net.sourceforge.phpeclipse.ui/src/net/sourceforge/phpeclipse/xml/ui/internal/text/XMLPartitionScanner.java

   1 /*
   2  * Copyright (c) 2002-2004 Widespace, OU and others.
   3  * All rights reserved. This program and the accompanying materials
   4  * are made available under the terms of the Common Public License v1.0
   5  * which accompanies this distribution, and is available at
   6  * http://www.eclipse.org/legal/cpl-v10.html
   7  *
   8  * Contributors:
   9  *     Igor Malinin - initial contribution
  10  *
  11  * $Id: XMLPartitionScanner.java,v 1.5 2006-10-21 23:14:13 pombredanne Exp $
  12  */
  13
  14 package net.sourceforge.phpeclipse.xml.ui.internal.text;
  15
  16 import java.util.HashMap;
  17 import java.util.Map;
  18
  19 //incastrix
  20 //import org.eclipse.jface.text.Assert;
  21 import org.eclipse.core.runtime.Assert;
  22 import org.eclipse.jface.text.BadLocationException;
  23 import org.eclipse.jface.text.IDocument;
  24 import org.eclipse.jface.text.rules.ICharacterScanner;
  25 import org.eclipse.jface.text.rules.IPartitionTokenScanner;
  26 import org.eclipse.jface.text.rules.IToken;
  27 import org.eclipse.jface.text.rules.Token;
  28
  29 /**
  30  *
  31  *
  32  * @author Igor Malinin
  33  */
  34 public class XMLPartitionScanner implements IPartitionTokenScanner {
  35         public static final String XML_PI = "__xml_processing_instruction";
  36
  37         public static final String XML_COMMENT = "__xml_comment";
  38
  39         public static final String XML_DECL = "__xml_declaration";
  40
  41         public static final String XML_TAG = "__xml_tag";
  42
  43         public static final String XML_ATTRIBUTE = "__xml_attribute";
  44
  45         public static final String XML_CDATA = "__xml_cdata";
  46
  47         public static final String DTD_INTERNAL = "__dtd_internal";
  48
  49         public static final String DTD_INTERNAL_PI = "__dtd_internal_pi";
  50
  51         public static final String DTD_INTERNAL_COMMENT = "__dtd_internal_comment";
  52
  53         public static final String DTD_INTERNAL_DECL = "__dtd_internal_declaration";
  54
  55         public static final String DTD_CONDITIONAL = "__dtd_conditional";
  56
  57         public static final int STATE_DEFAULT = 0;
  58
  59         public static final int STATE_TAG = 1;
  60
  61         public static final int STATE_DECL = 2;
  62
  63         public static final int STATE_CDATA = 4;
  64
  65         public static final int STATE_INTERNAL = 8;
  66
  67         protected IDocument document;
  68
  69         protected int end;
  70
  71         protected int offset;
  72
  73         protected int length;
  74
  75         protected int position;
  76
  77         protected int state;
  78
  79         protected boolean parsedtd;
  80
  81         protected Map tokens = new HashMap();
  82
  83         public XMLPartitionScanner(boolean parsedtd) {
  84                 this.parsedtd = parsedtd;
  85         }
  86
  87         /*
  88          * @see org.eclipse.jface.text.rules.ITokenScanner#nextToken()
  89          */
  90         public IToken nextToken() {
  91                 offset += length;
  92
  93                 switch (state) {
  94                 case STATE_TAG:
  95                         return nextTagToken();
  96
  97                 case STATE_DECL:
  98                         return nextDeclToken();
  99
 100                 case STATE_CDATA:
 101                         return nextCDATAToken();
 102                 }
 103
 104                 switch (read()) {
 105                 case ICharacterScanner.EOF:
 106                         state = STATE_DEFAULT;
 107                         return getToken(null);
 108
 109                 case '<':
 110                         switch (read()) {
 111                         case ICharacterScanner.EOF:
 112                                 if (parsedtd || isInternal()) {
 113                                         break;
 114                                 }
 115
 116                                 state = STATE_DEFAULT;
 117                                 return getToken(XML_TAG);
 118
 119                         case '?': // <? <?PI
 120                                 return nextPIToken();
 121
 122                         case '!': // <! <!DEFINITION or <![CDATA[ or <!--COMMENT
 123                                 switch (read()) {
 124                                 case ICharacterScanner.EOF:
 125                                         state = STATE_DEFAULT;
 126                                         return getToken(XML_TAG);
 127
 128                                 case '-': // <!- <!--COMMENT
 129                                         switch (read()) {
 130                                         case ICharacterScanner.EOF:
 131                                                 return nextDeclToken();
 132
 133                                         case '-': // <!--
 134                                                 return nextCommentToken();
 135                                         }
 136
 137                                 case '[': // <![ <![CDATA[ or <![%cond;[
 138                                         if (parsedtd) {
 139                                                 return nextConditionalToken();
 140                                         }
 141
 142                                         if (!isInternal()) {
 143                                                 return nextCDATAToken();
 144                                         }
 145                                 }
 146
 147                                 return nextDeclToken();
 148                         }
 149
 150                         if (parsedtd || isInternal()) {
 151                                 break;
 152                         }
 153
