Changeset 5817

Timestamp:

2003-11-11T11:17:34+13:00 (20 years ago)

Author:

kjdon

Message:

changed the metadata parsing to look over multiple lines if starts with a quote but doesn't end with a quote - needed for legacy collections. also made it always write out a [l=en] qualifier for metadata even if its english.

File:

• trunk/gli/src/org/greenstone/gatherer/cdm/CollectionConfiguration.java (modified) (6 diffs)

trunk/gli/src/org/greenstone/gatherer/cdm/CollectionConfiguration.java

@@ -757 +757 @@
         String language_str = command_element.getAttribute(LANGUAGE_ATTRIBUTE);
         // If this is element is in english, and it is the first one found, we don't need to write the language argument.
-        if(!language_str.equals(ENGLISH_LANGUAGE_STR) || known_metadata == null || known_metadata.contains(name_str)) {
-            text.append(LBRACKET_CHARACTER);
-            text.append(LANGUAGE_ARGUMENT);
-            text.append(language_str);
-            text.append(RBRACKET_CHARACTER);
-            text.append(SPACE_CHARACTER);
-        }
+        // we should always write the language string
+        //if(!language_str.equals(ENGLISH_LANGUAGE_STR) || known_metadata == null || known_metadata.contains(name_str)) {
+        text.append(LBRACKET_CHARACTER);
+        text.append(LANGUAGE_ARGUMENT);
+        text.append(language_str);
+        text.append(RBRACKET_CHARACTER);
+        text.append(SPACE_CHARACTER);
+            //}
         if(known_metadata != null) {
             known_metadata.add(name_str);
@@ -797 +798 @@
     try {
         Element collect_cfg_element = document.getDocumentElement();
-        // Read in the file command at a time.
+        // Read in the file one command at a time.
         FileReader in_reader = new FileReader(collect_config_file);
         BufferedReader in = new BufferedReader(in_reader);
@@ -856 +857 @@
             }
             if(command_element == null && command_type.equals(COLLECTIONMETADATA_STR)) {
-            command_element = parseMetadata(command_str);
+            // collectionmeta may go over more than one line, so 
+            // pass in the reader
+            command_element = parseMetadata(command_str, in);
             }
             if(command_element == null && (command_type.equals(COLLECTIONMETADATA_BETA_STR) || command_type.equals(COLLECTIONMETADATA_PUBLIC_STR) || command_type.equals(COLLECTIONMETADATA_CREATOR_STR) || command_type.equals(COLLECTIONMETADATA_MAINTAINER_STR))) {
@@ -1135 +1138 @@
     }
 
-    private Element parseMetadata(String command_str) {
+    private Element parseMetadata(String command_str, BufferedReader in) {
     Element command_element = null;
     try {
@@ -1145 +1148 @@
         String name_str = tokenizer.nextToken();
         String value_str = tokenizer.nextToken();
-        String language_str = "en"; // By default
+        String language_str = "en"; // By default - why do we assume English???
         // Check if the value string is actually a language string
         if(value_str.startsWith(LBRACKET_CHARACTER) && value_str.endsWith(RBRACKET_CHARACTER)) {
@@ -1151 +1154 @@
             value_str = tokenizer.nextToken();
         }
-        // Remove any speech marks from value
-        if(value_str.startsWith(SPEECH_CHARACTER) && value_str.endsWith(SPEECH_CHARACTER)) {
-            value_str = value_str.substring(1, value_str.length() - 1);
-        }
+
+        // now we need to handle the case where the value is enclosed in quotes (single or double) and may extend across multiple lines
+        String start_string = value_str.substring(0,1);
+        if (start_string.equals("\"") || start_string.equals("\'")) {
+
+            if (value_str.endsWith(start_string) && value_str.length()!=1) {
+            // we remove the quotes from the ends
+            value_str = value_str.substring(1, value_str.length() - 1);
+            } else {
+            
+            Gatherer.println("cdm.CollectionConfiguration.parseMEtadata: we have found a starting quote but not an ending quote, so now read lines until we get to an end quote");
+            StringBuffer value_raw = new StringBuffer(value_str.substring(1));
+            int pos = value_raw.indexOf(start_string);
+            int old_pos = 0;
+            while (pos != -1 && value_raw.charAt(pos-1)=='\\') {
+                old_pos = pos+1;
+                pos = value_raw.indexOf(start_string, old_pos);
+            }
+            while(pos == -1) {
+                String next_line = in.readLine();
+                if(next_line != null) {
+                value_raw.append(next_line);
+                value_raw.append(StaticStrings.NEW_LINE_CHAR);
+                }
+                next_line = null;
+                pos = value_raw.indexOf(start_string, old_pos);
+                while (pos != -1 && value_raw.charAt(pos-1)=='\\') {
+                old_pos = pos+1;
+                pos = value_raw.indexOf(start_string, old_pos);
+                }
+            }
+            
+            value_str = value_raw.substring(0, value_raw.lastIndexOf(start_string));
+            value_raw = null;
+            
+            } // else
+        } // if starts with a quote
+                    
         if(value_str != null) {
             // Ready the value str (which can contain all sorts of funky characters) for writing as a DOM value