ext/scintilla/lexers/LexMySQL.cxx

   1 /**
   2  * Scintilla source code edit control
   3  * @file LexMySQL.cxx
   4  * Lexer for MySQL
   5  *
   6  * Improved by Mike Lischke <mike.lischke@sun.com>
   7  * Adopted from LexSQL.cxx by Anders Karlsson <anders@mysql.com>
   8  * Original work by Neil Hodgson <neilh@scintilla.org>
   9  * Copyright 1998-2005 by Neil Hodgson <neilh@scintilla.org>
  10  * The License.txt file describes the conditions under which this software may be distributed.
  11  */
  12
  13 #include <stdlib.h>
  14 #include <string.h>
  15 #include <stdio.h>
  16 #include <stdarg.h>
  17 #include <assert.h>
  18 #include <ctype.h>
  19
  20 #include "ILexer.h"
  21 #include "Scintilla.h"
  22 #include "SciLexer.h"
  23
  24 #include "WordList.h"
  25 #include "LexAccessor.h"
  26 #include "Accessor.h"
  27 #include "StyleContext.h"
  28 #include "CharacterSet.h"
  29 #include "LexerModule.h"
  30
  31 #ifdef SCI_NAMESPACE
  32 using namespace Scintilla;
  33 #endif
  34
  35 static inline bool IsAWordChar(int ch) {
  36         return (ch < 0x80) && (isalnum(ch) || ch == '_');
  37 }
  38
  39 static inline bool IsAWordStart(int ch) {
  40         return (ch < 0x80) && (isalpha(ch) || ch == '_');
  41 }
  42
  43 static inline bool IsADoxygenChar(int ch) {
  44         return (islower(ch) || ch == '$' || ch == '@' ||
  45                 ch == '\\' || ch == '&' || ch == '<' ||
  46                 ch == '>' || ch == '#' || ch == '{' ||
  47                 ch == '}' || ch == '[' || ch == ']');
  48 }
  49
  50 static inline bool IsANumberChar(int ch) {
  51         // Not exactly following number definition (several dots are seen as OK, etc.)
  52         // but probably enough in most cases.
  53         return (ch < 0x80) &&
  54                 (isdigit(ch) || toupper(ch) == 'E' ||
  55              ch == '.' || ch == '-' || ch == '+');
  56 }
  57
  58 //--------------------------------------------------------------------------------------------------
  59
  60 /**
  61  * Check if the current content context represent a keyword and set the context state if so.
  62  */
  63 static void CheckForKeyword(StyleContext& sc, WordList* keywordlists[])
  64 {
  65   int length = sc.LengthCurrent() + 1; // +1 for the next char
  66   char* s = new char[length];
  67   sc.GetCurrentLowered(s, length);
  68   if (keywordlists[0]->InList(s))
  69     sc.ChangeState(SCE_MYSQL_MAJORKEYWORD);
  70   else
  71     if (keywordlists[1]->InList(s))
  72       sc.ChangeState(SCE_MYSQL_KEYWORD);
  73     else
  74       if (keywordlists[2]->InList(s))
  75         sc.ChangeState(SCE_MYSQL_DATABASEOBJECT);
  76       else
  77         if (keywordlists[3]->InList(s))
  78           sc.ChangeState(SCE_MYSQL_FUNCTION);
  79         else
  80           if (keywordlists[5]->InList(s))
  81             sc.ChangeState(SCE_MYSQL_PROCEDUREKEYWORD);
  82           else
  83             if (keywordlists[6]->InList(s))
  84               sc.ChangeState(SCE_MYSQL_USER1);
  85             else
  86               if (keywordlists[7]->InList(s))
  87                 sc.ChangeState(SCE_MYSQL_USER2);
  88               else
  89                 if (keywordlists[8]->InList(s))
  90                   sc.ChangeState(SCE_MYSQL_USER3);
  91   delete [] s;
  92 }
  93
  94 //--------------------------------------------------------------------------------------------------
  95
  96 static void ColouriseMySQLDoc(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],
  97                             Accessor &styler)
  98 {
  99         StyleContext sc(startPos, length, initStyle, styler);
 100
 101         for (; sc.More(); sc.Forward())
 102   {
 103                 // Determine if the current state should terminate.
