ext/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 #ifdef SCI_NAMESPACE
  30 using namespace Scintilla;
  31 #endif
  32
  33 static inline bool IsAWordChar(const int ch) {
  34         return (isalnum(ch) || ch == '?' || ch == '!' || ch == '.' || ch == '\'' || ch == '+' || ch == '-' || ch == '*' || ch == '&' || ch == '|' || ch == '=' || ch == '_' || ch == '~');
  35 }
  36
  37 static inline bool IsAWordStart(const int ch, const int ch2) {
  38         return ((ch == '+' || ch == '-' || ch == '.') && !isdigit(ch2)) ||
  39                 (isalpha(ch) || ch == '?' || ch == '!' || ch == '\'' || ch == '*' || ch == '&' || ch == '|' || ch == '=' || ch == '_' || ch == '~');
  40 }
  41
  42 static inline bool IsAnOperator(const int ch, const int ch2, const int ch3) {
  43         // One char operators
  44         if (IsASpaceOrTab(ch2)) {
  45                 return ch == '+' || ch == '-' || ch == '*' || ch == '/' || ch == '<' || ch == '>' || ch == '=' || ch == '?';
  46         }
  47
  48         // Two char operators
  49         if (IsASpaceOrTab(ch3)) {
  50                 return (ch == '*' && ch2 == '*') ||
  51                            (ch == '/' && ch2 == '/') ||
  52                            (ch == '<' && (ch2 == '=' || ch2 == '>')) ||
  53                            (ch == '>' && ch2 == '=') ||
  54                            (ch == '=' && (ch2 == '=' || ch2 == '?')) ||
  55                            (ch == '?' && ch2 == '?');
  56         }
  57
  58         return false;
  59 }
  60
  61 static inline bool IsBinaryStart(const int ch, const int ch2, const int ch3, const int ch4) {
  62         return (ch == '#' && ch2 == '{') ||
  63                    (IsADigit(ch) && ch2 == '#' && ch3 == '{' ) ||
  64                    (IsADigit(ch) && IsADigit(ch2) && ch3 == '#' && ch4 == '{' );
  65 }
  66
  67
  68 static void ColouriseRebolDoc(unsigned int startPos, int length, int initStyle, WordList *keywordlists[], Accessor &styler) {
  69
  70         WordList &keywords = *keywordlists[0];
  71         WordList &keywords2 = *keywordlists[1];
  72         WordList &keywords3 = *keywordlists[2];
  73         WordList &keywords4 = *keywordlists[3];
  74         WordList &keywords5 = *keywordlists[4];
  75         WordList &keywords6 = *keywordlists[5];
  76         WordList &keywords7 = *keywordlists[6];
  77         WordList &keywords8 = *keywordlists[7];
  78
  79         int currentLine = styler.GetLine(startPos);
  80         // Initialize the braced string {.. { ... } ..} nesting level, if we are inside such a string.
  81         int stringLevel = 0;
  82         if (initStyle == SCE_REBOL_BRACEDSTRING || initStyle == SCE_REBOL_COMMENTBLOCK) {
  83                 stringLevel = styler.GetLineState(currentLine - 1);
  84         }
  85
  86         bool blockComment = initStyle == SCE_REBOL_COMMENTBLOCK;
  87         int dotCount = 0;
  88
  89         // Do not leak onto next line
  90         if (initStyle == SCE_REBOL_COMMENTLINE) {
  91                 initStyle = SCE_REBOL_DEFAULT;
  92         }
  93
  94         StyleContext sc(startPos, length, initStyle, styler);
  95         if (startPos == 0) {
  96                 sc.SetState(SCE_REBOL_PREFACE);
  97         }
  98         for (; sc.More(); sc.Forward()) {
  99
 100                 //--- What to do at line end ?
