ext/scintilla/lexers/LexVB.cxx

   1 // Scintilla source code edit control
   2 /** @file LexVB.cxx
   3  ** Lexer for Visual Basic and VBScript.
   4  **/
   5 // Copyright 1998-2005 by Neil Hodgson <neilh@scintilla.org>
   6 // The License.txt file describes the conditions under which this software may be distributed.
   7
   8 #include <stdlib.h>
   9 #include <string.h>
  10 #include <stdio.h>
  11 #include <stdarg.h>
  12 #include <assert.h>
  13 #include <ctype.h>
  14
  15 #include "ILexer.h"
  16 #include "Scintilla.h"
  17 #include "SciLexer.h"
  18
  19 #include "WordList.h"
  20 #include "LexAccessor.h"
  21 #include "Accessor.h"
  22 #include "StyleContext.h"
  23 #include "CharacterSet.h"
  24 #include "LexerModule.h"
  25
  26 using namespace Scintilla;
  27
  28 // Internal state, highlighted as number
  29 #define SCE_B_FILENUMBER SCE_B_DEFAULT+100
  30
  31
  32 static bool IsVBComment(Accessor &styler, Sci_Position pos, Sci_Position len) {
  33         return len > 0 && styler[pos] == '\'';
  34 }
  35
  36 static inline bool IsTypeCharacter(int ch) {
  37         return ch == '%' || ch == '&' || ch == '@' || ch == '!' || ch == '#' || ch == '$';
  38 }
  39
  40 // Extended to accept accented characters
  41 static inline bool IsAWordChar(int ch) {
  42         return ch >= 0x80 ||
  43                (isalnum(ch) || ch == '.' || ch == '_');
  44 }
  45
  46 static inline bool IsAWordStart(int ch) {
  47         return ch >= 0x80 ||
  48                (isalpha(ch) || ch == '_');
  49 }
  50
  51 static inline bool IsANumberChar(int ch) {
  52         // Not exactly following number definition (several dots are seen as OK, etc.)
  53         // but probably enough in most cases.
  54         return (ch < 0x80) &&
  55                 (isdigit(ch) || toupper(ch) == 'E' ||
  56              ch == '.' || ch == '-' || ch == '+');
  57 }
  58
  59 static void ColouriseVBDoc(Sci_PositionU startPos, Sci_Position length, int initStyle,
  60                            WordList *keywordlists[], Accessor &styler, bool vbScriptSyntax) {
  61
  62         WordList &keywords = *keywordlists[0];
  63         WordList &keywords2 = *keywordlists[1];
  64         WordList &keywords3 = *keywordlists[2];
  65         WordList &keywords4 = *keywordlists[3];
  66
  67         styler.StartAt(startPos);
  68
  69         int visibleChars = 0;
  70         int fileNbDigits = 0;
  71
  72         // Do not leak onto next line
  73         if (initStyle == SCE_B_STRINGEOL || initStyle == SCE_B_COMMENT || initStyle == SCE_B_PREPROCESSOR) {
  74                 initStyle = SCE_B_DEFAULT;
  75         }
  76
  77         StyleContext sc(startPos, length, initStyle, styler);
  78
  79         for (; sc.More(); sc.Forward()) {
  80
  81                 if (sc.state == SCE_B_OPERATOR) {
  82                         sc.SetState(SCE_B_DEFAULT);
  83                 } else if (sc.state == SCE_B_IDENTIFIER) {
  84                         if (!IsAWordChar(sc.ch)) {
  85                                 // In Basic (except VBScript), a variable name or a function name
  86                                 // can end with a special character indicating the type of the value
  87                                 // held or returned.
  88                                 bool skipType = false;
  89                                 if (!vbScriptSyntax && IsTypeCharacter(sc.ch)) {
  90                                         sc.Forward();   // Skip it
  91                                         skipType = true;
  92                                 }
  93                                 if (sc.ch == ']') {
  94                                         sc.Forward();
  95                                 }
  96                                 char s[100];
  97                                 sc.GetCurrentLowered(s, sizeof(s));
  98                                 if (skipType) {
  99                                         s[strlen(s) - 1] = '\0';
 100                                 }
 101                                 if (strcmp(s, "rem") == 0) {
 102                                         sc.ChangeState(SCE_B_COMMENT);
 103                                 } else {
 104                                         if (keywords.InList(s)) {
 105                                                 sc.ChangeState(SCE_B_KEYWORD);
 106                                         } else if (keywords2.InList(s)) {
 107                                                 sc.ChangeState(SCE_B_KEYWORD2);
 108                                         } else if (keywords3.InList(s)) {
 109                                                 sc.ChangeState(SCE_B_KEYWORD3);
 110                                         } else if (keywords4.InList(s)) {
 111                                                 sc.ChangeState(SCE_B_KEYWORD4);
 112                                         }       // Else, it is really an identifier...
