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

   1 /**********************************************************************
   2  Copyright (c) 2002  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://solareclipse.sourceforge.net/legal/cpl-v10.html
   7
   8  Contributors:
   9  Igor Malinin - initial contribution
  10
  11  $Id: PHPPartitionScanner.java,v 1.35 2007-03-17 14:07:31 axelcl Exp $
  12  **********************************************************************/
  13 package net.sourceforge.phpeclipse.phpeditor.php;
  14
  15 import java.util.HashMap;
  16 import java.util.Map;
  17
  18 import net.sourceforge.phpdt.internal.compiler.parser.Scanner;
  19 import net.sourceforge.phpeclipse.ui.text.rules.AbstractPartitioner;
  20
  21 import org.eclipse.jface.text.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 PHPPartitionScanner implements IPartitionTokenScanner {
  35         public static final String PHP_SCRIPTING_AREA = "__php_scripting_area ";
  36
  37         public static final int STATE_DEFAULT = 0;
  38
  39         // public static final int STATE_TAG = 1;
  40         // public static final int STATE_SCRIPT = 2;
  41
  42         private IDocument document;
  43
  44         // private int begin;
  45
  46         private int end;
  47
  48         private int offset;
  49
  50         private int length;
  51
  52         private int position;
  53
  54         // private int state;
  55
  56         private Map tokens = new HashMap();
  57
  58         public PHPPartitionScanner() {
  59         }
  60
  61         /*
  62          * @see org.eclipse.jface.text.rules.ITokenScanner#nextToken()
  63          */
  64         public IToken nextToken() {
  65                 offset += length;
  66
  67                 /*
  68                  * switch (state) { case STATE_TAG: return nextTagToken(); }
  69                  */
  70
  71                 switch (read()) {
  72                 case ICharacterScanner.EOF:
  73                         // state = STATE_DEFAULT;
  74                         return getToken(null);
  75
  76                 case '<':
  77                         switch (read()) {
  78                         case ICharacterScanner.EOF:
  79                                 // state = STATE_DEFAULT;
  80                                 return getToken(null);
  81
  82                         case '?': // <?
  83                                 // int ch = read();
  84                                 //
  85                                 // switch (ch) {
  86                                 // case ICharacterScanner.EOF:
  87                                 // state = STATE_DEFAULT;
  88                                 // return getToken(PHP_SCRIPTING_AREA);
  89                                 // }
  90                                 return scanUntilPHPEndToken(PHP_SCRIPTING_AREA);
  91                         }
  92
  93                         unread();
  94                 }
  95
  96                 loop: while (true) {
  97                         switch (read()) {
  98                         case ICharacterScanner.EOF:
  99                                 // state = STATE_DEFAULT;
 100                                 return getToken(null);
 101
 102                         case '<':
 103                                 switch (read()) {
 104                                 case ICharacterScanner.EOF:
 105                                         // state = STATE_DEFAULT;
 106                                         return getToken(null);
 107
 108                                 case '?':
 109                                         unread();
 110                                         break;
 111
 112                                 case '<':
 113                                         unread();
 114
 115                                 default:
 116                                         continue loop;
 117                                 }
 118
 119                                 unread();
 120
 121                                 // state = STATE_DEFAULT;
 122                                 return getToken(null);
 123                         }
 124                 }
 125         }
 126
 127         private IToken scanUntilPHPEndToken(String token) {
 128                 int ch = read();
 129                 while (true) {
 130                         switch (ch) {
 131                         case ICharacterScanner.EOF:
 132                                 // state = STATE_DEFAULT;
 133                                 return getToken(token);
 134                         case '"': // double quoted string
 135                                 // read until end of double quoted string
 136                                 if (!readUntilEscapedDQ()) {
 137                                         // state = STATE_DEFAULT;
 138                                         return getToken(token);
 139                                 }
 140                                 break;
 141                         case '<': // heredoc string
 142                                 ch = read();
 143                                 switch (ch) {
 144                                 case ICharacterScanner.EOF:
 145                                         break;
 146                                 case '<':
 147                                         ch = read();
 148                                         switch (ch) {
 149                                         case ICharacterScanner.EOF:
 150                                                 break;
 151                                         case '<':
 152                                                 // read until end of heredoc string
 153                                                 if (!readUntilEscapedHEREDOC()) {
 154                                                         // state = STATE_DEFAULT;
 155                                                         return getToken(token);
 156                                                 }
 157                                         }
 158                                 }
 159                                 break;
 160                         case '\'': // single quoted string
 161                                 // read until end of single quoted string
 162                                 if (!readUntilEscapedSQ()) {
 163                                         // state = STATE_DEFAULT;
 164                                         return getToken(token);
 165                                 }
 166                                 break;
 167                         case '/': // comment start?
