| 1 | // Scintilla source code edit control |
| 2 | /** @file LexFlagship.cxx |
| 3 | ** Lexer for Harbour and FlagShip. |
| 4 | ** (Syntactically compatible to other xBase dialects, like Clipper, dBase, Clip, FoxPro etc.) |
| 5 | **/ |
| 6 | // Copyright 2005 by Randy Butler |
| 7 | // Copyright 2010 by Xavi <jarabal/at/gmail.com> (Harbour) |
| 8 | // Copyright 1998-2003 by Neil Hodgson <neilh@scintilla.org> |
| 9 | // The License.txt file describes the conditions under which this software may be distributed. |
| 10 | |
| 11 | #include <stdlib.h> |
| 12 | #include <string.h> |
| 13 | #include <stdio.h> |
| 14 | #include <stdarg.h> |
| 15 | #include <assert.h> |
| 16 | #include <ctype.h> |
| 17 | |
| 18 | #include <string> |
| 19 | #include <string_view> |
| 20 | |
| 21 | #include "ILexer.h" |
| 22 | #include "Scintilla.h" |
| 23 | #include "SciLexer.h" |
| 24 | |
| 25 | #include "WordList.h" |
| 26 | #include "LexAccessor.h" |
| 27 | #include "Accessor.h" |
| 28 | #include "StyleContext.h" |
| 29 | #include "CharacterSet.h" |
| 30 | #include "LexerModule.h" |
| 31 | |
| 32 | using namespace Lexilla; |
| 33 | |
| 34 | // Extended to accept accented characters |
| 35 | static inline bool IsAWordChar(int ch) |
| 36 | { |
| 37 | return ch >= 0x80 || |
| 38 | (isalnum(ch) || ch == '_'); |
| 39 | } |
| 40 | |
| 41 | static void ColouriseFlagShipDoc(Sci_PositionU startPos, Sci_Position length, int initStyle, |
| 42 | WordList *keywordlists[], Accessor &styler) |
| 43 | { |
| 44 | |
| 45 | WordList &keywords = *keywordlists[0]; |
| 46 | WordList &keywords2 = *keywordlists[1]; |
| 47 | WordList &keywords3 = *keywordlists[2]; |
| 48 | WordList &keywords4 = *keywordlists[3]; |
| 49 | WordList &keywords5 = *keywordlists[4]; |
| 50 | |
| 51 | // property lexer.flagship.styling.within.preprocessor |
| 52 | // For Harbour code, determines whether all preprocessor code is styled in the preprocessor style (0) or only from the |
| 53 | // initial # to the end of the command word(1, the default). It also determines how to present text, dump, and disabled code. |
| 54 | bool stylingWithinPreprocessor = styler.GetPropertyInt("lexer.flagship.styling.within.preprocessor" , 1) != 0; |
| 55 | |
| 56 | CharacterSet setDoxygen(CharacterSet::setAlpha, "$@\\&<>#{}[]" ); |
| 57 | |
| 58 | int visibleChars = 0; |
| 59 | int closeStringChar = 0; |
| 60 | int styleBeforeDCKeyword = SCE_FS_DEFAULT; |
| 61 | bool bEnableCode = initStyle < SCE_FS_DISABLEDCODE; |
| 62 | |
| 63 | StyleContext sc(startPos, length, initStyle, styler); |
| 64 | |
| 65 | for (; sc.More(); sc.Forward()) { |
| 66 | |
| 67 | // Determine if the current state should terminate. |
| 68 | switch (sc.state) { |
| 69 | case SCE_FS_OPERATOR: |
| 70 | case SCE_FS_OPERATOR_C: |
| 71 | case SCE_FS_WORDOPERATOR: |
| 72 | sc.SetState(bEnableCode ? SCE_FS_DEFAULT : SCE_FS_DEFAULT_C); |
| 73 | break; |
| 74 | case SCE_FS_IDENTIFIER: |
