net.sourceforge.phpeclipse.ui/src/net/sourceforge/phpeclipse/xml/ui/internal/text/PHPXMLPartitionScanner.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: PHPXMLPartitionScanner.java,v 1.3 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 import org.eclipse.jface.text.Assert;
  20 import org.eclipse.jface.text.BadLocationException;
  21 import org.eclipse.jface.text.IDocument;
  22 import org.eclipse.jface.text.rules.ICharacterScanner;
  23 import org.eclipse.jface.text.rules.IPartitionTokenScanner;
  24 import org.eclipse.jface.text.rules.IToken;
  25 import org.eclipse.jface.text.rules.Token;
  26
  27 /**
  28  *
  29  *
  30  * @author Igor Malinin
  31  */
  32 public class PHPXMLPartitionScanner implements IPartitionTokenScanner {
  33         // public static final String XML_PI = "__xml_processing_instruction";
  34         public static final String XML_COMMENT = "__xml_comment";
  35
  36         public static final String XML_DECL = "__xml_declaration";
  37
  38         public static final String XML_TAG = "__xml_tag";
  39
  40         public static final String XML_ATTRIBUTE = "__xml_attribute";
  41
  42         public static final String XML_CDATA = "__xml_cdata";
  43
  44         public static final String DTD_INTERNAL = "__dtd_internal";
  45
  46         // public static final String DTD_INTERNAL_PI = "__dtd_internal_pi";
  47         public static final String DTD_INTERNAL_COMMENT = "__dtd_internal_comment";
  48
  49         public static final String DTD_INTERNAL_DECL = "__dtd_internal_declaration";
  50
  51         public static final String DTD_CONDITIONAL = "__dtd_conditional";
  52
  53         public static final int STATE_DEFAULT = 0;
  54
  55         public static final int STATE_TAG = 1;
  56
  57         public static final int STATE_DECL = 2;
  58
  59         public static final int STATE_CDATA = 4;
  60
  61         public static final int STATE_INTERNAL = 8;
  62
  63         protected IDocument document;
  64
  65         protected int end;
  66
  67         protected int offset;
  68
  69         protected int length;
  70
  71         protected int position;
  72
  73         protected int state;
  74
  75         protected boolean parsedtd;
  76
  77         protected Map tokens = new HashMap();
  78
  79         public PHPXMLPartitionScanner(boolean parsedtd) {
  80                 this.parsedtd = parsedtd;
  81         }
  82
  83         /*
  84          * @see org.eclipse.jface.text.rules.ITokenScanner#nextToken()
  85          */
  86         public IToken nextToken() {
  87                 offset += length;
  88
  89                 switch (state) {
  90                 case STATE_TAG:
  91                         return nextTagToken();
  92
  93                 case STATE_DECL:
  94                         return nextDeclToken();
  95
  96                 case STATE_CDATA:
  97                         return nextCDATAToken();
  98                 }
  99
 100                 switch (read()) {
 101                 case ICharacterScanner.EOF:
 102                         state = STATE_DEFAULT;
 103                         return getToken(null);
 104
 105                 case '<':
 106                         switch (read()) {
 107                         case ICharacterScanner.EOF:
 108                                 if (parsedtd || isInternal()) {
 109                                         break;
 110                                 }
 111
 112                                 state = STATE_DEFAULT;
 113                                 return getToken(XML_TAG);
 114
 115                         case '?': // <? <?PI
 116                                 unread();
 117                                 break;
 118                         // return nextPIToken();
 119
 120                         case '!': // <! <!DEFINITION or <![CDATA[ or <!--COMMENT
 121                                 switch (read()) {
 122                                 case ICharacterScanner.EOF:
 123                                         state = STATE_DEFAULT;
 124                                         return getToken(XML_TAG);
 125
 126                                 case '-': // <!- <!--COMMENT
 127                                         switch (read()) {
 128                                         case ICharacterScanner.EOF:
 129                                                 return nextDeclToken();
 130
 131                                         case '-': // <!--
 132                                                 return nextCommentToken();
 133                                         }
 134
 135                                 case '[': // <![ <![CDATA[ or <![%cond;[
 136                                         if (parsedtd) {
 137                                                 return nextConditionalToken();
 138                                         }
 139
 140                                         if (!isInternal()) {
 141                                                 return nextCDATAToken();
 142                                         }
 143                                 }
 144
 145                                 return nextDeclToken();
 146                         }
 147
 148                         if (parsedtd || isInternal()) {
 149                                 break;
 150                         }
 151
 152                         unread();
 153
 154                         return nextTagToken();
 155
 156                 case ']':
