Changeset 12999 for trunk/indexers/lucene-gs

Timestamp:

2006-10-04T17:16:37+13:00 (18 years ago)

Author:

mdewsnip

Message:

Code for caching query results (eg. in index/sidx/cache), disabled by default.

File:

• trunk/indexers/lucene-gs/src/org/greenstone/LuceneWrapper/GS2LuceneQuery.java (modified) (8 diffs)

trunk/indexers/lucene-gs/src/org/greenstone/LuceneWrapper/GS2LuceneQuery.java

@@ -40 +40 @@
     static private String TEXTFIELD = "TX";
 
+    // Fairly self-explanatory I should hope
+    static private boolean query_result_caching_enabled = false;
+
     // Use the standard set of English stop words by default
     static private String[] stop_words = StandardAnalyzer.STOP_WORDS;
@@ -45 +48 @@
     // Command-line options
     static private String fuzziness = null;
+    static private String filter_string = null;
     static private Filter filter = null;
+    static private String sort_string = null;
     static private Sort sorter = new Sort();
     static private String default_conjuction_operator = "OR";
@@ -60 +65 @@
 
         try {
-        Searcher searcher = new IndexSearcher(args[0]);
+        String index_directory = args[0];
+        Searcher searcher = new IndexSearcher(index_directory);
         IndexReader reader = ((IndexSearcher) searcher).getIndexReader();
+
+        // Prepare the index cache directory, if query result caching is enabled
+        if (query_result_caching_enabled) {
+        // Make the index cache directory if it doesn't already exist
+        File index_cache_directory = new File(index_directory, "cache");
+        if (!index_cache_directory.exists()) {
+            index_cache_directory.mkdir();
+        }
+
+        // Disable caching if the index cache directory isn't available
+        if (!index_cache_directory.exists() || !index_cache_directory.isDirectory()) {
+            query_result_caching_enabled = false;
+        }
+        }
 
         // Create one query parser with the standard set of stop words, and one with none
@@ -73 +93 @@
         if (args[i].equals("-sort")) {
             i++;
-            sorter = new Sort(args[i]);
+            sort_string = args[i];
+            sorter = new Sort(sort_string);
         }
         else if (args[i].equals("-filter")) {
             i++;
-            filter = parseFilterString(args[i]);
+            filter_string = args[i];
+            filter = parseFilterString(filter_string);
         }
         else if (args[i].equals("-dco")) {
@@ -112 +134 @@
         // The query string has been specified as a command-line argument
         if (query_string != null) {
-        runQuery(searcher, reader, query_parser, query_parser_no_stop_words, query_string);
+        runQuery(index_directory, searcher, reader, query_parser, query_parser_no_stop_words, query_string);
         }
 
@@ -125 +147 @@
             }
 
-            runQuery(searcher, reader, query_parser, query_parser_no_stop_words, query_string);
+            runQuery(index_directory, searcher, reader, query_parser, query_parser_no_stop_words, query_string);
         }
         }
@@ -137 +159 @@
 
 
-    private static void runQuery(Searcher searcher, IndexReader reader, QueryParser query_parser, QueryParser query_parser_no_stop_words, String query_string)
+    private static void runQuery(String index_directory, Searcher searcher, IndexReader reader, QueryParser query_parser, QueryParser query_parser_no_stop_words, String query_string)
     throws IOException
     {
     StringBuffer query_results_xml = new StringBuffer();
 
-    query_results_xml.append("<ResultSet>\n");
+    // Check if this query result has been cached from a previous search (if it's enabled)
+    File query_result_cache_file = null;
+    if (query_result_caching_enabled) {
+        // Generate the cache file name from the query options
+        String query_result_cache_file_name = query_string + "-";
+        query_result_cache_file_name += ((fuzziness != null) ? fuzziness : "") + "-";
+        query_result_cache_file_name += ((filter_string != null) ? filter_string : "") + "-";
+        query_result_cache_file_name += ((sort_string != null) ? sort_string : "") + "-";
+        query_result_cache_file_name += default_conjuction_operator + "-";
+        query_result_cache_file_name += start_results + "-" + end_results;
+        query_result_cache_file_name = fileSafe(query_result_cache_file_name);
+
+        // If the query result cache file exists, just return its contents and we're done
+        File index_cache_directory = new File(index_directory, "cache");
+        query_result_cache_file = new File(index_cache_directory, query_result_cache_file_name);
+        if (query_result_cache_file.exists() && query_result_cache_file.isFile()) {
+        FileInputStream fis = new FileInputStream(query_result_cache_file);
+        InputStreamReader isr = new InputStreamReader(fis, "UTF-8");
+        BufferedReader buffered_reader = new BufferedReader(isr);
+        String line = "";
+        while ((line = buffered_reader.readLine()) != null) {
+            query_results_xml.append(line + "\n");
+        }
+        String query_results_xml_string = query_results_xml.toString();
+        query_results_xml_string = query_results_xml_string.replaceFirst("cached=\"false\"", "cached=\"true\"");
+        System.out.print(query_results_xml_string);
+        return;
+        }
+    }
+
+    query_results_xml.append("<ResultSet cached=\"false\">\n");
     query_results_xml.append("  <QueryString>" + xmlSafe(query_string) + "</QueryString>\n");
     if (filter != null) {
@@ -227 +279 @@
 
     System.out.print(query_results_xml);
+
+    // Cache this query result, if desired
+    if (query_result_caching_enabled) {
+        FileWriter query_result_cache_file_writer = new FileWriter(query_result_cache_file);
+        query_result_cache_file_writer.write(query_results_xml.toString());
+        query_result_cache_file_writer.close();
+    }
+    }
+
+
+    private static String fileSafe(String text)
+    {
+    StringBuffer file_safe_text = new StringBuffer();
+    for (int i = 0; i < text.length(); i++) {
+        char character = text.charAt(i);
+        if ((character >= 'A' && character <= 'Z') || (character >= 'a' && character <= 'z') || (character >= '0' && character <= '9') || character == '-') {
+        file_safe_text.append(character);
+        }
+        else {
+        file_safe_text.append('%');
+        file_safe_text.append((int) character);
+        }
+    }
+    return file_safe_text.toString();
     }