Improved xml scanner for this bug
[phpeclipse.git] / net.sourceforge.phpeclipse / src / net / sourceforge / phpeclipse / phpeditor / php / PHPPartitionScanner.java
index 54eaa7d..a4581f8 100644 (file)
-/**
- * This program and the accompanying materials
- * are made available under the terms of the Common Public License v1.0
- * which accompanies this distribution, and is available at
- * http://www.eclipse.org/legal/cpl-v10.html
- * Created on 05.03.2003
- *
- * @author Stefan Langer (musk)
- * @version $Revision: 1.15 $
- */
+/**********************************************************************
+ Copyright (c) 2002  Widespace, OU  and others.
+ All rights reserved.   This program and the accompanying materials
+ are made available under the terms of the Common Public License v1.0
+ which accompanies this distribution, and is available at
+ http://solareclipse.sourceforge.net/legal/cpl-v10.html
+
+ Contributors:
+ Igor Malinin - initial contribution
+
+ $Id: PHPPartitionScanner.java,v 1.29 2005-05-15 23:24:41 axelcl Exp $
+ **********************************************************************/
 package net.sourceforge.phpeclipse.phpeditor.php;
 
-import java.util.*;
+import java.util.HashMap;
+import java.util.Map;
 
-import org.eclipse.jface.text.*;
-import org.eclipse.jface.text.rules.*;
+import net.sourceforge.phpeclipse.ui.text.rules.AbstractPartitioner;
+
+import org.eclipse.jface.text.Assert;
+import org.eclipse.jface.text.BadLocationException;
+import org.eclipse.jface.text.IDocument;
+import org.eclipse.jface.text.rules.ICharacterScanner;
+import org.eclipse.jface.text.rules.IPartitionTokenScanner;
+import org.eclipse.jface.text.rules.IToken;
+import org.eclipse.jface.text.rules.Token;
 
 /**
  * 
+ * 
+ * @author Igor Malinin
  */
 public class PHPPartitionScanner implements IPartitionTokenScanner {
-  private static final boolean DEBUG = false;
-  private IDocument fDocument = null;
-  private int fOffset = -1;
-  private String fContentType = IPHPPartitionScannerConstants.HTML;
-  private String fPrevContentType;
-
-  private boolean partitionBorder = false;
-  private int fTokenOffset;
-  private int fEnd = -1;
-  private int fLength;
+  public static final String PHP_SCRIPTING_AREA = "__php_scripting_area ";
+
+  public static final int STATE_DEFAULT = 0;
+
+  //   public static final int STATE_TAG = 1;
+  //   public static final int STATE_SCRIPT = 2;
+
+  private IDocument document;
+
+  private int begin;
+
+  private int end;
+
+  private int offset;
+
+  private int length;
+
+  private int position;
+
+  private int state;
+
   private Map tokens = new HashMap();
 
   public PHPPartitionScanner() {
-    this.tokens.put(
-      IPHPPartitionScannerConstants.PHP,
-      new Token(IPHPPartitionScannerConstants.PHP));
-    this.tokens.put(
-      IPHPPartitionScannerConstants.PHP_MULTILINE_COMMENT,
-      new Token(IPHPPartitionScannerConstants.PHP_MULTILINE_COMMENT));
-    this.tokens.put(
-      IPHPPartitionScannerConstants.HTML,
-      new Token(IPHPPartitionScannerConstants.HTML));
-    this.tokens.put(
-      IPHPPartitionScannerConstants.HTML_MULTILINE_COMMENT,
-      new Token(IPHPPartitionScannerConstants.HTML_MULTILINE_COMMENT));
-    this.tokens.put(
-      IDocument.DEFAULT_CONTENT_TYPE,
-      new Token(IDocument.DEFAULT_CONTENT_TYPE));
   }
 
