src/stc/scintilla/lexers/LexHaskell.cxx

   1 /******************************************************************
   2  *    LexHaskell.cxx
   3  *
   4  *    A haskell lexer for the scintilla code control.
   5  *    Some stuff "lended" from LexPython.cxx and LexCPP.cxx.
   6  *    External lexer stuff inspired from the caml external lexer.
   7  *
   8  *    Written by Tobias Engvall - tumm at dtek dot chalmers dot se
   9  *
  10  *    Several bug fixes by Krasimir Angelov - kr.angelov at gmail.com
  11  *
  12  *    TODO:
  13  *    * Implement a folder :)
  14  *    * Nice Character-lexing (stuff inside '\''), LexPython has
  15  *      this.
  16  *
  17  *
  18  *****************************************************************/
  19
  20 #include <stdlib.h>
  21 #include <string.h>
  22 #include <stdio.h>
  23 #include <stdarg.h>
  24 #include <assert.h>
  25 #include <ctype.h>
  26
  27 #include "ILexer.h"
  28 #include "Scintilla.h"
  29 #include "SciLexer.h"
  30
  31 #include "PropSetSimple.h"
  32 #include "WordList.h"
  33 #include "LexAccessor.h"
  34 #include "Accessor.h"
  35 #include "StyleContext.h"
  36 #include "CharacterSet.h"
  37 #include "LexerModule.h"
  38
  39 #ifdef SCI_NAMESPACE
  40 using namespace Scintilla;
  41 #endif
  42
  43 #ifdef BUILD_AS_EXTERNAL_LEXER
  44
  45 #include "ExternalLexer.h"
  46 #include "WindowAccessor.h"
  47
  48 #define BUILD_EXTERNAL_LEXER 0
  49
  50 #endif
  51
  52 #define HA_MODE_DEFAULT     0
  53 #define HA_MODE_IMPORT1     1
  54 #define HA_MODE_IMPORT2     2
  55 #define HA_MODE_IMPORT3     3
  56 #define HA_MODE_MODULE      4
  57 #define HA_MODE_FFI         5
  58 #define HA_MODE_TYPE        6
  59
  60 static inline bool IsNewline(const int ch) {
  61    return (ch == '\n' || ch == '\r');
  62 }
  63
  64 static inline bool IsWhitespace(const int ch) {
  65    return (  ch == ' '
  66           || ch == '\t'
  67           || IsNewline(ch) );
  68 }
  69
  70 static inline bool IsAWordStart(const int ch) {
  71    return (ch < 0x80) && (isalnum(ch) || ch == '_');
  72 }
  73
  74 static inline bool IsAWordChar(const int ch) {
  75    return (ch < 0x80) && (isalnum(ch) || ch == '.' || ch == '_' || ch == '\'');
  76 }
  77
  78 static void ColorizeHaskellDoc(unsigned int startPos, int length, int initStyle,
  79                                WordList *keywordlists[], Accessor &styler) {
  80
  81    WordList &keywords = *keywordlists[0];
  82    WordList &ffi      = *keywordlists[1];
  83
  84    StyleContext sc(startPos, length, initStyle, styler);
  85
  86    int lineCurrent = styler.GetLine(startPos);
  87    int state = lineCurrent ? styler.GetLineState(lineCurrent-1)
  88                            : HA_MODE_DEFAULT;
  89    int mode  = state & 0xF;
  90    int xmode = state >> 4;
  91
  92    while (sc.More()) {
  93       // Check for state end
  94
  95          // Operator
  96       if (sc.state == SCE_HA_OPERATOR) {
  97          if (isascii(sc.ch) && isoperator(static_cast<char>(sc.ch))) {
  98             sc.Forward();
  99          } else {
 100             styler.ColourTo(sc.currentPos - 1, sc.state);
 101             sc.ChangeState(SCE_HA_DEFAULT);
 102          }
 103       }
 104          // String
 105       else if (sc.state == SCE_HA_STRING) {
 106          if (sc.ch == '\"') {
 107                         sc.Forward();
 108             styler.ColourTo(sc.currentPos-1, sc.state);
 109             sc.ChangeState(SCE_HA_DEFAULT);
 110          } else if (sc.ch == '\\') {
 111             sc.Forward(2);
 112          } else if (sc.atLineEnd) {
 113                         styler.ColourTo(sc.currentPos-1, sc.state);
 114                         sc.ChangeState(SCE_HA_DEFAULT);
 115                  } else {
 116                         sc.Forward();
 117                  }
 118       }
 119          // Char
 120       else if (sc.state == SCE_HA_CHARACTER) {
 121          if (sc.ch == '\'') {
 122                         sc.Forward();
 123             styler.ColourTo(sc.currentPos-1, sc.state);
 124             sc.ChangeState(SCE_HA_DEFAULT);
 125          } else if (sc.ch == '\\') {
 126             sc.Forward(2);
 127          } else if (sc.atLineEnd) {
 128                         styler.ColourTo(sc.currentPos-1, sc.state);
 129                         sc.ChangeState(SCE_HA_DEFAULT);