 154                         unread();
 155
 156                         return nextTagToken();
 157
 158                 case ']':
 159                         if (isInternal()) {
 160                                 unread();
 161
 162                                 state = STATE_DECL;
 163                                 length = 0;
 164                                 return nextToken();
 165                         }
 166                         break;
 167                 default:
 168                         unread();
 169                 }
 170
 171                 loop: while (true) {
 172                         switch (read()) {
 173                         case ICharacterScanner.EOF:
 174                                 state = STATE_DEFAULT;
 175                                 return getToken(null);
 176
 177                         case '<':
 178                                 if (parsedtd || isInternal()) {
 179                                         switch (read()) {
 180                                         case ICharacterScanner.EOF:
 181                                                 state = STATE_DEFAULT;
 182                                                 return getToken(null);
 183
 184                                         case '!':
 185                                         case '?':
 186                                                 unread();
 187                                                 break;
 188
 189                                         default:
 190                                                 continue loop;
 191                                         }
 192                                 }
 193
 194                                 unread();
 195
 196                                 state &= STATE_INTERNAL;
 197                                 return getToken(isInternal() ? DTD_INTERNAL : null);
 198
 199                         case ']':
 200                                 if (isInternal()) {
 201                                         unread();
 202
 203                                         state = STATE_DECL;
 204                                         if (position == offset) {
 205                                                 // nothing between
 206                                                 length = 0;
 207                                                 return nextToken();
 208                                         }
 209
 210                                         return getToken(DTD_INTERNAL);
 211                                 }
 212                         }
 213                 }
 214         }
 215
 216         private IToken nextTagToken() {
 217                 int quot = read();
 218
 219                 switch (quot) {
 220                 case ICharacterScanner.EOF:
 221                 case '>':
 222                         state = STATE_DEFAULT;
 223                         return getToken(XML_TAG);
 224
 225                 case '"':
 226                 case '\'':
 227                         while (true) {
 228                                 int ch = read();
 229
 230                                 if (ch == quot) {
 231                                         state = STATE_TAG;
 232                                         return getToken(XML_ATTRIBUTE);
 233                                 }
 234
 235                                 switch (ch) {
 236                                 case '<':
 237                                         unread();
 238
 239                                 case ICharacterScanner.EOF:
 240                                         state = STATE_DEFAULT;
 241                                         return getToken(XML_ATTRIBUTE);
 242                                 }
 243                         }
 244                 default:
 245                         unread();
 246                 }
 247
 248                 while (true) {
 249                         switch (read()) {
 250                         case '<':
 251                                 unread();
 252
 253                         case ICharacterScanner.EOF:
 254                         case '>':
 255                                 state = STATE_DEFAULT;
 256                                 return getToken(XML_TAG);
 257
 258                         case '"':
 259                         case '\'':
 260                                 unread();
 261
 262                                 state = STATE_TAG;
 263                                 return getToken(XML_TAG);
 264                         }
 265                 }
 266         }
 267
 268         private IToken nextDeclToken() {
 269                 loop: while (true) {
 270                         switch (read()) {
 271                         case ICharacterScanner.EOF:
 272                                 state = STATE_DEFAULT;
 273                                 return getToken(isInternal() ? DTD_INTERNAL_DECL : XML_DECL);
 274
 275                         case '<':
 276                                 if (parsedtd || isInternal()) {
 277                                         switch (read()) {
 278                                         case ICharacterScanner.EOF:
 279                                                 state = STATE_DEFAULT;
 280                                                 return getToken(isInternal() ? DTD_INTERNAL : null);
 281
 282                                         case '!':
 283                                         case '?':
 284                                                 unread();
 285                                                 break;
 286
 287                                         default:
 288                                                 continue loop;
 289                                         }
 290                                 }
 291
