scintilla/lexers/LexLua.cxx

   1 // Scintilla source code edit control
   2 /** @file LexLua.cxx
   3  ** Lexer for Lua language.
   4  **
   5  ** Written by Paul Winwood.
   6  ** Folder by Alexey Yutkin.
   7  ** Modified by Marcos E. Wurzius & Philippe Lhoste
   8  **/
   9
  10 #include <stdlib.h>
  11 #include <string.h>
  12 #include <stdio.h>
  13 #include <stdarg.h>
  14 #include <assert.h>
  15 #include <ctype.h>
  16
  17 #include "ILexer.h"
  18 #include "Scintilla.h"
  19 #include "SciLexer.h"
  20
  21 #include "WordList.h"
  22 #include "LexAccessor.h"
  23 #include "Accessor.h"
  24 #include "StyleContext.h"
  25 #include "CharacterSet.h"
  26 #include "LexerModule.h"
  27
  28 #ifdef SCI_NAMESPACE
  29 using namespace Scintilla;
  30 #endif
  31
  32 // Test for [=[ ... ]=] delimiters, returns 0 if it's only a [ or ],
  33 // return 1 for [[ or ]], returns >=2 for [=[ or ]=] and so on.
  34 // The maximum number of '=' characters allowed is 254.
  35 static int LongDelimCheck(StyleContext &sc) {
  36         int sep = 1;
  37         while (sc.GetRelative(sep) == '=' && sep < 0xFF)
  38                 sep++;
  39         if (sc.GetRelative(sep) == sc.ch)
  40                 return sep;
  41         return 0;
  42 }
  43
  44 static void ColouriseLuaDoc(
  45         Sci_PositionU startPos,
  46         Sci_Position length,
  47         int initStyle,
  48         WordList *keywordlists[],
  49         Accessor &styler) {
  50
  51         WordList &keywords = *keywordlists[0];
  52         WordList &keywords2 = *keywordlists[1];
  53         WordList &keywords3 = *keywordlists[2];
  54         WordList &keywords4 = *keywordlists[3];
  55         WordList &keywords5 = *keywordlists[4];
  56         WordList &keywords6 = *keywordlists[5];
  57         WordList &keywords7 = *keywordlists[6];
  58         WordList &keywords8 = *keywordlists[7];
  59
  60         // Accepts accented characters
  61         CharacterSet setWordStart(CharacterSet::setAlpha, "_", 0x80, true);
  62         CharacterSet setWord(CharacterSet::setAlphaNum, "_", 0x80, true);
  63         // Not exactly following number definition (several dots are seen as OK, etc.)
  64         // but probably enough in most cases. [pP] is for hex floats.
  65         CharacterSet setNumber(CharacterSet::setDigits, ".-+abcdefpABCDEFP");
  66         CharacterSet setExponent(CharacterSet::setNone, "eEpP");
  67         CharacterSet setLuaOperator(CharacterSet::setNone, "*/-+()={}~[];<>,.^%:#&|");
  68         CharacterSet setEscapeSkip(CharacterSet::setNone, "\"'\\");
  69
  70         Sci_Position currentLine = styler.GetLine(startPos);
  71         // Initialize long string [[ ... ]] or block comment --[[ ... ]] nesting level,
  72         // if we are inside such a string. Block comment was introduced in Lua 5.0,
  73         // blocks with separators [=[ ... ]=] in Lua 5.1.
  74         // Continuation of a string (\z whitespace escaping) is controlled by stringWs.