-  private IToken getToken(String type) {
-    fLength = fOffset - fTokenOffset;
-    IToken token = (IToken) this.tokens.get(type);
-    Assert.isNotNull(token, "Token for type \"" + type + "\" not found!");
-    if (DEBUG) {
-      System.out.println(
-        "Partition: fTokenOffset="
-          + fTokenOffset
-          + " fContentType="
-          + fContentType
-          + " fLength="
-          + fLength);
+  /*
+   * @see org.eclipse.jface.text.rules.ITokenScanner#nextToken()
+   */
+  public IToken nextToken() {
+    offset += length;
+
+    /*
+     * switch (state) { case STATE_TAG: return nextTagToken(); }
+     */
+
+    switch (read()) {
+    case ICharacterScanner.EOF:
+      state = STATE_DEFAULT;
+      return getToken(null);
+
+    case '<':
+      switch (read()) {
+      case ICharacterScanner.EOF:
+        state = STATE_DEFAULT;
+        return getToken(null);
+
+      case '?': // <?
+//        int ch = read();
+//
+//        switch (ch) {
+//        case ICharacterScanner.EOF:
+//          state = STATE_DEFAULT;
+//          return getToken(PHP_SCRIPTING_AREA);
+//        }
+        return scanUntilPHPEndToken(PHP_SCRIPTING_AREA);
+      }
+
+      unread();
     }
-    return token;
-  }
 
-/* (non-Javadoc)
- * @see org.eclipse.jface.text.rules.IPartitionTokenScanner#setPartialRange(org.eclipse.jface.text.IDocument, int, int, java.lang.String, int)
- */
-public void setPartialRange(
-    IDocument document,
-    int offset,
-    int length,
-    String contentType,
-    int partitionOffset)
-{
-    this.setRange(document, offset, length);
-    if (DEBUG)
-    {
-        System.out.println(
-            "PartialRange: contentType="
-                + contentType
-                + " partitionOffset="
-                + partitionOffset);
+    loop: while (true) {
+      switch (read()) {
+      case ICharacterScanner.EOF:
+        state = STATE_DEFAULT;
+        return getToken(null);
+
+      case '<':
+        switch (read()) {
+        case ICharacterScanner.EOF:
+          state = STATE_DEFAULT;
+          return getToken(null);
+
+        case '?':
+          unread();
+          break;
+
+        case '<':
+          unread();
+
+        default:
+          continue loop;
+        }
+
+        unread();
+
+        state = STATE_DEFAULT;
+        return getToken(null);
+      }
     }
-       
-       if(offset == partitionOffset)
-       {
-               try
-        {
-            fContentType = fDocument.getContentType(offset);
+  }
+
+  private IToken scanUntilPHPEndToken(String token) {
+    int ch = read();
+    while (true) {
+      switch (ch) {
+      case ICharacterScanner.EOF:
+        state = STATE_DEFAULT;
+        return getToken(token);
+      case '"': // double quoted string
+        // read until end of double quoted string
+        if (!readUntilEscapedDQ()) {
+          state = STATE_DEFAULT;
+          return getToken(token);
         }
-        catch (BadLocationException e)
-        {
-            //should never happen
+        break;
+      case '\'': // single quoted string
+        // read until end of single quoted string
+        if (!readUntilEscapedSQ()) {
+          state = STATE_DEFAULT;
+          return getToken(token);
         }
-       }
-    else if (this.tokens.containsKey(contentType))
-        fContentType = contentType;
-    // TODO Calculate previouse contenttype
-    if (partitionOffset > -1)
-    {
-        partitionBorder = false;
-        fTokenOffset = partitionOffset;
-    }
-}
+        break;
+      case '/': // comment start?
+        ch = read();
+        switch (ch) {
+        case ICharacterScanner.EOF:
+          break;
+        case '/':
+          // read until end of line
+          if (!readSingleLine()) {
+            state = STATE_DEFAULT;
+            return getToken(token);
+          }
+          break;
+        case '*':
+          // read until end of comment
+          if (!readMultiLineComment()) {
+            state = STATE_DEFAULT;
+            return getToken(token);
+          }
+          break;
+        default:
+          continue;
+        }
+        break;
+      case '#': // line comment
+        // read until end of line
+        if (!readSingleLine()) {
+          state = STATE_DEFAULT;
+          return getToken(token);
+        }
+        break;
+      case '?':
+        ch = read();
+        switch (ch) {
+        case ICharacterScanner.EOF:
+        case '>':
+          state = STATE_DEFAULT;
+          return getToken(token);
 
