net.sourceforge.phpeclipse/src/net/sourceforge/phpeclipse/phpeditor/php/PHPPartitionScanner.java

   1 /**
   2  * This program and the accompanying materials
   3  * are made available under the terms of the Common Public License v1.0
   4  * which accompanies this distribution, and is available at
   5  * http://www.eclipse.org/legal/cpl-v10.html
   6  * Created on 05.03.2003
   7  *
   8  * @author Stefan Langer (musk)
   9  * @version $Revision: 1.17 $
  10  */
  11 package net.sourceforge.phpeclipse.phpeditor.php;
  12
  13 import java.util.*;
  14
  15 import org.eclipse.jface.text.*;
  16 import org.eclipse.jface.text.rules.*;
  17
  18 /**
  19  *
  20  */
  21 public class PHPPartitionScanner implements IPartitionTokenScanner
  22 {
  23     private static final boolean DEBUG = true;
  24     private boolean fInString = false;
  25     private boolean fInDoubString = false;
  26     private IDocument fDocument = null;
  27     private int fOffset = -1;
  28     private String fContentType = IPHPPartitionScannerConstants.HTML;
  29     private String fPrevContentType = IPHPPartitionScannerConstants.HTML;
  30     private boolean partitionBorder = false;
  31     private int fTokenOffset;
  32     private int fEnd = -1;
  33     private int fLength;
  34     private int fCurrentLength;
  35     private Map tokens = new HashMap();
  36
  37     public PHPPartitionScanner()
  38     {
  39         this.tokens.put(
  40             IPHPPartitionScannerConstants.PHP,
  41             new Token(IPHPPartitionScannerConstants.PHP));
  42         this.tokens.put(
  43             IPHPPartitionScannerConstants.PHP_MULTILINE_COMMENT,
  44             new Token(IPHPPartitionScannerConstants.PHP_MULTILINE_COMMENT));
  45         this.tokens.put(
  46             IPHPPartitionScannerConstants.HTML,
  47             new Token(IPHPPartitionScannerConstants.HTML));
  48         this.tokens.put(
  49             IPHPPartitionScannerConstants.HTML_MULTILINE_COMMENT,
  50             new Token(IPHPPartitionScannerConstants.HTML_MULTILINE_COMMENT));
  51         this.tokens.put(
  52             IDocument.DEFAULT_CONTENT_TYPE,
  53             new Token(IDocument.DEFAULT_CONTENT_TYPE));
  54     }
  55
  56     private IToken getToken(String type)
  57     {
  58         fLength = fCurrentLength;
  59         if (DEBUG)
  60         {
  61
  62             try
  63             {
  64                 if (fLength <= 0)
  65                 {
  66                     int line = fDocument.getLineOfOffset(fOffset);
  67                     System.err.println(
  68                         "Error at "
  69                             + line
  70                             + " offset:"
  71                             + String.valueOf(
  72                                 fOffset - fDocument.getLineOffset(line)));
  73                 }
  74             }
  75             catch (BadLocationException e)
  76             {   // should never happen
  77                 // TODO Write stacktrace to log
  78                 e.printStackTrace();
  79             }
  80         }
  81         Assert.isTrue(fLength > 0, "Partition length <= 0!");
  82         fCurrentLength = 0;
  83         // String can never cross partition borders so reset string detection
  84         fInString = false;
  85         fInDoubString = false;
  86         IToken token = (IToken) this.tokens.get(type);
  87         Assert.isNotNull(token, "Token for type \"" + type + "\" not found!");
  88         if (DEBUG)
  89         {
  90             System.out.println(
  91                 "Partition: fTokenOffset="
  92                     + fTokenOffset
  93                     + " fContentType="
  94                     + type
  95                     + " fLength="
  96                     + fLength);
  97         }
  98         return token;
  99     }
 100
 101     /* (non-Javadoc)
 102      * @see org.eclipse.jface.text.rules.IPartitionTokenScanner#setPartialRange(org.eclipse.jface.text.IDocument, int, int, java.lang.String, int)
 103      */
 104     public void setPartialRange(
 105         IDocument document,
 106         int offset,
 107         int length,
 108         String contentType,
 109         int partitionOffset)
 110     {
 111         if (DEBUG)
 112         {
 113             System.out.println(
 114                 "PartialRange: contentType="
 115                     + contentType
 116                     + " partitionOffset="
 117                     + partitionOffset);
 118         }
 119
 120         try
 121         {
 122             if (partitionOffset > -1)
 123             {
 124                 partitionBorder = false;
 125                 // because of strings we have to parse the whole partition
 126                 this.setRange(
 127                     document,
 128                     partitionOffset,
 129                     offset - partitionOffset + length);
 130                 // sometimes we get a wrong partition so we retrieve the partition
 131                 // directly from the document
 132                 fContentType = fDocument.getContentType(partitionOffset);
 133             }
 134             else
 135                 this.setRange(document, offset, length);