 101                 if (sc.atLineEnd) {
 102                         // Can be either inside a {} string or simply at eol
 103                         if (sc.state != SCE_REBOL_BRACEDSTRING && sc.state != SCE_REBOL_COMMENTBLOCK &&
 104                                 sc.state != SCE_REBOL_BINARY && sc.state != SCE_REBOL_PREFACE)
 105                                 sc.SetState(SCE_REBOL_DEFAULT);
 106
 107                         // Update the line state, so it can be seen by next line
 108                         currentLine = styler.GetLine(sc.currentPos);
 109                         switch (sc.state) {
 110                         case SCE_REBOL_BRACEDSTRING:
 111                         case SCE_REBOL_COMMENTBLOCK:
 112                                 // Inside a braced string, we set the line state
 113                                 styler.SetLineState(currentLine, stringLevel);
 114                                 break;
 115                         default:
 116                                 // Reset the line state
 117                                 styler.SetLineState(currentLine, 0);
 118                                 break;
 119                         }
 120
 121                         // continue with next char
 122                         continue;
 123                 }
 124
 125                 //--- What to do on white-space ?
 126                 if (IsASpaceOrTab(sc.ch))
 127                 {
 128                         // Return to default if any of these states
 129                         if (sc.state == SCE_REBOL_OPERATOR || sc.state == SCE_REBOL_CHARACTER ||
 130                                 sc.state == SCE_REBOL_NUMBER || sc.state == SCE_REBOL_PAIR ||
 131                                 sc.state == SCE_REBOL_TUPLE || sc.state == SCE_REBOL_FILE ||
 132                                 sc.state == SCE_REBOL_DATE || sc.state == SCE_REBOL_TIME ||
 133                                 sc.state == SCE_REBOL_MONEY || sc.state == SCE_REBOL_ISSUE ||
 134                                 sc.state == SCE_REBOL_URL || sc.state == SCE_REBOL_EMAIL) {
 135                                 sc.SetState(SCE_REBOL_DEFAULT);
 136                         }
 137                 }
 138
 139                 //--- Specialize state ?
 140                 // URL, Email look like identifier
 141                 if (sc.state == SCE_REBOL_IDENTIFIER)
 142                 {
 143                         if (sc.ch == ':' && !IsASpace(sc.chNext)) {
 144                                 sc.ChangeState(SCE_REBOL_URL);
 145                         } else if (sc.ch == '@') {
 146                                 sc.ChangeState(SCE_REBOL_EMAIL);
 147                         } else if (sc.ch == '$') {
 148                                 sc.ChangeState(SCE_REBOL_MONEY);
 149                         }
 150                 }
 151                 // Words look like identifiers
 152                 if (sc.state == SCE_REBOL_IDENTIFIER || (sc.state >= SCE_REBOL_WORD && sc.state <= SCE_REBOL_WORD8)) {
 153                         // Keywords ?
 154                         if (!IsAWordChar(sc.ch) || sc.Match('/')) {
 155                                 char s[100];
 156                                 sc.GetCurrentLowered(s, sizeof(s));
 157                                 blockComment = strcmp(s, "comment") == 0;
 158                                 if (keywords8.InList(s)) {
 159                                         sc.ChangeState(SCE_REBOL_WORD8);
 160                                 } else if (keywords7.InList(s)) {
 161                                         sc.ChangeState(SCE_REBOL_WORD7);
 162                                 } else if (keywords6.InList(s)) {
 163                                         sc.ChangeState(SCE_REBOL_WORD6);
 164                                 } else if (keywords5.InList(s)) {
 165                                         sc.ChangeState(SCE_REBOL_WORD5);
 166                                 } else if (keywords4.InList(s)) {
 167                                         sc.ChangeState(SCE_REBOL_WORD4);
 168                                 } else if (keywords3.InList(s)) {