 157                         if (isInternal()) {
 158                                 unread();
 159
 160                                 state = STATE_DECL;
 161                                 length = 0;
 162                                 return nextToken();
 163                         }
 164                         break;
 165                 default:
 166                         unread();
 167                 }
 168
 169                 loop: while (true) {
 170                         switch (read()) {
 171                         case ICharacterScanner.EOF:
 172                                 state = STATE_DEFAULT;
 173                                 return getToken(null);
 174
 175                         case '<':
 176                                 if (parsedtd || isInternal()) {
 177                                         switch (read()) {
 178                                         case ICharacterScanner.EOF:
 179                                                 state = STATE_DEFAULT;
 180                                                 return getToken(null);
 181
 182                                         case '!':
 183                                         case '?':
 184                                                 unread();
 185                                                 break;
 186
 187                                         default:
 188                                                 continue loop;
 189                                         }
 190                                 }
 191
 192                                 unread();
 193
 194                                 state &= STATE_INTERNAL;
 195                                 return getToken(isInternal() ? DTD_INTERNAL : null);
 196
 197                         case ']':
 198                                 if (isInternal()) {
 199                                         unread();
 200
 201                                         state = STATE_DECL;
 202                                         if (position == offset) {
 203                                                 // nothing between
 204                                                 length = 0;
 205                                                 return nextToken();
 206                                         }
 207
 208                                         return getToken(DTD_INTERNAL);
 209                                 }
 210                         }
 211                 }
 212         }
 213
 214         private IToken nextTagToken() {
 215                 int quot = read();
 216
 217                 switch (quot) {
 218                 case ICharacterScanner.EOF:
 219                 case '>':
 220                         state = STATE_DEFAULT;
 221                         return getToken(XML_TAG);
 222
 223                 case '"':
 224                 case '\'':
 225                         while (true) {
 226                                 int ch = read();
 227
 228                                 if (ch == quot) {
 229                                         state = STATE_TAG;
 230                                         return getToken(XML_ATTRIBUTE);
 231                                 }
 232
 233                                 switch (ch) {
 234                                 case '<':
 235                                         unread();
 236
 237                                 case ICharacterScanner.EOF:
 238                                         state = STATE_DEFAULT;
 239                                         return getToken(XML_ATTRIBUTE);
 240                                 }
 241                         }
 242                 default:
 243                         unread();
 244                 }
 245
 246                 while (true) {
 247                         switch (read()) {
 248                         case '<':
 249                                 unread();
 250
 251                         case ICharacterScanner.EOF:
 252                         case '>':
 253                                 state = STATE_DEFAULT;
 254                                 return getToken(XML_TAG);
 255
 256                         case '"':
 257                         case '\'':
 258                                 unread();
 259
 260                                 state = STATE_TAG;
 261                                 return getToken(XML_TAG);
 262                         }
 263                 }
 264         }
 265
 266         private IToken nextDeclToken() {
 267                 loop: while (true) {
 268                         switch (read()) {
 269                         case ICharacterScanner.EOF:
 270                                 state = STATE_DEFAULT;
 271                                 return getToken(isInternal() ? DTD_INTERNAL_DECL : XML_DECL);
 272
 273                         case '<':
 274                                 if (parsedtd || isInternal()) {
 275                                         switch (read()) {
 276                                         case ICharacterScanner.EOF:
 277                                                 state = STATE_DEFAULT;
 278                                                 return getToken(isInternal() ? DTD_INTERNAL : null);
 279
 280                                         case '!':
 281                                         case '?':
 282                                                 unread();
 283                                                 break;
 284
 285                                         default:
