contrib/src/stc/scintilla/src/LexRebol.cxx

   1 // Scintilla source code edit control
   2 /** @file LexRebol.cxx
   3  ** Lexer for REBOL.
   4  ** Written by Pascal Hurni, inspired from LexLua by Paul Winwood & Marcos E. Wurzius & Philippe Lhoste
   5  **
   6  ** History:
   7  **             2005-04-07      First release.
   8  **             2005-04-10      Closing parens and brackets go now in default style
   9  **                                     String and comment nesting should be more safe
  10  **/
  11 // Copyright 2005 by Pascal Hurni <pascal_hurni@fastmail.fm>
  12 // The License.txt file describes the conditions under which this software may be distributed.
  13
  14 #include <stdlib.h>
  15 #include <string.h>
  16 #include <ctype.h>
  17 #include <stdio.h>
  18 #include <stdarg.h>
  19
  20 #include "Platform.h"
  21
  22 #include "PropSet.h"
  23 #include "Accessor.h"
  24 #include "KeyWords.h"
  25 #include "Scintilla.h"
  26 #include "SciLexer.h"
  27 #include "StyleContext.h"
  28
  29
  30 static inline bool IsAWordChar(const int ch) {
  31         return (isalnum(ch) || ch == '?' || ch == '!' || ch == '.' || ch == '\'' || ch == '+' || ch == '-' || ch == '*' || ch == '&' || ch == '|' || ch == '=' || ch == '_' || ch == '~');
  32 }
  33
  34 static inline bool IsAWordStart(const int ch, const int ch2) {
  35         return ((ch == '+' || ch == '-' || ch == '.') && !isdigit(ch2)) ||
  36                 (isalpha(ch) || ch == '?' || ch == '!' || ch == '\'' || ch == '*' || ch == '&' || ch == '|' || ch == '=' || ch == '_' || ch == '~');
  37 }
  38
  39 static inline bool IsAnOperator(const int ch, const int ch2, const int ch3) {
  40         // One char operators
  41         if (IsASpaceOrTab(ch2)) {
  42                 return ch == '+' || ch == '-' || ch == '*' || ch == '/' || ch == '<' || ch == '>' || ch == '=' || ch == '?';
  43         }
  44
  45         // Two char operators
  46         if (IsASpaceOrTab(ch3)) {
  47                 return (ch == '*' && ch2 == '*') ||
  48                            (ch == '/' && ch2 == '/') ||
  49                            (ch == '<' && (ch2 == '=' || ch2 == '>')) ||
  50                            (ch == '>' && ch2 == '=') ||
  51                            (ch == '=' && (ch2 == '=' || ch2 == '?')) ||
  52                            (ch == '?' && ch2 == '?');
  53         }
  54
  55         return false;
  56 }
  57
  58 static inline bool IsBinaryStart(const int ch, const int ch2, const int ch3, const int ch4) {
  59         return (ch == '#' && ch2 == '{') ||
  60                    (IsADigit(ch) && ch2 == '#' && ch3 == '{' ) ||
  61                    (IsADigit(ch) && IsADigit(ch2) && ch3 == '#' && ch4 == '{' );
  62 }
  63
  64
  65 static void ColouriseRebolDoc(unsigned int startPos, int length, int initStyle, WordList *keywordlists[], Accessor &styler) {
  66
  67         WordList &keywords = *keywordlists[0];
  68         WordList &keywords2 = *keywordlists[1];
  69         WordList &keywords3 = *keywordlists[2];
  70         WordList &keywords4 = *keywordlists[3];
  71         WordList &keywords5 = *keywordlists[4];
  72         WordList &keywords6 = *keywordlists[5];
  73         WordList &keywords7 = *keywordlists[6];
  74         WordList &keywords8 = *keywordlists[7];
  75
  76         int currentLine = styler.GetLine(startPos);
  77         // Initialize the braced string {.. { ... } ..} nesting level, if we are inside such a string.