 292                                 unread();
 293
 294                         case '>':
 295                                 state &= STATE_INTERNAL;
 296                                 return getToken(isInternal() ? DTD_INTERNAL_DECL : XML_DECL);
 297
 298                         case '[': // <!DOCTYPE xxx [dtd]>
 299                                 if (!isInternal()) {
 300                                         state = STATE_INTERNAL;
 301                                         return getToken(XML_DECL);
 302                                 }
 303                         }
 304                 }
 305         }
 306
 307         private IToken nextCommentToken() {
 308                 state &= STATE_INTERNAL;
 309
 310                 loop: while (true) {
 311                         switch (read()) {
 312                         case ICharacterScanner.EOF:
 313                                 break loop;
 314
 315                         case '-': // - -->
 316                                 switch (read()) {
 317                                 case ICharacterScanner.EOF:
 318                                         break loop;
 319
 320                                 case '-': // -- -->
 321                                         switch (read()) {
 322                                         case ICharacterScanner.EOF:
 323                                         case '>':
 324                                                 break loop;
 325                                         }
 326
 327                                         unread();
 328                                         continue loop;
 329                                 }
 330                         }
 331                 }
 332
 333                 return getToken(isInternal() ? DTD_INTERNAL_COMMENT : XML_COMMENT);
 334         }
 335
 336         private IToken nextPIToken() {
 337                 state &= STATE_INTERNAL;
 338
 339                 loop: while (true) {
 340                         switch (read()) {
 341                         case ICharacterScanner.EOF:
 342                                 break loop;
 343
 344                         case '?': // ? ?>
 345                                 switch (read()) {
 346                                 case ICharacterScanner.EOF:
 347                                 case '>':
 348                                         break loop;
 349                                 }
 350
 351                                 unread();
 352                         }
 353                 }
 354
 355                 return getToken(isInternal() ? DTD_INTERNAL_PI : XML_PI);
 356         }
 357
 358         private IToken nextCDATAToken() {
 359                 state = STATE_DEFAULT;
 360
 361                 loop: while (true) {
 362                         switch (read()) {
 363                         case ICharacterScanner.EOF:
 364                                 break loop;
 365
 366                         case ']': // ] ]]>
 367                                 switch (read()) {
 368                                 case ICharacterScanner.EOF:
 369                                         break loop;
 370
 371                                 case ']': // ]] ]]>
 372                                         switch (read()) {
 373                                         case ICharacterScanner.EOF:
 374                                         case '>': // ]]>
 375                                                 break loop;
 376                                         }
 377
 378                                         unread();
 379                                         unread();
 380                                         continue loop;
 381                                 }
 382                         }
 383                 }
 384
 385                 return getToken(XML_CDATA);
 386         }
 387
 388         private IToken nextConditionalToken() {
 389                 state = STATE_DEFAULT;
 390
 391                 int level = 1;
 392
 393                 loop: while (true) {
 394                         switch (read()) {
 395                         case ICharacterScanner.EOF:
 396                                 break loop;
 397
 398                         case '<': // - -->
 399                                 switch (read()) {
 400                                 case ICharacterScanner.EOF:
 401                                         break loop;
 402
 403                                 case '!': // -- -->
 404                                         switch (read()) {
 405                                         case ICharacterScanner.EOF:
 406                                                 break loop;
 407
 408                                         case '[':
 409                                                 ++level;
 410                                                 continue loop;
 411                                         }
 412
 413                                         unread();
 414                                         continue loop;
 415                                 }
 416
 417                                 unread();
 418                                 continue loop;
 419
 420                         case ']': // - -->
 421                                 switch (read()) {
 422                                 case ICharacterScanner.EOF:
 423                                         break loop;
 424
 425                                 case ']': // -- -->
 426                                         switch (read()) {
 427                                         case ICharacterScanner.EOF:
 428                                         case '>':
 429                                                 if (--level == 0) {
 430                                                         break loop;
