src/stc/scintilla/src/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 <ctype.h>
  13 #include <stdarg.h>
  14 #include <stdio.h>
  15
  16 #include "Platform.h"
  17
  18 #include "PropSet.h"
  19 #include "Accessor.h"
  20 #include "StyleContext.h"
  21 #include "KeyWords.h"
  22 #include "Scintilla.h"
  23 #include "SciLexer.h"
  24
  25 // Extended to accept accented characters
  26 static inline bool IsAWordChar(int ch) {
  27         return ch >= 0x80 ||
  28                (isalnum(ch) || ch == '.' || ch == '_');
  29 }
  30
  31 static inline bool IsAWordStart(int ch) {
  32         return ch >= 0x80 ||
  33                (isalpha(ch) || ch == '_');
  34 }
  35
  36 static inline bool IsANumberChar(int ch) {
  37         // Not exactly following number definition (several dots are seen as OK, etc.)
  38         // but probably enough in most cases.
  39         return (ch < 0x80) &&
  40                 (isdigit(ch) || toupper(ch) == 'E' ||
  41              ch == '.' || ch == '-' || ch == '+');
  42 }
  43
  44 static inline bool IsLuaOperator(int ch) {
  45         if (ch >= 0x80 || isalnum(ch)) {
  46                 return false;
  47         }
  48         // '.' left out as it is used to make up numbers
  49         if (ch == '*' || ch == '/' || ch == '-' || ch == '+' ||
  50                 ch == '(' || ch == ')' || ch == '=' ||
  51                 ch == '{' || ch == '}' || ch == '~' ||
  52                 ch == '[' || ch == ']' || ch == ';' ||
  53                 ch == '<' || ch == '>' || ch == ',' ||
  54                 ch == '.' || ch == '^' || ch == '%' || ch == ':') {
  55                 return true;
  56         }
  57         return false;
  58 }
  59
  60 // Test for [=[ ... ]=] delimiters, returns 0 if it's only a [ or ],
  61 // return 1 for [[ or ]], returns >=2 for [=[ or ]=] and so on.
  62 // The maximum number of '=' characters allowed is 254.
  63 static int LongDelimCheck(StyleContext &sc) {
  64         int sep = 1;
  65         while (sc.GetRelative(sep) == '=' && sep < 0xFF)
  66                 sep++;
  67         if (sc.GetRelative(sep) == sc.ch)
  68                 return sep;
  69         return 0;
  70 }
  71
  72 static void ColouriseLuaDoc(
  73         unsigned int startPos,
  74         int length,
  75         int initStyle,
  76         WordList *keywordlists[],
  77         Accessor &styler) {
  78
  79         WordList &keywords = *keywordlists[0];
  80         WordList &keywords2 = *keywordlists[1];
  81         WordList &keywords3 = *keywordlists[2];
  82         WordList &keywords4 = *keywordlists[3];
  83         WordList &keywords5 = *keywordlists[4];
  84         WordList &keywords6 = *keywordlists[5];
  85         WordList &keywords7 = *keywordlists[6];
  86         WordList &keywords8 = *keywordlists[7];
  87
  88         int currentLine = styler.GetLine(startPos);
  89         // Initialize long string [[ ... ]] or block comment --[[ ... ]] nesting level,
  90         // if we are inside such a string. Block comment was introduced in Lua 5.0,
  91         // blocks with separators [=[ ... ]=] in Lua 5.1.
