a junit Testcase to test the parser
[phpeclipse.git] / net.sourceforge.phpeclipse / src / net / sourceforge / phpeclipse / phpeditor / php / PHPPartitionScanner.java
index 8e2485d..52f8def 100644 (file)
@@ -6,37 +6,41 @@
  * Created on 05.03.2003
  *
  * @author Stefan Langer (musk)
- * @version $Revision: 1.10 $
+ * @version $Revision: 1.19 $
  */
 package net.sourceforge.phpeclipse.phpeditor.php;
 
-import java.util.*;
 
-import org.eclipse.jface.text.*;
-import org.eclipse.jface.text.rules.*;
+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.Map;
+
+import org.eclipse.jface.text.Assert;
+import org.eclipse.jface.text.BadLocationException;
+import org.eclipse.jface.text.IDocument;
+import org.eclipse.jface.text.ITypedRegion;
+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;
 
 /**
  * 
  */
 public class PHPPartitionScanner implements IPartitionTokenScanner
 {
-//    private final int HTML = 0;
-//    private final int PHP = 1;
-//    private final int JS = 2;
-//    private final int CSS = 4;
-//    private final int COMMENT = 5;
-//    private final int HTMLCOMMENT = 6;
-
+    private static final boolean DEBUG = false;
+    private boolean fInString = false;
+    private boolean fInDoubString = false;
     private IDocument fDocument = null;
     private int fOffset = -1;
-//    private int fLastOffset = -1;
     private String fContentType = IPHPPartitionScannerConstants.HTML;
-       
-       private boolean partitionBorder = false;
+    private String fPrevContentType = IPHPPartitionScannerConstants.HTML;
+    private boolean partitionBorder = false;
     private int fTokenOffset;
     private int fEnd = -1;
     private int fLength;
-    //private int fState = HTML;
+    private int fCurrentLength;
     private Map tokens = new HashMap();
 
     public PHPPartitionScanner()
@@ -53,35 +57,102 @@ public class PHPPartitionScanner implements IPartitionTokenScanner
         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!");
-               return token;
+        fLength = fCurrentLength;
+        if (DEBUG)
+        {
+
+            try
+            {
+                if (fLength <= 0)
+                {
+                    int line = fDocument.getLineOfOffset(fOffset);
+                    System.err.println(
+                        "Error at "
+                            + line
+                            + " offset:"
+                            + String.valueOf(
+                                fOffset - fDocument.getLineOffset(line)));
+                }
+            }
+            catch (BadLocationException e)
+            {  // should never happen
+                // TODO Write stacktrace to log
+                e.printStackTrace();
+            }
+        }
+        Assert.isTrue(fLength > 0, "Partition length <= 0!");
+        fCurrentLength = 0;
+        // String can never cross partition borders so reset string detection
+        fInString = false;
+        fInDoubString = false;
+        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="
+                    + type
+                    + " fLength="
+                    + fLength);
+        }
+        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(this.tokens.containsKey(contentType))
-               fContentType = contentType;
-    if (partitionOffset > -1)
+    /* (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)
     {
-       partitionBorder = false;
-        fTokenOffset = partitionOffset;
+        if (DEBUG)
+        {
+            System.out.println(
+                "PartialRange: contentType="
+                    + contentType
+                    + " partitionOffset="
+                    + partitionOffset);
+        }
+
+        try
+        {
+            if (partitionOffset > -1)
+            {
+                partitionBorder = false;
+                // because of strings we have to parse the whole partition
+                this.setRange(
+                    document,
+                    partitionOffset,
+                    offset - partitionOffset + length);
+                // sometimes we get a wrong partition so we retrieve the partition
+                // directly from the document
+                fContentType = fDocument.getContentType(partitionOffset);
+            }
+            else
+                this.setRange(document, offset, length);
+
+        }
+        catch (BadLocationException e)
+        {
+            // should never happen
+            // TODO print stack trace to log
+            // fall back just scan the whole document again
+            this.setRange(document, 0, fDocument.getLength());
+        }
+
     }
-}
 
     /* (non-Javadoc)
      * @see org.eclipse.jface.text.rules.ITokenScanner#getTokenLength()
@@ -95,8 +166,8 @@ public void setPartialRange(
      * @see org.eclipse.jface.text.rules.ITokenScanner#getTokenOffset()
      */
     public int getTokenOffset()
-    {  
-       return fTokenOffset;
+    {
+        return fTokenOffset;
     }
 
