htmlrewriter.java

jetspeed源代码

        private boolean removeHead = true;
        private boolean openInNewWindow = false;
        
        // remove the onClick=, onBlur=, etc. - Attributes
        private boolean removeOnSomething = true;
        
        private boolean inScript = false;
        private boolean inStyle = false;
        
        private StringWriter result = new StringWriter();
        
        private Callback () {
        }
        
        
        private Callback addToResult(Object txt)
        {
            // to allow for implementation using Stringbuffer or StringWriter
            // I don't know yet, which one is better in this case
            if (ignoreLevel > 0) return this;

            try {
                result.write(txt.toString());
            } catch (Exception e) { /* ignore */ }
            return this;
        }

        private Callback addToResult(char[] txt)
        {
            if (ignoreLevel > 0) return this;

            try {
                result.write(txt);
            } catch (Exception e) { /* ignore */ }
            return this;
        }
        
        /** Accessor to the Callback's content-String
         * @return Cleaned and rewritten HTML-Content
         */        
        public String getResult() {
            try {
                result.flush();
            } catch (Exception e) { /* ignore */ }
            
            // WARNING: doesn't work, if you remove " " + ... but don't know why
            String res = " " + result.toString(); 

            return res;
        }
        
       
        public void flush() throws javax.swing.text.BadLocationException {
            // nothing to do here ...
        }

        /** 
         * Because Scripts and Stlyle sometimes are defined in comments, thoese
         * will be written. Otherwise comments are removed
         */
        public void handleComment(char[] values,int param) {
            if ( !( inStyle || inScript))
                return;

            try {
                result.write("<!--");
                result.write(values);
                result.write("-->");
            } catch (Exception e) { /* ignore */ }
          // we ignore them 
        }

        public void handleEndOfLineString(java.lang.String str) {
            addToResult("\n");
        }

        public void handleError(java.lang.String str,int param) {
            // ignored
        }

        public void handleSimpleTag(HTML.Tag tag,MutableAttributeSet attrs,int param) {
            if (removeMeta && (tag == HTML.Tag.META)) {
                return;
            }            
            appendTagToResult(tag,attrs);        
        }

        public void handleStartTag(HTML.Tag tag,  MutableAttributeSet attrs, int position) {
            appendTagToResult(tag,attrs);
        }

        public void handleEndTag(HTML.Tag tag, int position) {
            if ((tag ==HTML.Tag.FORM) && (inForm)) { 
                // form handling seems to be buggy
                addToResult("</").addToResult(tag).addToResult(">");
                inForm = false;
            } else if (tag == HTML.Tag.FORM) {
                // do nothing! ... i.e. we are now outside of any <FORM>, so a
                // closing </form> is not really needed ... 
            } else {
                addToResult("</").addToResult(tag).addToResult(">");
            }
            
            
            if ( (removeScript == false) && (tag == HTML.Tag.SCRIPT)) {
                inScript = false;
            } else if ( (removeStyle == false) && (tag == HTML.Tag.STYLE)) {
                inStyle = false;
            }

            if ( removeScript && (tag == HTML.Tag.SCRIPT)) {
                ignoreLevel --;
            } else if ( removeStyle && (tag == HTML.Tag.STYLE)) {
                ignoreLevel --;
            } else if ( removeHead && (tag == HTML.Tag.HEAD)) {
                ignoreLevel --;
            } else if ( removeApplet && (tag == HTML.Tag.APPLET)) {
                ignoreLevel --;
            } else if ( removeObject && (tag == HTML.Tag.OBJECT)) {
                ignoreLevel --;
            } else if ( removeNoScript && (tag.toString().equalsIgnoreCase("NOSCRIPT"))) {
                ignoreLevel --;
            }
        }
  
        private void appendTagToResult(HTML.Tag tag, MutableAttributeSet attrs) {

            if (tag.toString().equalsIgnoreCase("__ENDOFLINETAG__")) {
                // jdk 1.2.2 places a tag <__ENDOFLINETAG__> in the result ...
                // we don't want this one
                return;
            }
            
