || sc.state == SCE_CSS_EXTENDED_PSEUDOELEMENT ||\r
 221                         sc.state == SCE_CSS_UNKNOWN_PSEUDOCLASS ||\r
 222                         sc.state == SCE_CSS_IMPORTANT\r
 223                 )) {\r
 224                         char s[100];\r
 225                         sc.GetCurrentLowered(s, sizeof(s));\r
 226                         char *s2 = s;\r
 227                         while (*s2 && !IsAWordChar(*s2))\r
 228                                 s2++;\r
 229                         switch (sc.state) {\r
 230                         case SCE_CSS_IDENTIFIER:\r
 231                         case SCE_CSS_IDENTIFIER2:\r
 232                         case SCE_CSS_IDENTIFIER3:\r
 233                         case SCE_CSS_EXTENDED_IDENTIFIER:\r
 234                         case SCE_CSS_UNKNOWN_IDENTIFIER:\r
 235                                 if (css1Props.InList(s2))\r
 236                                         sc.ChangeState(SCE_CSS_IDENTIFIER);\r
 237                                 else if (css2Props.InList(s2))\r
 238                                         sc.ChangeState(SCE_CSS_IDENTIFIER2);\r
 239                                 else if (css3Props.InList(s2))\r
 240                                         sc.ChangeState(SCE_CSS_IDENTIFIER3);\r
 241                                 else if (exProps.InList(s2))\r
 242                                         sc.ChangeState(SCE_CSS_EXTENDED_IDENTIFIER);\r
 243                                 else\r
 244                                         sc.ChangeState(SCE_CSS_UNKNOWN_IDENTIFIER);\r
 245                                 break;\r
 246                         case SCE_CSS_PSEUDOCLASS:\r
 247                         case SCE_CSS_PSEUDOELEMENT:\r
 248                         case SCE_CSS_EXTENDED_PSEUDOCLASS:\r
 249                         case SCE_CSS_EXTENDED_PSEUDOELEMENT:\r
 250                         case SCE_CSS_UNKNOWN_PSEUDOCLASS:\r
 251                                 if (op == ':' && opPrev != ':' && pseudoClasses.InList(s2))\r
 252                                         sc.ChangeState(SCE_CSS_PSEUDOCLASS);\r
 253                                 else if (opPrev == ':' && pseudoElements.InList(s2))\r
 254                                         sc.ChangeState(SCE_CSS_PSEUDOELEMENT);\r
 255                                 else if ((op == ':' || (op == '(' && lastState == SCE_CSS_EXTENDED_PSEUDOCLASS)) && opPrev != ':' && exPseudoClasses.InList(s2))\r
 256                                         sc.ChangeState(SCE_CSS_EXTENDED_PSEUDOCLASS);\r
 257                                 else if (opPrev == ':' && exPseudoElements.InList(s2))\r
 258                                         sc.ChangeState(SCE_CSS_EXTENDED_PSEUDOELEMENT);\r
 259                                 else\r
 260                                         sc.ChangeState(SCE_CSS_UNKNOWN_PSEUDOCLASS);\r
 261                                 break;\r
 262                         case SCE_CSS_IMPORTANT:\r
 263                                 if (strcmp(s2, "important") != 0)\r
 264                                         sc.ChangeState(SCE_CSS_VALUE);\r
 265                                 break;\r
 266                         }\r
 267                 }\r
 268 \r
 269                 if (sc.ch != '.' && sc.ch != ':' && sc.ch != '#' && (\r
 270                         sc.state == SCE_CSS_CLASS || sc.state == SCE_CSS_ID ||\r
 271                         (sc.ch != '(' && sc.ch != ')' && ( /* This line of the condition makes it possible to extend pseudo-classes with parentheses */\r
 272                                 sc.state == SCE_CSS_PSEUDOCLASS || sc.state == SCE_CSS_PSEUDOELEMENT ||\r
 273                                 sc.state == SCE_CSS_EXTENDED_PSEUDOCLASS || sc.state == SCE_CSS_EXTENDED_PSEUDOELEMENT ||\r
 274                                 sc.state == SCE_CSS_UNKNOWN_PSEUDOCLASS\r
 275                         ))\r
 276                 ))\r
 277                         sc.SetState(SCE_CSS_TAG);\r
 278 \r
 279                 if (sc.Match('/', '*')) {\r
 280                         lastStateC = sc.state;\r
 281                         sc.SetState(SCE_CSS_COMMENT);\r
 282                         sc.Forward();\r
