Merge decaabf into 632a19b

.editorconfig

@@ -23,6 +23,7 @@ indent_size = 4
 known_first_party = invenio_sipstore
 multi_line_output = 2
 default_section = THIRDPARTY
+skip = .eggs
 
 # RST files (used by sphinx)
 [*.rst]

invenio_sipstore/api.py

@@ -205,7 +205,7 @@ def sip(self):
 
     @classmethod
     def create(cls, pid, record, archivable, create_sip_files=True,
-               user_id=None, agent=None):
+               user_id=None, agent=None, sip_metadata_type=None):
         """Create a SIP, from the PID and the Record.
 
         Apart from the SIP itself, it also creates ``RecordSIP`` for the
@@ -223,11 +223,20 @@ def create(cls, pid, record, archivable, create_sip_files=True,
         :param bool archivable: tells if the record should be archived.
             Usefull when ``Invenio-Archivematica`` is installed.
         :param bool create_sip_files: If True the SIPFiles will be created.
+        :param str sip_metadata_type: Used to fetch the
+            :py:class:`invenio_sipstore.models.SIPMetadataType` by ``name``.
+            If not provided, the ``$schema`` attribute of ``record`` will be
+            used to determine the
+            :py:class:`invenio_sipstore.models.SIPMetadataType`.
+            (default: ``None``)
         :returns: RecordSIP object.
         :rtype: :py:class:`invenio_sipstore.api.RecordSIP`
         """
         files = record.files if create_sip_files else None
-        mtype = SIPMetadataType.get_from_schema(record['$schema'])
+        if sip_metadata_type:
+            mtype = SIPMetadataType.get_from_name(sip_metadata_type)
+        else:
+            mtype = SIPMetadataType.get_from_schema(record['$schema'])
         metadata = {mtype.name: json.dumps(record.dumps())}
         with db.session.begin_nested():
             sip = SIP.create(archivable, files=files, metadata=metadata,

invenio_sipstore/archivers/bagit_archiver.py

@@ -16,6 +16,8 @@
 from flask import current_app
 from invenio_db import db
 from jsonschema import validate
+from six import string_types
+from werkzeug.utils import import_string
 
 from invenio_sipstore.api import SIP
 from invenio_sipstore.archivers import BaseArchiver
@@ -172,6 +174,26 @@ def _generate_bagging_date(self):
         """Generate the bagging date timestamp."""
         return datetime.utcnow().strftime("%Y-%m-%d %H:%M:%S.%f")
 
+    @staticmethod
+    def _generate_agent_tags(agent):
+        """Generate the agent info tags.
+
+        This method can be changed in the config to suit your needs, see
+        :py:data:`invenio_sipstore.config.SIPSTORE_AGENT_TAGS_FACTORY`
+
+        :return: agent information tags.
+        :rtype: list[(str, str)]
+        """
+        def _convert_key(key):
+            """Convert from snake_case."""
+            return ('-').join([x.capitalize() for x in key.split('_')])
+
+        exclude = ['$schema']
+
+        return [('X-Agent-{0}'.format(_convert_key(k)), v)
+                for k, v in sorted(agent.items())
+                if isinstance(v, string_types) and k not in exclude]
+
     def get_baginfo_file(self, filesinfo):
         """Create the bag-info.txt file from the tags."""
         # Include some auto-generated tags
@@ -185,6 +207,15 @@ def get_baginfo_file(self, filesinfo):
                 t_value = '{0}/{1}'.format(self.sip.id, self.archiver_version)
             content.append("{0}: {1}".format(t_name, t_value))
 
+        # Include agent tags
+        if self.sip.agent:
+            agent_tags_factory = import_string(
+                    current_app.config['SIPSTORE_AGENT_TAGS_FACTORY'])
+            agent_tags = agent_tags_factory(self.sip.agent)
+
+            for k, v in agent_tags:
+                content.append("{0}: {1}".format(k, v))
+
         content = "\n".join(content)
         return self._generate_extra_info(content, 'bag-info.txt')
 
@@ -243,7 +274,7 @@ def get_all_files(self):
         extra_files = self._get_extra_files(data_files, metadata_files)
 
         bagit_files = []
-        fetched_fi = [fi for fi in data_files if self._is_fetched(fi)]
+        fetched_fi = [f for f in data_files if self._is_fetched(f)]
         if fetched_fi:
             bagit_files.append(self.get_fetch_file(fetched_fi))
         bagit_files.append(self.get_baginfo_file(data_files + metadata_files +

invenio_sipstore/config.py

@@ -24,6 +24,10 @@
 SIPSTORE_AGENT_FACTORY = 'invenio_sipstore.api.SIP._build_agent_info'
 """Factory to build the agent, stored for the information about the SIP."""
 
+SIPSTORE_AGENT_TAGS_FACTORY = \
+    'invenio_sipstore.archivers.BagItArchiver._generate_agent_tags'
+"""Factory to build agent information tags."""
+
 SIPSTORE_FILEPATH_MAX_LEN = 1024
 """Max filepath length."""
 

setup.py

@@ -24,7 +24,7 @@
     'pytest-cov>=1.8.0',
     'pytest-mock>=1.6.0',
     'pytest-pep8>=1.0.6',
-    'pytest>=2.8.0',
+    'pytest>=3.6.0',
 ]
 
 extras_require = {

tests/conftest.py

@@ -146,7 +146,12 @@ def sips(db, locations, sip_metadata_types):
     SIP-3: File2(renamed on SIPFile, but same FileInstance), File3
     SIP-4: File4, File5, File6
     """
-    sip1 = SIP.create()
+    # A SIP with agent info
+    sip1 = SIP.create(agent={
+        'email': 'spiderpig@invenio.org',
+        'orcid': '1111-1111-1111-1111',
+        'ip_address': '1.1.1.1'
+    })
     sip1api = SIPApi(sip1)
     sip1api.attach_metadata('marcxml-test', '<p>XML 1</p>')
     sip1api.attach_metadata('json-test', '{"title": "JSON 1"}')

tests/test_api.py

@@ -250,5 +250,19 @@ def test_RecordSIP_create(db, mocker):
     assert SIPMetadata.query.count() == 2
     assert len(rsip.sip.files) == 0
     assert len(rsip.sip.metadata) == 1
+
+    # try with specific SIP metadata type
+    mtype = SIPMetadataType(title='JSON Test 2', name='json-test-2',
+                            format='json', schema=None)  # no schema
+    db.session.add(mtype)
+    db.session.commit()
+
+    rsip = RecordSIP.create(pid, record, True, create_sip_files=False,
+                            user_id=user.id, agent=agent,
+                            sip_metadata_type='json-test-2')
+    assert SIPMetadata.query.count() == 3
+    assert len(rsip.sip.metadata) == 1
+    assert rsip.sip.metadata[0].type.id == mtype.id
+
     # finalization
     rmtree(tmppath)

tests/test_bagit_archiver.py

@@ -193,7 +193,10 @@ def test_write_patched(mocker, sips, archive_fs,
             "Bagging-Date: {0}\n".format(dt) +
             "Payload-Oxum: 93.4\n"
             "External-Identifier: {0}/SIPBagIt-v1.0.0\n".format(sips[0].id) +
-            "External-Description: BagIt archive of SIP."
+            "External-Description: BagIt archive of SIP.\n"
+            "X-Agent-Email: spiderpig@invenio.org\n"
+            "X-Agent-Ip-Address: 1.1.1.1\n"
+            "X-Agent-Orcid: 1111-1111-1111-1111"
         )),
     ]
     expected_sip2 = [