     /* (non-Javadoc)
@@ -127,14 +198,20 @@ public void setPartialRange(
             switch (c)
             {
                 case '<' :
-                    if (fContentType != IPHPPartitionScannerConstants.PHP_MULTILINE_COMMENT && checkPattern(new char[] { '?', 'p', 'h', 'p' }, true))
+                    if (!isInString(IPHPPartitionScannerConstants.PHP)
+                        && fContentType
+                            != IPHPPartitionScannerConstants.PHP_MULTILINE_COMMENT
+                        && checkPattern(new char[] { '?', 'p', 'h', 'p' }, true))
                     {
-
                         if (fContentType != IPHPPartitionScannerConstants.PHP
-                            && fOffset - 5 > 0)
+                            && fCurrentLength > 5)
                         {
-                            fOffset -= 5;
+                            unread(5);
                             IToken token = getToken(fContentType);
+                            // save previouse contenttype
+                            //TODO build stack for previouse contenttype 
+                            fPrevContentType = fContentType;
+
                             fContentType = IPHPPartitionScannerConstants.PHP;
 
                             return token;
@@ -144,16 +221,41 @@ public void setPartialRange(
 
                         // remember offset of this partition
                         fTokenOffset = fOffset - 5;
+                        fCurrentLength = 5;
                     }
-                    else if (checkPattern(new char[] { '!', '-', '-' }))
+                    else if (
+                        !isInString(IPHPPartitionScannerConstants.PHP)
+                            && fContentType
+                                != IPHPPartitionScannerConstants
+                                    .PHP_MULTILINE_COMMENT
+                            && checkPattern(new char[] { '?' }, false))
                     {
-                        // return previouse partition
+                        if (fContentType != IPHPPartitionScannerConstants.PHP
+                            && fCurrentLength > 2)
+                        {
+                            unread(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;
+                        fCurrentLength = 2;
+                    }
+                    else if (
+                        !isInString(IPHPPartitionScannerConstants.PHP)
+                            && checkPattern(new char[] { '!', '-', '-' }))
+                    { // return previouse partition
                         if (fContentType
                             != IPHPPartitionScannerConstants
                                 .HTML_MULTILINE_COMMENT
-                            && fOffset - 4 > 0)
+                            && fCurrentLength > 4)
                         {
-                            fOffset -= 4;
+                            unread(4);
                             IToken token = getToken(fContentType);
                             fContentType =
                                 IPHPPartitionScannerConstants
@@ -164,26 +266,34 @@ public void setPartialRange(
                             fContentType =
                                 IPHPPartitionScannerConstants
                                     .HTML_MULTILINE_COMMENT;
-
+                                    
                         fTokenOffset = fOffset - 4;
+                        fCurrentLength = 4;
                     }
                     break;
                 case '?' :
-                    if (fContentType == IPHPPartitionScannerConstants.PHP)
+                    if (!isInString(IPHPPartitionScannerConstants.PHP)
+                        && fContentType == IPHPPartitionScannerConstants.PHP)
                     {
                         if ((c = read()) == '>')
-                        {
-                            fContentType = IPHPPartitionScannerConstants.HTML;
+                        { 
+                            if (fPrevContentType != null)
+                                fContentType = fPrevContentType;
+                            else
+                                fContentType =
+                                    IPHPPartitionScannerConstants.HTML;
                             partitionBorder = true;
                             return getToken(IPHPPartitionScannerConstants.PHP);
                         }
-                        else if(c != ICharacterScanner.EOF)
+                        else if (c != ICharacterScanner.EOF)
                             unread();
                     }
                     break;
                 case '-' :
-                    if (fContentType
-                        == IPHPPartitionScannerConstants.HTML_MULTILINE_COMMENT
+                    if (!isInString(IPHPPartitionScannerConstants.PHP)
+                        && fContentType
+                            == IPHPPartitionScannerConstants
+                                .HTML_MULTILINE_COMMENT
                         && checkPattern(new char[] { '-', '>' }))
                     {
                         fContentType = IPHPPartitionScannerConstants.HTML;
@@ -194,17 +304,16 @@ public void setPartialRange(
                     }
                     break;
                 case '/' :
-                    if ((c=read()) == '*')
+                    if (!isInString(IPHPPartitionScannerConstants.PHP) && (c = read()) == '*')
                     { // MULTINE COMMENT JAVASCRIPT, CSS, PHP
                         if (fContentType == IPHPPartitionScannerConstants.PHP
-                            && fOffset - 2 > 0)
+                            && fCurrentLength > 2)
                         {
-                            fOffset -= 2;
+                            unread(2);
                             IToken token = getToken(fContentType);
                             fContentType =
                                 IPHPPartitionScannerConstants
                                     .PHP_MULTILINE_COMMENT;
-
                             return token;
                         }
                         else if (
@@ -214,14 +323,15 @@ public void setPartialRange(
                         {
 
                             fTokenOffset = fOffset - 2;
+                            fCurrentLength = 2;
                         }
 
