Merge branch 'release-v29.2'

dpla-attic · Jul 17, 2014 · d33a2b2 · d33a2b2
2 parents 1ac9da7 + 69caebe
commit d33a2b2
Show file tree

Hide file tree

Showing 7 changed files with 23 additions and 17 deletions.
diff --git a/lib/akamod/primo-to-dpla.py b/lib/akamod/primo-to-dpla.py
@@ -12,8 +12,8 @@
 
 GEOPROP = None
 RECORD = "PrimoNMBib/record/"
-LINKS = "LINKS/"
-URL = "http://thoth.library.utah.edu:1701/primo_library/libweb/action/dlDisplay.do?vid=MWDL&afterPDS=true&docId="
+LINKS = "sear:LINKS/"
+URL = "http://utah-primoprod.hosted.exlibrisgroup.com/primo_library/libweb/action/dlDisplay.do?vid=MWDL&afterPDS=true&docId="
 
 #FIXME not format specific, move to generic module
 CONTEXT = {
@@ -81,7 +81,7 @@ def multi_transform(d, key, props, return_format="str"):
     RECORD + "display/lds09"        : lambda d, p: {"temporal": getprop(d, p)},
     RECORD + "facets/rsrctype"      : lambda d, p: {"type": getprop(d, p)},
     RECORD + "search/lsr03"         : lambda d, p: {"stateLocatedIn": getprop(d, p)},
-    LINKS + "linktorsrc"             : lambda d, p: {"identifier": getprop(d, p)}
+    LINKS + "sear:linktorsrc"       : lambda d, p: {"identifier": getprop(d, p)}
 }
 
 AGGREGATION_TRANSFORMER = {
@@ -91,7 +91,7 @@ def multi_transform(d, key, props, return_format="str"):
     "ingestType"                 : lambda d, p: {"ingestType": getprop(d, p)},
     "ingestDate"                 : lambda d, p: {"ingestDate": getprop(d, p)},
     RECORD + "control/recordid"  : lambda d, p: {"isShownAt": URL + getprop(d, p)},
-    LINKS + "thumbnail"          : lambda d, p: {"object": getprop(d, p)},
+    LINKS + "sear:thumbnail"     : lambda d, p: {"object": getprop(d, p)},
     "provider"                   : lambda d, p: {"provider": getprop(d, p)}
 }
 

diff --git a/lib/fetchers/mwdl_fetcher.py b/lib/fetchers/mwdl_fetcher.py
@@ -29,9 +29,11 @@ def retry_fetches(self):
     def mwdl_extract_records(self, content):
         error = None
         if not self.total_records:
-            total_records_prop = "SEGMENTS/JAGROOT/RESULT/DOCSET/TOTALHITS"
+            total_records_prop = "sear:SEGMENTS/sear:JAGROOT/sear:RESULT" \
+                                 "/sear:DOCSET/TOTALHITS"
             self.total_records = getprop(content, total_records_prop)
-        records = getprop(content, "SEGMENTS/JAGROOT/RESULT/DOCSET/DOC")
+        records = getprop(content, "sear:SEGMENTS/sear:JAGROOT/sear:RESULT"
+                                   "/sear:DOCSET/sear:DOC")
 
         if records:
             records = iterify(records)
@@ -40,6 +42,7 @@ def mwdl_extract_records(self, content):
                                         "PrimoNMBib/record/control/recordid")
         else:
             records = []
+            # Elements in the error response are not namespaced
             error = getprop(content, "SEGMENTS/JAGROOT/RESULT/ERROR/MESSAGE")
             if not error:
                 error = "No records found in MWDL content: %s" % content

diff --git a/lib/mappers/digitalnc_mapper.py b/lib/mappers/digitalnc_mapper.py
@@ -98,16 +98,17 @@ def map_description(self):
         prop = self.root_key + "note"
 
         if exists(self.provider_data, prop):
+            description = []
             for s in iterify(getprop(self.provider_data, prop)):
                 try:
                     if s.get("type") == "content":
-                        self.update_source_resource({"description":
-                                                     s.get("#text")})
-                        break
+                        description.append(s.get("#text"))
                 except:
                     logger.error("Error getting note/type from record %s" %
                                  self.provider_data["_id"])
 