  92         int nestLevel = 0;
  93         int sepCount = 0;
  94         if (initStyle == SCE_LUA_LITERALSTRING || initStyle == SCE_LUA_COMMENT) {
  95                 int lineState = styler.GetLineState(currentLine - 1);
  96                 nestLevel = lineState >> 8;
  97                 sepCount = lineState & 0xFF;
  98         }
  99
 100         // Do not leak onto next line
 101         if (initStyle == SCE_LUA_STRINGEOL || initStyle == SCE_LUA_COMMENTLINE || initStyle == SCE_LUA_PREPROCESSOR) {
 102                 initStyle = SCE_LUA_DEFAULT;
 103         }
 104
 105         StyleContext sc(startPos, length, initStyle, styler);
 106         if (startPos == 0 && sc.ch == '#') {
 107                 // shbang line: # is a comment only if first char of the script
 108                 sc.SetState(SCE_LUA_COMMENTLINE);
 109         }
 110         for (; sc.More(); sc.Forward()) {
 111                 if (sc.atLineEnd) {
 112                         // Update the line state, so it can be seen by next line
 113                         currentLine = styler.GetLine(sc.currentPos);
 114                         switch (sc.state) {
 115                         case SCE_LUA_LITERALSTRING:
 116                         case SCE_LUA_COMMENT:
 117                                 // Inside a literal string or block comment, we set the line state
 118                                 styler.SetLineState(currentLine, (nestLevel << 8) | sepCount);
 119                                 break;
 120                         default:
 121                                 // Reset the line state
 122                                 styler.SetLineState(currentLine, 0);
 123                                 break;
 124                         }
 125                 }
 126                 if (sc.atLineStart && (sc.state == SCE_LUA_STRING)) {
 127                         // Prevent SCE_LUA_STRINGEOL from leaking back to previous line
 128                         sc.SetState(SCE_LUA_STRING);
 129                 }
 130
 131                 // Handle string line continuation
 132                 if ((sc.state == SCE_LUA_STRING || sc.state == SCE_LUA_CHARACTER) &&
 133                         sc.ch == '\\') {
 134                         if (sc.chNext == '\n' || sc.chNext == '\r') {
 135                                 sc.Forward();
 136                                 if (sc.ch == '\r' && sc.chNext == '\n') {
 137                                         sc.Forward();
 138                                 }
 139                                 continue;
 140                         }
 141                 }
 142
 143                 // Determine if the current state should terminate.
 144                 if (sc.state == SCE_LUA_OPERATOR) {
 145                         sc.SetState(SCE_LUA_DEFAULT);
 146                 } else if (sc.state == SCE_LUA_NUMBER) {
 147                         // We stop the number definition on non-numerical non-dot non-eE non-sign char
 148                         if (!IsANumberChar(sc.ch)) {
 149                                 sc.SetState(SCE_LUA_DEFAULT);
 150                         }
 151                 } else if (sc.state == SCE_LUA_IDENTIFIER) {
 152                         if (!IsAWordChar(sc.ch) || sc.Match('.', '.')) {
 153                                 char s[100];
 154                                 sc.GetCurrent(s, sizeof(s));
 155                                 if (keywords.InList(s)) {
 156                                         sc.ChangeState(SCE_LUA_WORD);
 157                                 } else if (keywords2.InList(s)) {
 158                                         sc.ChangeState(SCE_LUA_WORD2);
 159                                 } else if (keywords3.InList(s)) {
 160                                         sc.ChangeState(SCE_LUA_WORD3);
 161                                 } else if (keywords4.InList(s)) {
 162                                         sc.ChangeState(SCE_LUA_WORD4);
 163                                 } else if (keywords5.InList(s)) {
 164                                         sc.ChangeState(SCE_LUA_WORD5);
 165                                 } else if (keywords6.InList(s)) {
 166                                         sc.ChangeState(SCE_LUA_WORD6);
 167                                 } else if (keywords6.InList(s)) {
 168                                         sc.ChangeState(SCE_LUA_WORD6);
 169                                 } else if (keywords7.InList(s)) {
 170                                         sc.ChangeState(SCE_LUA_WORD7);
 171                                 } else if (keywords8.InList(s)) {
 172                                         sc.ChangeState(SCE_LUA_WORD8);
 173                                 }
 174                                 sc.SetState(SCE_LUA_DEFAULT);
 175                         }