 168                                 ch = read();
 169                                 switch (ch) {
 170                                 case ICharacterScanner.EOF:
 171                                         break;
 172                                 case '/':
 173                                         // read until end of line
 174                                         if (!readSingleLine()) {
 175                                                 // state = STATE_DEFAULT;
 176                                                 return getToken(token);
 177                                         }
 178                                         break;
 179                                 case '*':
 180                                         // read until end of comment
 181                                         if (!readMultiLineComment()) {
 182                                                 // state = STATE_DEFAULT;
 183                                                 return getToken(token);
 184                                         }
 185                                         break;
 186                                 default:
 187                                         continue;
 188                                 }
 189                                 break;
 190                         case '#': // line comment
 191                                 // read until end of line
 192                                 if (!readSingleLine()) {
 193                                         // state = STATE_DEFAULT;
 194                                         return getToken(token);
 195                                 }
 196                                 break;
 197                         case '?':
 198                                 ch = read();
 199                                 switch (ch) {
 200                                 case ICharacterScanner.EOF:
 201                                 case '>':
 202                                         // state = STATE_DEFAULT;
 203                                         return getToken(token);
 204
 205                                 case '?':
 206                                         continue;
 207                                 default:
 208                                         continue;
 209                                 }
 210                         }
 211
 212                         ch = read();
 213                 }
 214         }
 215
 216         private IToken getToken(String type) {
 217                 length = position - offset;
 218
 219                 if (length == 0) {
 220                         return Token.EOF;
 221                 }
 222
 223                 // if (length<0) {
 224                 // try {
 225                 // System.out.println("Length<0:"+document.get(offset,5)+""+length);
 226                 // } catch (BadLocationException e) {
 227                 // e.printStackTrace();
 228                 // }
 229                 // }
 230
 231                 if (type == null) {
 232                         return Token.UNDEFINED;
 233                 }
 234
 235                 IToken token = (IToken) tokens.get(type);
 236                 if (token == null) {
 237                         token = new Token(type);
 238                         tokens.put(type, token);
 239                 }
 240
 241                 return token;
 242         }
 243
 244         private int read() {
 245                 if (position >= end) {
 246                         return ICharacterScanner.EOF;
 247                 }
 248
 249                 try {
 250                         return document.getChar(position++);
 251                 } catch (BadLocationException e) {
 252                         --position;
 253                         return ICharacterScanner.EOF;
 254                 }
 255         }
 256
 257         private boolean readUntilEscapedDQ() {
 258                 // search last double quoted character
 259                 try {
 260                         char ch;
 261                         while (true) {
 262                                 if (position >= end) {
 263                                         return false;
 264                                 }
 265                                 ch = document.getChar(position++);
 266                                 if (ch == '\\') {
 267                                         if (position >= end) {
 268                                                 return false;
 269                                         }
 270                                         ch = document.getChar(position++); // ignore escaped
 271                                         // character
 272                                 } else if (ch == '"') {
 273                                         return true;
 274                                 }
 275                         }
 276                 } catch (BadLocationException e) {
 277                         --position;
 278                 }
 279                 return false;
 280         }
 281
 282         private boolean readUntilEscapedSQ() {
 283                 // search last single quoted character
 284                 try {
 285                         char ch;
 286                         while (true) {
 287                                 if (position >= end) {
 288                                         return false;
 289                                 }
 290                                 ch = document.getChar(position++);
 291                                 if (ch == '\\') {
 292                                         if (position >= end) {
 293                                                 return false;
 294                                         }
 295                                         ch = document.getChar(position++); // ignore escaped
 296                                         // character
 297                                 } else if (ch == '\'') {
 298                                         return true;
 299                                 }
 300                         }
 301                 } catch (BadLocationException e) {
 302                         --position;
 303                 }
 304                 return false;
 305         }
 306
 307         private boolean readUntilEscapedHEREDOC() {
 308                 // search until heredoc ends
 309                 try {
 310                         char ch;
 311                         StringBuffer buf = new StringBuffer();
 312                         char[] heredocIdent;
 313                         if (position >= end) {
 314                                 return false;
 315                         }
