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 #ifdef SCI_NAMESPACE
  26 using namespace Scintilla;
  27 #endif
  28
  29 // Extended to accept accented characters
  30 static inline bool IsAWordChar(int ch) {
  31         return ch >= 0x80 ||
  32                (isalnum(ch) || ch == '.' || ch == '_');
  33 }
  34
  35 static inline bool IsAWordStart(int ch) {
  36         return ch >= 0x80 ||
  37                (isalpha(ch) || ch == '_');
  38 }
  39
  40 static inline bool IsANumberChar(int ch) {
  41         // Not exactly following number definition (several dots are seen as OK, etc.)
  42         // but probably enough in most cases.
  43         return (ch < 0x80) &&
  44                 (isdigit(ch) || toupper(ch) == 'E' ||
  45                 ch == '.' || ch == '-' || ch == '+' ||
  46                 (ch >= 'a' && ch <= 'f') || (ch >= 'A' && ch <= 'F'));
  47 }
  48
  49 static inline bool IsLuaOperator(int ch) {
  50         if (ch >= 0x80 || isalnum(ch)) {
  51                 return false;
  52         }
  53         // '.' left out as it is used to make up numbers
  54         if (ch == '*' || ch == '/' || ch == '-' || ch == '+' ||
  55                 ch == '(' || ch == ')' || ch == '=' ||
  56                 ch == '{' || ch == '}' || ch == '~' ||
  57                 ch == '[' || ch == ']' || ch == ';' ||
  58                 ch == '<' || ch == '>' || ch == ',' ||
  59                 ch == '.' || ch == '^' || ch == '%' || ch == ':' ||
  60                 ch == '#') {
  61                 return true;
  62         }
  63         return false;
  64 }
  65
  66 // Test for [=[ ... ]=] delimiters, returns 0 if it's only a [ or ],
  67 // return 1 for [[ or ]], returns >=2 for [=[ or ]=] and so on.
  68 // The maximum number of '=' characters allowed is 254.
  69 static int LongDelimCheck(StyleContext &sc) {
  70         int sep = 1;
  71         while (sc.GetRelative(sep) == '=' && sep < 0xFF)
  72                 sep++;
  73         if (sc.GetRelative(sep) == sc.ch)
  74                 return sep;
  75         return 0;
  76 }
  77
  78 static void ColouriseLuaDoc(
  79         unsigned int startPos,
  80         int length,
  81         int initStyle,
  82         WordList *keywordlists[],
  83         Accessor &styler) {
  84
  85         WordList &keywords = *keywordlists[0];
  86         WordList &keywords2 = *keywordlists[1];
  87         WordList &keywords3 = *keywordlists[2];
  88         WordList &keywords4 = *keywordlists[3];
  89         WordList &keywords5 = *keywordlists[4];
  90         WordList &keywords6 = *keywordlists[5];
  91         WordList &keywords7 = *keywordlists[6];
  92         WordList &keywords8 = *keywordlists[7];
  93
  94         int currentLine = styler.GetLine(startPos);
  95         // Initialize long string [[ ... ]] or block comment --[[ ... ]] nesting level,
  96         // if we are inside such a string. Block comment was introduced in Lua 5.0,
  97         // blocks with separators [=[ ... ]=] in Lua 5.1.