 169                                         sc.ChangeState(SCE_REBOL_WORD3);
 170                                 } else if (keywords2.InList(s)) {
 171                                         sc.ChangeState(SCE_REBOL_WORD2);
 172                                 } else if (keywords.InList(s)) {
 173                                         sc.ChangeState(SCE_REBOL_WORD);
 174                                 }
 175                                 // Keep same style if there are refinements
 176                                 if (!sc.Match('/')) {
 177                                         sc.SetState(SCE_REBOL_DEFAULT);
 178                                 }
 179                         }
 180                 // special numbers
 181                 } else if (sc.state == SCE_REBOL_NUMBER) {
 182                         switch (sc.ch) {
 183                         case 'x':       sc.ChangeState(SCE_REBOL_PAIR);
 184                                                 break;
 185                         case ':':       sc.ChangeState(SCE_REBOL_TIME);
 186                                                 break;
 187                         case '-':
 188                         case '/':       sc.ChangeState(SCE_REBOL_DATE);
 189                                                 break;
 190                         case '.':       if (++dotCount >= 2) sc.ChangeState(SCE_REBOL_TUPLE);
 191                                                 break;
 192                         }
 193                 }
 194
 195                 //--- Determine if the current state should terminate
 196                 if (sc.state == SCE_REBOL_QUOTEDSTRING || sc.state == SCE_REBOL_CHARACTER) {
 197                         if (sc.ch == '^' && sc.chNext == '\"') {
 198                                 sc.Forward();
 199                         } else if (sc.ch == '\"') {
 200                                 sc.ForwardSetState(SCE_REBOL_DEFAULT);
 201                         }
 202                 } else if (sc.state == SCE_REBOL_BRACEDSTRING || sc.state == SCE_REBOL_COMMENTBLOCK) {
 203                         if (sc.ch == '}') {
 204                                 if (--stringLevel == 0) {
 205                                         sc.ForwardSetState(SCE_REBOL_DEFAULT);
 206                                 }
 207                         } else if (sc.ch == '{') {
 208                                 stringLevel++;
 209                         }
 210                 } else if (sc.state == SCE_REBOL_BINARY) {
 211                         if (sc.ch == '}') {
 212                                 sc.ForwardSetState(SCE_REBOL_DEFAULT);
 213                         }
 214                 } else if (sc.state == SCE_REBOL_TAG) {
 215                         if (sc.ch == '>') {
 216                                 sc.ForwardSetState(SCE_REBOL_DEFAULT);
 217                         }
 218                 } else if (sc.state == SCE_REBOL_PREFACE) {
 219                         if (sc.MatchIgnoreCase("rebol"))
 220                         {
 221                                 int i;
 222                                 for (i=5; IsASpaceOrTab(styler.SafeGetCharAt(sc.currentPos+i, 0)); i++);
 223                                 if (sc.GetRelative(i) == '[')
 224                                         sc.SetState(SCE_REBOL_DEFAULT);
 225                         }
 226                 }
 227
 228                 //--- Parens and bracket changes to default style when the current is a number
 229                 if (sc.state == SCE_REBOL_NUMBER || sc.state == SCE_REBOL_PAIR || sc.state == SCE_REBOL_TUPLE ||
 230                         sc.state == SCE_REBOL_MONEY || sc.state == SCE_REBOL_ISSUE || sc.state == SCE_REBOL_EMAIL ||
 231                         sc.state == SCE_REBOL_URL || sc.state == SCE_REBOL_DATE || sc.state == SCE_REBOL_TIME) {
 232                         if (sc.ch == '(' || sc.ch == '[' || sc.ch == ')' || sc.ch == ']') {
 233                                 sc.SetState(SCE_REBOL_DEFAULT);
 234                         }
 235                 }
 236
 237                 //--- Determine if a new state should be entered.