 286                                                 continue loop;
 287                                         }
 288                                 }
 289
 290                                 unread();
 291
 292                         case '>':
 293                                 state &= STATE_INTERNAL;
 294                                 return getToken(isInternal() ? DTD_INTERNAL_DECL : XML_DECL);
 295
 296                         case '[': // <!DOCTYPE xxx [dtd]>
 297                                 if (!isInternal()) {
 298                                         state = STATE_INTERNAL;
 299                                         return getToken(XML_DECL);
 300                                 }
 301                         }
 302                 }
 303         }
 304
 305         private IToken nextCommentToken() {
 306                 state &= STATE_INTERNAL;
 307
 308                 loop: while (true) {
 309                         switch (read()) {
 310                         case ICharacterScanner.EOF:
 311                                 break loop;
 312
 313                         case '-': // - -->
 314                                 switch (read()) {
 315                                 case ICharacterScanner.EOF:
 316                                         break loop;
 317
 318                                 case '-': // -- -->
 319                                         switch (read()) {
 320                                         case ICharacterScanner.EOF:
 321                                         case '>':
 322                                                 break loop;
 323                                         }
 324
 325                                         unread();
 326                                         continue loop;
 327                                 }
 328                         }
 329                 }
 330
 331                 return getToken(isInternal() ? DTD_INTERNAL_COMMENT : XML_COMMENT);
 332         }
 333
 334         private IToken nextCDATAToken() {
 335                 state = STATE_DEFAULT;
 336
 337                 loop: while (true) {
 338                         switch (read()) {
 339                         case ICharacterScanner.EOF:
 340                                 break loop;
 341
 342                         case ']': // ] ]]>
 343                                 switch (read()) {
 344                                 case ICharacterScanner.EOF:
 345                                         break loop;
 346
 347                                 case ']': // ]] ]]>
 348                                         switch (read()) {
 349                                         case ICharacterScanner.EOF:
 350                                         case '>': // ]]>
 351                                                 break loop;
 352                                         }
 353
 354                                         unread();
 355                                         unread();
 356                                         continue loop;
 357                                 }
 358                         }
 359                 }
 360
 361                 return getToken(XML_CDATA);
 362         }
 363
 364         private IToken nextConditionalToken() {
 365                 state = STATE_DEFAULT;
 366
 367                 int level = 1;
 368
 369                 loop: while (true) {
 370                         switch (read()) {
 371                         case ICharacterScanner.EOF:
 372                                 break loop;
 373
 374                         case '<': // - -->
 375                                 switch (read()) {
 376                                 case ICharacterScanner.EOF:
 377                                         break loop;
 378
 379                                 case '!': // -- -->
 380                                         switch (read()) {
 381                                         case ICharacterScanner.EOF:
 382                                                 break loop;
 383
 384                                         case '[':
 385                                                 ++level;
 386                                                 continue loop;
 387                                         }
 388
 389                                         unread();
 390                                         continue loop;
 391                                 }
 392
 393                                 unread();
 394                                 continue loop;
 395
 396                         case ']': // - -->
 397                                 switch (read()) {
 398                                 case ICharacterScanner.EOF:
 399                                         break loop;
 400
 401                                 case ']': // -- -->
 402                                         switch (read()) {
 403                                         case ICharacterScanner.EOF:
 404                                         case '>':
 405                                                 if (--level == 0) {
 406                                                         break loop;
 407                                                 }
 408
 409                                                 continue loop;
 410                                         }
 411
 412                                         unread();
 413                                         unread();