  98         int nestLevel = 0;
  99         int sepCount = 0;
 100         if (initStyle == SCE_LUA_LITERALSTRING || initStyle == SCE_LUA_COMMENT) {
 101                 int lineState = styler.GetLineState(currentLine - 1);
 102                 nestLevel = lineState >> 8;
 103                 sepCount = lineState & 0xFF;
 104         }
 105
 106         // Do not leak onto next line
 107         if (initStyle == SCE_LUA_STRINGEOL || initStyle == SCE_LUA_COMMENTLINE || initStyle == SCE_LUA_PREPROCESSOR) {
 108                 initStyle = SCE_LUA_DEFAULT;
 109         }
 110
 111         StyleContext sc(startPos, length, initStyle, styler);
 112         if (startPos == 0 && sc.ch == '#') {
 113                 // shbang line: # is a comment only if first char of the script
 114                 sc.SetState(SCE_LUA_COMMENTLINE);
 115         }
 116         for (; sc.More(); sc.Forward()) {
 117                 if (sc.atLineEnd) {
 118                         // Update the line state, so it can be seen by next line
 119                         currentLine = styler.GetLine(sc.currentPos);
 120                         switch (sc.state) {
 121                         case SCE_LUA_LITERALSTRING:
 122                         case SCE_LUA_COMMENT:
 123                                 // Inside a literal string or block comment, we set the line state
 124                                 styler.SetLineState(currentLine, (nestLevel << 8) | sepCount);
 125                                 break;
 126                         default:
 127                                 // Reset the line state
 128                                 styler.SetLineState(currentLine, 0);
 129                                 break;
 130                         }
 131                 }
 132                 if (sc.atLineStart && (sc.state == SCE_LUA_STRING)) {
 133                         // Prevent SCE_LUA_STRINGEOL from leaking back to previous line
 134                         sc.SetState(SCE_LUA_STRING);
 135                 }
 136
 137                 // Handle string line continuation
 138                 if ((sc.state == SCE_LUA_STRING || sc.state == SCE_LUA_CHARACTER) &&
 139                         sc.ch == '\\') {
 140                         if (sc.chNext == '\n' || sc.chNext == '\r') {
 141                                 sc.Forward();
 142                                 if (sc.ch == '\r' && sc.chNext == '\n') {
 143                                         sc.Forward();
 144                                 }
 145                                 continue;
 146                         }
 147                 }
 148
 149                 // Determine if the current state should terminate.
 150                 if (sc.state == SCE_LUA_OPERATOR) {
 151                         sc.SetState(SCE_LUA_DEFAULT);
 152                 } else if (sc.state == SCE_LUA_NUMBER) {
 153                         // We stop the number definition on non-numerical non-dot non-eE non-sign non-hexdigit char
 154                         if (!IsANumberChar(sc.ch)) {
 155                                 sc.SetState(SCE_LUA_DEFAULT);
 156                         } else if (sc.ch == '-' || sc.ch == '+') {
 157                                 if (sc.chPrev != 'E' && sc.chPrev != 'e')
 158                                         sc.SetState(SCE_LUA_DEFAULT);
 159                         }
 160                 } else if (sc.state == SCE_LUA_IDENTIFIER) {
 161                         if (!IsAWordChar(sc.ch) || sc.Match('.', '.')) {
 162                                 char s[100];
 163                                 sc.GetCurrent(s, sizeof(s));
 164                                 if (keywords.InList(s)) {
 165                                         sc.ChangeState(SCE_LUA_WORD);
 166                                 } else if (keywords2.InList(s)) {
 167                                         sc.ChangeState(SCE_LUA_WORD2);
 168                                 } else if (keywords3.InList(s)) {
 169                                         sc.ChangeState(SCE_LUA_WORD3);
 170                                 } else if (keywords4.InList(s)) {
 171                                         sc.ChangeState(SCE_LUA_WORD4);
 172                                 } else if (keywords5.InList(s)) {
 173                                         sc.ChangeState(SCE_LUA_WORD5);
 174                                 } else if (keywords6.InList(s)) {
 175                                         sc.ChangeState(SCE_LUA_WORD6);
 176                                 } else if (keywords7.InList(s)) {
 177                                         sc.ChangeState(SCE_LUA_WORD7);
 178                                 } else if (keywords8.InList(s)) {
 179                                         sc.ChangeState(SCE_LUA_WORD8);
 180                                 }
 181                                 sc.SetState(SCE_LUA_DEFAULT);
 182                         }