-  /* (non-Javadoc)
-   * @see org.eclipse.jface.text.rules.ITokenScanner#getTokenLength()
-   */
-  public int getTokenLength() {
-    return fLength;
-  }
+        case '?':
+          continue;
+        default:
+          continue;
+        }
+      }
 
-  /* (non-Javadoc)
-   * @see org.eclipse.jface.text.rules.ITokenScanner#getTokenOffset()
-   */
-  public int getTokenOffset() {
-    return fTokenOffset;
+      ch = read();
+    }
   }
 
-/* (non-Javadoc)
- * @see org.eclipse.jface.text.rules.ITokenScanner#nextToken()
- */
-public IToken nextToken()
-{
-    int c;
-
-    // check if we are not allready at the end of the
-    // file
-    if ((c = read()) == ICharacterScanner.EOF)
-    {
-        partitionBorder = false;
-        return Token.EOF;
+  private IToken getToken(String type) {
+    length = position - offset;
+
+    if (length == 0) {
+      return Token.EOF;
     }
-    else
-        unread();
 
-    if (partitionBorder)
-    {
-        fTokenOffset = fOffset;
-        partitionBorder = false;
+//    if (length<0) {
+//      try {
+//        System.out.println("Length<0:"+document.get(offset,5)+""+length);
+//      } catch (BadLocationException e) {
+//        e.printStackTrace();
+//      }
+//    }
+    
+    if (type == null) {
+      return Token.UNDEFINED;
     }
 
-    while ((c = read()) != ICharacterScanner.EOF)
-    {
-        switch (c)
-        {
-            case '<' :
-                if (fContentType
-                    != IPHPPartitionScannerConstants.PHP_MULTILINE_COMMENT
-                    && checkPattern(new char[] { '?', 'p', 'h', 'p' }, true))
-                {
-                    if (fContentType != IPHPPartitionScannerConstants.PHP
-                        && fOffset - 5 > 0)
-                    {
-                        fOffset -= 5;
-                        IToken token = getToken(fContentType);
-                        // save previouse contenttype
-                        fPrevContentType = fContentType;
-
-                        fContentType = IPHPPartitionScannerConstants.PHP;
-
-                        return token;
-                    }
-                    else
-                        fContentType = IPHPPartitionScannerConstants.PHP;
-
-                    // remember offset of this partition
-                    fTokenOffset = fOffset - 5;
-                }
-                else if (
-                    fContentType
-                        != IPHPPartitionScannerConstants.PHP_MULTILINE_COMMENT
-                    && checkPattern(new char[] { '?' }, false))
-                {
-                    if (fContentType != IPHPPartitionScannerConstants.PHP
-                        && fOffset - 2 > 0)
-                    {
-                        fOffset -= 2;
-                            IToken token = getToken(fContentType);
-                // save previouse contenttype
-                        fPrevContentType =
-                        fContentType;
-                            fContentType = IPHPPartitionScannerConstants.PHP;
-                            return token;
-                            }
-                    else
-                        fContentType = IPHPPartitionScannerConstants.PHP;
-            // remember offset of this partition
-                        fTokenOffset = fOffset - 2;
-                            }
-                else if (checkPattern(new char[] { '!', '-', '-' }))
-                { // return previouse partition
-                if (
-                    fContentType
-                        != IPHPPartitionScannerConstants.HTML_MULTILINE_COMMENT
-                    && fOffset - 4 > 0)
-                    {
-                        fOffset -= 4;
-                            IToken token = getToken(fContentType);
-                            fContentType =
-                                IPHPPartitionScannerConstants
-                                    .HTML_MULTILINE_COMMENT;
-                            return token;
-                            }
-                    else
-                        fContentType =
-                            IPHPPartitionScannerConstants
-                                .HTML_MULTILINE_COMMENT;
-                            fTokenOffset = fOffset - 4;
-                            }
-                break; case '?' :
-                if (fContentType == IPHPPartitionScannerConstants.PHP)
-                {
-                    if ((c = read()) == '>')
-                    { // TODO Actually calculate the previouse contenttype from the document
-                    if (
-                        fPrevContentType
-                        != null)
-                            fContentType = fPrevContentType; else
-                                fContentType =
-                                    IPHPPartitionScannerConstants.HTML;
-                                    partitionBorder = true;
-                                    return getToken(
-                                        IPHPPartitionScannerConstants.PHP);
-                                    }
-                    else if (c != ICharacterScanner.EOF)
-                        unread(); }
-                break; case '-' :
-                if (fContentType
-                    == IPHPPartitionScannerConstants.HTML_MULTILINE_COMMENT
-                    && checkPattern(new char[] { '-', '>' }))
-                {
-                    fContentType = IPHPPartitionScannerConstants.HTML;
-                        partitionBorder = true;
-                        return getToken(
-                            IPHPPartitionScannerConstants
-                                .HTML_MULTILINE_COMMENT);
-                        }
-                break; case '/' :
-                if ((c = read()) == '*')
-                { // MULTINE COMMENT JAVASCRIPT, CSS, PHP
-                if (
-                    fContentType == IPHPPartitionScannerConstants.PHP
-                    && fOffset - 2 > 0)
-                    {
-                        fOffset -= 2;
-                            IToken token = getToken(fContentType);
-                            fContentType =
-                                IPHPPartitionScannerConstants
-                                    .PHP_MULTILINE_COMMENT;
-                            return token;
-                            }
-                    else if (
-                        fContentType
-                            == IPHPPartitionScannerConstants
-                                .PHP_MULTILINE_COMMENT)
-                    {
-
-                        fTokenOffset = fOffset - 2; }
-
-                }
-                else if (c != ICharacterScanner.EOF)
-                    unread(); break; case '*' :
-                        if ((c = read()) == '/')
-                        {
-                            if (fContentType
-                                == IPHPPartitionScannerConstants
-                                    .PHP_MULTILINE_COMMENT)
-                            {
-                                fContentType =
-                                    IPHPPartitionScannerConstants.PHP;
-                                    partitionBorder = true;
-                                    return getToken(
-                                        IPHPPartitionScannerConstants
-                                            .PHP_MULTILINE_COMMENT);
-                                    }
-                            else if (
-                                fContentType
-                                    == IPHPPartitionScannerConstants
-                                        .CSS_MULTILINE_COMMENT)
-                            {
-                            }
-                            else if (
-                                fContentType
-                                    == IPHPPartitionScannerConstants
-                                        .JS_MULTILINE_COMMENT)
-                            {
-                            }
-                        }
-                        else if (c != ICharacterScanner.EOF)
-                            unread(); break; }
-    } // end of file reached but we have to return the
-    // last partition.
-    return getToken(fContentType);
-        }
-  /* (non-Javadoc)
-   * @see org.eclipse.jface.text.rules.ITokenScanner#setRange(org.eclipse.jface.text.IDocument, int, int)
-   */
-  public void setRange(IDocument document, int offset, int length) {
-    if (DEBUG) {
-      System.out.println("SET RANGE: offset=" + offset + " length=" + length);
+    IToken token = (IToken) tokens.get(type);
+    if (token == null) {
+      token = new Token(type);
+      tokens.put(type, token);
     }
 