                     }
-                    else if(c != ICharacterScanner.EOF)
+                    else if (!isInString(IPHPPartitionScannerConstants.PHP) && c != ICharacterScanner.EOF)
                         unread();
                     break;
                 case '*' :
-                    if ((c = read()) == '/')
+                    if (!isInString(IPHPPartitionScannerConstants.PHP) && (c = read()) == '/')
                     {
                         if (fContentType
                             == IPHPPartitionScannerConstants
@@ -246,13 +356,20 @@ public void setPartialRange(
                         {
                         }
                     }
-                    else if(c != ICharacterScanner.EOF)
+                    else if (!isInString(IPHPPartitionScannerConstants.PHP) && c != ICharacterScanner.EOF)
                         unread();
                     break;
+                case '\'' :
+                    if (!fInDoubString)
+                        fInString = !fInString;
+                    break;
+                case '"' :
+                    // toggle String mode
+                    if (!fInString)
+                        fInDoubString = !fInDoubString;
+                    break;
             }
-        }
-
-        // end of file reached but we have to return the
+        } // end of file reached but we have to return the
         // last partition.
         return getToken(fContentType);
     }
@@ -261,23 +378,34 @@ public void setPartialRange(
      */
     public void setRange(IDocument document, int offset, int length)
     {
+        if (DEBUG)
+        {
+            System.out.println(
+                "SET RANGE: offset=" + offset + " length=" + length);
+        }
+
         fDocument = document;
         fOffset = offset;
         fTokenOffset = offset;
-        fLength = 0;    
+        fCurrentLength = 0;
+        fLength = 0;
         fEnd = fOffset + length;
-               //partitionBorder = false;
+        fInString = false;
+        fInDoubString = false;
+        fContentType = IPHPPartitionScannerConstants.HTML;
+//        String[] prev = getPartitionStack(offset);
     }
 
     private int read()
     {
         try
         {
-                       if (fOffset < fEnd)
+            if (fOffset < fEnd)
             {
-                               return fDocument.getChar(fOffset++);
+               fCurrentLength++;
+                return fDocument.getChar(fOffset++);
             }
-                       return ICharacterScanner.EOF;
+            return ICharacterScanner.EOF;
         }
         catch (BadLocationException e)
         {
@@ -291,6 +419,13 @@ public void setPartialRange(
     private void unread()
     {
         --fOffset;
+        --fCurrentLength;
+    }
+    
+    private void unread(int num)
+    {
+       fOffset -= num;
+       fCurrentLength -= num;
     }
 
     private boolean checkPattern(char[] pattern)
@@ -308,31 +443,93 @@ public void setPartialRange(
      */
     private boolean checkPattern(char[] pattern, boolean ignoreCase)
     {
-       int prevOffset = fOffset;
+        int prevOffset = fOffset;
+        int prevLength = fCurrentLength;
         for (int i = 0; i < pattern.length; i++)
         {
             int c = read();
-                       
-                       if(c == ICharacterScanner.EOF || !letterEquals(c, pattern[i], ignoreCase))
+
+            if (c == ICharacterScanner.EOF
+                || !letterEquals(c, pattern[i], ignoreCase))
             {
-               fOffset = prevOffset;
+                fOffset = prevOffset;
+                fCurrentLength = prevLength;
                 return false;
             }
         }
 
         return true;
     }
-    
+
     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;
+        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;
+
+        return false;
+    }
+    
+    /**
+     * Checks wether the offset is in a <code>String</code> and the specified 
+     * contenttype is the current content type.
+     * Strings are delimited, mutual exclusive, by a " or by a '.
+     * 
+     * @param contentType The contenttype to check.
+     * @return <code>true</code> if the current offset is in a string else 
+     *                         returns false.
+     */
+    private  boolean isInString(String contentType)
+    {
+       if(fContentType == contentType)
+               return (fInString || fInDoubString);
+       else
+               return false;           
+    }
+    
+    /**
+     * Returns the previouse partition stack for the given offset.
+     * 
+     * @param offset The offset to return the previouse partitionstack for.
+     * 
+     * @return The stack as a string array.
+     */
+    private String[] getPartitionStack(int offset)
+    {
+       ArrayList types = new ArrayList();
+       int tmpOffset = 0;
+       try
+        {
+            ITypedRegion region = fDocument.getPartition(offset);
+            tmpOffset = region.getOffset();
+            while(tmpOffset-1 > 0)
+            {
+               region = fDocument.getPartition(tmpOffset-1);
+               tmpOffset = region.getOffset();
+               types.add(0, region.getType());
+            }
+        }
+        catch (BadLocationException e)
+        {
+           if(DEBUG)
+           {
+                       e.printStackTrace();
+           }
+        }
+       
+               String[] retVal = new String[types.size()];
        
-       return false;
+       retVal = (String[])types.toArray(retVal);
+       return retVal;
     }
     
 }