Changeset 18353 for gli/branches/rtl-gli/src/org/greenstone/gatherer/util/XMLTools.java

Timestamp:

2009-01-12T11:17:33+13:00 (15 years ago)

Author:

kjdon

Message:

updated the rtl-gli branch with files from trunk. Result of a merge 14807:18318

File:

• gli/branches/rtl-gli/src/org/greenstone/gatherer/util/XMLTools.java (modified) (12 diffs)

gli/branches/rtl-gli/src/org/greenstone/gatherer/util/XMLTools.java

@@ -261 +261 @@
     static final public String NOTWELLFORMED= "not well-formed";
     static final private String HEADER = "<?xml version='1.0' encoding='UTF-8'?><collectionConfig xmlns:gsf='http://www.greenstone.org/greenstone3/schema/ConfigFormat' xmlns:xsl='http://www.w3.org/1999/XSL/Transform'>";
-    static final private String FOOTER = "</collectionConfig>";
+    static final private String FOOTER = "</collectionConfig>";
        
 public static String parse (String xml_str) {
@@ -332 +332 @@
             SAXParser parser = factory.newSAXParser ();
             FileReader r = new FileReader(xml_file);
-            InputSource iSource = new InputSource(r);
+            InputSource iSource = new InputSource(r);
             XMLReader reader = parser.getXMLReader ();
             reader.setContentHandler(new DefaultHandler());
@@ -372 +372 @@
     } // getLocationString(SAXParseException):String
 
+
     /** Parse an XML document from a given file path */
     static public Document parseXMLFile (String xml_file_path, boolean use_class_loader) {
@@ -409 +410 @@
         try {
             InputStreamReader isr = new InputStreamReader (xml_input_stream, "UTF-8");
-            Reader xml_reader = new BufferedReader (isr);
-            document = parseXML (xml_reader);
+            document = parseXML(isr);
             isr.close ();
             xml_input_stream.close ();
@@ -425 +425 @@
     static public Document parseXML (Reader xml_reader) {
         Document document = null;
-        
-        try {
-            InputSource isc       = new InputSource (xml_reader);
-            DOMParser parser      = new DOMParser ();
-            parser.setFeature ("http://xml.org/sax/features/validation", false);
-            parser.setFeature ("http://apache.org/xml/features/nonvalidating/load-external-dtd", false);
-            // May or may not be ignored, the documentation for Xerces is contradictory. If it works then parsing -should- be faster.
-            parser.setFeature ("http://apache.org/xml/features/dom/defer-node-expansion", true);
-            parser.setFeature ("http://apache.org/xml/features/dom/include-ignorable-whitespace", false);
-            parser.parse (isc);
-            document = parser.getDocument ();
-        }
-        catch (SAXException exception) {
-            System.err.println ("SAX exception: " + exception.getMessage ());
-            DebugStream.printStackTrace (exception);
+
+    // If debugging, the following will store the XML contents to be parsed, 
+    // which can then be inspected upon encountering a SAXException (need to run GLI with -debug on)
+    String xmlContents = "";
+
+    try {
+        Reader reader = null;
+
+        // (1) By default, GLI will remove any contents preceeding (and invalidating) 
+        // the XML and present these lines separately to the user
+        if(!DebugStream.isDebuggingEnabled()) {
+            try {
+                reader = new BufferedReader( new RemoveContentBeforeRootElementXMLReader(xml_reader) );
+            } catch ( Exception e ) {
+                System.err.println( "Exception while wrapping the reader in parseXML(Reader)" );
+                e.printStackTrace();
+            }
+        } 
+
+        // (2) If we are running GLI in debug mode:
+        // In case parsing exceptions are thrown (SAX Exceptions), we want to get some
+        // idea of where things went wrong. This will print the "XML" contents to either
+        // system.out (if debugging is off) or to the DebugStream otherwise.
+        // We need to read the XML twice to know the line where things went wrong, so
+        // do the additional reading only if we're debugging 
+        else {
+            StringBuffer buf = new StringBuffer();
+            char[] buffer = new char[500];
+            int numCharsRead = xml_reader.read(buffer, 0, buffer.length); 
+            while(numCharsRead != -1) {
+                buf.append(buffer, 0, numCharsRead);
+                numCharsRead = xml_reader.read(buffer, 0, buffer.length); 
+            }
+            xmlContents = buf.toString();
+            xml_reader.close(); // closing the old Reader
+            xml_reader = null;
+            buffer = null;
+            buf = null;
+            // we need a Reader to parse the same contents as the Reader that was just closed
+            reader = new BufferedReader(new StringReader(xmlContents));
+        }
+        
+        // (2) The actual XML parsing
+        InputSource isc       = new InputSource (reader);
+        DOMParser parser      = new DOMParser ();
+        parser.setFeature ("http://xml.org/sax/features/validation", false);
+        parser.setFeature ("http://apache.org/xml/features/nonvalidating/load-external-dtd", false);
+        // May or may not be ignored, the documentation for Xerces is contradictory. If it works then parsing -should- be faster.
+        parser.setFeature ("http://apache.org/xml/features/dom/defer-node-expansion", true);
+        parser.setFeature ("http://apache.org/xml/features/dom/include-ignorable-whitespace", false);
+        parser.parse (isc);
+        document = parser.getDocument ();
+        
+    } catch(SAXParseException e) {
+        showXMLParseFailureLine(e, xmlContents);
+    } catch (SAXException exception) {
+        System.err.println ("SAX exception: " + exception.getMessage ());
+        if(DebugStream.isDebuggingEnabled()) {
+        DebugStream.println("Encountered a SAX exception when parsing the following:\n*********START\n" 
+                    + xmlContents + "\n************END\n");
+        // Exit to let the user view the erroneous line/xml before it goes past the screen buffer?
+        DebugStream.println("Debug mode: Exiting the program as there was trouble parsing the XML..."); 
+        System.exit(-1);
+        } 
+        // else, not running in debug mode, so don't exit after exception  
+        System.out.println("***Turn debugging on (run GLI with -debug) to view the XML contents that could not be parsed.");
+        DebugStream.printStackTrace (exception);
         }
         catch (Exception exception) {
@@ -447 +499 @@
         return document;
     }
-    
+
+    /** Displays the line (string) where the SAXParseException occurred, given a String of the 
+     * entire xml that was being parsed and the SAXParseException object that was caught. 
+     * The messages are printed to DebugStream, so run GLI/FLI with -debug to view this output. 
+     * @param xmlContents is the entire xml that was being parsed when the exception occurred
+     * @param e is the SAXParseException object that was thrown upon parsing the xmlContents.
+     */
+    public static void showXMLParseFailureLine(SAXParseException e, String xmlContents) {
+    
+        // There should be no characters at all that preceed the <?xml>... bit. 
+        // The first check is for starting spaces:
+        if(xmlContents.startsWith("\n") || xmlContents.startsWith(" ") || xmlContents.startsWith("\t")) {
+            DebugStream.println("ERROR: illegal start of XML. Space/tab/newline should not preceed xml declaration.\n");
+            DebugStream.println("xmlContents (length is " + xmlContents.length() + "):\n" + xmlContents);
+            return; // nothing more to do, first error identified
+        }
+    
+        // the actual line (String literal) where parsing failed and the SAXParseException occurred.
+        String line = "";
+        int linenumber = e.getLineNumber();
+        DebugStream.println("\n****SAXParseException on LINE NUMBER: " + linenumber);
+        if(DebugStream.isDebuggingEnabled()) {
+            if(linenumber != -1) { 
+                // find the line in xmlContents string (xmlContents is only set if GLI is run with debugging turned on)
+                int start = 0;
+                int end = xmlContents.length();
+                for(int i = 1; i <= linenumber; i++) {
+                    end = xmlContents.indexOf("\n");
+                    if(end > 0) {
+                        line = xmlContents.substring(start, end);
+                    }
+                    start = end+1;
+                }
+                DebugStream.println("The parsing error occurred on this line:\n***********START\n" + line + "\n***********END");
+                DebugStream.println("SAXParseException message: " + e.getMessage() + "\n");
+            } else { // no particular line number, print out all the xml so debugger can inspect it
+                DebugStream.println("Encountered a SAX exception when parsing the following:\n*********START\n" 
+                    + xmlContents + "\n************END\n");
+            }
+            // Exit to let the user view the erroneous line/xml before it goes past the screen buffer?
+            DebugStream.println("\nDebug mode: Exiting the program as there was trouble parsing the XML..."); 
+            System.exit(-1);
+        } else { // not running in debug mode
+            System.out.println("***Turn debugging on (run GLI with -debug) to view the XML contents/line that could not be parsed.");
+        }
+    }
+
     