 136
 137         }
 138         catch (BadLocationException e)
 139         {
 140             // should never happen
 141             // TODO print stack trace to log
 142             // fall back just scan the whole document again
 143             this.setRange(document, 0, fDocument.getLength());
 144         }
 145
 146     }
 147
 148     /* (non-Javadoc)
 149      * @see org.eclipse.jface.text.rules.ITokenScanner#getTokenLength()
 150      */
 151     public int getTokenLength()
 152     {
 153         return fLength;
 154     }
 155
 156     /* (non-Javadoc)
 157      * @see org.eclipse.jface.text.rules.ITokenScanner#getTokenOffset()
 158      */
 159     public int getTokenOffset()
 160     {
 161         return fTokenOffset;
 162     }
 163
 164     /* (non-Javadoc)
 165      * @see org.eclipse.jface.text.rules.ITokenScanner#nextToken()
 166      */
 167     public IToken nextToken()
 168     {
 169         int c;
 170
 171         // check if we are not allready at the end of the
 172         // file
 173         if ((c = read()) == ICharacterScanner.EOF)
 174         {
 175             partitionBorder = false;
 176             return Token.EOF;
 177         }
 178         else
 179             unread();
 180
 181         if (partitionBorder)
 182         {
 183             fTokenOffset = fOffset;
 184             partitionBorder = false;
 185         }
 186
 187         while ((c = read()) != ICharacterScanner.EOF)
 188         {
 189             switch (c)
 190             {
 191                 case '<' :
 192                     if (!isInString(IPHPPartitionScannerConstants.PHP)
 193                         && fContentType
 194                             != IPHPPartitionScannerConstants.PHP_MULTILINE_COMMENT
 195                         && checkPattern(new char[] { '?', 'p', 'h', 'p' }, true))
 196                     {
 197                         if (fContentType != IPHPPartitionScannerConstants.PHP
 198                             && fCurrentLength > 5)
 199                         {
 200                             unread(5);
 201                             IToken token = getToken(fContentType);
 202                             // save previouse contenttype
 203                             //TODO build stack for previouse contenttype
 204                             fPrevContentType = fContentType;
 205
 206                             fContentType = IPHPPartitionScannerConstants.PHP;
 207
 208                             return token;
 209                         }
 210                         else
 211                             fContentType = IPHPPartitionScannerConstants.PHP;
 212
 213                         // remember offset of this partition
 214                         fTokenOffset = fOffset - 5;
 215                         fCurrentLength = 5;
 216                     }
 217                     else if (
 218                         !isInString(IPHPPartitionScannerConstants.PHP)
 219                             && fContentType
 220                                 != IPHPPartitionScannerConstants
 221                                     .PHP_MULTILINE_COMMENT
 222                             && checkPattern(new char[] { '?' }, false))
 223                     {
 224                         if (fContentType != IPHPPartitionScannerConstants.PHP
 225                             && fCurrentLength > 2)
 226                         {
 227                             unread(2);
 228                             IToken token = getToken(fContentType);
 229                             // save previouse contenttype
 230                             fPrevContentType = fContentType;
 231                             fContentType = IPHPPartitionScannerConstants.PHP;
 232                             return token;
 233                         }
 234                         else
 235                             fContentType = IPHPPartitionScannerConstants.PHP;
 236                         // remember offset of this partition
 237                         fTokenOffset = fOffset - 2;
 238                         fCurrentLength = 2;
 239                     }
 240                     else if (
 241                         !isInString(IPHPPartitionScannerConstants.PHP)
 242                             && checkPattern(new char[] { '!', '-', '-' }))
 243                     { // return previouse partition
 244                         if (fContentType
 245                             != IPHPPartitionScannerConstants
 246                                 .HTML_MULTILINE_COMMENT
 247                             && fCurrentLength > 4)
 248                         {
 249                             unread(4);
 250                             IToken token = getToken(fContentType);
 251                             fContentType =
 252                                 IPHPPartitionScannerConstants
 253                                     .HTML_MULTILINE_COMMENT;
 254                             return token;
 255                         }
 256                         else
 257                             fContentType =
 258                                 IPHPPartitionScannerConstants
