Merge pull request #13 from NatLibFi/feature/simplye-209/language-filter

Implement language facet for OPDS feed

api/lanes.py

@@ -1356,6 +1356,7 @@ class CrawlableFacets(Facets):
         Facets.COLLECTION_FACET_GROUP_NAME: Facets.COLLECTION_FULL,
         Facets.DISTRIBUTOR_FACETS_GROUP_NAME: Facets.DISTRIBUTOR_ALL,
         Facets.COLLECTION_NAME_FACETS_GROUP_NAME: Facets.COLLECTION_NAME_ALL,
+        Facets.LANGUAGE_FACET_GROUP_NAME: Facets.LANGUAGE_ALL,
     }
 
     @classmethod

core/configuration/library.py

@@ -264,6 +264,34 @@ class LibrarySettings(BaseSettings):
             skip=True,
         ),
     )
+    # Finland
+    facets_enabled_language: list[str] = FormField(
+        FacetConstants.DEFAULT_ENABLED_FACETS[FacetConstants.LANGUAGE_FACET_GROUP_NAME],
+        form=LibraryConfFormItem(
+            label="Allow patrons to filter language to",
+            type=ConfigurationFormItemType.MENU,
+            options={
+                facet: FacetConstants.FACET_DISPLAY_TITLES[facet]
+                for facet in FacetConstants.LANGUAGE_FACETS
+            },
+            category="Lanes & Filters",
+            paired="facets_default_language",
+            level=Level.SYS_ADMIN_OR_MANAGER,
+        ),
+    )
+    facets_default_language: str = FormField(
+        FacetConstants.LANGUAGE_ALL,
+        form=LibraryConfFormItem(
+            label="Default Language",
+            type=ConfigurationFormItemType.SELECT,
+            options={
+                facet: FacetConstants.FACET_DISPLAY_TITLES[facet]
+                for facet in FacetConstants.LANGUAGE_FACETS
+            },
+            category="Lanes & Filters",
+            skip=True,
+        ),
+    )
     library_description: str | None = FormField(
         None,
         form=LibraryConfFormItem(

core/external_search.py

@@ -1768,8 +1768,10 @@ def from_worklist(cls, _db, worklist, facets):
         excluded_audiobook_data_sources = [DataSource.lookup(_db, x) for x in excluded]
         if library is None:
             allow_holds = True
+            facets_enabled_language = FacetConstants.LANGUAGE_ALL
         else:
             allow_holds = library.settings.allow_holds
+            facets_enabled_language = library.settings.facets_enabled_language
         return cls(
             collections,
             media,
@@ -1784,6 +1786,7 @@ def from_worklist(cls, _db, worklist, facets):
             allow_holds=allow_holds,
             license_datasource=license_datasource_id,
             lane_building=True,
+            facets_enabled_language=facets_enabled_language,
         )
 
     def __init__(
@@ -1938,6 +1941,10 @@ def __init__(
 
         self.lane_building = kwargs.pop("lane_building", False)
 
+        self.facets_enabled_language = kwargs.pop(
+            "facets_enabled_language", FacetConstants.LANGUAGE_ALL
+        )
+
         # At this point there should be no keyword arguments -- you can't pass
         # whatever you want into this method.
         if kwargs:
@@ -2065,7 +2072,22 @@ def build(self, _chain_filters=None):
         if self.media:
             f = chain(f, Terms(medium=scrub_list(self.media)))
 
-        if self.languages:
+        # Finland, logic for LANGUAGE_OTHERS
+        if self.languages == [FacetConstants.LANGUAGE_OTHERS]:
+            excluded_terms = [
+                language
+                for language in self.facets_enabled_language
+                if language
+                not in {
+                    FacetConstants.LANGUAGE_ALL,
+                    FacetConstants.LANGUAGE_OTHERS,
+                }
+            ]
+            exclusion_query = Bool(
+                must_not=[Terms(language=scrub_list(excluded_terms))]
+            )
+            f = chain(f, exclusion_query)
+        elif self.languages:
             f = chain(f, Terms(language=scrub_list(self.languages)))
 
         if self.fiction is not None:

core/facets.py

@@ -30,6 +30,22 @@ class FacetConstants:
         AVAILABLE_OPEN_ACCESS,
     ]
 