-    fDocument = document;
-    fOffset = offset;
-    fTokenOffset = offset;
-    fLength = 0;
-    fEnd = fOffset + length;
-    //partitionBorder = false;
+    return token;
   }
 
   private int read() {
-    try {
-      if (fOffset < fEnd) {
-        return fDocument.getChar(fOffset++);
-      }
+    if (position >= end) {
       return ICharacterScanner.EOF;
+    }
+
+    try {
+      return document.getChar(position++);
     } catch (BadLocationException e) {
-      // should never happen
-      // TODO write stacktrace to log
-      fOffset = fEnd;
+      --position;
       return ICharacterScanner.EOF;
     }
   }
 
-  private void unread() {
-    --fOffset;
+  private boolean readUntilEscapedDQ() {
+    // search last double quoted character
+    try {
+      char ch;
+      while (true) {
+        if (position >= end) {
+          return false;
+        }
+        ch = document.getChar(position++);
+        if (ch == '\\') {
+          if (position >= end) {
+            return false;
+          }
+          ch = document.getChar(position++); // ignore escaped character
+        } else if (ch == '"') {
+          return true;
+        }
+      }
+    } catch (BadLocationException e) {
+      --position;
+    }
+    return false;
+  }
+  
+  private boolean readUntilEscapedSQ() {
+    // search last single quoted character
+    try {  
+      char ch;
+      while (true) {
+        if (position >= end) {
+          return false;
+        }
+        ch = document.getChar(position++); 
+        if (ch == '\\') {
+          if (position >= end) {
+            return false;
+          }
+          ch = document.getChar(position++); // ignore escaped character
+        } else if (ch == '\'') {
+          return true;
+        }
+      }
+    } catch (BadLocationException e) {
+      --position;
+    }
+    return false;
   }
 