     static public StringBuffer readXMLStream (InputStream input_stream) {
@@ -540 +638 @@
             f.setLineWidth (0); // Why isn't this working!
             f.setPreserveSpace (false);
-            
-            f.setNonEscapingElements (nonEscapingTagNames);
+            if (nonEscapingTagNames != null) {
+          f.setNonEscapingElements (nonEscapingTagNames);
+        }
             // Create the necessary writer stream for serialization.
             OutputStreamWriter osw = new OutputStreamWriter (os, "UTF-8");
@@ -560 +659 @@
     /** Write an XML document to a given file */
     static public void writeXMLFile (File xml_file, Document document) {
-        try {
-            OutputStream os = new FileOutputStream (xml_file);
-            // Create an output format for our document.
-            OutputFormat f = new OutputFormat (document);
-            f.setEncoding ("UTF-8");
-            f.setIndenting (true);
-            f.setLineWidth (0); // Why isn't this working!
-            f.setPreserveSpace (false);
-            // Create the necessary writer stream for serialization.
-            OutputStreamWriter osw = new OutputStreamWriter (os, "UTF-8");
-            Writer w               = new BufferedWriter (osw);
-            // Generate a new serializer from the above.
-            XMLSerializer s        = new XMLSerializer (w, f);
-            s.asDOMSerializer ();
-            // Finally serialize the document to file.
-            s.serialize (document);
-            // And close.
-            os.close ();
-        }
-        catch (Exception exception) {
-            DebugStream.printStackTrace (exception);
-        }
-    }
-    
+      writeXMLFile(xml_file, document, null);
+    }
+
     public static void printXMLNode (Node e) {
         printXMLNode (e, 0) ;
@@ -686 +764 @@
         
     }
+
+  public static String xmlNodeToStringWithoutIndenting (Node e) {
+    StringBuffer sb = new StringBuffer ("");
+    xmlNodeToStringWithoutNewline(sb, e, -1);
+    return sb.toString();
+  }
     public static String xmlNodeToStringWithoutNewline (Node e){
         StringBuffer sb = new StringBuffer ("");
@@ -696 +780 @@
         for (int i=0 ; i<depth ; i++)
             sb.append (' ') ;
-        
+      
         if (e.getNodeType () == Node.TEXT_NODE){
             if (e.getNodeValue () != "") {
@@ -727 +811 @@
             int len = children.getLength ();
             for (int i = 0; i < len; i++) {
+          if (depth >= 0) {
                 xmlNodeToStringWithoutNewline (sb,children.item (i), depth + 1);
+          } else {
+        xmlNodeToStringWithoutNewline (sb,children.item (i), depth);
+          }
             }
             
@@ -735 +823 @@
             sb.append ("</" + e.getNodeName () + ">");
         }
-        
-        
-    }
-    
+    }
 }