 104                 switch (sc.state)
 105     {
 106       case SCE_MYSQL_OPERATOR:
 107         sc.SetState(SCE_MYSQL_DEFAULT);
 108         break;
 109       case SCE_MYSQL_NUMBER:
 110         // We stop the number definition on non-numerical non-dot non-eE non-sign char.
 111         if (!IsANumberChar(sc.ch))
 112           sc.SetState(SCE_MYSQL_DEFAULT);
 113         break;
 114       case SCE_MYSQL_IDENTIFIER:
 115         // Switch from identifier to keyword state and open a new state for the new char.
 116         if (!IsAWordChar(sc.ch))
 117         {
 118           CheckForKeyword(sc, keywordlists);
 119
 120           // Additional check for function keywords needed.
 121           // A function name must be followed by an opening parenthesis.
 122           if (sc.state == SCE_MYSQL_FUNCTION && sc.ch != '(')
 123             sc.ChangeState(SCE_MYSQL_DEFAULT);
 124
 125           sc.SetState(SCE_MYSQL_DEFAULT);
 126         }
 127         break;
 128       case SCE_MYSQL_VARIABLE:
 129         if (!IsAWordChar(sc.ch))
 130           sc.SetState(SCE_MYSQL_DEFAULT);
 131         break;
 132       case SCE_MYSQL_SYSTEMVARIABLE:
 133         if (!IsAWordChar(sc.ch))
 134         {
 135           int length = sc.LengthCurrent() + 1;
 136           char* s = new char[length];
 137           sc.GetCurrentLowered(s, length);
 138
 139           // Check for known system variables here.
 140           if (keywordlists[4]->InList(&s[2]))
 141             sc.ChangeState(SCE_MYSQL_KNOWNSYSTEMVARIABLE);
 142           delete [] s;
 143
 144           sc.SetState(SCE_MYSQL_DEFAULT);
 145         }
 146         break;
 147       case SCE_MYSQL_QUOTEDIDENTIFIER:
 148         if (sc.ch == '`')
 149         {
 150           if (sc.chNext == '`')
 151             sc.Forward();       // Ignore it
 152           else
 153             sc.ForwardSetState(SCE_MYSQL_DEFAULT);
 154                                 }
 155                         break;
 156       case SCE_MYSQL_COMMENT:
 157       case SCE_MYSQL_HIDDENCOMMAND:
 158         if (sc.Match('*', '/'))
 159         {
 160           sc.Forward();
 161           sc.ForwardSetState(SCE_MYSQL_DEFAULT);
 162         }
 163         break;
 164       case SCE_MYSQL_COMMENTLINE:
 165         if (sc.atLineStart)
 166           sc.SetState(SCE_MYSQL_DEFAULT);
 167         break;
 168       case SCE_MYSQL_SQSTRING:
 169         if (sc.ch == '\\')
 170           sc.Forward(); // Escape sequence
 171         else
 172           if (sc.ch == '\'')
 173           {
 174             // End of single quoted string reached?
 175             if (sc.chNext == '\'')
 176               sc.Forward();
 177             else
 178               sc.ForwardSetState(SCE_MYSQL_DEFAULT);
 179           }
 180         break;
 181       case SCE_MYSQL_DQSTRING:
 182         if (sc.ch == '\\')
 183           sc.Forward(); // Escape sequence
 184         else
 185           if (sc.ch == '\"')
 186           {
 187             // End of single quoted string reached?
 188             if (sc.chNext == '\"')
 189               sc.Forward();
 190             else
 191               sc.ForwardSetState(SCE_MYSQL_DEFAULT);
 192           }
 193         break;
 194     }
 195
 196     // Determine if a new state should be entered.
 197     if (sc.state == SCE_MYSQL_DEFAULT)
 198     {
 199       switch (sc.ch)
 200       {
 201         case '@':
 202           if (sc.chNext == '@')
 203           {
 204             sc.SetState(SCE_MYSQL_SYSTEMVARIABLE);
 205             sc.Forward(2); // Skip past @@.