 238                 if (sc.state == SCE_REBOL_DEFAULT) {
 239                         if (IsAnOperator(sc.ch, sc.chNext, sc.GetRelative(2))) {
 240                                 sc.SetState(SCE_REBOL_OPERATOR);
 241                         } else if (IsBinaryStart(sc.ch, sc.chNext, sc.GetRelative(2), sc.GetRelative(3))) {
 242                                 sc.SetState(SCE_REBOL_BINARY);
 243                         } else if (IsAWordStart(sc.ch, sc.chNext)) {
 244                                 sc.SetState(SCE_REBOL_IDENTIFIER);
 245                         } else if (IsADigit(sc.ch) || sc.ch == '+' || sc.ch == '-' || /*Decimal*/ sc.ch == '.' || sc.ch == ',') {
 246                                 dotCount = 0;
 247                                 sc.SetState(SCE_REBOL_NUMBER);
 248                         } else if (sc.ch == '\"') {
 249                                 sc.SetState(SCE_REBOL_QUOTEDSTRING);
 250                         } else if (sc.ch == '{') {
 251                                 sc.SetState(blockComment ? SCE_REBOL_COMMENTBLOCK : SCE_REBOL_BRACEDSTRING);
 252                                 ++stringLevel;
 253                         } else if (sc.ch == ';') {
 254                                 sc.SetState(SCE_REBOL_COMMENTLINE);
 255                         } else if (sc.ch == '$') {
 256                                 sc.SetState(SCE_REBOL_MONEY);
 257                         } else if (sc.ch == '%') {
 258                                 sc.SetState(SCE_REBOL_FILE);
 259                         } else if (sc.ch == '<') {
 260                                 sc.SetState(SCE_REBOL_TAG);
 261                         } else if (sc.ch == '#' && sc.chNext == '"') {
 262                                 sc.SetState(SCE_REBOL_CHARACTER);
 263                                 sc.Forward();
 264                         } else if (sc.ch == '#' && sc.chNext != '"' && sc.chNext != '{' ) {
 265                                 sc.SetState(SCE_REBOL_ISSUE);
 266                         }
 267                 }
 268         }
 269         sc.Complete();
 270 }
 271
 272
 273 static void FoldRebolDoc(unsigned int startPos, int length, int /* initStyle */, WordList *[],
 274                             Accessor &styler) {
 275         unsigned int lengthDoc = startPos + length;
 276         int visibleChars = 0;
 277         int lineCurrent = styler.GetLine(startPos);
 278         int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
 279         int levelCurrent = levelPrev;
 280         char chNext = styler[startPos];
 281         int styleNext = styler.StyleAt(startPos);
 282         for (unsigned int i = startPos; i < lengthDoc; i++) {
 283                 char ch = chNext;
 284                 chNext = styler.SafeGetCharAt(i + 1);
 285                 int style = styleNext;
 286                 styleNext = styler.StyleAt(i + 1);
 287                 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
 288                 if (style == SCE_REBOL_DEFAULT) {
 289                         if (ch == '[') {
 290                                 levelCurrent++;
 291                         } else if (ch == ']') {
 292                                 levelCurrent--;
 293                         }
 294                 }
 295                 if (atEOL) {
 296                         int lev = levelPrev;
 297                         if (visibleChars == 0)
 298                                 lev |= SC_FOLDLEVELWHITEFLAG;
 299                         if ((levelCurrent > levelPrev) && (visibleChars > 0))
 300                                 lev |= SC_FOLDLEVELHEADERFLAG;
 301                         if (lev != styler.LevelAt(lineCurrent)) {
 302                                 styler.SetLevel(lineCurrent, lev);
 303                         }
 304                         lineCurrent++;
 305                         levelPrev = levelCurrent;
 306                         visibleChars = 0;
 307                 }
 308                 if (!isspacechar(ch))
 309                         visibleChars++;
 310         }
 311         // Fill in the real level of the next line, keeping the current flags as they will be filled in later
 312         int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;
 313         styler.SetLevel(lineCurrent, levelPrev | flagsNext);
 314 }
 315
 316 static const char * const rebolWordListDesc[] = {
 317         "Keywords",
 318         0
 319 };
 320
 321 LexerModule lmREBOL(SCLEX_REBOL, ColouriseRebolDoc, "rebol", FoldRebolDoc, rebolWordListDesc);
 322