 176                 } else if (sc.state == SCE_LUA_COMMENTLINE || sc.state == SCE_LUA_PREPROCESSOR) {
 177                         if (sc.atLineEnd) {
 178                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 179                         }
 180                 } else if (sc.state == SCE_LUA_STRING) {
 181                         if (sc.ch == '\\') {
 182                                 if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\') {
 183                                         sc.Forward();
 184                                 }
 185                         } else if (sc.ch == '\"') {
 186                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 187                         } else if (sc.atLineEnd) {
 188                                 sc.ChangeState(SCE_LUA_STRINGEOL);
 189                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 190                         }
 191                 } else if (sc.state == SCE_LUA_CHARACTER) {
 192                         if (sc.ch == '\\') {
 193                                 if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\') {
 194                                         sc.Forward();
 195                                 }
 196                         } else if (sc.ch == '\'') {
 197                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 198                         } else if (sc.atLineEnd) {
 199                                 sc.ChangeState(SCE_LUA_STRINGEOL);
 200                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 201                         }
 202                 } else if (sc.state == SCE_LUA_LITERALSTRING || sc.state == SCE_LUA_COMMENT) {
 203                         if (sc.ch == '[') {
 204                                 int sep = LongDelimCheck(sc);
 205                                 if (sep == 1 && sepCount == 1) {    // [[-only allowed to nest
 206                                         nestLevel++;
 207                                         sc.Forward();
 208                                 }
 209                         } else if (sc.ch == ']') {
 210                                 int sep = LongDelimCheck(sc);
 211                                 if (sep == 1 && sepCount == 1) {    // un-nest with ]]-only
 212                                         nestLevel--;
 213                                         sc.Forward();
 214                                         if (nestLevel == 0) {
 215                                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 216                                         }
 217                                 } else if (sep > 1 && sep == sepCount) {   // ]=]-style delim
 218                                         sc.Forward(sep);
 219                                         sc.ForwardSetState(SCE_LUA_DEFAULT);
 220                                 }
 221                         }
 222                 }
 223
 224                 // Determine if a new state should be entered.
 225                 if (sc.state == SCE_LUA_DEFAULT) {
 226                         if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
 227                                 sc.SetState(SCE_LUA_NUMBER);
 228                         } else if (IsAWordStart(sc.ch)) {
 229                                 sc.SetState(SCE_LUA_IDENTIFIER);
 230                         } else if (sc.ch == '\"') {
 231                                 sc.SetState(SCE_LUA_STRING);
 232                         } else if (sc.ch == '\'') {
 233                                 sc.SetState(SCE_LUA_CHARACTER);
 234                         } else if (sc.ch == '[') {
 235                                 sepCount = LongDelimCheck(sc);
 236                                 if (sepCount == 0) {
 237                                         sc.SetState(SCE_LUA_OPERATOR);
 238                                 } else {
 239                                         nestLevel = 1;
 240                                         sc.SetState(SCE_LUA_LITERALSTRING);
 241                                         sc.Forward(sepCount);
 242                                 }
 243                         } else if (sc.Match('-', '-')) {
 244                                 sc.SetState(SCE_LUA_COMMENTLINE);
 245                                 if (sc.Match("--[")) {
 246                                         sc.Forward(2);
 247                                         sepCount = LongDelimCheck(sc);
 248                                         if (sepCount > 0) {
 249                                                 nestLevel = 1;
 250                                                 sc.ChangeState(SCE_LUA_COMMENT);
 251                                                 sc.Forward(sepCount);
 252                                         }
 253                                 } else {