 206           }
 207           else
 208             if (IsAWordStart(sc.ch))
 209             {
 210               sc.SetState(SCE_MYSQL_VARIABLE);
 211               sc.Forward(); // Skip past @.
 212             }
 213             else
 214               sc.SetState(SCE_MYSQL_OPERATOR);
 215           break;
 216         case '`':
 217           sc.SetState(SCE_MYSQL_QUOTEDIDENTIFIER);
 218           break;
 219         case '#':
 220           sc.SetState(SCE_MYSQL_COMMENTLINE);
 221           break;
 222         case '\'':
 223           sc.SetState(SCE_MYSQL_SQSTRING);
 224           break;
 225         case '\"':
 226           sc.SetState(SCE_MYSQL_DQSTRING);
 227           break;
 228         default:
 229           if (IsADigit(sc.ch) || (sc.ch == '.' && IsADigit(sc.chNext)))
 230             sc.SetState(SCE_MYSQL_NUMBER);
 231           else
 232             if (IsAWordStart(sc.ch))
 233               sc.SetState(SCE_MYSQL_IDENTIFIER);
 234             else
 235               if (sc.Match('/', '*'))
 236               {
 237                 sc.SetState(SCE_MYSQL_COMMENT);
 238
 239                 // Skip comment introducer and check for hidden command.
 240                 sc.Forward(2);
 241                 if (sc.ch == '!')
 242                 {
 243                   sc.ChangeState(SCE_MYSQL_HIDDENCOMMAND);
 244                   sc.Forward();
 245                 }
 246               }
 247               else
 248                 if (sc.Match("--"))
 249                 {
 250                   // Special MySQL single line comment.
 251                   sc.SetState(SCE_MYSQL_COMMENTLINE);
 252                   sc.Forward(2);
 253
 254                   // Check the third character too. It must be a space or EOL.
 255                   if (sc.ch != ' ' && sc.ch != '\n' && sc.ch != '\r')
 256                     sc.ChangeState(SCE_MYSQL_OPERATOR);
 257                 }
 258                 else
 259                   if (isoperator(static_cast<char>(sc.ch)))
 260                     sc.SetState(SCE_MYSQL_OPERATOR);
 261       }
 262     }
 263   }
 264
 265   // Do a final check for keywords if we currently have an identifier, to highlight them
 266   // also at the end of a line.
 267   if (sc.state == SCE_MYSQL_IDENTIFIER)
 268   {
 269     CheckForKeyword(sc, keywordlists);
 270
 271     // Additional check for function keywords needed.
 272     // A function name must be followed by an opening parenthesis.
 273     if (sc.state == SCE_MYSQL_FUNCTION && sc.ch != '(')
 274       sc.ChangeState(SCE_MYSQL_DEFAULT);
 275   }
 276
 277   sc.Complete();
 278 }
 279
 280 //--------------------------------------------------------------------------------------------------
 281
 282 /**
 283  * Helper function to determine if we have a foldable comment currently.
 284  */
 285 static bool IsStreamCommentStyle(int style)
 286 {
 287         return style == SCE_MYSQL_COMMENT;
 288 }
 289
 290 //--------------------------------------------------------------------------------------------------
 291
 292 /**
 293  * Code copied from StyleContext and modified to work here. Should go into Accessor as a
 294  * companion to Match()...
 295  */
 296 bool MatchIgnoreCase(Accessor &styler, int currentPos, const char *s)
 297 {
 298   for (int n = 0; *s; n++)
 299   {
 300     if (*s != tolower(styler.SafeGetCharAt(currentPos + n)))
 301       return false;
 302     s++;
 303   }
 304   return true;
 305 }
 306
 307 //--------------------------------------------------------------------------------------------------
 308
 309 // Store both the current line's fold level and the next lines in the
 310 // level store to make it easy to pick up with each increment.
