scintilla/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  *
  11  *    TODO:
  12  *    * Implement a folder :)
  13  *    * Nice Character-lexing (stuff inside '\''), LexPython has
  14  *      this.
  15  *
  16  *
  17  *****************************************************************/
  18
  19 #include <stdlib.h>
  20 #include <string.h>
  21 #include <ctype.h>
  22 #include <stdio.h>
  23 #include <stdarg.h>
  24
  25 #include "Platform.h"
  26
  27 #include "PropSet.h"
  28 #include "PropSetSimple.h"
  29 #include "Accessor.h"
  30 #include "StyleContext.h"
  31 #include "KeyWords.h"
  32 #include "Scintilla.h"
  33 #include "SciLexer.h"
  34
  35 #ifdef SCI_NAMESPACE
  36 using namespace Scintilla;
  37 #endif
  38
  39 #ifdef BUILD_AS_EXTERNAL_LEXER
  40
  41 #include "ExternalLexer.h"
  42 #include "WindowAccessor.h"
  43
  44 #define BUILD_EXTERNAL_LEXER 0
  45
  46 #endif
  47
  48 // Max level of nested comments
  49 #define SCE_HA_COMMENTMAX SCE_HA_COMMENTBLOCK3
  50
  51
  52 enum kwType { kwOther, kwClass, kwData, kwInstance, kwImport, kwModule, kwType};
  53
  54 static inline bool IsNewline(const int ch) {
  55    return (ch == '\n' || ch == '\r');
  56 }
  57
  58 static inline bool IsWhitespace(const int ch) {
  59    return (  ch == ' '
  60           || ch == '\t'
  61           || IsNewline(ch) );
  62 }
  63
  64 static inline bool IsAWordStart(const int ch) {
  65    return (ch < 0x80) && (isalnum(ch) || ch == '_');
  66 }
  67
  68 static inline bool IsAWordChar(const int ch) {
  69    return (ch < 0x80) && (isalnum(ch) || ch == '.' || ch == '_' || ch == '\'');
  70 }
  71
  72 static void ColorizeHaskellDoc(unsigned int startPos, int length, int initStyle,
  73                                WordList *keywordlists[], Accessor &styler) {
  74
  75    WordList &keywords = *keywordlists[0];
  76
  77    int kwLast = kwOther;
  78
  79    StyleContext sc(startPos, length, initStyle, styler);
  80
  81    for (; sc.More(); sc.Forward()) {
  82
  83       // Check for state end
  84          // Operator
  85       if (sc.state == SCE_HA_OPERATOR) {
  86          kwLast = kwOther;
  87          sc.SetState(SCE_HA_DEFAULT);
  88       }
  89          // String
  90       else if (sc.state == SCE_HA_STRING) {
  91          if (sc.ch == '\"') {
  92             sc.ForwardSetState(SCE_HA_DEFAULT);
  93          } else if (sc.ch == '\\') {
  94             sc.Forward();
  95          }
  96       }
  97          // Char
  98       else if (sc.state == SCE_HA_CHARACTER) {
  99          if (sc.ch == '\'') {
 100             sc.ForwardSetState(SCE_HA_DEFAULT);
 101          } else if (sc.ch == '\\') {
 102             sc.Forward();
 103          }
 104       }
 105          // Number
 106       else if (sc.state == SCE_HA_NUMBER) {
 107          if (!IsADigit(sc.ch)) {
 108             sc.SetState(SCE_HA_DEFAULT);
 109          }
 110       }
 111          // Types, constructors, etc.
 112       else if (sc.state == SCE_HA_CAPITAL) {
 113          if (!IsAWordChar(sc.ch) || sc.ch == '.') {
 114             sc.SetState(SCE_HA_DEFAULT);
 115          }
 116       }
 117          // Identifier
 118       else if (sc.state == SCE_HA_IDENTIFIER) {
 119          if (!IsAWordChar(sc.ch)) {
 120             char s[100];
 121             sc.GetCurrent(s, sizeof(s));
 122             int style = SCE_HA_IDENTIFIER;
 123             if ((kwLast == kwImport) || (strcmp(s,"qualified") == 0) || (strcmp(s,"as") == 0)) {
 124                style = SCE_HA_IMPORT;
 125             } else if (keywords.InList(s)) {
 126                style = SCE_HA_KEYWORD;
 127             } else if (kwLast == kwData) {
 128                style = SCE_HA_DATA;
 129             } else if (kwLast == kwClass) {
 130                style = SCE_HA_CLASS;
 131             } else if (kwLast == kwModule) {
 132                style = SCE_HA_MODULE;
 133             } else if (isupper(s[0])) {
 134                style = SCE_HA_CAPITAL;
 135             }
 136             sc.ChangeState(style);
 137             sc.SetState(SCE_HA_DEFAULT);
 138             if (style == SCE_HA_KEYWORD) {
 139                if (0 == strcmp(s, "class"))
 140                   kwLast = kwClass;
 141                else if (0 == strcmp(s, "data"))
 142                   kwLast = kwData;
 143                else if (0 == strcmp(s, "instance"))
 144                   kwLast = kwInstance;
 145                else if (0 == strcmp(s, "import"))
 146                   kwLast = kwImport;
 147                else if (0 == strcmp(s, "module"))
 148                   kwLast = kwModule;
 149                else
 150                   kwLast = kwOther;
 151             } else if (style == SCE_HA_CLASS || style == SCE_HA_IMPORT ||
 152                        style == SCE_HA_MODULE || style == SCE_HA_CAPITAL ||
 153                        style == SCE_HA_DATA || style == SCE_HA_INSTANCE) {
 154                kwLast = kwOther;
 155             }
 156          }
 157       }
 158          // Comments
 159             // Oneliner
 160       else if (sc.state == SCE_HA_COMMENTLINE) {
 161          if (IsNewline(sc.ch))
 162             sc.SetState(SCE_HA_DEFAULT);
 163       }
 164             // Nested
 165       else if (sc.state >= SCE_HA_COMMENTBLOCK) {
 166          if (sc.Match("{-")) {
 167             if (sc.state < SCE_HA_COMMENTMAX)
 168                sc.SetState(sc.state + 1);
 169          }
 170          else if (sc.Match("-}")) {
 171             sc.Forward();
 172             if (sc.state == SCE_HA_COMMENTBLOCK)
 173                sc.ForwardSetState(SCE_HA_DEFAULT);
 174             else
 175                sc.ForwardSetState(sc.state - 1);
 176          }
 177       }
 178       // New state?