 431                                                 }
 432
 433                                                 continue loop;
 434                                         }
 435
 436                                         unread();
 437                                         unread();
 438                                         continue loop;
 439                                 }
 440                         }
 441                 }
 442
 443                 return getToken(DTD_CONDITIONAL);
 444         }
 445
 446         private IToken getToken(String type) {
 447                 length = position - offset;
 448
 449                 if (length == 0) {
 450                         return Token.EOF;
 451                 }
 452
 453                 if (type == null) {
 454                         return Token.UNDEFINED;
 455                 }
 456
 457                 IToken token = (IToken) tokens.get(type);
 458                 if (token == null) {
 459                         token = new Token(type);
 460                         tokens.put(type, token);
 461                 }
 462
 463                 return token;
 464         }
 465
 466         private boolean isInternal() {
 467                 return (state & STATE_INTERNAL) != 0;
 468         }
 469
 470         private int read() {
 471                 if (position >= end) {
 472                         return ICharacterScanner.EOF;
 473                 }
 474
 475                 try {
 476                         return document.getChar(position++);
 477                 } catch (BadLocationException e) {
 478                         --position;
 479                         return ICharacterScanner.EOF;
 480                 }
 481         }
 482
 483         private void unread() {
 484                 --position;
 485         }
 486
 487         /*
 488          * @see org.eclipse.jface.text.rules.ITokenScanner#getTokenOffset()
 489          */
 490         public int getTokenOffset() {
 491                 Assert.isTrue(offset >= 0, Integer.toString(offset));
 492                 return offset;
 493         }
 494
 495         /*
 496          * @see org.eclipse.jface.text.rules.ITokenScanner#getTokenLength()
 497          */
 498         public int getTokenLength() {
 499                 return length;
 500         }
 501
 502         /*
 503          * @see org.eclipse.jface.text.rules.ITokenScanner#setRange(IDocument, int,
 504          *      int)
 505          */
 506         public void setRange(IDocument document, int offset, int length) {
 507                 this.document = document;
 508                 this.end = offset + length;
 509
 510                 this.offset = offset;
 511                 this.position = offset;
 512                 this.length = 0;
 513
 514                 this.state = STATE_DEFAULT;
 515         }
 516
 517         /*
 518          * @see org.eclipse.jface.text.rules.IPartitionTokenScanner
 519          */
 520         // public void setPartialRange(IDocument document, int offset, int length,
 521         // String contentType, int partitionOffset) {
 522         // state = STATE_DEFAULT;
 523         // if (partitionOffset > -1) {
 524         // int delta = offset - partitionOffset;
 525         // if (delta > 0) {
 526         // setRange(document, partitionOffset, length + delta);
 527         // return;
 528         // }
 529         // }
 530         // setRange(document, partitionOffset, length);
 531         // }
 532         /*
 533          * @see org.eclipse.jface.text.rules.IPartitionTokenScanner
 534          */
 535         public void setPartialRange(IDocument document, int offset, int length,
 536                         String contentType, int partitionOffset) {
 537                 // boolean flag = false;
 538                 this.document = document;
 539                 this.end = offset + length;
 540
 541                 // NB! Undocumented value: -1
 542                 if (partitionOffset >= 0) {
 543                         offset = partitionOffset;
 544                         // flag = true;
 545                 }
 546
 547                 this.offset = offset;
 548                 this.position = offset;
 549                 this.length = 0;
 550
 551                 // if (flag) {
 552                 // state = STATE_DEFAULT;
 553                 // return;
 554                 // }
 555                 if (contentType == XML_ATTRIBUTE) {
 556                         state = STATE_TAG;
 557                         return;
 558                 }
 559
 560                 if (contentType == XML_TAG) {
 561                         state = isContinuationPartition() ? STATE_TAG : STATE_DEFAULT;
 562                         return;
 563                 }
 564
 565                 if (contentType == XML_DECL) {
 566                         state = isContinuationPartition() ? STATE_DECL : STATE_DEFAULT;
 567                         return;
 568                 }
 569
 570                 if (contentType == DTD_INTERNAL || contentType == DTD_INTERNAL_PI
 571                                 || contentType == DTD_INTERNAL_DECL
 572                                 || contentType == DTD_INTERNAL_COMMENT) {
 573                         state = STATE_INTERNAL;
 574                         return;
 575                 }
 576
 577                 state = STATE_DEFAULT;
 578         }
 579
 580         private boolean isContinuationPartition() {
 581                 try {
 582                         String type = document.getContentType(offset - 1);
 583
 584                         if (type != IDocument.DEFAULT_CONTENT_TYPE) {
 585                                 return true;
 586                         }
 587                 } catch (BadLocationException e) {
 588                 }
 589
 590                 return false;
 591         }
 592 }