 113                                         sc.SetState(SCE_B_DEFAULT);
 114                                 }
 115                         }
 116                 } else if (sc.state == SCE_B_NUMBER) {
 117                         // We stop the number definition on non-numerical non-dot non-eE non-sign char
 118                         // Also accepts A-F for hex. numbers
 119                         if (!IsANumberChar(sc.ch) && !(tolower(sc.ch) >= 'a' && tolower(sc.ch) <= 'f')) {
 120                                 sc.SetState(SCE_B_DEFAULT);
 121                         }
 122                 } else if (sc.state == SCE_B_STRING) {
 123                         // VB doubles quotes to preserve them, so just end this string
 124                         // state now as a following quote will start again
 125                         if (sc.ch == '\"') {
 126                                 if (sc.chNext == '\"') {
 127                                         sc.Forward();
 128                                 } else {
 129                                         if (tolower(sc.chNext) == 'c') {
 130                                                 sc.Forward();
 131                                         }
 132                                         sc.ForwardSetState(SCE_B_DEFAULT);
 133                                 }
 134                         } else if (sc.atLineEnd) {
 135                                 visibleChars = 0;
 136                                 sc.ChangeState(SCE_B_STRINGEOL);
 137                                 sc.ForwardSetState(SCE_B_DEFAULT);
 138                         }
 139                 } else if (sc.state == SCE_B_COMMENT) {
 140                         if (sc.atLineEnd) {
 141                                 visibleChars = 0;
 142                                 sc.ForwardSetState(SCE_B_DEFAULT);
 143                         }
 144                 } else if (sc.state == SCE_B_PREPROCESSOR) {
 145                         if (sc.atLineEnd) {
 146                                 visibleChars = 0;
 147                                 sc.ForwardSetState(SCE_B_DEFAULT);
 148                         }
 149                 } else if (sc.state == SCE_B_FILENUMBER) {
 150                         if (IsADigit(sc.ch)) {
 151                                 fileNbDigits++;
 152                                 if (fileNbDigits > 3) {
 153                                         sc.ChangeState(SCE_B_DATE);
 154                                 }
 155                         } else if (sc.ch == '\r' || sc.ch == '\n' || sc.ch == ',') {
 156                                 // Regular uses: Close #1; Put #1, ...; Get #1, ... etc.
 157                                 // Too bad if date is format #27, Oct, 2003# or something like that...
 158                                 // Use regular number state
 159                                 sc.ChangeState(SCE_B_NUMBER);
 160                                 sc.SetState(SCE_B_DEFAULT);
 161                         } else if (sc.ch == '#') {
 162                                 sc.ChangeState(SCE_B_DATE);
 163                                 sc.ForwardSetState(SCE_B_DEFAULT);
 164                         } else {
 165                                 sc.ChangeState(SCE_B_DATE);
 166                         }
 167                         if (sc.state != SCE_B_FILENUMBER) {
 168                                 fileNbDigits = 0;
 169                         }
 170                 } else if (sc.state == SCE_B_DATE) {
 171                         if (sc.atLineEnd) {
 172                                 visibleChars = 0;
 173                                 sc.ChangeState(SCE_B_STRINGEOL);
 174                                 sc.ForwardSetState(SCE_B_DEFAULT);
 175                         } else if (sc.ch == '#') {
 176                                 sc.ForwardSetState(SCE_B_DEFAULT);
 177                         }
 178                 }
 179
 180                 if (sc.state == SCE_B_DEFAULT) {
 181                         if (sc.ch == '\'') {
 182                                 sc.SetState(SCE_B_COMMENT);
 183                         } else if (sc.ch == '\"') {
 184                                 sc.SetState(SCE_B_STRING);
 185                         } else if (sc.ch == '#' && visibleChars == 0) {
 186                                 // Preprocessor commands are alone on their line
 187                                 sc.SetState(SCE_B_PREPROCESSOR);
 188                         } else if (sc.ch == '#') {
 189                                 // It can be a date literal, ending with #, or a file number, from 1 to 511
 190                                 // The date literal depends on the locale, so anything can go between #'s.
 191                                 // Can be #January 1, 1993# or #1 Jan 93# or #05/11/2003#, etc.
 192                                 // So we set the FILENUMBER state, and switch to DATE if it isn't a file number
 193                                 sc.SetState(SCE_B_FILENUMBER);
 194                         } else if (sc.ch == '&' && tolower(sc.chNext) == 'h') {
 195                                 // Hexadecimal number
 196                                 sc.SetState(SCE_B_NUMBER);
 197                                 sc.Forward();
 198                         } else if (sc.ch == '&' && tolower(sc.chNext) == 'o') {
 199                                 // Octal number
 200                                 sc.SetState(SCE_B_NUMBER);
 201                                 sc.Forward();
 202                         } else if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
 203                                 sc.SetState(SCE_B_NUMBER);
 204                         } else if (IsAWordStart(sc.ch) || (sc.ch == '[')) {
 205                                 sc.SetState(SCE_B_IDENTIFIER);
 206                         } else if (isoperator(static_cast<char>(sc.ch)) || (sc.ch == '\\')) {   // Integer division
 207                                 sc.SetState(SCE_B_OPERATOR);
 208                         }
 209                 }
 210
 211                 if (sc.atLineEnd) {
 212                         visibleChars = 0;
 213                 }
 214                 if (!IsASpace(sc.ch)) {
 215                         visibleChars++;
 216                 }
 217         }
 218
 219         if (sc.state == SCE_B_IDENTIFIER && !IsAWordChar(sc.ch)) {
 220                 // In Basic (except VBScript), a variable name or a function name
 221                 // can end with a special character indicating the type of the value
 222                 // held or returned.
