Changeset 31296 for other-projects/hathitrust/wcsa/vol-checker

Timestamp:

2017-01-18T17:54:56+13:00 (7 years ago)

Author:

davidb

Message:

Make loading in of ID file more portable

File:

• other-projects/hathitrust/wcsa/vol-checker/src/org/hathitrust/extractedfeatures/VolumeCheck.java (modified) (3 diffs)

other-projects/hathitrust/wcsa/vol-checker/src/org/hathitrust/extractedfeatures/VolumeCheck.java

@@ -1 +1 @@
 package org.hathitrust.extractedfeatures;
 
+import java.io.BufferedInputStream;
 import java.io.BufferedReader;
 import java.io.FileReader;
 import java.io.IOException;
+import java.io.InputStream;
+import java.io.InputStreamReader;
 import java.io.PrintWriter;
+import java.io.UnsupportedEncodingException;
 import java.util.ArrayList;
 import java.util.HashMap;
@@ -28 +32 @@
     
     public VolumeCheck() {
-        if (id_check_ == null) {
-            id_check_ = new HashMap<String,Boolean>(HASHMAP_INIT_SIZE);
-            storeIDs(DATA_DIR + "/full-listing.txt");
-        }
+        
     }
 
-    protected void storeIDs(String filename)
+    protected void storeIDs(BufferedReader br)
     {       
+        long line_num = 1;
+        String line;
+        
+        String opt_file_ext = ".json.bz2";
+
         try {
-            FileReader fr = new FileReader(filename);
-            BufferedReader br = new BufferedReader(fr);
-            
-            long line_num = 1;
-            String line;
-            System.err.println("Loading in volume IDS: " + filename);
-            
-            String opt_file_ext = ".json.bz2";
-            
-            while ((line = br.readLine()) != null) {
-                
-                String file_id = line.substring(line.lastIndexOf("/")+1);
-                String id = null;
-                if (file_id.endsWith(opt_file_ext)) {
-                    id = file_id.substring(0,file_id.lastIndexOf(opt_file_ext));
-                }
-                else {
-                    id = file_id;
-                }
-                
-                id = id.replaceAll("\\+", ":").replaceAll("=", "/");
-                         
-                id_check_.put(id, true);
-                
-                if ((line_num % 100000) == 0) {
-                    System.err.println("sample id = " + id);
-                    System.err.println("Passed line: " + line_num);
-                }
-                line_num++;
-                
-            }
-            
-            br.close();
+
+            while ((line = br.readLine()) != null) {
+
+                String file_id = line.substring(line.lastIndexOf("/")+1);
+                String id = null;
+                if (file_id.endsWith(opt_file_ext)) {
+                    id = file_id.substring(0,file_id.lastIndexOf(opt_file_ext));
+                }
+                else {
+                    id = file_id;
+                }
+
+                id = id.replaceAll("\\+", ":").replaceAll("=", "/");
+
+                id_check_.put(id, true);
+
+                if ((line_num % 100000) == 0) {
+                    System.err.println("sample id = " + id);
+                    System.err.println("Passed line: " + line_num);
+                }
+                line_num++;
+
+            }
         }
         catch (Exception e) {
@@ -80 +76 @@
      */
     public void init(ServletConfig config) throws ServletException {
-        // currently everything we need to happen at start up can be done in the constructor
+        super.init(config);
+        
+        if (id_check_ == null) {
+            id_check_ = new HashMap<String,Boolean>(HASHMAP_INIT_SIZE);
+            
+            String htrc_list_file = "htrc-ef-all-files.txt";
+            InputStream is = getServletContext().getResourceAsStream("/WEB-INF/" + htrc_list_file);
+            
+            try {
+                System.err.println("INFO: Loading in volume IDS: " + htrc_list_file);
+                
+                InputStreamReader isr = new InputStreamReader(is, "UTF-8");
+                BufferedReader br = new BufferedReader(isr);
+
+                storeIDs(br);
+                br.close();
+            }
+            catch (Exception e) {
+                e.printStackTrace();
+            }
+        }
+        
+        
+         
+
+         
     }