  75         int nestLevel = 0;
  76         int sepCount = 0;
  77         int stringWs = 0;
  78         if (initStyle == SCE_LUA_LITERALSTRING || initStyle == SCE_LUA_COMMENT ||
  79                 initStyle == SCE_LUA_STRING || initStyle == SCE_LUA_CHARACTER) {
  80                 int lineState = styler.GetLineState(currentLine - 1);
  81                 nestLevel = lineState >> 9;
  82                 sepCount = lineState & 0xFF;
  83                 stringWs = lineState & 0x100;
  84         }
  85
  86         // Do not leak onto next line
  87         if (initStyle == SCE_LUA_STRINGEOL || initStyle == SCE_LUA_COMMENTLINE || initStyle == SCE_LUA_PREPROCESSOR) {
  88                 initStyle = SCE_LUA_DEFAULT;
  89         }
  90
  91         StyleContext sc(startPos, length, initStyle, styler);
  92         if (startPos == 0 && sc.ch == '#') {
  93                 // shbang line: # is a comment only if first char of the script
  94                 sc.SetState(SCE_LUA_COMMENTLINE);
  95         }
  96         for (; sc.More(); sc.Forward()) {
  97                 if (sc.atLineEnd) {
  98                         // Update the line state, so it can be seen by next line
  99                         currentLine = styler.GetLine(sc.currentPos);
 100                         switch (sc.state) {
 101                         case SCE_LUA_LITERALSTRING:
 102                         case SCE_LUA_COMMENT:
 103                         case SCE_LUA_STRING:
 104                         case SCE_LUA_CHARACTER:
 105                                 // Inside a literal string, block comment or string, we set the line state
 106                                 styler.SetLineState(currentLine, (nestLevel << 9) | stringWs | sepCount);
 107                                 break;
 108                         default:
 109                                 // Reset the line state
 110                                 styler.SetLineState(currentLine, 0);
 111                                 break;
 112                         }
 113                 }
 114                 if (sc.atLineStart && (sc.state == SCE_LUA_STRING)) {
 115                         // Prevent SCE_LUA_STRINGEOL from leaking back to previous line
 116                         sc.SetState(SCE_LUA_STRING);
 117                 }
 118
 119                 // Handle string line continuation
 120                 if ((sc.state == SCE_LUA_STRING || sc.state == SCE_LUA_CHARACTER) &&
 121                                 sc.ch == '\\') {
 122                         if (sc.chNext == '\n' || sc.chNext == '\r') {
 123                                 sc.Forward();
 124                                 if (sc.ch == '\r' && sc.chNext == '\n') {
 125                                         sc.Forward();
 126                                 }
 127                                 continue;
 128                         }
 129                 }
 130
 131                 // Determine if the current state should terminate.
 132                 if (sc.state == SCE_LUA_OPERATOR) {
 133                         if (sc.ch == ':' && sc.chPrev == ':') { // :: <label> :: forward scan
 134                                 sc.Forward();
 135                                 Sci_Position ln = 0;
 136                                 while (IsASpaceOrTab(sc.GetRelative(ln)))       // skip over spaces/tabs
 137                                         ln++;
 138                                 Sci_Position ws1 = ln;
 139                                 if (setWordStart.Contains(sc.GetRelative(ln))) {
 140                                         int c, i = 0;
 141                                         char s[100];
 142                                         while (setWord.Contains(c = sc.GetRelative(ln))) {      // get potential label
 143                                                 if (i < 90)
 144                                                         s[i++] = static_cast<char>(c);
 145                                                 ln++;
 146                                         }
 147                                         s[i] = '\0'; Sci_Position lbl = ln;
 148                                         if (!keywords.InList(s)) {
 149                                                 while (IsASpaceOrTab(sc.GetRelative(ln)))       // skip over spaces/tabs
 150                                                         ln++;
 151                                                 Sci_Position ws2 = ln - lbl;
 152                                                 if (sc.GetRelative(ln) == ':' && sc.GetRelative(ln + 1) == ':') {
 153                                                         // final :: found, complete valid label construct
 154                                                         sc.ChangeState(SCE_LUA_LABEL);
 155                                                         if (ws1) {
 156                                                                 sc.SetState(SCE_LUA_DEFAULT);
 157                                                                 sc.ForwardBytes(ws1);
 158                                                         }
 159                                                         sc.SetState(SCE_LUA_LABEL);
 160                                                         sc.ForwardBytes(lbl - ws1);