 259                                     .HTML_MULTILINE_COMMENT;
 260
 261                         fTokenOffset = fOffset - 4;
 262                         fCurrentLength = 4;
 263                     }
 264                     break;
 265                 case '?' :
 266                     if (!isInString(IPHPPartitionScannerConstants.PHP)
 267                         && fContentType == IPHPPartitionScannerConstants.PHP)
 268                     {
 269                         if ((c = read()) == '>')
 270                         {
 271                             if (fPrevContentType != null)
 272                                 fContentType = fPrevContentType;
 273                             else
 274                                 fContentType =
 275                                     IPHPPartitionScannerConstants.HTML;
 276                             partitionBorder = true;
 277                             return getToken(IPHPPartitionScannerConstants.PHP);
 278                         }
 279                         else if (c != ICharacterScanner.EOF)
 280                             unread();
 281                     }
 282                     break;
 283                 case '-' :
 284                     if (!isInString(IPHPPartitionScannerConstants.PHP)
 285                         && fContentType
 286                             == IPHPPartitionScannerConstants
 287                                 .HTML_MULTILINE_COMMENT
 288                         && checkPattern(new char[] { '-', '>' }))
 289                     {
 290                         fContentType = IPHPPartitionScannerConstants.HTML;
 291                         partitionBorder = true;
 292                         return getToken(
 293                             IPHPPartitionScannerConstants
 294                                 .HTML_MULTILINE_COMMENT);
 295                     }
 296                     break;
 297                 case '/' :
 298                     if (!isInString(IPHPPartitionScannerConstants.PHP) && (c = read()) == '*')
 299                     { // MULTINE COMMENT JAVASCRIPT, CSS, PHP
 300                         if (fContentType == IPHPPartitionScannerConstants.PHP
 301                             && fCurrentLength > 2)
 302                         {
 303                             unread(2);
 304                             IToken token = getToken(fContentType);
 305                             fContentType =
 306                                 IPHPPartitionScannerConstants
 307                                     .PHP_MULTILINE_COMMENT;
 308                             return token;
 309                         }
 310                         else if (
 311                             fContentType
 312                                 == IPHPPartitionScannerConstants
 313                                     .PHP_MULTILINE_COMMENT)
 314                         {
 315
 316                             fTokenOffset = fOffset - 2;
 317                             fCurrentLength = 2;
 318                         }
 319
 320                     }
 321                     else if (!isInString(IPHPPartitionScannerConstants.PHP) && c != ICharacterScanner.EOF)
 322                         unread();
 323                     break;
 324                 case '*' :
 325                     if (!isInString(IPHPPartitionScannerConstants.PHP) && (c = read()) == '/')
 326                     {
 327                         if (fContentType
 328                             == IPHPPartitionScannerConstants
 329                                 .PHP_MULTILINE_COMMENT)
 330                         {
 331                             fContentType = IPHPPartitionScannerConstants.PHP;
 332                             partitionBorder = true;
 333                             return getToken(
 334                                 IPHPPartitionScannerConstants
 335                                     .PHP_MULTILINE_COMMENT);
 336                         }
 337                         else if (
 338                             fContentType
 339                                 == IPHPPartitionScannerConstants
 340                                     .CSS_MULTILINE_COMMENT)
 341                         {
 342                         }
 343                         else if (
 344                             fContentType
 345                                 == IPHPPartitionScannerConstants
 346                                     .JS_MULTILINE_COMMENT)
 347                         {
 348                         }
 349                     }
 350                     else if (!isInString(IPHPPartitionScannerConstants.PHP) && c != ICharacterScanner.EOF)
 351                         unread();
 352                     break;
 353                 case '\'' :
 354                     if (!fInDoubString)
 355                         fInString = !fInString;
 356                     break;
 357                 case '"' :
 358                     // toggle String mode
 359                     if (!fInString)
 360                         fInDoubString = !fInDoubString;
 361                     break;
 362             }
 363         } // end of file reached but we have to return the
 364         // last partition.
