Improve licenses detection in harvesting

CHANGELOG.md

@@ -17,6 +17,7 @@
 - Properly serialize empty geometries for zones missing it and prevent leaflet crash on invalid bounds [#1188](https://github.com/opendatateam/udata/pull/1188)
 - Start validating some configuration parameters [#1197](https://github.com/opendatateam/udata/pull/1197)
 - Migration to remove resources without title or url [#1200](https://github.com/opendatateam/udata/pull/1200)
+- Improve harvesting licenses detection [#1203](https://github.com/opendatateam/udata/pull/1203)
 
 ## 1.1.8 (2017-09-28)
 

udata/core/dataset/models.py

@@ -111,13 +111,30 @@ def __unicode__(self):
         return self.title
 
     @classmethod
-    def guess(cls, text):
+    def guess(cls, *strings, **kwargs):
+        '''
+        Try to guess a license from a list of strings.
+
+        Accept a `default` keyword argument which will be
+        the default fallback license.
+        '''
+        license = None
+        for string in strings:
+            license = cls.guess_one(string)
+            if license:
+                break
+        return license or kwargs.get('default')
+
+    @classmethod
+    def guess_one(cls, text):
         '''
         Try to guess license from a string.
 
         Try to exact match on identifier then slugified title
         and fallback on edit distance ranking (after slugification)
         '''
+        if not text:
+            return
         qs = cls.objects
         text = text.strip().lower()  # Stored identifiers are lower case
         slug = cls.slug.slugify(text)  # Use slug as it normalize string
@@ -132,6 +149,10 @@ def guess(cls, text):
                 license = candidates[0]
         return license
 
+    @classmethod
+    def default(cls):
+        return cls.objects(id=DEFAULT_LICENSE['id']).first()
+
 
 class DatasetQuerySet(db.OwnedQuerySet):
     def visible(self):

udata/core/dataset/rdf.py

@@ -372,10 +372,7 @@ def dataset_from_rdf(graph, dataset=None):
             elif isinstance(value, RdfResource):
                 licenses.add(value.identifier.toPython())
 
-    for text in licenses:
-        license = License.guess(text)
-        if license:
-            dataset.license = license
-            break
+    default_license = dataset.license or License.default()
+    dataset.license = License.guess(*licenses, default=default_license)
 
     return dataset

udata/harvest/backends/ckan.py

@@ -77,6 +77,7 @@
     'title': basestring,
     'notes': Any(All(basestring, normalize_string), None),
     'license_id': All(DefaultTo('not-specified'), basestring),
+    'license_title': Any(basestring, None),
     'tags': [tag],
 
     'metadata_created': All(basestring, to_date),
@@ -159,8 +160,13 @@ def process(self, item):
             dataset.slug = data['name']
         dataset.title = data['title']
         dataset.description = data['notes']
-        dataset.license = License.objects(id=data['license_id']).first()
-        # dataset.license = license or License.objects.get(id='notspecified')
+
+        # Detect license
+        default_license = dataset.license or License.default()
+        dataset.license = License.guess(data['license_id'],
+                                        data['license_title'],
+                                        default=default_license)
+
         dataset.tags = [t['name'] for t in data['tags'] if t['name']]
 
         dataset.created_at = data['metadata_created']

udata/harvest/backends/ods.py

@@ -115,10 +115,12 @@ def process(self, item):
 
         dataset.tags = list(tags)
 
-        ods_license_id = ods_metadata.get('license')
-        if ods_license_id and ods_license_id in self.LICENSES:
-            license_id = self.LICENSES[ods_license_id]
-            dataset.license = License.objects.get(id=license_id)
+        # Detect license
+        default_license = dataset.license or License.default()
+        license_id = ods_metadata.get('license')
+        dataset.license = License.guess(license_id,
+                                        self.LICENSES.get(license_id),
+                                        default=default_license)
 
         dataset.resources = []
 

udata/tests/dataset/test_dataset_model.py

@@ -247,6 +247,19 @@ def test_not_found(self):
         found = License.guess('should not be found')
         self.assertIsNone(found)
 
+    def test_not_found_with_default(self):
+        license = LicenseFactory()
+        found = License.guess('should not be found', default=license)
+        self.assertEqual(found.id, license.id)
+
+    def test_none(self):
+        found = License.guess(None)
+        self.assertIsNone(found)
+
+    def test_empty_string(self):
+        found = License.guess('')
+        self.assertIsNone(found)
+
     def test_exact_match_by_id(self):
         license = LicenseFactory()
         found = License.guess(license.id)
@@ -294,3 +307,9 @@ def test_match_by_title_with_mismatching_case(self):
         found = License.guess('License ODBL')
         self.assertIsInstance(found, License)
         self.assertEqual(license.id, found.id)
+
+    def test_multiple_strings(self):
+        license = LicenseFactory()
+        found = License.guess('should not match', license.id)
+        self.assertIsInstance(found, License)
+        self.assertEqual(license.id, found.id)