 283                 } else if (sc.state == SCE_CSS_VALUE && (sc.ch == '\"' || sc.ch == '\'')) {\r
 284                         sc.SetState((sc.ch == '\"' ? SCE_CSS_DOUBLESTRING : SCE_CSS_SINGLESTRING));\r
 285                 } else if (IsCssOperator(sc.ch)\r
 286                         && (sc.state != SCE_CSS_ATTRIBUTE || sc.ch == ']')\r
 287                         && (sc.state != SCE_CSS_VALUE || sc.ch == ';' || sc.ch == '}' || sc.ch == '!')\r
 288                         && (sc.state != SCE_CSS_DIRECTIVE || sc.ch == ';' || sc.ch == '{')\r
 289                 ) {\r
 290                         if (sc.state != SCE_CSS_OPERATOR)\r
 291                                 lastState = sc.state;\r
 292                         sc.SetState(SCE_CSS_OPERATOR);\r
 293                         op = sc.ch;\r
 294                         opPrev = sc.chPrev;\r
 295                 }\r
 296         }\r
 297 \r
 298         sc.Complete();\r
 299 }\r
 300 \r
 301 static void FoldCSSDoc(unsigned int startPos, int length, int, WordList *[], Accessor &styler) {\r
 302         bool foldComment = styler.GetPropertyInt("fold.comment") != 0;\r
 303         bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;\r
 304         unsigned int endPos = startPos + length;\r
 305         int visibleChars = 0;\r
 306         int lineCurrent = styler.GetLine(startPos);\r
 307         int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;\r
 308         int levelCurrent = levelPrev;\r
 309         char chNext = styler[startPos];\r
 310         bool inComment = (styler.StyleAt(startPos-1) == SCE_CSS_COMMENT);\r
 311         for (unsigned int i = startPos; i < endPos; i++) {\r
 312                 char ch = chNext;\r
 313                 chNext = styler.SafeGetCharAt(i + 1);\r
 314                 int style = styler.StyleAt(i);\r
 315                 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');\r
 316                 if (foldComment) {\r
 317                         if (!inComment && (style == SCE_CSS_COMMENT))\r
 318                                 levelCurrent++;\r
 319                         else if (inComment && (style != SCE_CSS_COMMENT))\r
 320                                 levelCurrent--;\r
 321                         inComment = (style == SCE_CSS_COMMENT);\r
 322                 }\r
 323                 if (style == SCE_CSS_OPERATOR) {\r
 324                         if (ch == '{') {\r
 325                                 levelCurrent++;\r
 326                         } else if (ch == '}') {\r
 327                                 levelCurrent--;\r
 328                         }\r
 329                 }\r
 330                 if (atEOL) {\r
 331                         int lev = levelPrev;\r
 332                         if (visibleChars == 0 && foldCompact)\r
 333                                 lev |= SC_FOLDLEVELWHITEFLAG;\r
 334                         if ((levelCurrent > levelPrev) && (visibleChars > 0))\r
 335                                 lev |= SC_FOLDLEVELHEADERFLAG;\r
 336                         if (lev != styler.LevelAt(lineCurrent)) {\r
 337                                 styler.SetLevel(lineCurrent, lev);\r
 338                         }\r
 339                         lineCurrent++;\r
 340                         levelPrev = levelCurrent;\r
 341                         visibleChars = 0;\r
 342                 }\r
 343                 if (!isspacechar(ch))\r
 344                         visibleChars++;\r
 345         }\r
 346         // Fill in the real level of the next line, keeping the current flags as they will be filled in later\r
 347         int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;\r
 348         styler.SetLevel(lineCurrent, levelPrev | flagsNext);\r
 349 }\r
 350 \r
 351 static const char * const cssWordListDesc[] = {\r
 352         "CSS1 Properties",\r
 353         "Pseudo-classes",\r
 354         "CSS2 Properties",\r
 355         "CSS3 Properties",\r
 356         "Pseudo-elements",\r
 357         "Browser-Specific CSS Properties",\r
 358         "Browser-Specific Pseudo-classes",\r
 359         "Browser-Specific Pseudo-elements",\r
 360         0\r
 361 };\r
 362 \r
 363 LexerModule lmCss(SCLEX_CSS, ColouriseCssDoc, "css", FoldCSSDoc, cssWordListDesc);\r