+            self.update_source_resource({"description": description})
+
     def map_title(self):
         prop = self.root_key + "titleInfo"
         _dict = {"title": []}

diff --git a/lib/mappers/primo_mapper.py b/lib/mappers/primo_mapper.py
@@ -9,8 +9,9 @@ def __init__(self, provider_data):
             self.root_key = "PrimoNMBib/record/"
         else:
             self.root_key = ""
-        self.links_key = "LINKS/"
-        self.is_shown_at_url = "http://thoth.library.utah.edu:1701/" + \
+        self.links_key = "sear:LINKS/"
+        self.is_shown_at_url = "http://utah-primoprod.hosted." + \
+                               "exlibrisgroup.com/" + \
                                "primo_library/libweb/action/" + \
                                "dlDisplay.do?vid=MWDL&afterPDS=true&docId="
 
@@ -66,7 +67,7 @@ def map_state_located_in(self):
 
     def map_identifier(self):
         self._map_source_resource_prop("identifier",
-                                       self.links_key + "linktorsrc")
+                                       self.links_key + "sear:linktorsrc")
 
     def map_spatial(self):
         prop = self.root_key + "display/lds08"
@@ -109,8 +110,8 @@ def map_is_shown_at(self):
                                                   record_id})
 
     def map_object(self):
-        prop = self.links_key + "thumbnail"
-        if exists(self.provier_data, prop):
+        prop = self.links_key + "sear:thumbnail"
+        if exists(self.provider_data, prop):
             self.mapped_data.update({"object":
                                      getprop(self.provider_data, prop)})
 

diff --git a/lib/mapv3.py b/lib/mapv3.py
@@ -71,7 +71,7 @@ def array_or_item(obj):
                 "oneOf": array_or_item({"$ref": "#/definitions/hasView"})
             },
             "id": {"type": "string"},
-            "ingestionSequence": {"type": "number"},
+            "ingestionSequence": {"type": ["number", "null"]},
             # do not check ingestDate as the 'date-time' json-schema
             # format, since it requires RFC 3339 conformance and not
             # ISO 8601 conformance
@@ -164,6 +164,7 @@ def array_or_item(obj):
                     "extent": {"$ref": "#/definitions/arrayOrString"},
                     "format": {"$ref": "#/definitions/arrayOrString"},
                     "identifier": {"$ref": "#/definitions/arrayOrString"},
+                    "isPartOf": {"$ref": "#/definitions/arrayOrString"},
                     "language": {
                         "type": "array",
                         "items": {

diff --git a/profiles/mwdl.pjs b/profiles/mwdl.pjs
@@ -1,7 +1,7 @@
 {
     "name": "mwdl",
     "type": "mwdl",
-    "endpoint_url": "http://utah-primoprod.hosted.exlibrisgroup.com/PrimoWebServices/xservice/search/brief?institution=MWDL&indx=1&loc=local,scope:%28mw%29&query=facet_tlevel,exact,online_resources&bulkSize=1000&query_exc=facet_rtype,exact,collections",
+    "endpoint_url": "http://utah-primoprod.hosted.exlibrisgroup.com/PrimoWebServices/xservice/search/brief?institution=MWDL&loc=local,scope:%28mw%29&query=facet_tlevel,exact,online_resources&query_exc=facet_rtype,exact,collections",
     "endpoint_url_params": {
         "indx": 1,
         "bulkSize": 500

diff --git a/setup.py b/setup.py
@@ -31,7 +31,7 @@
 from distutils.core import setup
 
 setup( name = 'ingestion',
-       version = '29.1',
+       version = '29.2',
        description='DPLA Ingestion Subsystem',
        author='Digital Public Library of America',
        author_email='[email protected]',