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