| 75 | case SCE_FS_IDENTIFIER_C: |
| 76 | if (!IsAWordChar(sc.ch)) { |
| 77 | char s[64]; |
| 78 | sc.GetCurrentLowered(s, sizeof(s)); |
| 79 | if (keywords.InList(s)) { |
| 80 | sc.ChangeState(bEnableCode ? SCE_FS_KEYWORD : SCE_FS_KEYWORD_C); |
| 81 | } else if (keywords2.InList(s)) { |
| 82 | sc.ChangeState(bEnableCode ? SCE_FS_KEYWORD2 : SCE_FS_KEYWORD2_C); |
| 83 | } else if (bEnableCode && keywords3.InList(s)) { |
| 84 | sc.ChangeState(SCE_FS_KEYWORD3); |
| 85 | } else if (bEnableCode && keywords4.InList(s)) { |
| 86 | sc.ChangeState(SCE_FS_KEYWORD4); |
| 87 | }// Else, it is really an identifier... |
| 88 | sc.SetState(bEnableCode ? SCE_FS_DEFAULT : SCE_FS_DEFAULT_C); |
| 89 | } |
| 90 | break; |
| 91 | case SCE_FS_NUMBER: |
| 92 | if (!IsAWordChar(sc.ch) && !(sc.ch == '.' && IsADigit(sc.chNext))) { |
| 93 | sc.SetState(SCE_FS_DEFAULT); |
| 94 | } |
| 95 | break; |
| 96 | case SCE_FS_NUMBER_C: |
| 97 | if (!IsAWordChar(sc.ch) && sc.ch != '.') { |
| 98 | sc.SetState(SCE_FS_DEFAULT_C); |
| 99 | } |
| 100 | break; |
| 101 | case SCE_FS_CONSTANT: |
| 102 | if (!IsAWordChar(sc.ch)) { |
| 103 | sc.SetState(SCE_FS_DEFAULT); |
| 104 | } |
| 105 | break; |
| 106 | case SCE_FS_STRING: |
| 107 | case SCE_FS_STRING_C: |
| 108 | if (sc.ch == closeStringChar) { |
| 109 | sc.ForwardSetState(bEnableCode ? SCE_FS_DEFAULT : SCE_FS_DEFAULT_C); |
| 110 | } else if (sc.atLineEnd) { |
| 111 | sc.ChangeState(bEnableCode ? SCE_FS_STRINGEOL : SCE_FS_STRINGEOL_C); |
| 112 | } |
| 113 | break; |
| 114 | case SCE_FS_STRINGEOL: |
| 115 | case SCE_FS_STRINGEOL_C: |
| 116 | if (sc.atLineStart) { |
| 117 | sc.SetState(bEnableCode ? SCE_FS_DEFAULT : SCE_FS_DEFAULT_C); |
| 118 | } |
| 119 | break; |
| 120 | case SCE_FS_COMMENTDOC: |
| 121 | case SCE_FS_COMMENTDOC_C: |
| 122 | if (sc.Match('*', '/')) { |
| 123 | sc.Forward(); |
| 124 | sc.ForwardSetState(bEnableCode ? SCE_FS_DEFAULT : SCE_FS_DEFAULT_C); |
| 125 | } else if (sc.ch == '@' || sc.ch == '\\') { // JavaDoc and Doxygen support |
| 126 | // Verify that we have the conditions to mark a comment-doc-keyword |
| 127 | if ((IsASpace(sc.chPrev) || sc.chPrev == '*') && (!IsASpace(sc.chNext))) { |
| 128 | styleBeforeDCKeyword = bEnableCode ? SCE_FS_COMMENTDOC : SCE_FS_COMMENTDOC_C; |
| 129 | sc.SetState(SCE_FS_COMMENTDOCKEYWORD); |
| 130 | } |
| 131 | } |
| 132 | break; |
| 133 | case SCE_FS_COMMENT: |
| 134 | case SCE_FS_COMMENTLINE: |
| 135 | if (sc.atLineStart) { |
| 136 | sc.SetState(SCE_FS_DEFAULT); |
| 137 | } |
| 138 | break; |
| 139 | case SCE_FS_COMMENTLINEDOC: |
| 140 | case SCE_FS_COMMENTLINEDOC_C: |
| 141 | if (sc.atLineStart) { |
| 142 | sc.SetState(bEnableCode ? SCE_FS_DEFAULT : SCE_FS_DEFAULT_C); |
| 143 | } else if (sc.ch == '@' || sc.ch == '\\') { // JavaDoc and Doxygen support |
| 144 | // Verify that we have the conditions to mark a comment-doc-keyword |
| 145 | if ((IsASpace(sc.chPrev) || sc.chPrev == '/' || sc.chPrev == '!') && (!IsASpace(sc.chNext))) { |