 161                                                         if (ws2) {
 162                                                                 sc.SetState(SCE_LUA_DEFAULT);
 163                                                                 sc.ForwardBytes(ws2);
 164                                                         }
 165                                                         sc.SetState(SCE_LUA_LABEL);
 166                                                         sc.ForwardBytes(2);
 167                                                 }
 168                                         }
 169                                 }
 170                         }
 171                         sc.SetState(SCE_LUA_DEFAULT);
 172                 } else if (sc.state == SCE_LUA_NUMBER) {
 173                         // We stop the number definition on non-numerical non-dot non-eEpP non-sign non-hexdigit char
 174                         if (!setNumber.Contains(sc.ch)) {
 175                                 sc.SetState(SCE_LUA_DEFAULT);
 176                         } else if (sc.ch == '-' || sc.ch == '+') {
 177                                 if (!setExponent.Contains(sc.chPrev))
 178                                         sc.SetState(SCE_LUA_DEFAULT);
 179                         }
 180                 } else if (sc.state == SCE_LUA_IDENTIFIER) {
 181                         if (!(setWord.Contains(sc.ch) || sc.ch == '.') || sc.Match('.', '.')) {
 182                                 char s[100];
 183                                 sc.GetCurrent(s, sizeof(s));
 184                                 if (keywords.InList(s)) {
 185                                         sc.ChangeState(SCE_LUA_WORD);
 186                                         if (strcmp(s, "goto") == 0) {   // goto <label> forward scan
 187                                                 sc.SetState(SCE_LUA_DEFAULT);
 188                                                 while (IsASpaceOrTab(sc.ch) && !sc.atLineEnd)
 189                                                         sc.Forward();
 190                                                 if (setWordStart.Contains(sc.ch)) {
 191                                                         sc.SetState(SCE_LUA_LABEL);
 192                                                         sc.Forward();
 193                                                         while (setWord.Contains(sc.ch))
 194                                                                 sc.Forward();
 195                                                         sc.GetCurrent(s, sizeof(s));
 196                                                         if (keywords.InList(s))
 197                                                                 sc.ChangeState(SCE_LUA_WORD);
 198                                                 }
 199                                                 sc.SetState(SCE_LUA_DEFAULT);
 200                                         }
 201                                 } else if (keywords2.InList(s)) {
 202                                         sc.ChangeState(SCE_LUA_WORD2);
 203                                 } else if (keywords3.InList(s)) {
 204                                         sc.ChangeState(SCE_LUA_WORD3);
 205                                 } else if (keywords4.InList(s)) {
 206                                         sc.ChangeState(SCE_LUA_WORD4);
 207                                 } else if (keywords5.InList(s)) {
 208                                         sc.ChangeState(SCE_LUA_WORD5);
 209                                 } else if (keywords6.InList(s)) {
 210                                         sc.ChangeState(SCE_LUA_WORD6);
 211                                 } else if (keywords7.InList(s)) {
 212                                         sc.ChangeState(SCE_LUA_WORD7);
 213                                 } else if (keywords8.InList(s)) {
 214                                         sc.ChangeState(SCE_LUA_WORD8);
 215                                 }
 216                                 sc.SetState(SCE_LUA_DEFAULT);
 217                         }
 218                 } else if (sc.state == SCE_LUA_COMMENTLINE || sc.state == SCE_LUA_PREPROCESSOR) {
 219                         if (sc.atLineEnd) {
 220                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 221                         }
 222                 } else if (sc.state == SCE_LUA_STRING) {
 223                         if (stringWs) {
 224                                 if (!IsASpace(sc.ch))
 225                                         stringWs = 0;
 226                         }
 227                         if (sc.ch == '\\') {
 228                                 if (setEscapeSkip.Contains(sc.chNext)) {
 229                                         sc.Forward();
 230                                 } else if (sc.chNext == 'z') {
 231                                         sc.Forward();
 232                                         stringWs = 0x100;
 233                                 }
 234                         } else if (sc.ch == '\"') {
 235                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 236                         } else if (stringWs == 0 && sc.atLineEnd) {
 237                                 sc.ChangeState(SCE_LUA_STRINGEOL);