 130                  } else {
 131                         sc.Forward();
 132                  }
 133       }
 134          // Number
 135       else if (sc.state == SCE_HA_NUMBER) {
 136          if (IsADigit(sc.ch, xmode)) {
 137             sc.Forward();
 138          } else if ((xmode == 10) &&
 139                     (sc.ch == 'e' || sc.ch == 'E') &&
 140                     (IsADigit(sc.chNext) || sc.chNext == '+' || sc.chNext == '-')) {
 141                         sc.Forward();
 142                         if (sc.ch == '+' || sc.ch == '-')
 143                                 sc.Forward();
 144          } else {
 145             styler.ColourTo(sc.currentPos - 1, sc.state);
 146             sc.ChangeState(SCE_HA_DEFAULT);
 147          }
 148       }
 149          // Identifier
 150       else if (sc.state == SCE_HA_IDENTIFIER) {
 151          if (IsAWordChar(sc.ch)) {
 152             sc.Forward();
 153          } else {
 154             char s[100];
 155             sc.GetCurrent(s, sizeof(s));
 156             int style = sc.state;
 157             int new_mode = 0;
 158             if (keywords.InList(s)) {
 159                style = SCE_HA_KEYWORD;
 160             } else if (isupper(s[0])) {
 161                if (mode >= HA_MODE_IMPORT1 && mode <= HA_MODE_IMPORT3) {
 162                   style    = SCE_HA_MODULE;
 163                   new_mode = HA_MODE_IMPORT2;
 164                } else if (mode == HA_MODE_MODULE)
 165                   style = SCE_HA_MODULE;
 166                else
 167                   style = SCE_HA_CAPITAL;
 168             } else if (mode == HA_MODE_IMPORT1 &&
 169                        strcmp(s,"qualified") == 0) {
 170                 style    = SCE_HA_KEYWORD;
 171                 new_mode = HA_MODE_IMPORT1;
 172             } else if (mode == HA_MODE_IMPORT2) {
 173                 if (strcmp(s,"as") == 0) {
 174                    style    = SCE_HA_KEYWORD;
 175                    new_mode = HA_MODE_IMPORT3;
 176                } else if (strcmp(s,"hiding") == 0) {
 177                    style     = SCE_HA_KEYWORD;
 178                }
 179             } else if (mode == HA_MODE_FFI) {
 180                            if (ffi.InList(s)) {
 181                   style = SCE_HA_KEYWORD;
 182                   new_mode = HA_MODE_FFI;
 183                }
 184             }
 185             else if (mode == HA_MODE_TYPE) {
 186                if (strcmp(s,"family") == 0)
 187                   style    = SCE_HA_KEYWORD;
 188                         }
 189             styler.ColourTo(sc.currentPos - 1, style);
 190             if (strcmp(s,"import") == 0 && mode != HA_MODE_FFI)
 191                new_mode = HA_MODE_IMPORT1;
 192             else if (strcmp(s,"module") == 0)
 193                new_mode = HA_MODE_MODULE;
 194             else if (strcmp(s,"foreign") == 0)
 195                new_mode = HA_MODE_FFI;
 196             else if (strcmp(s,"type") == 0)
 197                new_mode = HA_MODE_TYPE;
 198             sc.ChangeState(SCE_HA_DEFAULT);
 199             mode = new_mode;
 200          }
 201       }
 202
 203          // Comments
 204             // Oneliner
 205       else if (sc.state == SCE_HA_COMMENTLINE) {
 206          if (sc.atLineEnd) {
 207             styler.ColourTo(sc.currentPos - 1, sc.state);
 208             sc.ChangeState(SCE_HA_DEFAULT);
 209          } else {
 210             sc.Forward();
 211          }
 212       }
 213             // Nested
 214       else if (sc.state == SCE_HA_COMMENTBLOCK) {
 215          if (sc.Match("{-")) {
 216             sc.Forward(2);
 217             xmode++;
 218          }
 219          else if (sc.Match("-}")) {
 220             sc.Forward(2);
 221             xmode--;
 222             if (xmode == 0) {
 223                styler.ColourTo(sc.currentPos - 1, sc.state);
 224                sc.ChangeState(SCE_HA_DEFAULT);
 225             }
 226          } else {
 227             if (sc.atLineEnd) {
 228                                 // Remember the line state for future incremental lexing
 229                                 styler.SetLineState(lineCurrent, (xmode << 4) | mode);
 230                                 lineCurrent++;
 231                         }
 232             sc.Forward();
 233          }
 234       }
 235       // New state?