-  private boolean checkPattern(char[] pattern) {
-    return checkPattern(pattern, false);
+  private boolean readSingleLine() {
+    try {
+      do {
+        if (position >= end) {
+          return false;
+        }
+      } while (document.getChar(position++) != '\n'); 
+      return true;
+    } catch (BadLocationException e) {
+      --position;
+    }
+    return false;
   }
 
-  /**
-   * Check if next character sequence read from document is equals to 
-   * the provided pattern. Pattern is read from left to right until the 
-   * first character read doesn't match. If this happens all read characters are
-   * unread.
-   * @param pattern The pattern to check.
-   * @return <code>true</code> if pattern is equals else returns <code>false</code>.
-   */
-  private boolean checkPattern(char[] pattern, boolean ignoreCase) {
-    int prevOffset = fOffset;
-    for (int i = 0; i < pattern.length; i++) {
-      int c = read();
-
-      if (c == ICharacterScanner.EOF
-        || !letterEquals(c, pattern[i], ignoreCase)) {
-        fOffset = prevOffset;
-        return false;
+  private boolean readMultiLineComment() {
+    try {
+      char ch;
+      while (true) {
+        if (position >= end) {
+          return false;
+        }
+        ch = document.getChar(position++);
+        if (ch == '*') {
+          if (position >= end) {
+            return false;
+          }
+          if (document.getChar(position) == '/') {
+            position++;
+            return true;
+          }
+        }
       }
+    } catch (BadLocationException e) {
+      --position;
     }
+    return false;
+  }
 
-    return true;
+  private void unread() {
+    --position;
   }
 
-  private boolean letterEquals(int test, char letter, boolean ignoreCase) {
-    if (test == letter)
-      return true;
-    else if (
-      ignoreCase
-        && Character.isLowerCase(letter)
-        && test == Character.toUpperCase(letter))
-      return true;
-    else if (
-      ignoreCase
-        && Character.isUpperCase(letter)
-        && test == Character.toLowerCase(letter))
-      return true;
+  /*
+   * @see org.eclipse.jface.text.rules.ITokenScanner#getTokenOffset()
+   */
+  public int getTokenOffset() {
+    if (AbstractPartitioner.DEBUG) {
+      Assert.isTrue(offset >= 0, Integer.toString(offset));
+    }
+    return offset;
+  }
 
-    return false;
+  /*
+   * @see org.eclipse.jface.text.rules.ITokenScanner#getTokenLength()
+   */
+  public int getTokenLength() {
+    return length;
+  }
+
+  /*
+   * @see org.eclipse.jface.text.rules.ITokenScanner#setRange(IDocument, int, int)
+   */
+  public void setRange(IDocument document, int offset, int length) {
+    this.document = document;
+    this.begin = offset;
+    this.end = offset + length;
+
+    this.offset = offset;
+    this.position = offset;
+    this.length = 0;
+  }
+
+  /*
+   * @see org.eclipse.jface.text.rules.IPartitionTokenScanner
+   */
+  public void setPartialRange(IDocument document, int offset, int length, String contentType, int partitionOffset) {
+    state = STATE_DEFAULT;
+    if (partitionOffset > -1) {
+      int delta = offset - partitionOffset;
+      if (delta > 0) {
+        setRange(document, partitionOffset, length + delta);
+        return;
+      }
+    }
+    setRange(document, partitionOffset, length);
   }
 
-}
+  //   private boolean isContinuationPartition(IDocument document, int offset) {
+  //           try {
+  //                   String type = document.getContentType(offset - 1);
+  //
+  //                   if (type != IDocument.DEFAULT_CONTENT_TYPE) {
+  //                           return true;
+  //                   }
+  //           } catch (BadLocationException e) {}
+  //
+  //           return false;
+  //   }
+}
\ No newline at end of file