Changeset 31505 for other-projects/hathitrust/wcsa/extracted-features-solr/trunk/solr-ingest/src/main/java/org/hathitrust/extractedfeatures/SolrDocJSON.java

Timestamp:

2017-03-13T15:31:40+13:00 (7 years ago)

Author:

davidb

Message:

Added in storing of top-level document metadata as separate solr-doc

File:

• other-projects/hathitrust/wcsa/extracted-features-solr/trunk/solr-ingest/src/main/java/org/hathitrust/extractedfeatures/SolrDocJSON.java (modified) (3 diffs)

other-projects/hathitrust/wcsa/extracted-features-solr/trunk/solr-ingest/src/main/java/org/hathitrust/extractedfeatures/SolrDocJSON.java

@@ -28 +28 @@
 public class SolrDocJSON {
 
+    protected static JSONObject generateToplevelMetadataSolrDocJSON(String volume_id, JSONObject ef_metadata)
+    {
+        JSONObject solr_update_json = null;
+        
+        
+        String [] metadata_single = new String[] {
+                "accessProfile",
+                "rightsAttributes",
+                "hathitrustRecordNumber",
+                "title",
+                "imprint",
+                "pubDate",
+                "pubPlace",
+                "language",
+                "issuance",
+                "typeOfResource"
+        };
+
+        String [] metadata_multiple = new String[] {
+                "oclc",
+                "isbn",
+                "issn",
+                "lccn",
+                "genre",        
+                "names"
+        };
+
+        if (ef_metadata != null) {
+            
+            // For JSON Solr format see:
+            //   https://cwiki.apache.org/confluence/display/solr/Uploading+Data+with+Index+Handlers
+                
+            //String title= ef_metadata.getString("title");
+            JSONObject solr_add_json = new JSONObject();
+            
+            JSONObject solr_doc_json = new JSONObject();
+            solr_doc_json.put("id", volume_id);
+            
+            for (String metaname: metadata_single) {
+                String metavalue = ef_metadata.getString(metaname);
+                if (metavalue != null) {
+                    solr_doc_json.put(metaname+"_t",metavalue);
+                }
+            }
+            
+            for (String metaname: metadata_multiple) {
+                JSONArray metavalues = ef_metadata.getJSONArray(metaname);
+                if (metavalues != null) {
+                    solr_doc_json.put(metaname+"_t",metavalues);
+                }
+            }
+            
+            solr_add_json.put("commitWithin", 5000);
+            solr_add_json.put("doc", solr_doc_json);
+
+            solr_update_json = new JSONObject();
+            solr_update_json.put("add",solr_add_json);
+        
+        }
+        else {
+            System.err.println("Warning: null metadata for '" + volume_id + "'");
+        }
+
+        return solr_update_json;
+    }
+
+
     
     protected static ArrayList<String> getTokenPosCountWords(JSONObject ef_token_pos_count, String page_id,
@@ -376 +443 @@
         }
     }
+    
     protected static JSONObject generateSolrDocJSON(String volume_id, String page_id, JSONObject ef_page,
                                                     WhitelistBloomFilter whitelist_bloomfilter, 
@@ -618 +686 @@
             e.printStackTrace();
         }
-        
     }
 }