Added provenance information/<about> section

src/oaipmh/common.py

@@ -46,6 +46,46 @@ def getField(self, name):
 
     __getitem__ = getField
 
+class About(object):
+    def __init__(self, element, baseURL, identifier, datestamp, metadataNamespace, harvestDate, repositoryID=None, repositoryName=None):
+        self._element = element
+        # force identifier to be a string, it might be 
+        # an lxml.etree._ElementStringResult...
+        try:
+            self._identifier = str(identifier)
+        except UnicodeEncodeError:
+            self._identifier = unicode(identifier)
+        self._datestamp = datestamp
+        self._baseURL = baseURL
+        self._metadataNamespace = metadataNamespace
+        self._harvestDate = harvestDate
+        self._repositoryID = repositoryID
+        self._repositoryName = repositoryName
+
+    def element(self):
+        return self._element
+
+    def identifier(self):
+        return self._identifier
+
+    def datestamp(self):
+        return self._datestamp
+
+    def baseURL(self):
+        return self._baseURL
+
+    def metadataNamespace(self):
+        return self._metadataNamespace
+
+    def harvestDate(self):
+        return self._harvestDate
+
+    def repositoryID(self):
+        return self._repositoryID
+
+    def repositoryName(self):
+        return self._repositoryName
+
 class Identify(object):
     def __init__(self, repositoryName, baseURL, protocolVersion, adminEmails,
                  earliestDatestamp, deletedRecord, granularity, compression,

src/oaipmh/server.py

@@ -1,6 +1,8 @@
 from lxml.etree import ElementTree, Element, SubElement
 from lxml import etree
 from datetime import datetime
+import base64
+
 try:
     from urllib.parse import urlencode, quote, unquote
 except ImportError:
@@ -48,6 +50,8 @@ def getRecord(self, **kw):
         self._outputHeader(e_record, header)   
         if not header.isDeleted():
             self._outputMetadata(e_record, kw['metadataPrefix'], metadata)
+            if about:
+                self._outputAbout(e_record, about)
         return envelope
 
     def getMetadata(self, **kw):
@@ -131,7 +135,8 @@ def outputFunc(element, records, token_kw):
                 self._outputHeader(e_record, header)
                 if not header.isDeleted():
                     self._outputMetadata(e_record, metadataPrefix, metadata)
-                # XXX about
+                    if about:
+                        self._outputAbout(e_record, about)
         self._outputResuming(
             e_listRecords,
             self._server.listRecords,
@@ -243,6 +248,29 @@ def _outputMetadata(self, element, metadata_prefix, metadata):
         self._metadata_registry.writeMetadata(
             metadata_prefix, e_metadata, metadata)
 
+    def _outputAbout(self, element, about):
+        if about.baseURL():
+            e_about = SubElement(element, nsoai('about'))
+            e_provenance = SubElement(e_about, nsoai('provenance'))
+            e_provenance.set ('xmlns', 'http://www.openarchives.org/OAI/2.0/provenance')
+            e_provenance.set ('{%s}schemaLocation' % NS_XSI, 'http://www.openarchives.org/OAI/2.0/provenance http://www.openarchives.org/OAI/2.0/provenance.xsd')
+            e_originDescription = SubElement(e_provenance, nsoai('originDescription'))
+            e_originDescription.set ('harvestDate', datetime_to_datestamp(about.harvestDate()))
+            e_originDescription.set ('altered', "true")
+            e_baseURL = SubElement(e_originDescription, nsoai('baseURL'))
+            e_baseURL.text = about.baseURL()
+            e_identifier = SubElement(e_originDescription, nsoai('identifier'))
+            e_identifier.text = about.identifier()
+            e_datestamp = SubElement(e_originDescription, nsoai('datestamp'))
+#           e_datestamp.text = datetime_to_datestamp(about.datestamp())
+            e_datestamp.text = about.datestamp()
+            e_metadataNamespace = SubElement(e_originDescription, nsoai('metadataNamespace'))
+            e_metadataNamespace.text = about.metadataNamespace()
+            e_repositoryID = SubElement(e_originDescription, nsoai('repositoryID'))
+            e_repositoryID.text = about.repositoryID()
+            e_repositoryName = SubElement(e_originDescription, nsoai('repositoryName'))
+            e_repositoryName.text = about.repositoryName()
+
 class ServerBase(common.ResumptionOAIPMH):
     """A server that responds to messages by returning OAI-PMH compliant XML.
 
@@ -452,11 +480,13 @@ def encodeResumptionToken(kw, cursor):
     until = kw.get('until')
     if until is not None:
         kw['until'] = datetime_to_datestamp(until)
-    return quote(urlencode(kw))
+#    return quote(urlencode(kw))
+    return base64.b64encode(bytes(urlencode(kw), encoding="utf-8"))
 
 def decodeResumptionToken(token):
-    token = str(unquote(token))
-
+#    token = str(unquote(token))
+    token = base64.b64decode(token).decode('utf-8')
+
     try:
         kw = parse_qs(token, True, True)
     except ValueError:
@@ -468,6 +498,7 @@ def decodeResumptionToken(token):
         if key == 'from_' or key == 'until':
             value = datestamp_to_datetime(value)
         result[key] = value
+
     try:
         cursor = int(result.pop('cursor'))
     except (KeyError, ValueError):