  78         int stringLevel = 0;
  79         if (initStyle == SCE_REBOL_BRACEDSTRING || initStyle == SCE_REBOL_COMMENTBLOCK) {
  80                 stringLevel = styler.GetLineState(currentLine - 1);
  81         }
  82
  83         bool blockComment = initStyle == SCE_REBOL_COMMENTBLOCK;
  84         int dotCount = 0;
  85
  86         // Do not leak onto next line
  87         if (initStyle == SCE_REBOL_COMMENTLINE) {
  88                 initStyle = SCE_REBOL_DEFAULT;
  89         }
  90
  91         StyleContext sc(startPos, length, initStyle, styler);
  92         if (startPos == 0) {
  93                 sc.SetState(SCE_REBOL_PREFACE);
  94         }
  95         for (; sc.More(); sc.Forward()) {
  96
  97                 //--- What to do at line end ?
  98                 if (sc.atLineEnd) {
  99                         // Can be either inside a {} string or simply at eol
 100                         if (sc.state != SCE_REBOL_BRACEDSTRING && sc.state != SCE_REBOL_COMMENTBLOCK &&
 101                                 sc.state != SCE_REBOL_BINARY && sc.state != SCE_REBOL_PREFACE)
 102                                 sc.SetState(SCE_REBOL_DEFAULT);
 103
 104                         // Update the line state, so it can be seen by next line
 105                         currentLine = styler.GetLine(sc.currentPos);
 106                         switch (sc.state) {
 107                         case SCE_REBOL_BRACEDSTRING:
 108                         case SCE_REBOL_COMMENTBLOCK:
 109                                 // Inside a braced string, we set the line state
 110                                 styler.SetLineState(currentLine, stringLevel);
 111                                 break;
 112                         default:
 113                                 // Reset the line state
 114                                 styler.SetLineState(currentLine, 0);
 115                                 break;
 116                         }
 117
 118                         // continue with next char
 119                         continue;
 120                 }
 121
 122                 //--- What to do on white-space ?
 123                 if (IsASpaceOrTab(sc.ch))
 124                 {
 125                         // Return to default if any of these states
 126                         if (sc.state == SCE_REBOL_OPERATOR || sc.state == SCE_REBOL_CHARACTER ||
 127                                 sc.state == SCE_REBOL_NUMBER || sc.state == SCE_REBOL_PAIR ||
 128                                 sc.state == SCE_REBOL_TUPLE || sc.state == SCE_REBOL_FILE ||
 129                                 sc.state == SCE_REBOL_DATE || sc.state == SCE_REBOL_TIME ||
 130                                 sc.state == SCE_REBOL_MONEY || sc.state == SCE_REBOL_ISSUE ||
 131                                 sc.state == SCE_REBOL_URL || sc.state == SCE_REBOL_EMAIL) {
 132                                 sc.SetState(SCE_REBOL_DEFAULT);
 133                         }
 134                 }
 135
 136                 //--- Specialize state ?
 137                 // URL, Email look like identifier
 138                 if (sc.state == SCE_REBOL_IDENTIFIER)
 139                 {
 140                         if (sc.ch == ':' && !IsASpace(sc.chNext)) {
 141                                 sc.ChangeState(SCE_REBOL_URL);
 142                         } else if (sc.ch == '@') {
 143                                 sc.ChangeState(SCE_REBOL_EMAIL);
 144                         } else if (sc.ch == '$') {
 145                                 sc.ChangeState(SCE_REBOL_MONEY);
 146                         }
 147                 }
 148                 // Words look like identifiers
 149                 if (sc.state == SCE_REBOL_IDENTIFIER || (sc.state >= SCE_REBOL_WORD && sc.state <= SCE_REBOL_WORD8)) {
 150                         // Keywords ?