 179       if (sc.state == SCE_HA_DEFAULT) {
 180          // Digit
 181          if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
 182             sc.SetState(SCE_HA_NUMBER);
 183             if (sc.ch == '0' && (sc.chNext == 'X' || sc.chNext == 'x')) { // Match anything starting with "0x" or "0X", too
 184                sc.Forward(1);
 185             }
 186          }
 187          // Comment line
 188          else if (sc.Match("--")) {
 189             sc.SetState(SCE_HA_COMMENTLINE);
 190          // Comment block
 191          }
 192          else if (sc.Match("{-")) {
 193             sc.SetState(SCE_HA_COMMENTBLOCK);
 194          }
 195          // String
 196          else if (sc.Match('\"')) {
 197             sc.SetState(SCE_HA_STRING);
 198          }
 199          // Character
 200          else if (sc.Match('\'')) {
 201             sc.SetState(SCE_HA_CHARACTER);
 202          }
 203          // Stringstart
 204          else if (sc.Match('\"')) {
 205             sc.SetState(SCE_HA_STRING);
 206          }
 207          // Operator
 208          else if (isascii(sc.ch) && isoperator(static_cast<char>(sc.ch))) {
 209             sc.SetState(SCE_HA_OPERATOR);
 210          }
 211          // Keyword
 212          else if (IsAWordStart(sc.ch)) {
 213                sc.SetState(SCE_HA_IDENTIFIER);
 214          }
 215
 216       }
 217    }
 218    sc.Complete();
 219 }
 220
 221 // External stuff - used for dynamic-loading, not implemented in wxStyledTextCtrl yet.
 222 // Inspired by the caml external lexer - Credits to Robert Roessler - http://www.rftp.com
 223 #ifdef BUILD_EXTERNAL_LEXER
 224 static const char* LexerName = "haskell";
 225
 226 void EXT_LEXER_DECL Lex(unsigned int lexer, unsigned int startPos, int length, int initStyle,
 227                         char *words[], WindowID window, char *props)
 228 {
 229    PropSetSimple ps;
 230    ps.SetMultiple(props);
 231    WindowAccessor wa(window, ps);
 232
 233    int nWL = 0;
 234    for (; words[nWL]; nWL++) ;
 235    WordList** wl = new WordList* [nWL + 1];
 236    int i = 0;
 237    for (; i<nWL; i++)
 238    {
 239       wl[i] = new WordList();
 240       wl[i]->Set(words[i]);
 241    }
 242    wl[i] = 0;
 243
 244    ColorizeHaskellDoc(startPos, length, initStyle, wl, wa);
 245    wa.Flush();
 246    for (i=nWL-1;i>=0;i--)
 247       delete wl[i];
 248    delete [] wl;
 249 }
 250
 251 void EXT_LEXER_DECL Fold (unsigned int lexer, unsigned int startPos, int length, int initStyle,
 252                         char *words[], WindowID window, char *props)
 253 {
 254
 255 }
 256
 257 int EXT_LEXER_DECL GetLexerCount()
 258 {
 259    return 1;
 260 }
 261
 262 void EXT_LEXER_DECL GetLexerName(unsigned int Index, char *name, int buflength)
 263 {
 264    if (buflength > 0) {
 265       buflength--;
 266       int n = strlen(LexerName);
 267       if (n > buflength)
 268          n = buflength;
 269       memcpy(name, LexerName, n), name[n] = '\0';
 270    }
 271 }
 272 #endif
 273
 274 LexerModule lmHaskell(SCLEX_HASKELL, ColorizeHaskellDoc, "haskell");
 275