 311 static void FoldMySQLDoc(unsigned int startPos, int length, int initStyle, WordList *[], Accessor &styler)
 312 {
 313         bool foldComment = styler.GetPropertyInt("fold.comment") != 0;
 314         bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;
 315         bool foldOnlyBegin = styler.GetPropertyInt("fold.sql.only.begin", 0) != 0;
 316
 317         int visibleChars = 0;
 318         int lineCurrent = styler.GetLine(startPos);
 319         int levelCurrent = SC_FOLDLEVELBASE;
 320         if (lineCurrent > 0)
 321                 levelCurrent = styler.LevelAt(lineCurrent - 1) >> 16;
 322         int levelNext = levelCurrent;
 323
 324         int styleNext = styler.StyleAt(startPos);
 325         int style = initStyle;
 326
 327   bool endPending = false;
 328         bool whenPending = false;
 329         bool elseIfPending = false;
 330
 331   char nextChar = styler.SafeGetCharAt(startPos);
 332   for (unsigned int i = startPos; length > 0; i++, length--)
 333   {
 334                 int stylePrev = style;
 335                 style = styleNext;
 336                 styleNext = styler.StyleAt(i + 1);
 337
 338     char currentChar = nextChar;
 339     nextChar = styler.SafeGetCharAt(i + 1);
 340                 bool atEOL = (currentChar == '\r' && nextChar != '\n') || (currentChar == '\n');
 341
 342     switch (style)
 343     {
 344       case SCE_MYSQL_COMMENT:
 345         if (foldComment)
 346         {
 347           // Multiline comment style /* .. */.
 348           if (IsStreamCommentStyle(style))
 349           {
 350             // Increase level if we just start a foldable comment.
 351             if (!IsStreamCommentStyle(stylePrev))
 352               levelNext++;
 353             else
 354               // If we are in the middle of a foldable comment check if it ends now.
 355               // Don't end at the line end, though.
 356               if (!IsStreamCommentStyle(styleNext) && !atEOL)
 357                 levelNext--;
 358           }
 359         }
 360         break;
 361       case SCE_MYSQL_COMMENTLINE:
 362         if (foldComment)
 363         {
 364           // Not really a standard, but we add support for single line comments
 365           // with special curly braces syntax as foldable comments too.
 366           // MySQL needs -- comments to be followed by space or control char
 367           if (styler.Match(i, "--"))
 368           {
 369             char chNext2 = styler.SafeGetCharAt(i + 2);
 370             char chNext3 = styler.SafeGetCharAt(i + 3);
 371             if (chNext2 == '{' || chNext3 == '{')
 372               levelNext++;
 373             else
 374               if (chNext2 == '}' || chNext3 == '}')
 375                 levelNext--;
 376           }
 377         }
 378         break;
 379       case SCE_MYSQL_HIDDENCOMMAND:
 380         if (endPending)
 381         {
 382           // A conditional command is not a white space so it should end the current block
 383           // before opening a new one.
 384           endPending = false;
 385           levelNext--;
 386           if (levelNext < SC_FOLDLEVELBASE)
 387             levelNext = SC_FOLDLEVELBASE;
 388         }
 389         if (style != stylePrev)
 390           levelNext++;
 391         else
 392           if (style != styleNext)
 393           {
 394             levelNext--;
 395             if (levelNext < SC_FOLDLEVELBASE)
 396               levelNext = SC_FOLDLEVELBASE;
 397           }
 398         break;
 399       case SCE_MYSQL_OPERATOR:
 400         if (endPending)
 401         {
 402           endPending = false;
 403           levelNext--;
 404           if (levelNext < SC_FOLDLEVELBASE)
 405             levelNext = SC_FOLDLEVELBASE;
 406         }
 407         if (currentChar == '(')
 408           levelNext++;
 409         else
 410           if (currentChar == ')')
 411           {
 412             levelNext--;
 413             if (levelNext < SC_FOLDLEVELBASE)
 414               levelNext = SC_FOLDLEVELBASE;
 415           }
 416         break;
 417       case SCE_MYSQL_MAJORKEYWORD:
 418       case SCE_MYSQL_KEYWORD:
 419       case SCE_MYSQL_FUNCTION:
 420       case SCE_MYSQL_PROCEDUREKEYWORD:
 421         // Reserved and other keywords.