 151                         if (!IsAWordChar(sc.ch) || sc.Match('/')) {
 152                                 char s[100];
 153                                 sc.GetCurrentLowered(s, sizeof(s));
 154                                 blockComment = strcmp(s, "comment") == 0;
 155                                 if (keywords8.InList(s)) {
 156                                         sc.ChangeState(SCE_REBOL_WORD8);
 157                                 } else if (keywords7.InList(s)) {
 158                                         sc.ChangeState(SCE_REBOL_WORD7);
 159                                 } else if (keywords6.InList(s)) {
 160                                         sc.ChangeState(SCE_REBOL_WORD6);
 161                                 } else if (keywords5.InList(s)) {
 162                                         sc.ChangeState(SCE_REBOL_WORD5);
 163                                 } else if (keywords4.InList(s)) {
 164                                         sc.ChangeState(SCE_REBOL_WORD4);
 165                                 } else if (keywords3.InList(s)) {
 166                                         sc.ChangeState(SCE_REBOL_WORD3);
 167                                 } else if (keywords2.InList(s)) {
 168                                         sc.ChangeState(SCE_REBOL_WORD2);
 169                                 } else if (keywords.InList(s)) {
 170                                         sc.ChangeState(SCE_REBOL_WORD);
 171                                 }
 172                                 // Keep same style if there are refinements
 173                                 if (!sc.Match('/')) {
 174                                         sc.SetState(SCE_REBOL_DEFAULT);
 175                                 }
 176                         }
 177                 // special numbers
 178                 } else if (sc.state == SCE_REBOL_NUMBER) {
 179                         switch (sc.ch) {
 180                         case 'x':       sc.ChangeState(SCE_REBOL_PAIR);
 181                                                 break;
 182                         case ':':       sc.ChangeState(SCE_REBOL_TIME);
 183                                                 break;
 184                         case '-':
 185                         case '/':       sc.ChangeState(SCE_REBOL_DATE);
 186                                                 break;
 187                         case '.':       if (++dotCount >= 2) sc.ChangeState(SCE_REBOL_TUPLE);
 188                                                 break;
 189                         }
 190                 }
 191
 192                 //--- Determine if the current state should terminate
 193                 if (sc.state == SCE_REBOL_QUOTEDSTRING || sc.state == SCE_REBOL_CHARACTER) {
 194                         if (sc.ch == '^' && sc.chNext == '\"') {
 195                                 sc.Forward();
 196                         } else if (sc.ch == '\"') {
 197                                 sc.ForwardSetState(SCE_REBOL_DEFAULT);
 198                         }
 199                 } else if (sc.state == SCE_REBOL_BRACEDSTRING || sc.state == SCE_REBOL_COMMENTBLOCK) {
 200                         if (sc.ch == '}') {
 201                                 if (--stringLevel == 0) {
 202                                         sc.ForwardSetState(SCE_REBOL_DEFAULT);
 203                                 }
 204                         } else if (sc.ch == '{') {
 205                                 stringLevel++;
 206                         }
 207                 } else if (sc.state == SCE_REBOL_BINARY) {
 208                         if (sc.ch == '}') {
 209                                 sc.ForwardSetState(SCE_REBOL_DEFAULT);
 210                         }
 211                 } else if (sc.state == SCE_REBOL_TAG) {
 212                         if (sc.ch == '>') {
 213                                 sc.ForwardSetState(SCE_REBOL_DEFAULT);
 214                         }
 215                 } else if (sc.state == SCE_REBOL_PREFACE) {
 216                         if (sc.MatchIgnoreCase("rebol"))
 217                         {
 218                                 int i;
 219                                 for (i=5; IsASpaceOrTab(styler.SafeGetCharAt(sc.currentPos+i, 0)); i++);
 220                                 if (sc.GetRelative(i) == '[')
 221                                         sc.SetState(SCE_REBOL_DEFAULT);
 222                         }
 223                 }
 224
 225                 //--- Parens and bracket changes to default style when the current is a number
 226                 if (sc.state == SCE_REBOL_NUMBER || sc.state == SCE_REBOL_PAIR || sc.state == SCE_REBOL_TUPLE ||
 227                         sc.state == SCE_REBOL_MONEY || sc.state == SCE_REBOL_ISSUE || sc.state == SCE_REBOL_EMAIL ||
 228                         sc.state == SCE_REBOL_URL || sc.state == SCE_REBOL_DATE || sc.state == SCE_REBOL_TIME) {
 229                         if (sc.ch == '(' || sc.ch == '[' || sc.ch == ')' || sc.ch == ']') {
 230                                 sc.SetState(SCE_REBOL_DEFAULT);
 231                         }
 232                 }
 233
 234                 //--- Determine if a new state should be entered.
