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