 238                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 239                         }
 240                 } else if (sc.state == SCE_LUA_CHARACTER) {
 241                         if (stringWs) {
 242                                 if (!IsASpace(sc.ch))
 243                                         stringWs = 0;
 244                         }
 245                         if (sc.ch == '\\') {
 246                                 if (setEscapeSkip.Contains(sc.chNext)) {
 247                                         sc.Forward();
 248                                 } else if (sc.chNext == 'z') {
 249                                         sc.Forward();
 250                                         stringWs = 0x100;
 251                                 }
 252                         } else if (sc.ch == '\'') {
 253                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 254                         } else if (stringWs == 0 && sc.atLineEnd) {
 255                                 sc.ChangeState(SCE_LUA_STRINGEOL);
 256                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 257                         }
 258                 } else if (sc.state == SCE_LUA_LITERALSTRING || sc.state == SCE_LUA_COMMENT) {
 259                         if (sc.ch == '[') {
 260                                 int sep = LongDelimCheck(sc);
 261                                 if (sep == 1 && sepCount == 1) {    // [[-only allowed to nest
 262                                         nestLevel++;
 263                                         sc.Forward();
 264                                 }
 265                         } else if (sc.ch == ']') {
 266                                 int sep = LongDelimCheck(sc);
 267                                 if (sep == 1 && sepCount == 1) {    // un-nest with ]]-only
 268                                         nestLevel--;
 269                                         sc.Forward();
 270                                         if (nestLevel == 0) {
 271                                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 272                                         }
 273                                 } else if (sep > 1 && sep == sepCount) {   // ]=]-style delim
 274                                         sc.Forward(sep);
 275                                         sc.ForwardSetState(SCE_LUA_DEFAULT);
 276                                 }
 277                         }
 278                 }
 279
 280                 // Determine if a new state should be entered.
 281                 if (sc.state == SCE_LUA_DEFAULT) {
 282                         if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
 283                                 sc.SetState(SCE_LUA_NUMBER);
 284                                 if (sc.ch == '0' && toupper(sc.chNext) == 'X') {
 285                                         sc.Forward();
 286                                 }
 287                         } else if (setWordStart.Contains(sc.ch)) {
 288                                 sc.SetState(SCE_LUA_IDENTIFIER);
 289                         } else if (sc.ch == '\"') {
 290                                 sc.SetState(SCE_LUA_STRING);
 291                                 stringWs = 0;
 292                         } else if (sc.ch == '\'') {
 293                                 sc.SetState(SCE_LUA_CHARACTER);
 294                                 stringWs = 0;
 295                         } else if (sc.ch == '[') {
 296                                 sepCount = LongDelimCheck(sc);
 297                                 if (sepCount == 0) {
 298                                         sc.SetState(SCE_LUA_OPERATOR);
 299                                 } else {
 300                                         nestLevel = 1;
 301                                         sc.SetState(SCE_LUA_LITERALSTRING);
 302                                         sc.Forward(sepCount);
 303                                 }
 304                         } else if (sc.Match('-', '-')) {
 305                                 sc.SetState(SCE_LUA_COMMENTLINE);
 306                                 if (sc.Match("--[")) {
 307                                         sc.Forward(2);
 308                                         sepCount = LongDelimCheck(sc);
 309                                         if (sepCount > 0) {
 310                                                 nestLevel = 1;
 311                                                 sc.ChangeState(SCE_LUA_COMMENT);
 312                                                 sc.Forward(sepCount);
 313                                         }
 314                                 } else {
 315                                         sc.Forward();
 316                                 }
 317                         } else if (sc.atLineStart && sc.Match('$')) {
 318                                 sc.SetState(SCE_LUA_PREPROCESSOR);      // Obsolete since Lua 4.0, but still in old code
 319                         } else if (setLuaOperator.Contains(sc.ch)) {
 320                                 sc.SetState(SCE_LUA_OPERATOR);
 321                         }
 322                 }
 323         }
 324
 325         if (setWord.Contains(sc.chPrev) || sc.chPrev == '.') {
 326                 char s[100];
 327                 sc.GetCurrent(s, sizeof(s));