 414                                         continue loop;
 415                                 }
 416                         }
 417                 }
 418
 419                 return getToken(DTD_CONDITIONAL);
 420         }
 421
 422         private IToken getToken(String type) {
 423                 length = position - offset;
 424
 425                 if (length == 0) {
 426                         return Token.EOF;
 427                 }
 428
 429                 if (type == null) {
 430                         return Token.UNDEFINED;
 431                 }
 432
 433                 IToken token = (IToken) tokens.get(type);
 434                 if (token == null) {
 435                         token = new Token(type);
 436                         tokens.put(type, token);
 437                 }
 438
 439                 return token;
 440         }
 441
 442         private boolean isInternal() {
 443                 return (state & STATE_INTERNAL) != 0;
 444         }
 445
 446         private int read() {
 447                 if (position >= end) {
 448                         return ICharacterScanner.EOF;
 449                 }
 450
 451                 try {
 452                         return document.getChar(position++);
 453                 } catch (BadLocationException e) {
 454                         --position;
 455                         return ICharacterScanner.EOF;
 456                 }
 457         }
 458
 459         private void unread() {
 460                 --position;
 461         }
 462
 463         /*
 464          * @see org.eclipse.jface.text.rules.ITokenScanner#getTokenOffset()
 465          */
 466         public int getTokenOffset() {
 467                 Assert.isTrue(offset >= 0, Integer.toString(offset));
 468                 return offset;
 469         }
 470
 471         /*
 472          * @see org.eclipse.jface.text.rules.ITokenScanner#getTokenLength()
 473          */
 474         public int getTokenLength() {
 475                 return length;
 476         }
 477
 478         /*
 479          * @see org.eclipse.jface.text.rules.ITokenScanner#setRange(IDocument, int,
 480          *      int)
 481          */
 482         public void setRange(IDocument document, int offset, int length) {
 483                 this.document = document;
 484                 this.end = offset + length;
 485
 486                 this.offset = offset;
 487                 this.position = offset;
 488                 this.length = 0;
 489
 490                 this.state = STATE_DEFAULT;
 491         }
 492
 493         /*
 494          * @see org.eclipse.jface.text.rules.IPartitionTokenScanner
 495          */
 496         // public void setPartialRange(IDocument document, int offset, int length,
 497         // String contentType, int partitionOffset) {
 498         // state = STATE_DEFAULT;
 499         // if (partitionOffset > -1) {
 500         // int delta = offset - partitionOffset;
 501         // if (delta > 0) {
 502         // setRange(document, partitionOffset, length + delta);
 503         // return;
 504         // }
 505         // }
 506         // setRange(document, partitionOffset, length);
 507         // }
 508         /*
 509          * @see org.eclipse.jface.text.rules.IPartitionTokenScanner
 510          */
 511         public void setPartialRange(IDocument document, int offset, int length,
 512                         String contentType, int partitionOffset) {
 513                 // boolean flag = false;
 514                 this.document = document;
 515                 this.end = offset + length;
 516
 517                 // NB! Undocumented value: -1
 518                 if (partitionOffset >= 0) {
 519                         offset = partitionOffset;
 520                         // flag = true;
 521                 }
 522
 523                 this.offset = offset;
 524                 this.position = offset;
 525                 this.length = 0;
 526
 527                 // if (flag) {
 528                 // state = STATE_DEFAULT;
 529                 // return;
 530                 // }
 531                 if (contentType == XML_ATTRIBUTE) {
 532                         state = STATE_TAG;
 533                         return;
 534                 }
 535
 536                 if (contentType == XML_TAG) {
 537                         state = isContinuationPartition() ? STATE_TAG : STATE_DEFAULT;
 538                         return;
 539                 }
 540
 541                 if (contentType == XML_DECL) {
 542                         state = isContinuationPartition() ? STATE_DECL : STATE_DEFAULT;
 543                         return;
 544                 }
 545
 546                 if (contentType == DTD_INTERNAL || contentType == DTD_INTERNAL_DECL
 547                                 || contentType == DTD_INTERNAL_COMMENT) {
 548                         state = STATE_INTERNAL;
 549                         return;
 550                 }
 551
 552                 state = STATE_DEFAULT;
 553         }
 554
 555         private boolean isContinuationPartition() {
 556                 try {
 557                         String type = document.getContentType(offset - 1);
 558
 559                         if (type != IDocument.DEFAULT_CONTENT_TYPE) {
 560                                 return true;
 561                         }
 562                 } catch (BadLocationException e) {
 563                 }
 564
 565                 return false;
 566         }
 567 }