            if (tag.toString().equalsIgnoreCase("__IMPLIED__")) {
                // jdk 1.3 places a tag <__IMPLIED__> in the result ...
                // we don't want this one
                return;
            }
            
            convertURLS(tag,attrs);
            Enumeration e = attrs.getAttributeNames();
            if (tag == HTML.Tag.BASE)
                return;
            
            addToResult("<").addToResult(tag);
            while (e.hasMoreElements()) {
                Object attr = e.nextElement();
                String attrName = attr.toString();
                String value = attrs.getAttribute(attr).toString();

                // include attribute only when Not(RemoveOnSomething = True and starts with "on")
                if (!(removeOnSomething
                && attrName.toLowerCase().startsWith("on")
                && (attrName.length() > 2))) {
                    // Attribute included
                    addToResult(" ").addToResult(attr).addToResult("=\"")
                    .addToResult(value).addToResult("\"");
                }
            }
            addToResult(">");
        }
                   
        /** Here the magic happens.
         *
         * If someone wants new types of URLs to be rewritten, add them here
         * @param tag TAG from the Callback-Interface
         * @param attrs Attribute-Set from the Callback-Interface
         */
        
        private void convertURLS( HTML.Tag tag, MutableAttributeSet attrs ) {

           // first we do an URL-rewrite on different tags
            
            if (tag == HTML.Tag.A) {
                if (attrs.getAttribute(HTML.Attribute.HREF) != null) {
                    // ---- CHECKING <A HREF
                    addConvertedAttribute( HTML.Attribute.HREF,
                    attrs );
                }
                if ((attrs.getAttribute(HTML.Attribute.TARGET) == null) && cb.openInNewWindow) {
                    attrs.addAttribute(HTML.Attribute.TARGET, "_BLANK");
                }
            } else if (tag == HTML.Tag.AREA) {
                if (attrs.getAttribute(HTML.Attribute.HREF) != null) {
                    // ---- CHECKING <A HREF
                    addConvertedAttribute( HTML.Attribute.HREF,
                    attrs );
                }
                if ((attrs.getAttribute(HTML.Attribute.TARGET) == null) && cb.openInNewWindow) {
                    attrs.addAttribute(HTML.Attribute.TARGET, "_BLANK");
                }
            } else if (((tag == HTML.Tag.IMG) || (tag == HTML.Tag.INPUT) || (tag == HTML.Tag.SCRIPT))
                         && (attrs.getAttribute(HTML.Attribute.SRC) != null)) {
                // ---- CHECKING <IMG SRC & <INPUT SRC
                addConvertedAttribute( HTML.Attribute.SRC,
                                       attrs );
            } else if (tag == HTML.Tag.LINK) {
                if (attrs.getAttribute(HTML.Attribute.HREF) != null) {
                    // ---- CHECKING <LINK HREF
                    addConvertedAttribute( HTML.Attribute.HREF,
                    attrs );
                }
            } else if ( tag == HTML.Tag.APPLET ) {
                // ---- CHECKING <APPLET CODEBASE=
                if (attrs.getAttribute(HTML.Attribute.CODEBASE) == null) {
                    int endOfPath = baseUrl.toString().lastIndexOf("/");
                    attrs.addAttribute(HTML.Attribute.CODEBASE, 
                                       baseUrl.toString().substring(0,endOfPath +1));
                } else {
                    addConvertedAttribute( HTML.Attribute.CODEBASE, attrs );
                }
            } else if (tag == HTML.Tag.OBJECT) {
                // ---- CHECKING <OBJECT CODEBASE=
                if (attrs.getAttribute(HTML.Attribute.CODEBASE) == null) {
                    int endOfPath = baseUrl.toString().lastIndexOf("/");
                    attrs.addAttribute(HTML.Attribute.CODEBASE, 
                                       baseUrl.toString().substring(0,endOfPath +1));
                } else {
                    addConvertedAttribute( HTML.Attribute.CODEBASE, attrs );
                }
            } else if (tag == HTML.Tag.BODY) {
                if (attrs.getAttribute(HTML.Attribute.BACKGROUND) != null) {
                    // background images are applied to the ENTIRE page, this remove them!
                    attrs.removeAttribute( HTML.Attribute.BACKGROUND);
                }
            } else if (tag == HTML.Tag.BASE) {
                if (attrs.getAttribute(HTML.Attribute.HREF) != null) {
                    try {
                        baseUrl = new URL(attrs.getAttribute(HTML.Attribute.HREF).toString());
                    } catch (Throwable t) {
                        logger.error( "HTMLRewriter: Setting BASE=" 
                        + attrs.getAttribute(HTML.Attribute.HREF).toString()
                        + t.getMessage());
                    }
                    attrs.removeAttribute(HTML.Attribute.HREF);
                }
            } else if (tag == HTML.Tag.FORM) {
                // ---- CHECKING <FORM ACTION=
                  inForm = true; // buggy <form> handling in jdk 1.3 
                  if (attrs.getAttribute(HTML.Attribute.ACTION) == null) {
                      //self referencing <FORM>
                       attrs.addAttribute(HTML.Attribute.ACTION,
                                          baseUrl.toString());
                  } else {
                        addConvertedAttribute( HTML.Attribute.ACTION,
                                               attrs );
                  }
            } else if (tag == HTML.Tag.TD) {
                // ---- CHECKING <TD BACKGROUND=
                  if (! (attrs.getAttribute(HTML.Attribute.BACKGROUND) == null)) {
                      addConvertedAttribute( HTML.Attribute.BACKGROUND,
                                             attrs );
                  }
            }

