write MDO record to output dir

iromlab/cdworker.py

@@ -23,6 +23,7 @@
 from . import isobuster
 from . import dbpoweramp
 from . import verifyaudio
+from . import mdo
 
 
 def mediumLoaded(driveName):
@@ -107,6 +108,7 @@ def processDisc(carrierData):
     """Process one disc / job"""
 
     jobID = carrierData['jobID']
+    PPN = carrierData['PPN']
 
     logging.info(''.join(['### Job identifier: ', jobID]))
     logging.info(''.join(['PPN: ', carrierData['PPN']]))
@@ -292,6 +294,15 @@ def processDisc(carrierData):
             logging.info(''.join(['isolyzerSuccess: ', str(isolyzerSuccess)]))
             logging.info(''.join(['imageTruncated: ', str(imageTruncated)]))
 
+        # Fetch metadata from KBMDO and store as file
+        logging.info('*** Writing metadata from KB-MDO to file ***')
+        wroteMDORecord = mdo.writeMDORecord(PPN, dirOut)
+
+        if not wroteMDORecord:
+            success = False
+            reject = True
+            logging.error("Could not write metadata from KB-MDO")
+
         # Generate checksum file
         logging.info('*** Computing checksums ***')
         successChecksum = checksumDirectory(dirOut)

iromlab/mdo.py

@@ -0,0 +1,50 @@
+#! /usr/bin/env python
+"""
+Quick and dirty module for writing entire MDO record for a PPN to file
+using SRU query. Code based on kbapi.sru.py
+"""
+
+import os
+import io
+import requests
+import urllib
+
+def writeMDORecord(PPN, writeDirectory):
+    """Write MDO record for a PPN to file"""
+
+    fileOut = os.path.join(writeDirectory, "meta-kbmdo.xml")
+    sruSearchString = '"PPN=' + PPN + '"'
+
+    SRU_BASEURL = 'http://jsru.kb.nl/sru/sru'
+    SRU_BASEURL += '?version=1.2&maximumRecords=%i'
+    SRU_BASEURL += '&operation=searchRetrieve'
+    SRU_BASEURL += '&startRecord=%i'
+    SRU_BASEURL += '&recordSchema=%s'
+    SRU_BASEURL += '&x-collection=%s&query=%s'
+
+    maximumrecords = 1
+    startrecord = 1
+    recordschema = 'dcx'
+    collection = 'GGC'
+
+    query = urllib.parse.quote_plus(sruSearchString)
+
+    url = SRU_BASEURL % (maximumrecords, startrecord,
+                         recordschema, collection, query)
+    r = requests.get(url)
+
+    if r.status_code != 200:
+        # Status code: must be 200, otherwise an error occurred
+        wroteMDORecord = False
+    else:
+        record_data = r.content.decode("utf-8") 
+
+        try:
+            with io.open(fileOut, "w", encoding="utf-8") as fOut:
+                fOut.write(str(record_data))
+            fOut.close()
+            wroteMDORecord = True
+        except IOError:
+            wroteMDORecord = False
+
+    return wroteMDORecord