 422         if (style != stylePrev)
 423         {
 424           // END decreases the folding level, regardless which keyword follows.
 425           bool endFound = MatchIgnoreCase(styler, i, "end");
 426           if (endPending)
 427           {
 428             levelNext--;
 429             if (levelNext < SC_FOLDLEVELBASE)
 430               levelNext = SC_FOLDLEVELBASE;
 431           }
 432           else
 433             if (!endFound)
 434             {
 435               if (MatchIgnoreCase(styler, i, "begin"))
 436                 levelNext++;
 437               else
 438               {
 439                 if (!foldOnlyBegin)
 440                 {
 441                   bool whileFound = MatchIgnoreCase(styler, i, "while");
 442                   bool loopFound = MatchIgnoreCase(styler, i, "loop");
 443                   bool repeatFound = MatchIgnoreCase(styler, i, "repeat");
 444                   bool caseFound = MatchIgnoreCase(styler, i, "case");
 445
 446                   if (whileFound || loopFound || repeatFound || caseFound)
 447                     levelNext++;
 448                   else
 449                   {
 450                     // IF alone does not increase the fold level as it is also used in non-block'ed
 451                     // code like DROP PROCEDURE blah IF EXISTS.
 452                     // Instead THEN opens the new level (if not part of an ELSEIF or WHEN (case) branch).
 453                     if (MatchIgnoreCase(styler, i, "then"))
 454                     {
 455                       if (!elseIfPending && !whenPending)
 456                         levelNext++;
 457                       else
 458                       {
 459                         elseIfPending = false;
 460                         whenPending = false;
 461                       }
 462                     }
 463                     else
 464                     {
 465                       // Neither of if/then/while/loop/repeat/case, so check for
 466                       // sub parts of IF and CASE.
 467                       if (MatchIgnoreCase(styler, i, "elseif"))
 468                         elseIfPending = true;
 469                       if (MatchIgnoreCase(styler, i, "when"))
 470                         whenPending = true;
 471                     }
 472                   }
 473                 }
 474               }
 475             }
 476
 477           // Keep the current end state for the next round.
 478           endPending = endFound;
 479         }
 480         break;
 481
 482       default:
 483         if (!isspace(currentChar) && endPending)
 484         {
 485           // END followed by a non-whitespace character (not covered by other cases like identifiers)
 486           // also should end a folding block. Typical case: END followed by self defined delimiter.
 487           levelNext--;
 488           if (levelNext < SC_FOLDLEVELBASE)
 489             levelNext = SC_FOLDLEVELBASE;
 490         }
 491         break;
 492     }
 493
 494     if (atEOL)
 495     {
 496       // Apply the new folding level to this line.
 497       // Leave pending states as they are otherwise a line break will de-sync
 498       // code folding and valid syntax.
 499       int levelUse = levelCurrent;
 500       int lev = levelUse | levelNext << 16;
 501       if (visibleChars == 0 && foldCompact)
 502         lev |= SC_FOLDLEVELWHITEFLAG;
 503       if (levelUse < levelNext)
 504         lev |= SC_FOLDLEVELHEADERFLAG;
 505       if (lev != styler.LevelAt(lineCurrent))
 506         styler.SetLevel(lineCurrent, lev);
 507
 508       lineCurrent++;
 509       levelCurrent = levelNext;
 510       visibleChars = 0;
 511     }
 512
 513                 if (!isspacechar(currentChar))
 514       visibleChars++;
 515   }
 516 }
 517
 518 //--------------------------------------------------------------------------------------------------
 519
 520 static const char * const mysqlWordListDesc[] = {
 521         "Major Keywords",
 522         "Keywords",
 523         "Database Objects",
 524         "Functions",
 525         "System Variables",
 526         "Procedure keywords",
 527         "User Keywords 1",
 528         "User Keywords 2",
 529         "User Keywords 3",
 530         0
 531 };
 532
 533 LexerModule lmMySQL(SCLEX_MYSQL, ColouriseMySQLDoc, "mysql", FoldMySQLDoc, mysqlWordListDesc);