+    # Finland
+    # Subset the collection by language.
+    LANGUAGE_FACET_GROUP_NAME = "language"
+    LANGUAGE_ALL = "all"
+    LANGUAGE_FINNISH = "fin"
+    LANGUAGE_SWEDISH = "swe"
+    LANGUAGE_ENGLISH = "eng"
+    LANGUAGE_OTHERS = "others"
+    LANGUAGE_FACETS: list[str] = [
+        LANGUAGE_ALL,
+        LANGUAGE_FINNISH,
+        LANGUAGE_SWEDISH,
+        LANGUAGE_ENGLISH,
+        LANGUAGE_OTHERS,
+    ]
+
     # The names of the order facets.
     ORDER_FACET_GROUP_NAME = "order"
     ORDER_TITLE = "title"
@@ -66,6 +82,7 @@ class FacetConstants:
         COLLECTION_FACET_GROUP_NAME: COLLECTION_FACETS,
         AVAILABILITY_FACET_GROUP_NAME: AVAILABILITY_FACETS,
         ORDER_FACET_GROUP_NAME: ORDER_FACETS,
+        LANGUAGE_FACET_GROUP_NAME: LANGUAGE_FACETS,  # Finland
     }
 
     GROUP_DISPLAY_TITLES = {
@@ -74,6 +91,7 @@ class FacetConstants:
         COLLECTION_FACET_GROUP_NAME: _("Collection"),
         DISTRIBUTOR_FACETS_GROUP_NAME: _("Distributor"),
         COLLECTION_NAME_FACETS_GROUP_NAME: _("Collection Name"),
+        LANGUAGE_FACET_GROUP_NAME: _("Language"),  # Finland
     }
 
     GROUP_DESCRIPTIONS = {
@@ -84,6 +102,7 @@ class FacetConstants:
         COLLECTION_NAME_FACETS_GROUP_NAME: _(
             "Allow patrons to filter by collection name"
         ),
+        LANGUAGE_FACET_GROUP_NAME: _("Allow patrons to filter by language"),  # Finland
     }
 
     FACET_DISPLAY_TITLES = {
@@ -98,6 +117,11 @@ class FacetConstants:
         AVAILABLE_OPEN_ACCESS: _("Yours to keep"),
         COLLECTION_FULL: _("Everything"),
         COLLECTION_FEATURED: _("Popular Books"),
+        LANGUAGE_ALL: _("All"),
+        LANGUAGE_FINNISH: _("Finnish"),
+        LANGUAGE_SWEDISH: _("Swedish"),
+        LANGUAGE_ENGLISH: _("English"),
+        LANGUAGE_OTHERS: _("Others"),
     }
 
     # For titles generated based on some runtime value
@@ -118,6 +142,7 @@ class FacetConstants:
         COLLECTION_FACET_GROUP_NAME: [COLLECTION_FULL, COLLECTION_FEATURED],
         DISTRIBUTOR_FACETS_GROUP_NAME: [DISTRIBUTOR_ALL],
         COLLECTION_NAME_FACETS_GROUP_NAME: [COLLECTION_NAME_ALL],
+        LANGUAGE_FACET_GROUP_NAME: LANGUAGE_FACETS,
     }
 
     # Unless a library offers an alternate configuration, these
@@ -128,6 +153,7 @@ class FacetConstants:
         COLLECTION_FACET_GROUP_NAME: COLLECTION_FULL,
         DISTRIBUTOR_FACETS_GROUP_NAME: DISTRIBUTOR_ALL,
         COLLECTION_NAME_FACETS_GROUP_NAME: COLLECTION_NAME_ALL,
+        LANGUAGE_FACET_GROUP_NAME: LANGUAGE_ALL,
     }
 
     SORT_ORDER_TO_OPENSEARCH_FIELD_NAME = {

core/lane.py

@@ -324,6 +324,7 @@ def default(
         entrypoint=None,
         distributor=None,
         collection_name=None,
+        language=None,
     ):
         return cls(
             library,
@@ -332,6 +333,7 @@ def default(
             order=order,
             distributor=distributor,
             collection_name=collection_name,
+            language=language,
             entrypoint=entrypoint,
         )
 
@@ -436,12 +438,21 @@ def _values_from_request(
                 400,
             )
 