| 146 | styleBeforeDCKeyword = bEnableCode ? SCE_FS_COMMENTLINEDOC : SCE_FS_COMMENTLINEDOC_C; |
| 147 | sc.SetState(SCE_FS_COMMENTDOCKEYWORD); |
| 148 | } |
| 149 | } |
| 150 | break; |
| 151 | case SCE_FS_COMMENTDOCKEYWORD: |
| 152 | if ((styleBeforeDCKeyword == SCE_FS_COMMENTDOC || styleBeforeDCKeyword == SCE_FS_COMMENTDOC_C) && |
| 153 | sc.Match('*', '/')) { |
| 154 | sc.ChangeState(SCE_FS_COMMENTDOCKEYWORDERROR); |
| 155 | sc.Forward(); |
| 156 | sc.ForwardSetState(bEnableCode ? SCE_FS_DEFAULT : SCE_FS_DEFAULT_C); |
| 157 | } else if (!setDoxygen.Contains(sc.ch)) { |
| 158 | char s[64]; |
| 159 | sc.GetCurrentLowered(s, sizeof(s)); |
| 160 | if (!IsASpace(sc.ch) || !keywords5.InList(s + 1)) { |
| 161 | sc.ChangeState(SCE_FS_COMMENTDOCKEYWORDERROR); |
| 162 | } |
| 163 | sc.SetState(styleBeforeDCKeyword); |
| 164 | } |
| 165 | break; |
| 166 | case SCE_FS_PREPROCESSOR: |
| 167 | case SCE_FS_PREPROCESSOR_C: |
| 168 | if (sc.atLineEnd) { |
| 169 | if (!(sc.chPrev == ';' || sc.GetRelative(-2) == ';')) { |
| 170 | sc.SetState(bEnableCode ? SCE_FS_DEFAULT : SCE_FS_DEFAULT_C); |
| 171 | } |
| 172 | } else if (stylingWithinPreprocessor) { |
| 173 | if (IsASpaceOrTab(sc.ch)) { |
| 174 | sc.SetState(bEnableCode ? SCE_FS_DEFAULT : SCE_FS_DEFAULT_C); |
| 175 | } |
| 176 | } else if (sc.Match('/', '*') || sc.Match('/', '/') || sc.Match('&', '&')) { |
| 177 | sc.SetState(bEnableCode ? SCE_FS_DEFAULT : SCE_FS_DEFAULT_C); |
| 178 | } |
| 179 | break; |
| 180 | case SCE_FS_DISABLEDCODE: |
| 181 | if (sc.ch == '#' && visibleChars == 0) { |
| 182 | sc.SetState(bEnableCode ? SCE_FS_PREPROCESSOR : SCE_FS_PREPROCESSOR_C); |
| 183 | do { // Skip whitespace between # and preprocessor word |
| 184 | sc.Forward(); |
| 185 | } while (IsASpaceOrTab(sc.ch) && sc.More()); |
| 186 | if (sc.MatchIgnoreCase("pragma" )) { |
| 187 | sc.Forward(6); |
| 188 | do { // Skip more whitespace until keyword |
| 189 | sc.Forward(); |
| 190 | } while (IsASpaceOrTab(sc.ch) && sc.More()); |
| 191 | if (sc.MatchIgnoreCase("enddump" ) || sc.MatchIgnoreCase("__endtext" )) { |
| 192 | bEnableCode = true; |
| 193 | sc.SetState(SCE_FS_DISABLEDCODE); |
| 194 | sc.Forward(sc.ch == '_' ? 8 : 6); |
| 195 | sc.ForwardSetState(SCE_FS_DEFAULT); |
| 196 | } else { |
| 197 | sc.ChangeState(SCE_FS_DISABLEDCODE); |
| 198 | } |
| 199 | } else { |
| 200 | sc.ChangeState(SCE_FS_DISABLEDCODE); |
| 201 | } |
| 202 | } |
| 203 | break; |
| 204 | case SCE_FS_DATE: |
| 205 | if (sc.ch == '}') { |
| 206 | sc.ForwardSetState(SCE_FS_DEFAULT); |
| 207 | } else if (sc.atLineEnd) { |
| 208 | sc.ChangeState(SCE_FS_STRINGEOL); |
| 209 | } |
| 210 | } |
| 211 | |
| 212 | // Determine if a new state should be entered. |
| 213 | if (sc.state == SCE_FS_DEFAULT || sc.state == SCE_FS_DEFAULT_C) { |
| 214 | if (bEnableCode && |
| 215 | (sc.MatchIgnoreCase(".and." ) || sc.MatchIgnoreCase(".not." ))) { |