 316                         ch = document.getChar(position++);
 317                         // #1493165 start
 318                         while (ch == ' ') {
 319                                 if (position >= end) {
 320                                         return false;
 321                                 }
 322                                 ch = document.getChar(position++);
 323                         }
 324                         // #1493165 end
 325                         if (!Scanner.isPHPIdentifierStart(ch)) {
 326                                 return false;
 327                         }
 328                         while (Scanner.isPHPIdentifierPart(ch)) {
 329                                 buf.append(ch);
 330                                 if (position >= end) {
 331                                         return false;
 332                                 }
 333                                 ch = document.getChar(position++);
 334                         }
 335                         heredocIdent = buf.toString().toCharArray();
 336                         while (true) {
 337                                 if (position >= end) {
 338                                         return false;
 339                                 }
 340                                 ch = document.getChar(position++);
 341                                 if (ch == '\n') { // heredoc could end after a newline
 342                                         int pos = 0;
 343                                         while (true) {
 344                                                 if (position >= end) {
 345                                                         return false;
 346                                                 }
 347                                                 if (pos == heredocIdent.length) {
 348                                                         return true;
 349                                                 }
 350                                                 ch = document.getChar(position++); // ignore escaped
 351                                                 // character
 352                                                 if (ch != heredocIdent[pos]) {
 353                                                         break;
 354                                                 }
 355                                                 pos++;
 356                                         }
 357                                 }
 358                         }
 359                 } catch (BadLocationException e) {
 360                         --position;
 361                 }
 362                 return false;
 363         }
 364
 365         private boolean readSingleLine() {
 366                 try {
 367                         do {
 368                                 if (position >= end) {
 369                                         return false;
 370                                 }
 371                         } while (document.getChar(position++) != '\n');
 372                         return true;
 373                 } catch (BadLocationException e) {
 374                         --position;
 375                 }
 376                 return false;
 377         }
 378
 379         private boolean readMultiLineComment() {
 380                 try {
 381                         char ch;
 382                         while (true) {
 383                                 if (position >= end) {
 384                                         return false;
 385                                 }
 386                                 ch = document.getChar(position++);
 387                                 if (ch == '*') {
 388                                         if (position >= end) {
 389                                                 return false;
 390                                         }
 391                                         if (document.getChar(position) == '/') {
 392                                                 position++;
 393                                                 return true;
 394                                         }
 395                                 }
 396                         }
 397                 } catch (BadLocationException e) {
 398                         --position;
 399                 }
 400                 return false;
 401         }
 402
 403         private void unread() {
 404                 --position;
 405         }
 406
 407         /*
 408          * @see org.eclipse.jface.text.rules.ITokenScanner#getTokenOffset()
 409          */
 410         public int getTokenOffset() {
 411                 if (AbstractPartitioner.DEBUG) {
 412                         Assert.isTrue(offset >= 0, Integer.toString(offset));
 413                 }
 414                 return offset;
 415         }
 416
 417         /*
 418          * @see org.eclipse.jface.text.rules.ITokenScanner#getTokenLength()
 419          */
 420         public int getTokenLength() {
 421                 return length;
 422         }
 423
 424         /*
 425          * @see org.eclipse.jface.text.rules.ITokenScanner#setRange(IDocument, int,
 426          *      int)
 427          */
 428         public void setRange(IDocument document, int offset, int length) {
 429                 this.document = document;
 430                 // this.begin = offset;
 431                 this.end = offset + length;
 432
 433                 this.offset = offset;
 434                 this.position = offset;
 435                 this.length = 0;
 436         }
 437
 438         /*
 439          * @see org.eclipse.jface.text.rules.IPartitionTokenScanner
 440          */
 441         public void setPartialRange(IDocument document, int offset, int length,
 442                         String contentType, int partitionOffset) {
 443                 // state = STATE_DEFAULT;
 444                 if (partitionOffset > -1) {
 445                         int delta = offset - partitionOffset;
 446                         if (delta > 0) {
 447                                 setRange(document, partitionOffset, length + delta);
 448                                 return;
 449                         }
 450                 }
 451                 setRange(document, partitionOffset, length);
 452         }
 453
 454         // private boolean isContinuationPartition(IDocument document, int offset) {
 455         // try {
 456         // String type = document.getContentType(offset - 1);
 457         //
 458         // if (type != IDocument.DEFAULT_CONTENT_TYPE) {
 459         // return true;
 460         // }
 461         // } catch (BadLocationException e) {}
 462         //
 463         // return false;
 464         // }
 465 }