Changeset 31352 for other-projects

Timestamp:

2017-01-24T21:54:12+13:00 (7 years ago)

Author:

davidb

Message:

collection-to-workset now with id-check added to filter

File:

• other-projects/hathitrust/wcsa/vol-checker/src/org/hathitrust/extractedfeatures/VolumeCheck.java (modified) (2 diffs)

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

@@ -213 +213 @@
             BufferedReader reader = new BufferedReader(isr);
             
-            StringBuilder sb = new StringBuilder();
+            StringBuilder workset_friendly_sb = new StringBuilder();
+            StringBuilder workset_unfriendly_sb = new StringBuilder();
+            
             String line = null;
+            int ci = 0;
             while ((line = reader.readLine()) != null)
             {
-                sb.append(line + "\n");
+                if (ci==0) {
+                    workset_friendly_sb.append("#" + line + "\n");
+                }
+                else {
+                    int first_tab_pos=line.indexOf("\t");
+                    String id = (first_tab_pos>0) ? line.substring(0, first_tab_pos) : line;
+                    
+                    if (id_check_.containsKey(id)) {
+                        workset_friendly_sb.append(line + "\n");        
+                    }
+                    else {
+                        workset_unfriendly_sb.append("#" + line + "\n");
+                    }
+                }
+                
+                ci++;
             }
               
             response.setContentType("text/plain");
             PrintWriter pw = response.getWriter();
-            pw.append(sb.toString());
+            pw.append(workset_friendly_sb.toString());
+            pw.append("## The following volumes are not in the HTRC Extracted Feature dataset\n");
+            pw.append(workset_unfriendly_sb.toString());
         }
         catch (Exception e) {
@@ -242 +262 @@
         String cgi_convert_col = request.getParameter("convert-col");
         
-        System.err.println("**** cgi_convert_col = '" + cgi_convert_col + "'");
+        //System.err.println("**** cgi_convert_col = '" + cgi_convert_col + "'");
         
         if (cgi_ids != null) {