| 216 | sc.SetState(SCE_FS_WORDOPERATOR); |
| 217 | sc.Forward(4); |
| 218 | } else if (bEnableCode && sc.MatchIgnoreCase(".or." )) { |
| 219 | sc.SetState(SCE_FS_WORDOPERATOR); |
| 220 | sc.Forward(3); |
| 221 | } else if (bEnableCode && |
| 222 | (sc.MatchIgnoreCase(".t." ) || sc.MatchIgnoreCase(".f." ) || |
| 223 | (!IsAWordChar(sc.GetRelative(3)) && sc.MatchIgnoreCase("nil" )))) { |
| 224 | sc.SetState(SCE_FS_CONSTANT); |
| 225 | sc.Forward(2); |
| 226 | } else if (sc.Match('/', '*')) { |
| 227 | sc.SetState(bEnableCode ? SCE_FS_COMMENTDOC : SCE_FS_COMMENTDOC_C); |
| 228 | sc.Forward(); |
| 229 | } else if (bEnableCode && sc.Match('&', '&')) { |
| 230 | sc.SetState(SCE_FS_COMMENTLINE); |
| 231 | sc.Forward(); |
| 232 | } else if (sc.Match('/', '/')) { |
| 233 | sc.SetState(bEnableCode ? SCE_FS_COMMENTLINEDOC : SCE_FS_COMMENTLINEDOC_C); |
| 234 | sc.Forward(); |
| 235 | } else if (bEnableCode && sc.ch == '*' && visibleChars == 0) { |
| 236 | sc.SetState(SCE_FS_COMMENT); |
| 237 | } else if (sc.ch == '\"' || sc.ch == '\'') { |
| 238 | sc.SetState(bEnableCode ? SCE_FS_STRING : SCE_FS_STRING_C); |
| 239 | closeStringChar = sc.ch; |
| 240 | } else if (closeStringChar == '>' && sc.ch == '<') { |
| 241 | sc.SetState(bEnableCode ? SCE_FS_STRING : SCE_FS_STRING_C); |
| 242 | } else if (sc.ch == '#' && visibleChars == 0) { |
| 243 | sc.SetState(bEnableCode ? SCE_FS_PREPROCESSOR : SCE_FS_PREPROCESSOR_C); |
| 244 | do { // Skip whitespace between # and preprocessor word |
| 245 | sc.Forward(); |
| 246 | } while (IsASpaceOrTab(sc.ch) && sc.More()); |
| 247 | if (sc.atLineEnd) { |
| 248 | sc.SetState(bEnableCode ? SCE_FS_DEFAULT : SCE_FS_DEFAULT_C); |
| 249 | } else if (sc.MatchIgnoreCase("include" )) { |
| 250 | if (stylingWithinPreprocessor) { |
| 251 | closeStringChar = '>'; |
| 252 | } |
| 253 | } else if (sc.MatchIgnoreCase("pragma" )) { |
| 254 | sc.Forward(6); |
| 255 | do { // Skip more whitespace until keyword |
| 256 | sc.Forward(); |
| 257 | } while (IsASpaceOrTab(sc.ch) && sc.More()); |
| 258 | if (sc.MatchIgnoreCase("begindump" ) || sc.MatchIgnoreCase("__cstream" )) { |
| 259 | bEnableCode = false; |
| 260 | if (stylingWithinPreprocessor) { |
| 261 | sc.SetState(SCE_FS_DISABLEDCODE); |
| 262 | sc.Forward(8); |
| 263 | sc.ForwardSetState(SCE_FS_DEFAULT_C); |
| 264 | } else { |
| 265 | sc.SetState(SCE_FS_DISABLEDCODE); |
| 266 | } |
| 267 | } else if (sc.MatchIgnoreCase("enddump" ) || sc.MatchIgnoreCase("__endtext" )) { |
| 268 | bEnableCode = true; |
| 269 | sc.SetState(SCE_FS_DISABLEDCODE); |
| 270 | sc.Forward(sc.ch == '_' ? 8 : 6); |
| 271 | sc.ForwardSetState(SCE_FS_DEFAULT); |
| 272 | } |
| 273 | } |
| 274 | } else if (bEnableCode && sc.ch == '{') { |
| 275 | Sci_Position p = 0; |
| 276 | int chSeek; |
| 277 | Sci_PositionU endPos(startPos + length); |
| 278 | do { // Skip whitespace |
| 279 | chSeek = sc.GetRelative(++p); |
| 280 | } while (IsASpaceOrTab(chSeek) && (sc.currentPos + p < endPos)); |