 365         return getToken(fContentType);
 366     }
 367     /* (non-Javadoc)
 368      * @see org.eclipse.jface.text.rules.ITokenScanner#setRange(org.eclipse.jface.text.IDocument, int, int)
 369      */
 370     public void setRange(IDocument document, int offset, int length)
 371     {
 372         if (DEBUG)
 373         {
 374             System.out.println(
 375                 "SET RANGE: offset=" + offset + " length=" + length);
 376         }
 377
 378         fDocument = document;
 379         fOffset = offset;
 380         fTokenOffset = offset;
 381         fCurrentLength = 0;
 382         fLength = 0;
 383         fEnd = fOffset + length;
 384         fInString = false;
 385         fInDoubString = false;
 386         fContentType = IPHPPartitionScannerConstants.HTML;
 387 //        String[] prev = getPartitionStack(offset);
 388     }
 389
 390     private int read()
 391     {
 392         try
 393         {
 394             if (fOffset < fEnd)
 395             {
 396                 fCurrentLength++;
 397                 return fDocument.getChar(fOffset++);
 398             }
 399             return ICharacterScanner.EOF;
 400         }
 401         catch (BadLocationException e)
 402         {
 403             // should never happen
 404             // TODO write stacktrace to log
 405             fOffset = fEnd;
 406             return ICharacterScanner.EOF;
 407         }
 408     }
 409
 410     private void unread()
 411     {
 412         --fOffset;
 413         --fCurrentLength;
 414     }
 415
 416     private void unread(int num)
 417     {
 418         fOffset -= num;
 419         fCurrentLength -= num;
 420     }
 421
 422     private boolean checkPattern(char[] pattern)
 423     {
 424         return checkPattern(pattern, false);
 425     }
 426
 427     /**
 428      * Check if next character sequence read from document is equals to
 429      * the provided pattern. Pattern is read from left to right until the
 430      * first character read doesn't match. If this happens all read characters are
 431      * unread.
 432      * @param pattern The pattern to check.
 433      * @return <code>true</code> if pattern is equals else returns <code>false</code>.
 434      */
 435     private boolean checkPattern(char[] pattern, boolean ignoreCase)
 436     {
 437         int prevOffset = fOffset;
 438         int prevLength = fCurrentLength;
 439         for (int i = 0; i < pattern.length; i++)
 440         {
 441             int c = read();
 442
 443             if (c == ICharacterScanner.EOF
 444                 || !letterEquals(c, pattern[i], ignoreCase))
 445             {
 446                 fOffset = prevOffset;
 447                 fCurrentLength = prevLength;
 448                 return false;
 449             }
 450         }
 451
 452         return true;
 453     }
 454
 455     private boolean letterEquals(int test, char letter, boolean ignoreCase)
 456     {
 457         if (test == letter)
 458             return true;
 459         else if (
 460             ignoreCase
 461                 && Character.isLowerCase(letter)
 462                 && test == Character.toUpperCase(letter))
 463             return true;
 464         else if (
 465             ignoreCase
 466                 && Character.isUpperCase(letter)
 467                 && test == Character.toLowerCase(letter))
 468             return true;
 469
 470         return false;
 471     }
 472
 473     /**
 474      * Checks wether the offset is in a <code>String</code> and the specified
 475      * contenttype is the current content type.
 476      * Strings are delimited, mutual exclusive, by a " or by a '.
 477      *
 478      * @param contentType The contenttype to check.
 479      * @return <code>true</code> if the current offset is in a string else
 480      *                  returns false.
 481      */
 482     private  boolean isInString(String contentType)
 483     {
 484         if(fContentType == contentType)
 485                 return (fInString || fInDoubString);
 486         else
 487                 return false;
 488     }
 489
 490     /**
 491      * Returns the previouse partition stack for the given offset.
 492      *
 493      * @param offset The offset to return the previouse partitionstack for.
 494      *
 495      * @return The stack as a string array.
 496      */
 497     private String[] getPartitionStack(int offset)
 498     {
 499         ArrayList types = new ArrayList();
 500         int tmpOffset = 0;
 501         try
 502         {
 503             ITypedRegion region = fDocument.getPartition(offset);
 504             tmpOffset = region.getOffset();
 505             while(tmpOffset-1 > 0)
 506             {
 507                 region = fDocument.getPartition(tmpOffset-1);
 508                 tmpOffset = region.getOffset();
 509                 types.add(0, region.getType());
 510             }
 511         }
 512         catch (BadLocationException e)
 513         {
 514            if(DEBUG)
 515            {
 516                         e.printStackTrace();
 517            }
 518         }
 519
 520                 String[] retVal = new String[types.size()];
 521
 522         retVal = (String[])types.toArray(retVal);
 523         return retVal;
 524     }
 525
 526 }