 183                 } else if (sc.state == SCE_LUA_COMMENTLINE || sc.state == SCE_LUA_PREPROCESSOR) {
 184                         if (sc.atLineEnd) {
 185                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 186                         }
 187                 } else if (sc.state == SCE_LUA_STRING) {
 188                         if (sc.ch == '\\') {
 189                                 if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\') {
 190                                         sc.Forward();
 191                                 }
 192                         } else if (sc.ch == '\"') {
 193                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 194                         } else if (sc.atLineEnd) {
 195                                 sc.ChangeState(SCE_LUA_STRINGEOL);
 196                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 197                         }
 198                 } else if (sc.state == SCE_LUA_CHARACTER) {
 199                         if (sc.ch == '\\') {
 200                                 if (sc.chNext == '\"' || sc.chNext == '\'' || sc.chNext == '\\') {
 201                                         sc.Forward();
 202                                 }
 203                         } else if (sc.ch == '\'') {
 204                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 205                         } else if (sc.atLineEnd) {
 206                                 sc.ChangeState(SCE_LUA_STRINGEOL);
 207                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 208                         }
 209                 } else if (sc.state == SCE_LUA_LITERALSTRING || sc.state == SCE_LUA_COMMENT) {
 210                         if (sc.ch == '[') {
 211                                 int sep = LongDelimCheck(sc);
 212                                 if (sep == 1 && sepCount == 1) {    // [[-only allowed to nest
 213                                         nestLevel++;
 214                                         sc.Forward();
 215                                 }
 216                         } else if (sc.ch == ']') {
 217                                 int sep = LongDelimCheck(sc);
 218                                 if (sep == 1 && sepCount == 1) {    // un-nest with ]]-only
 219                                         nestLevel--;
 220                                         sc.Forward();
 221                                         if (nestLevel == 0) {
 222                                                 sc.ForwardSetState(SCE_LUA_DEFAULT);
 223                                         }
 224                                 } else if (sep > 1 && sep == sepCount) {   // ]=]-style delim
 225                                         sc.Forward(sep);
 226                                         sc.ForwardSetState(SCE_LUA_DEFAULT);
 227                                 }
 228                         }
 229                 }
 230
 231                 // Determine if a new state should be entered.
 232                 if (sc.state == SCE_LUA_DEFAULT) {
 233                         if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext))) {
 234                                 sc.SetState(SCE_LUA_NUMBER);
 235                                 if (sc.ch == '0' && toupper(sc.chNext) == 'X') {
 236                                         sc.Forward(1);
 237                                 }
 238                         } else if (IsAWordStart(sc.ch)) {
 239                                 sc.SetState(SCE_LUA_IDENTIFIER);
 240                         } else if (sc.ch == '\"') {
 241                                 sc.SetState(SCE_LUA_STRING);
 242                         } else if (sc.ch == '\'') {
 243                                 sc.SetState(SCE_LUA_CHARACTER);
 244                         } else if (sc.ch == '[') {
 245                                 sepCount = LongDelimCheck(sc);
 246                                 if (sepCount == 0) {
 247                                         sc.SetState(SCE_LUA_OPERATOR);
 248                                 } else {
 249                                         nestLevel = 1;
 250                                         sc.SetState(SCE_LUA_LITERALSTRING);
 251                                         sc.Forward(sepCount);
 252                                 }
 253                         } else if (sc.Match('-', '-')) {
 254                                 sc.SetState(SCE_LUA_COMMENTLINE);
 255                                 if (sc.Match("--[")) {
 256                                         sc.Forward(2);
 257                                         sepCount = LongDelimCheck(sc);
 258                                         if (sepCount > 0) {
 259                                                 nestLevel = 1;
 260                                                 sc.ChangeState(SCE_LUA_COMMENT);
 261                                                 sc.Forward(sepCount);