| 281 | if (chSeek == '^') { |
| 282 | sc.SetState(SCE_FS_DATE); |
| 283 | } else { |
| 284 | sc.SetState(SCE_FS_OPERATOR); |
| 285 | } |
| 286 | } else if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) { |
| 287 | sc.SetState(bEnableCode ? SCE_FS_NUMBER : SCE_FS_NUMBER_C); |
| 288 | } else if (IsAWordChar(sc.ch)) { |
| 289 | sc.SetState(bEnableCode ? SCE_FS_IDENTIFIER : SCE_FS_IDENTIFIER_C); |
| 290 | } else if (isoperator(static_cast<char>(sc.ch)) || (bEnableCode && sc.ch == '@')) { |
| 291 | sc.SetState(bEnableCode ? SCE_FS_OPERATOR : SCE_FS_OPERATOR_C); |
| 292 | } |
| 293 | } |
| 294 | |
| 295 | if (sc.atLineEnd) { |
| 296 | visibleChars = 0; |
| 297 | closeStringChar = 0; |
| 298 | } |
| 299 | if (!IsASpace(sc.ch)) { |
| 300 | visibleChars++; |
| 301 | } |
| 302 | } |
| 303 | sc.Complete(); |
| 304 | } |
| 305 | |
| 306 | static void FoldFlagShipDoc(Sci_PositionU startPos, Sci_Position length, int, |
| 307 | WordList *[], Accessor &styler) |
| 308 | { |
| 309 | |
| 310 | Sci_Position endPos = startPos + length; |
| 311 | |
| 312 | // Backtrack to previous line in case need to fix its fold status |
| 313 | Sci_Position lineCurrent = styler.GetLine(startPos); |
| 314 | if (startPos > 0 && lineCurrent > 0) { |
| 315 | lineCurrent--; |
| 316 | startPos = styler.LineStart(lineCurrent); |
| 317 | } |
| 318 | int spaceFlags = 0; |
| 319 | int indentCurrent = styler.IndentAmount(lineCurrent, &spaceFlags); |
| 320 | char chNext = styler[startPos]; |
| 321 | for (Sci_Position i = startPos; i < endPos; i++) { |
| 322 | char ch = chNext; |
| 323 | chNext = styler.SafeGetCharAt(i + 1); |
| 324 | |
| 325 | if ((ch == '\r' && chNext != '\n') || (ch == '\n') || (i == endPos-1)) { |
| 326 | int lev = indentCurrent; |
| 327 | int indentNext = styler.IndentAmount(lineCurrent + 1, &spaceFlags); |
| 328 | if (!(indentCurrent & SC_FOLDLEVELWHITEFLAG)) { |
| 329 | if ((indentCurrent & SC_FOLDLEVELNUMBERMASK) < (indentNext & SC_FOLDLEVELNUMBERMASK)) { |
| 330 | lev |= SC_FOLDLEVELHEADERFLAG; |
| 331 | } else if (indentNext & SC_FOLDLEVELWHITEFLAG) { |
| 332 | int spaceFlags2 = 0; |
| 333 | int indentNext2 = styler.IndentAmount(lineCurrent + 2, &spaceFlags2); |
| 334 | if ((indentCurrent & SC_FOLDLEVELNUMBERMASK) < (indentNext2 & SC_FOLDLEVELNUMBERMASK)) { |
| 335 | lev |= SC_FOLDLEVELHEADERFLAG; |
| 336 | } |
| 337 | } |
| 338 | } |
| 339 | indentCurrent = indentNext; |
| 340 | styler.SetLevel(lineCurrent, lev); |
| 341 | lineCurrent++; |
| 342 | } |
| 343 | } |
| 344 | } |
| 345 | |
| 346 | static const char * const FSWordListDesc[] = { |
| 347 | "Keywords Commands" , |
| 348 | "Std Library Functions" , |
| 349 | "Procedure, return, exit" , |
| 350 | "Class (oop)" , |
| 351 | "Doxygen keywords" , |
| 352 | 0 |
| 353 | }; |
| 354 | |
| 355 | LexerModule lmFlagShip(SCLEX_FLAGSHIP, ColouriseFlagShipDoc, "flagship" , FoldFlagShipDoc, FSWordListDesc); |
| 356 | |