Merge pull request #157 from tetherless-world/unrewrite_bnodes_on_nan…

…opub_load

Unrewrite bnodes on nanopub load

main.py

@@ -221,6 +221,24 @@ def run_importer(self, entity_name):
                                                       self,
                                                       update_listener=self.nanopub_update_listener)
 
+    _file_depot = None
+    @property
+    def file_depot(self):
+        if self._file_depot is None:
+            if DepotManager.get('files') is None:
+                DepotManager.configure('files', self.config['file_archive'])
+            self._file_depot = DepotManager.get('files')
+        return self._file_depot
+
+    _nanopub_depot = None
+    @property
+    def nanopub_depot(self):
+        if self._nanopub_depot is None:
+            if DepotManager.get('nanopublications') is None:
+                DepotManager.configure('nanopublications', self.config['nanopub_archive'])
+            self._nanopub_depot = DepotManager.get('nanopublications')
+        return self._nanopub_depot
+
     def configure_database(self):
         """
         Database configuration should be set here
@@ -244,15 +262,6 @@ def configure_database(self):
         self.security = Security(self, self.datastore,
                                  register_form=ExtendedRegisterForm)
 
-        self.file_depot = DepotManager.get('files')
-        if self.file_depot is None:
-            DepotManager.configure('files', self.config['file_archive'])
-            self.file_depot = DepotManager.get('files')
-        self.nanopub_depot = DepotManager.get('nanopublications')
-        if self.nanopub_depot is None:
-            DepotManager.configure('nanopublications', self.config['nanopub_archive'])
-            self.nanopub_depot = DepotManager.get('nanopublications')
-
     def __weighted_route(self, *args, **kwargs):
         """
         Override the match_compare_key function of the Rule created by invoking Flask.route.

tests/api/test_nanopub_crud.py

@@ -26,16 +26,24 @@ def test_create(self):
 
     def test_bnode_rewrite(self):
         self.login_new_user()
+        self.app.config['BNODE_REWRITE'] = True
         response = self.post_nanopub(data=PERSON_INSTANCE_BNODE_TURTLE,
                                         content_type="text/turtle",
                                         expected_headers=["Location"])
 
+        rewritten_bnode_subjects = [s for s,p,o,context in self.app.db.quads()
+                                    if isinstance(s, URIRef) and s.startswith('bnode:')]
+        bnode_subjects = [s for s,p,o,context in self.app.db.quads() if isinstance(s, BNode)]
+        self.assertEquals(len(rewritten_bnode_subjects),1)
+        self.assertEquals(len(bnode_subjects), 0)
+
+        # Nanopublication bnodes should, when retrieved, get turned back into bnodes
         nanopub = self.app.nanopub_manager.get(URIRef(response.headers['Location']))
         rewritten_bnode_subjects = [s for s,p,o,context in nanopub.quads()
                                     if isinstance(s, URIRef) and s.startswith('bnode:')]
         bnode_subjects = [s for s,p,o,context in nanopub.quads() if isinstance(s, BNode)]
-        self.assertEquals(len(rewritten_bnode_subjects),1)
-        self.assertEquals(len(bnode_subjects), 0)
+        self.assertEquals(len(rewritten_bnode_subjects),0)
+        self.assertEquals(len(bnode_subjects), 1)
 
     def test_no_bnode_rewrite(self):
         self.app.config['BNODE_REWRITE'] = False

tests/unit/whyis_test/autonomic/test_frir_agent.py

@@ -35,6 +35,11 @@ def test_basic_file_storage(self):
 
         content = self.client.get("/about",query_string={"uri":manifestation},
                                   follow_redirects=True)
+        # This isn't going to work until we untangle the app
+        # configured by manage.py from the app that's configured for
+        # running the tests. The get_entity blueprint seems to be
+        # using the manage.py app.
+
         self.assertEquals(content.mimetype, "application/n-quads")
         digest = hashlib.sha256(content.data).hexdigest().lstrip('0')
         self.assertEquals(pmanif[digest], manifestation)

whyis/blueprint/entity/get_entity.py

@@ -28,12 +28,12 @@ def view(name=None, format=None, view=None):
     # 'view' is the default view
     fileid = resource.value(current_app.NS.whyis.hasFileID)
     if fileid is not None and 'view' not in request.args:
-        print (resource.identifier, fileid)
+        fileid = fileid.value
         f = None
-        if current_app.file_depot.exists(fileid):
-            f = current_app.file_depot.get(fileid)
-        elif current_app.nanopub_depot.exists(fileid):
+        if current_app.nanopub_depot.exists(fileid):
             f = current_app.nanopub_depot.get(fileid)
+        elif current_app.file_depot.exists(fileid):
+            f = current_app.file_depot.get(fileid)
         if f is not None:
             fsa = FileServeApp(f, current_app.config["file_archive"].get("cache_max_age",3600*24*7))
             return fsa