 254                                         sc.Forward();
 255                                 }
 256                         } else if (sc.atLineStart && sc.Match('$')) {
 257                                 sc.SetState(SCE_LUA_PREPROCESSOR);      // Obsolete since Lua 4.0, but still in old code
 258                         } else if (IsLuaOperator(static_cast<char>(sc.ch))) {
 259                                 sc.SetState(SCE_LUA_OPERATOR);
 260                         }
 261                 }
 262         }
 263         sc.Complete();
 264 }
 265
 266 static void FoldLuaDoc(unsigned int startPos, int length, int /* initStyle */, WordList *[],
 267                        Accessor &styler) {
 268         unsigned int lengthDoc = startPos + length;
 269         int visibleChars = 0;
 270         int lineCurrent = styler.GetLine(startPos);
 271         int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
 272         int levelCurrent = levelPrev;
 273         char chNext = styler[startPos];
 274         bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;
 275         int styleNext = styler.StyleAt(startPos);
 276         char s[10];
 277
 278         for (unsigned int i = startPos; i < lengthDoc; i++) {
 279                 char ch = chNext;
 280                 chNext = styler.SafeGetCharAt(i + 1);
 281                 int style = styleNext;
 282                 styleNext = styler.StyleAt(i + 1);
 283                 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
 284                 if (style == SCE_LUA_WORD) {
 285                         if (ch == 'i' || ch == 'd' || ch == 'f' || ch == 'e' || ch == 'r' || ch == 'u') {
 286                                 for (unsigned int j = 0; j < 8; j++) {
 287                                         if (!iswordchar(styler[i + j])) {
 288                                                 break;
 289                                         }
 290                                         s[j] = styler[i + j];
 291                                         s[j + 1] = '\0';
 292                                 }
 293
 294                                 if ((strcmp(s, "if") == 0) || (strcmp(s, "do") == 0) || (strcmp(s, "function") == 0) || (strcmp(s, "repeat") == 0)) {
 295                                         levelCurrent++;
 296                                 }
 297                                 if ((strcmp(s, "end") == 0) || (strcmp(s, "elseif") == 0) || (strcmp(s, "until") == 0)) {
 298                                         levelCurrent--;
 299                                 }
 300                         }
 301                 } else if (style == SCE_LUA_OPERATOR) {
 302                         if (ch == '{' || ch == '(') {
 303                                 levelCurrent++;
 304                         } else if (ch == '}' || ch == ')') {
 305                                 levelCurrent--;
 306                         }
 307                 } else if (style == SCE_LUA_LITERALSTRING || style == SCE_LUA_COMMENT) {
 308                         if (ch == '[') {
 309                                 levelCurrent++;
 310                         } else if (ch == ']') {
 311                                 levelCurrent--;
 312                         }
 313                 }
 314
 315                 if (atEOL) {
 316                         int lev = levelPrev;
 317                         if (visibleChars == 0 && foldCompact) {
 318                                 lev |= SC_FOLDLEVELWHITEFLAG;
 319                         }
 320                         if ((levelCurrent > levelPrev) && (visibleChars > 0)) {
 321                                 lev |= SC_FOLDLEVELHEADERFLAG;
 322                         }
 323                         if (lev != styler.LevelAt(lineCurrent)) {
 324                                 styler.SetLevel(lineCurrent, lev);
 325                         }
 326                         lineCurrent++;
 327                         levelPrev = levelCurrent;
 328                         visibleChars = 0;
 329                 }
 330                 if (!isspacechar(ch)) {
 331                         visibleChars++;
 332                 }
 333         }
 334         // Fill in the real level of the next line, keeping the current flags as they will be filled in later
 335
 336         int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;
 337         styler.SetLevel(lineCurrent, levelPrev | flagsNext);
 338 }
 339
 340 static const char * const luaWordListDesc[] = {
 341         "Keywords",
 342         "Basic functions",
 343         "String, (table) & math functions",
 344         "(coroutines), I/O & system facilities",
 345         "user1",
 346         "user2",
 347         "user3",
 348         "user4",
 349         0
 350 };
 351
 352 LexerModule lmLua(SCLEX_LUA, ColouriseLuaDoc, "lua", FoldLuaDoc, luaWordListDesc);