Changeset 31911

Timestamp:

2017-08-21T15:38:19+12:00 (6 years ago)

Author:

ak19

Message:

First part of commit after using earliestDatestamp from oai-inf.db rather than build conf: 1. skip output of OAI records for internal 'OID' stored in oai-inf.db, whose entry represents the collection's earliest timestamp info. 2. Now prefer doc's lastmod date from oai-inf.db to doc's oailastmod date in indedb. So for all records, not just deleted records that only have entries in oaiinfdb and not indexdb.

Location:

main/trunk/greenstone3/src/java/org/greenstone/gsdl3

Files:

• service/OAIPMH.java (modified) (7 diffs)
• util/OAIXML.java (modified) (1 diff)

main/trunk/greenstone3/src/java/org/greenstone/gsdl3/service/OAIPMH.java

@@ -368 +368 @@
         logger.warn("OID: " + oid + " is not present in the collection's oai-inf database.");
     } else  {
+
+        // indexdb doesn't have info on deleted docs, only oaiinf db does.
+        // So only oaiinfdb has timestamps for deleted docs
+        // For non-deleted doc ids: also obtain timestamp from oaiinf db, 
+        // but if the oaiinf db doesn't exist, resort to oailastmodified fields of indexdb.
+        String timestamp = oai_info.getInfo(OAIXML.OAI_INF_TIMESTAMP); // stored in seconds, like oailastmodified in the collection index db
+        millis = Long.parseLong(timestamp)*1000; // in milliseconds     
+        
         String oaiinf_status = oai_info.getInfo(OAIXML.OAI_INF_STATUS);
         if(oaiinf_status != null && oaiinf_status.equals(OAIXML.OAI_INF_DELETED)) {
         OID_is_deleted = true;
-
-        // get the right timestamp for deletion: from oaiinf db
-        String timestamp = oai_info.getInfo(OAIXML.OAI_INF_TIMESTAMP); // in seconds presumably, like oailastmodified in the collection index db        
-        
-        millis = Long.parseLong(timestamp)*1000; // in milliseconds
         }
     }
@@ -384 +387 @@
     DBInfo info = this.coll_db.getInfo(oid);
     if (info == null) {
-      logger.error("OID: " + oid + " is not present in the collection database.");
-      //return OAIXML.createErrorResponse(OAIXML.ID_DOES_NOT_EXIST, ""); // not an error: may exist as deleted (marked 'D') in oai-inf db
-    }
-    else if (millis == -1) { // so !OID_is_deleted, get oailastmodified from collection's index db
-    ArrayList<String> keys = new ArrayList<String>(info.getKeys()); 
+    if(!OID_is_deleted) { // we don't expect to find entries for deleted docs in index db.
+        logger.error("OID: " + oid + " is not present in the collection index database.");
+        return OAIXML.createErrorResponse(OAIXML.ID_DOES_NOT_EXIST, "");
+    } // if doc deleted, id missing in indexdb is not an error: doc id would exist only in oai-inf db, marked as deleted 'D'
+    }
+    else if (millis == -1) { // so couldn't get doc lastmod from oaiinf db, get oailastmodified from collection's index db
     millis = getDateStampMillis(info);  
     }
@@ -494 +498 @@
     for(int i=0; i<oid_list.size(); i++) {
       String oid = oid_list.get(i);
+
+      if(oid.equals(OAIXML.OAI_EARLIEST_TIMESTAMP_OID)) { // internal id not doc id, so skip
+      continue;
+      }
+      
       boolean OID_is_deleted = false;
       long millis = -1;
@@ -503 +512 @@
           logger.warn("OID: " + oid + " is not present in the collection's oai-inf database.");
       } else  {
+          
+          // indexdb doesn't have info on deleted docs, only oaiinf db does.
+          // So only oaiinfdb has timestamps for deleted docs
+          // For non-deleted doc ids: also obtain timestamp from oaiinf db, 
+          // but if the oaiinf db doesn't exist, resort to oailastmodified fields of indexdb.
+          String timestamp = oai_info.getInfo(OAIXML.OAI_INF_TIMESTAMP); // stored in seconds like oailastmodified in the collection index db         
+          millis = Long.parseLong(timestamp)*1000; // in milliseconds
+          
           String oaiinf_status = oai_info.getInfo(OAIXML.OAI_INF_STATUS);
           if(oaiinf_status != null && oaiinf_status.equals(OAIXML.OAI_INF_DELETED)) {
           OID_is_deleted = true;
-          
-          // get the right timestamp for deletion: from oaiinf db
-          String timestamp = oai_info.getInfo(OAIXML.OAI_INF_TIMESTAMP); // in seconds presumably, like oailastmodified in the collection index db      
-          
-          millis = Long.parseLong(timestamp)*1000; // in milliseconds
           }
       }
@@ -516 +528 @@
       DBInfo info = this.coll_db.getInfo(oid);
       if (info == null) { // can happen if oid was deleted, in which case only oai_info keeps a record of the oid
-        logger.error("Collection database does not contain information about oid: " +oid);
-      }
-      else if (millis == -1) { // so !OID_is_deleted, get oailastmodified from collection's index db
+      if(!OID_is_deleted) { // we don't expect to find entries for deleted docs in index db.
+          logger.error("Collection database does not contain information about oid: " +oid);
+      }
+      }
+      else if (millis == -1) { // so couldn't get doc lastmod from oaiinf db, get oailastmodified from collection's index db
       
       millis = getDateStampMillis(info);
@@ -524 +538 @@
 
       Date this_date = null;
+      String oailastmodified = (millis == -1) ? "" : OAIXML.getTime(millis);
+      
       if (millis == -1) {
       if (from_date != null || until_date !=null) {
@@ -544 +560 @@
       
       //compose the header element, which we'll be appending no matter what
-      Element header = createHeaderElement(doc, oid, OAIXML.getTime(millis), OID_is_deleted);
+      Element header = createHeaderElement(doc, oid, oailastmodified, OID_is_deleted);
 
       if (include_metadata) { // doing ListRecords

main/trunk/greenstone3/src/java/org/greenstone/gsdl3/util/OAIXML.java

@@ -78 +78 @@
   public static final String RESPONSE_DATE = "responseDate";
   public static final String REQUEST = "request";
-  
+
+  // internal "OID" identifiers in the oai-inf db
+  /** represents the timestamp of the OAI collection (when its oai-inf db was first created) */
+  public static final String OAI_EARLIEST_TIMESTAMP_OID = "_earliesttimestamp";  
+    
   // Identify data
   public static final String ADMIN_EMAIL = "adminEmail";