whyis/nanopub/nanopublication_manager.py

@@ -210,7 +210,7 @@ def skolemize(x):
                 return x
             for store in stores:
                 for s, p, o, c in rdflib.ConjunctiveGraph(store).quads():
-                    if 'BNODE_REWRITE' not in self.app.config or self.app.config['BNODE_REWRITE']:
+                    if self.app.config.get('BNODE_REWRITE', False):
                         s = skolemize(s)
                         o = skolemize(o)
                         # predicates can't be bnodes, and contexts have already been rewritten.
@@ -244,6 +244,12 @@ def get(self, nanopub_uri, graph=None):
         ?np np:hasAssertion?|np:hasProvenance?|np:hasPublicationInfo? ?g.
         graph ?g { ?s ?p ?o}
         }''', initNs={'np':np}, initBindings={'np':nanopub_uri})
-        graph.addN(quads)
+        for s, p, o, g in quads:
+            if self.app.config.get('BNODE_REWRITE', False):
+                if isinstance(s, rdflib.URIRef) and s.startswith('bnode:'):
+                    s = rdflib.BNode(s.replace('bnode:','',1))
+                if isinstance(o, rdflib.URIRef) and o.startswith('bnode:'):
+                    o = rdflib.BNode(o.replace('bnode:','',1))
+            graph.add((s,p,o,g))
         nanopub = Nanopublication(store=graph.store, identifier=nanopub_uri)
         return nanopub

whyis/test/integration_test_case.py

@@ -1,34 +1,12 @@
 import urllib.request, urllib.error, urllib.parse
 from flask import Flask, g
-from flask_testing import TestCase
+from .test_case import TestCase
 from flask_login import login_user, current_user, current_app
 import requests
 
 
 class IntegrationTestCase(TestCase):
 
-    def create_app(self):
-        from whyis.app_factory import app_factory
-        from depot.manager import DepotManager
-        import config_defaults
-
-        if 'admin_queryEndpoint' in config_defaults.Test:
-            del config_defaults.Test['admin_queryEndpoint']
-            del config_defaults.Test['admin_updateEndpoint']
-            del config_defaults.Test['knowledge_queryEndpoint']
-            del config_defaults.Test['knowledge_updateEndpoint']
-
-        # Default port is 5000
-        config_defaults.Test['LIVESERVER_PORT'] = 8943
-        # Default timeout is 5 seconds
-        config_defaults.Test['LIVESERVER_TIMEOUT'] = 10
-
-        application = app_factory(config_defaults.Test, config_defaults.project_name)
-        application.config['TESTING'] = True
-        application.config['WTF_CSRF_ENABLED'] = False
-
-        return application
-
     def create_user(self, email, password, username="identifier", fn="First", ln="Last", roles='Admin'):
         import commands
         from uuid import uuid4

whyis/test/test_case.py

@@ -1,8 +1,10 @@
 import flask_testing
 
+from flask import current_app
 from flask import Response
 from rdflib import URIRef
 from typing import Optional, Dict
+from depot.manager import DepotManager
 
 class TestCase(flask_testing.TestCase):
 
@@ -35,23 +37,29 @@ def create_app(self):
             import config
         except:
             from whyis import config_defaults as config
-        
+
         if 'admin_queryEndpoint' in config.Test:
             del config.Test['admin_queryEndpoint']
             del config.Test['admin_updateEndpoint']
             del config.Test['knowledge_queryEndpoint']
             del config.Test['knowledge_updateEndpoint']
-
+        config.Test['TESTING'] = True
+        config.Test['WTF_CSRF_ENABLED'] = False
+        config.Test['nanopub_archive'] = {
+            'depot.backend' : 'depot.io.memory.MemoryFileStorage'
+        }
+        config.Test['DEFAULT_ANONYMOUS_READ'] = False
+        config.Test['file_archive'] = {
+            'depot.backend' : 'depot.io.memory.MemoryFileStorage'
+        }
         # Default port is 5000
         config.Test['LIVESERVER_PORT'] = 8943
         # Default timeout is 5 seconds
         config.Test['LIVESERVER_TIMEOUT'] = 10
-
+        
         from whyis.app_factory import app_factory
         application = app_factory(config.Test, config.project_name)
-        application.config['TESTING'] = True
-        application.config['WTF_CSRF_ENABLED'] = False
-
+
         return application
 
     def create_user(self, email, password, username="identifier", fn="First", ln="Last", roles='Admin'):