            
            // then we check for ignored tags ...
            // btw. I know, that this code could be written in a shorter way, but
            // I think it's more readable like this ...

            // don't forget to add changes to  handleEndTag() as well, else 
            // things will get screwed up!
            
            if ( (removeScript == false) && (tag == HTML.Tag.SCRIPT)) {
                inScript = true;
            } else if ( (removeStyle == false) && (tag == HTML.Tag.STYLE)) {
                inStyle = true;
            }

            if ( removeScript && (tag == HTML.Tag.SCRIPT)) {
                  ignoreLevel ++;
            } else if ( removeStyle && (tag == HTML.Tag.STYLE)) {
                  ignoreLevel ++;
            } else if ( removeHead && (tag == HTML.Tag.HEAD)) {
                  ignoreLevel ++;
            } else if ( removeApplet && (tag == HTML.Tag.APPLET)) {
                  ignoreLevel ++;
            } else if ( removeObject && (tag == HTML.Tag.OBJECT)) {
                  ignoreLevel ++;
            } else if (removeNoScript && (tag.toString().equalsIgnoreCase("NOSCRIPT"))) {
                  ignoreLevel ++;
            }
        }

        /**
         *
         * Converts the given attribute to base URL, if not null
         *
         */
        private void addConvertedAttribute( HTML.Attribute attr,
                                            MutableAttributeSet attrs ) {
            if( attrs.getAttribute( attr ) != null ) {
                String attrSource =  attrs.getAttribute( attr ).toString();
                attrs.addAttribute( attr,
                                    generateNewUrl( attrSource ) );
            }
        }
              
              
        private String generateNewUrl(String oldURL) {
            try {
                URL x = new URL(baseUrl,oldURL);
                return x.toString();
            } catch (Throwable t) {
                if (oldURL.toLowerCase().startsWith("javascript:")) {
                    return oldURL;
                }
                logger.error( "HTMLRewriter: Setting BASE="
                + baseUrl
                + " Old = "
                + oldURL
                + t.getMessage());
                return oldURL; // default behaviour ...
            }
        }

        public void handleText(char[] values,int param) {
            addToResult(values);
        }
    }
}