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