 262                                         }
 263                                 } else {
 264                                         sc.Forward();
 265                                 }
 266                         } else if (sc.atLineStart && sc.Match('$')) {
 267                                 sc.SetState(SCE_LUA_PREPROCESSOR);      // Obsolete since Lua 4.0, but still in old code
 268                         } else if (IsLuaOperator(static_cast<char>(sc.ch))) {
 269                                 sc.SetState(SCE_LUA_OPERATOR);
 270                         }
 271                 }
 272         }
 273         sc.Complete();
 274 }
 275
 276 static void FoldLuaDoc(unsigned int startPos, int length, int /* initStyle */, WordList *[],
 277                        Accessor &styler) {
 278         unsigned int lengthDoc = startPos + length;
 279         int visibleChars = 0;
 280         int lineCurrent = styler.GetLine(startPos);
 281         int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
 282         int levelCurrent = levelPrev;
 283         char chNext = styler[startPos];
 284         bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;
 285         int styleNext = styler.StyleAt(startPos);
 286         char s[10];
 287
 288         for (unsigned int i = startPos; i < lengthDoc; i++) {
 289                 char ch = chNext;
 290                 chNext = styler.SafeGetCharAt(i + 1);
 291                 int style = styleNext;
 292                 styleNext = styler.StyleAt(i + 1);
 293                 bool atEOL = (ch == '\r' && chNext != '\n') || (ch == '\n');
 294                 if (style == SCE_LUA_WORD) {
 295                         if (ch == 'i' || ch == 'd' || ch == 'f' || ch == 'e' || ch == 'r' || ch == 'u') {
 296                                 for (unsigned int j = 0; j < 8; j++) {
 297                                         if (!iswordchar(styler[i + j])) {
 298                                                 break;
 299                                         }
 300                                         s[j] = styler[i + j];
 301                                         s[j + 1] = '\0';
 302                                 }
 303
 304                                 if ((strcmp(s, "if") == 0) || (strcmp(s, "do") == 0) || (strcmp(s, "function") == 0) || (strcmp(s, "repeat") == 0)) {
 305                                         levelCurrent++;
 306                                 }
 307                                 if ((strcmp(s, "end") == 0) || (strcmp(s, "elseif") == 0) || (strcmp(s, "until") == 0)) {
 308                                         levelCurrent--;
 309                                 }
 310                         }
 311                 } else if (style == SCE_LUA_OPERATOR) {
 312                         if (ch == '{' || ch == '(') {
 313                                 levelCurrent++;
 314                         } else if (ch == '}' || ch == ')') {
 315                                 levelCurrent--;
 316                         }
 317                 } else if (style == SCE_LUA_LITERALSTRING || style == SCE_LUA_COMMENT) {
 318                         if (ch == '[') {
 319                                 levelCurrent++;
 320                         } else if (ch == ']') {
 321                                 levelCurrent--;
 322                         }
 323                 }
 324
 325                 if (atEOL) {
 326                         int lev = levelPrev;
 327                         if (visibleChars == 0 && foldCompact) {
 328                                 lev |= SC_FOLDLEVELWHITEFLAG;
 329                         }
 330                         if ((levelCurrent > levelPrev) && (visibleChars > 0)) {
 331                                 lev |= SC_FOLDLEVELHEADERFLAG;
 332                         }
 333                         if (lev != styler.LevelAt(lineCurrent)) {
 334                                 styler.SetLevel(lineCurrent, lev);
 335                         }
 336                         lineCurrent++;
 337                         levelPrev = levelCurrent;
 338                         visibleChars = 0;
 339                 }
 340                 if (!isspacechar(ch)) {
 341                         visibleChars++;
 342                 }
 343         }
 344         // Fill in the real level of the next line, keeping the current flags as they will be filled in later
 345
 346         int flagsNext = styler.LevelAt(lineCurrent) & ~SC_FOLDLEVELNUMBERMASK;
 347         styler.SetLevel(lineCurrent, levelPrev | flagsNext);
 348 }
 349
 350 static const char * const luaWordListDesc[] = {
 351         "Keywords",
 352         "Basic functions",
 353         "String, (table) & math functions",
 354         "(coroutines), I/O & system facilities",
 355         "user1",
 356         "user2",
 357         "user3",
 358         "user4",
 359         0
 360 };
 361
 362 LexerModule lmLua(SCLEX_LUA, ColouriseLuaDoc, "lua", FoldLuaDoc, luaWordListDesc);