 235                 if (sc.state == SCE_REBOL_DEFAULT) {
 236                         if (IsAnOperator(sc.ch, sc.chNext, sc.GetRelative(2))) {
 237                                 sc.SetState(SCE_REBOL_OPERATOR);
 238                         } else if (IsBinaryStart(sc.ch, sc.chNext, sc.GetRelative(2), sc.GetRelative(3))) {
 239                                 sc.SetState(SCE_REBOL_BINARY);
 240                         } else if (IsAWordStart(sc.ch, sc.chNext)) {
 241                                 sc.SetState(SCE_REBOL_IDENTIFIER);
 242                         } else if (IsADigit(sc.ch) || sc.ch == '+' || sc.ch == '-' || /*Decimal*/ sc.ch == '.' || sc.ch == ',') {
 243                                 dotCount = 0;
 244                                 sc.SetState(SCE_REBOL_NUMBER);
 245                         } else if (sc.ch == '\"') {
 246                                 sc.SetState(SCE_REBOL_QUOTEDSTRING);
 247                         } else if (sc.ch == '{') {
 248                                 sc.SetState(blockComment ? SCE_REBOL_COMMENTBLOCK : SCE_REBOL_BRACEDSTRING);
 249                                 ++stringLevel;
 250                         } else if (sc.ch == ';') {
 251                                 sc.SetState(SCE_REBOL_COMMENTLINE);
 252                         } else if (sc.ch == '$') {
 253                                 sc.SetState(SCE_REBOL_MONEY);
 254                         } else if (sc.ch == '%') {
 255                                 sc.SetState(SCE_REBOL_FILE);
 256                         } else if (sc.ch == '<') {
 257                                 sc.SetState(SCE_REBOL_TAG);
 258                         } else if (sc.ch == '#' && sc.chNext == '"') {
 259                                 sc.SetState(SCE_REBOL_CHARACTER);
 260                                 sc.Forward();
 261                         } else if (sc.ch == '#' && sc.chNext != '"' && sc.chNext != '{' ) {
 262                                 sc.SetState(SCE_REBOL_ISSUE);
 263                         }
 264                 }
 265         }
 266         sc.Complete();
 267 }
 268
 269
 270 static void FoldRebolDoc(unsigned int startPos, int length, int /* initStyle */, WordList *[],
 271                             Accessor &styler) {
 272         unsigned int lengthDoc = startPos + length;
 273         int visibleChars = 0;
 274         int lineCurrent = styler.GetLine(startPos);
 275         int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
 276         int levelCurrent = levelPrev;
 277         char chNext = styler[startPos];
 278         int styleNext = styler.StyleAt(startPos);
 279         for (unsigned int i = startPos; i < lengthDoc; i++) {
 280                 char ch = chNext;
 281                 chNext = styler.SafeGetCharAt(i + 1);
 282                 int style = styleNext;
 283                 styleNext = styler.StyleAt(i + 1);
 284                 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
 285                 if (style == SCE_REBOL_DEFAULT) {
 286                         if (ch == '[') {
 287                                 levelCurrent++;
 288                         } else if (ch == ']') {
 289                                 levelCurrent--;
 290                         }
 291                 }
 292                 if (atEOL) {
 293                         int lev = levelPrev;
 294                         if (visibleChars == 0)
 295                                 lev |= SC_FOLDLEVELWHITEFLAG;
 296                         if ((levelCurrent > levelPrev) && (visibleChars > 0))
 297                                 lev |= SC_FOLDLEVELHEADERFLAG;
 298                         if (lev != styler.LevelAt(lineCurrent)) {
 299                                 styler.SetLevel(lineCurrent, lev);
 300                         }
 301                         lineCurrent++;
 302                         levelPrev = levelCurrent;
 303                         visibleChars = 0;
 304                 }
 305                 if (!isspacechar(ch))
 306                         visibleChars++;
 307         }
 308         // Fill in the real level of the next line, keeping the current flags as they will be filled in later
 309         int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;
 310         styler.SetLevel(lineCurrent, levelPrev | flagsNext);
 311 }
 312
 313 static const char * const rebolWordListDesc[] = {
 314         "Keywords",
 315         0
 316 };
 317
 318 LexerModule lmREBOL(SCLEX_REBOL, ColouriseRebolDoc, "rebol", FoldRebolDoc, rebolWordListDesc);
 319