 223                 bool skipType = false;
 224                 if (!vbScriptSyntax && IsTypeCharacter(sc.ch)) {
 225                         sc.Forward();   // Skip it
 226                         skipType = true;
 227                 }
 228                 if (sc.ch == ']') {
 229                         sc.Forward();
 230                 }
 231                 char s[100];
 232                 sc.GetCurrentLowered(s, sizeof(s));
 233                 if (skipType) {
 234                         s[strlen(s) - 1] = '\0';
 235                 }
 236                 if (strcmp(s, "rem") == 0) {
 237                         sc.ChangeState(SCE_B_COMMENT);
 238                 } else {
 239                         if (keywords.InList(s)) {
 240                                 sc.ChangeState(SCE_B_KEYWORD);
 241                         } else if (keywords2.InList(s)) {
 242                                 sc.ChangeState(SCE_B_KEYWORD2);
 243                         } else if (keywords3.InList(s)) {
 244                                 sc.ChangeState(SCE_B_KEYWORD3);
 245                         } else if (keywords4.InList(s)) {
 246                                 sc.ChangeState(SCE_B_KEYWORD4);
 247                         }       // Else, it is really an identifier...
 248                         sc.SetState(SCE_B_DEFAULT);
 249                 }
 250         }
 251
 252         sc.Complete();
 253 }
 254
 255 static void FoldVBDoc(Sci_PositionU startPos, Sci_Position length, int,
 256                                                    WordList *[], Accessor &styler) {
 257         Sci_Position endPos = startPos + length;
 258
 259         // Backtrack to previous line in case need to fix its fold status
 260         Sci_Position lineCurrent = styler.GetLine(startPos);
 261         if (startPos > 0) {
 262                 if (lineCurrent > 0) {
 263                         lineCurrent--;
 264                         startPos = styler.LineStart(lineCurrent);
 265                 }
 266         }
 267         int spaceFlags = 0;
 268         int indentCurrent = styler.IndentAmount(lineCurrent, &spaceFlags, IsVBComment);
 269         char chNext = styler[startPos];
 270         for (Sci_Position i = startPos; i < endPos; i++) {
 271                 char ch = chNext;
 272                 chNext = styler.SafeGetCharAt(i + 1);
 273
 274                 if ((ch == '\r' && chNext != '\n') || (ch == '\n') || (i == endPos)) {
 275                         int lev = indentCurrent;
 276                         int indentNext = styler.IndentAmount(lineCurrent + 1, &spaceFlags, IsVBComment);
 277                         if (!(indentCurrent & SC_FOLDLEVELWHITEFLAG)) {
 278                                 // Only non whitespace lines can be headers
 279                                 if ((indentCurrent & SC_FOLDLEVELNUMBERMASK) < (indentNext & SC_FOLDLEVELNUMBERMASK)) {
 280                                         lev |= SC_FOLDLEVELHEADERFLAG;
 281                                 } else if (indentNext & SC_FOLDLEVELWHITEFLAG) {
 282                                         // Line after is blank so check the next - maybe should continue further?
 283                                         int spaceFlags2 = 0;
 284                                         int indentNext2 = styler.IndentAmount(lineCurrent + 2, &spaceFlags2, IsVBComment);
 285                                         if ((indentCurrent & SC_FOLDLEVELNUMBERMASK) < (indentNext2 & SC_FOLDLEVELNUMBERMASK)) {
 286                                                 lev |= SC_FOLDLEVELHEADERFLAG;
 287                                         }
 288                                 }
 289                         }
 290                         indentCurrent = indentNext;
 291                         styler.SetLevel(lineCurrent, lev);
 292                         lineCurrent++;
 293                 }
 294         }
 295 }
 296
 297 static void ColouriseVBNetDoc(Sci_PositionU startPos, Sci_Position length, int initStyle,
 298                            WordList *keywordlists[], Accessor &styler) {
 299         ColouriseVBDoc(startPos, length, initStyle, keywordlists, styler, false);
 300 }
 301
 302 static void ColouriseVBScriptDoc(Sci_PositionU startPos, Sci_Position length, int initStyle,
 303                            WordList *keywordlists[], Accessor &styler) {
 304         ColouriseVBDoc(startPos, length, initStyle, keywordlists, styler, true);
 305 }
 306
 307 static const char * const vbWordListDesc[] = {
 308         "Keywords",
 309         "user1",
 310         "user2",
 311         "user3",
 312         0
 313 };
 314
 315 LexerModule lmVB(SCLEX_VB, ColouriseVBNetDoc, "vb", FoldVBDoc, vbWordListDesc);
 316 LexerModule lmVBScript(SCLEX_VBSCRIPT, ColouriseVBScriptDoc, "vbscript", FoldVBDoc, vbWordListDesc);
 317