 236       if (sc.state == SCE_HA_DEFAULT) {
 237          // Digit
 238          if (IsADigit(sc.ch) ||
 239              (sc.ch == '.' && IsADigit(sc.chNext)) ||
 240              (sc.ch == '-' && IsADigit(sc.chNext))) {
 241             styler.ColourTo(sc.currentPos - 1, sc.state);
 242             sc.ChangeState(SCE_HA_NUMBER);
 243             if (sc.ch == '0' && (sc.chNext == 'X' || sc.chNext == 'x')) {
 244                                 // Match anything starting with "0x" or "0X", too
 245                                 sc.Forward(2);
 246                                 xmode = 16;
 247             } else if (sc.ch == '0' && (sc.chNext == 'O' || sc.chNext == 'o')) {
 248                                 // Match anything starting with "0x" or "0X", too
 249                                 sc.Forward(2);
 250                                 xmode = 8;
 251             } else {
 252                                 sc.Forward();
 253                                 xmode = 10;
 254                         }
 255             mode = HA_MODE_DEFAULT;
 256          }
 257          // Comment line
 258          else if (sc.Match("--")) {
 259             styler.ColourTo(sc.currentPos - 1, sc.state);
 260             sc.Forward(2);
 261             sc.ChangeState(SCE_HA_COMMENTLINE);
 262          // Comment block
 263          }
 264          else if (sc.Match("{-")) {
 265             styler.ColourTo(sc.currentPos - 1, sc.state);
 266             sc.Forward(2);
 267             sc.ChangeState(SCE_HA_COMMENTBLOCK);
 268             xmode = 1;
 269          }
 270          // String
 271          else if (sc.Match('\"')) {
 272             styler.ColourTo(sc.currentPos - 1, sc.state);
 273             sc.Forward();
 274             sc.ChangeState(SCE_HA_STRING);
 275          }
 276          // Character
 277          else if (sc.Match('\'')) {
 278             styler.ColourTo(sc.currentPos - 1, sc.state);
 279             sc.Forward();
 280             sc.ChangeState(SCE_HA_CHARACTER);
 281          }
 282          else if (sc.ch == '(' || sc.ch == ')' ||
 283                   sc.ch == '{' || sc.ch == '}' ||
 284                   sc.ch == '[' || sc.ch == ']') {
 285                         styler.ColourTo(sc.currentPos - 1, sc.state);
 286                         sc.Forward();
 287                         styler.ColourTo(sc.currentPos - 1, SCE_HA_OPERATOR);
 288                         mode = HA_MODE_DEFAULT;
 289                  }
 290          // Operator
 291          else if (isascii(sc.ch) && isoperator(static_cast<char>(sc.ch))) {
 292             styler.ColourTo(sc.currentPos - 1, sc.state);
 293             sc.Forward();
 294             sc.ChangeState(SCE_HA_OPERATOR);
 295             mode = HA_MODE_DEFAULT;
 296          }
 297          // Keyword
 298          else if (IsAWordStart(sc.ch)) {
 299             styler.ColourTo(sc.currentPos - 1, sc.state);
 300             sc.Forward();
 301             sc.ChangeState(SCE_HA_IDENTIFIER);
 302          } else {
 303             if (sc.atLineEnd) {
 304                                 // Remember the line state for future incremental lexing
 305                                 styler.SetLineState(lineCurrent, (xmode << 4) | mode);
 306                                 lineCurrent++;
 307                         }
 308             sc.Forward();
 309          }
 310       }
 311    }
 312    sc.Complete();
 313 }
 314
 315 // External stuff - used for dynamic-loading, not implemented in wxStyledTextCtrl yet.
 316 // Inspired by the caml external lexer - Credits to Robert Roessler - http://www.rftp.com
 317 #ifdef BUILD_EXTERNAL_LEXER
 318 static const char* LexerName = "haskell";
 319
 320 void EXT_LEXER_DECL Lex(unsigned int lexer, unsigned int startPos, int length, int initStyle,
 321                         char *words[], WindowID window, char *props)
 322 {
 323    PropSetSimple ps;
 324    ps.SetMultiple(props);
 325    WindowAccessor wa(window, ps);
 326
 327    int nWL = 0;
 328    for (; words[nWL]; nWL++) ;
 329    WordList** wl = new WordList* [nWL + 1];
 330    int i = 0;
 331    for (; i<nWL; i++)
 332    {
 333       wl[i] = new WordList();
 334       wl[i]->Set(words[i]);
 335    }
 336    wl[i] = 0;
 337
 338    ColorizeHaskellDoc(startPos, length, initStyle, wl, wa);
 339    wa.Flush();
 340    for (i=nWL-1;i>=0;i--)
 341       delete wl[i];
 342    delete [] wl;
 343 }
 344
 345 void EXT_LEXER_DECL Fold (unsigned int lexer, unsigned int startPos, int length, int initStyle,
 346                         char *words[], WindowID window, char *props)
 347 {
 348
 349 }
 350
 351 int EXT_LEXER_DECL GetLexerCount()
 352 {
 353    return 1;
 354 }
 355
 356 void EXT_LEXER_DECL GetLexerName(unsigned int Index, char *name, int buflength)
 357 {
 358    if (buflength > 0) {
 359       buflength--;
 360       int n = strlen(LexerName);
 361       if (n > buflength)
 362          n = buflength;
 363       memcpy(name, LexerName, n), name[n] = '\0';
 364    }
 365 }
 366 #endif
 367
 368 LexerModule lmHaskell(SCLEX_HASKELL, ColorizeHaskellDoc, "haskell");