 328                 if (keywords.InList(s)) {
 329                         sc.ChangeState(SCE_LUA_WORD);
 330                 } else if (keywords2.InList(s)) {
 331                         sc.ChangeState(SCE_LUA_WORD2);
 332                 } else if (keywords3.InList(s)) {
 333                         sc.ChangeState(SCE_LUA_WORD3);
 334                 } else if (keywords4.InList(s)) {
 335                         sc.ChangeState(SCE_LUA_WORD4);
 336                 } else if (keywords5.InList(s)) {
 337                         sc.ChangeState(SCE_LUA_WORD5);
 338                 } else if (keywords6.InList(s)) {
 339                         sc.ChangeState(SCE_LUA_WORD6);
 340                 } else if (keywords7.InList(s)) {
 341                         sc.ChangeState(SCE_LUA_WORD7);
 342                 } else if (keywords8.InList(s)) {
 343                         sc.ChangeState(SCE_LUA_WORD8);
 344                 }
 345         }
 346
 347         sc.Complete();
 348 }
 349
 350 static void FoldLuaDoc(Sci_PositionU startPos, Sci_Position length, int /* initStyle */, WordList *[],
 351                        Accessor &styler) {
 352         Sci_PositionU lengthDoc = startPos + length;
 353         int visibleChars = 0;
 354         Sci_Position lineCurrent = styler.GetLine(startPos);
 355         int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
 356         int levelCurrent = levelPrev;
 357         char chNext = styler[startPos];
 358         bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;
 359         int styleNext = styler.StyleAt(startPos);
 360
 361         for (Sci_PositionU i = startPos; i < lengthDoc; i++) {
 362                 char ch = chNext;
 363                 chNext = styler.SafeGetCharAt(i + 1);
 364                 int style = styleNext;
 365                 styleNext = styler.StyleAt(i + 1);
 366                 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
 367                 if (style == SCE_LUA_WORD) {
 368                         if (ch == 'i' || ch == 'd' || ch == 'f' || ch == 'e' || ch == 'r' || ch == 'u') {
 369                                 char s[10] = "";
 370                                 for (Sci_PositionU j = 0; j < 8; j++) {
 371                                         if (!iswordchar(styler[i + j])) {
 372                                                 break;
 373                                         }
 374                                         s[j] = styler[i + j];
 375                                         s[j + 1] = '\0';
 376                                 }
 377
 378                                 if ((strcmp(s, "if") == 0) || (strcmp(s, "do") == 0) || (strcmp(s, "function") == 0) || (strcmp(s, "repeat") == 0)) {
 379                                         levelCurrent++;
 380                                 }
 381                                 if ((strcmp(s, "end") == 0) || (strcmp(s, "elseif") == 0) || (strcmp(s, "until") == 0)) {
 382                                         levelCurrent--;
 383                                 }
 384                         }
 385                 } else if (style == SCE_LUA_OPERATOR) {
 386                         if (ch == '{' || ch == '(') {
 387                                 levelCurrent++;
 388                         } else if (ch == '}' || ch == ')') {
 389                                 levelCurrent--;
 390                         }
 391                 } else if (style == SCE_LUA_LITERALSTRING || style == SCE_LUA_COMMENT) {
 392                         if (ch == '[') {
 393                                 levelCurrent++;
 394                         } else if (ch == ']') {
 395                                 levelCurrent--;
 396                         }
 397                 }
 398
 399                 if (atEOL) {
 400                         int lev = levelPrev;
 401                         if (visibleChars == 0 && foldCompact) {
 402                                 lev |= SC_FOLDLEVELWHITEFLAG;
 403                         }
 404                         if ((levelCurrent > levelPrev) && (visibleChars > 0)) {
 405                                 lev |= SC_FOLDLEVELHEADERFLAG;
 406                         }
 407                         if (lev != styler.LevelAt(lineCurrent)) {
 408                                 styler.SetLevel(lineCurrent, lev);
 409                         }
 410                         lineCurrent++;
 411                         levelPrev = levelCurrent;
 412                         visibleChars = 0;
 413                 }
 414                 if (!isspacechar(ch)) {
 415                         visibleChars++;
 416                 }
 417         }
 418         // Fill in the real level of the next line, keeping the current flags as they will be filled in later
 419
 420         int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;
 421         styler.SetLevel(lineCurrent, levelPrev | flagsNext);
 422 }
 423
 424 static const char * const luaWordListDesc[] = {
 425         "Keywords",
 426         "Basic functions",
 427         "String, (table) & math functions",
 428         "(coroutines), I/O & system facilities",
 429         "user1",
 430         "user2",
 431         "user3",
 432         "user4",
 433         0
 434 };
 435
 436 LexerModule lmLua(SCLEX_LUA, ColouriseLuaDoc, "lua", FoldLuaDoc, luaWordListDesc);