+        # Finland
+        g = Facets.LANGUAGE_FACET_GROUP_NAME
+        language: str = get_argument(g, cls.default_facet(config, g))
+        language_facets = cls.available_facets(config, g)
+        # In the case of language, don't want to limit the input to only the
+        # available facet values as we might get searches from elsewhere with
+        # other languages.
+
         enabled = {
             Facets.ORDER_FACET_GROUP_NAME: order_facets,
             Facets.AVAILABILITY_FACET_GROUP_NAME: availability_facets,
             Facets.COLLECTION_FACET_GROUP_NAME: collection_facets,
             Facets.DISTRIBUTOR_FACETS_GROUP_NAME: distributor_facets,
             Facets.COLLECTION_NAME_FACETS_GROUP_NAME: collection_name_facets,
+            Facets.LANGUAGE_FACET_GROUP_NAME: language_facets,  # Finland
         }
 
         return dict(
@@ -451,6 +462,7 @@ def _values_from_request(
             distributor=distributor,
             collection_name=collection_name,
             enabled_facets=enabled,
+            language=language,  # Finland
         )
 
     @classmethod
@@ -484,6 +496,7 @@ def __init__(
         order,
         distributor,
         collection_name,
+        language,
         order_ascending=None,
         enabled_facets=None,
         entrypoint=None,
@@ -536,6 +549,9 @@ def __init__(
         self.collection_name = collection_name or self.default_facet(
             library, self.COLLECTION_NAME_FACETS_GROUP_NAME
         )
+        self.language: str = language or self.default_facet(
+            library, self.LANGUAGE_FACET_GROUP_NAME
+        )
         if order_ascending == self.ORDER_ASCENDING:
             order_ascending = True
         elif order_ascending == self.ORDER_DESCENDING:
@@ -551,6 +567,7 @@ def navigate(
         entrypoint=None,
         distributor=None,
         collection_name=None,
+        language=None,
     ):
         """Create a slightly different Facets object from this one."""
         return self.__class__(
@@ -560,6 +577,7 @@ def navigate(
             order=order or self.order,
             distributor=distributor or self.distributor,
             collection_name=collection_name or self.collection_name,
+            language=language or self.language,
             enabled_facets=self.facets_enabled_at_init,
             entrypoint=(entrypoint or self.entrypoint),
             entrypoint_is_default=False,
@@ -577,6 +595,8 @@ def items(self):
             yield (self.DISTRIBUTOR_FACETS_GROUP_NAME, self.distributor)
         if self.collection_name:
             yield (self.COLLECTION_NAME_FACETS_GROUP_NAME, self.collection_name)
+        if self.language:
+            yield (self.LANGUAGE_FACET_GROUP_NAME, self.language)
 
     @property
     def enabled_facets(self):
@@ -595,6 +615,7 @@ def enabled_facets(self):
                 self.COLLECTION_FACET_GROUP_NAME,
                 self.DISTRIBUTOR_FACETS_GROUP_NAME,
                 self.COLLECTION_NAME_FACETS_GROUP_NAME,
+                self.LANGUAGE_FACET_GROUP_NAME,
             ]
             for facet_type in facet_types:
                 yield self.facets_enabled_at_init.get(facet_type, [])
@@ -606,6 +627,7 @@ def enabled_facets(self):
                 Facets.COLLECTION_FACET_GROUP_NAME,
                 Facets.DISTRIBUTOR_FACETS_GROUP_NAME,
                 Facets.COLLECTION_NAME_FACETS_GROUP_NAME,
+                Facets.LANGUAGE_FACET_GROUP_NAME,
             ):
                 yield self.available_facets(self.library, group_name)
 
@@ -625,6 +647,7 @@ def facet_groups(self):
             collection_facets,
             distributor_facets,
             collection_name_facets,
+            language_facets,
         ) = self.enabled_facets
 
         def dy(new_value):
@@ -674,6 +697,13 @@ def dy(new_value):
                 facets = self.navigate(collection_name=facet)
                 yield (group, facet, facets, facet == current_value)
 
+        if len(language_facets) > 1:
+            for facet in language_facets:
+                group = self.LANGUAGE_FACET_GROUP_NAME
+                current_value = self.language
+                facets = self.navigate(language=facet)
+                yield (group, facet, facets, facet == current_value)
+
     def modify_search_filter(self, filter):
         """Modify the given external_search.Filter object
         so that it reflects the settings of this Facets object.
@@ -692,6 +722,10 @@ def modify_search_filter(self, filter):
         filter.availability = self.availability
         filter.subcollection = self.collection
 
+        # Finland
+        if self.language and self.language != self.LANGUAGE_ALL:
+            filter.languages = [self.language]
+
         # We can only have distributor and collection name facets if we have a library
         if self.library:
             _db = Session.object_session(self.library)
@@ -973,6 +1007,7 @@ def __init__(self, **kwargs):
         kwargs.setdefault("availability", None)
         kwargs.setdefault("distributor", None)
         kwargs.setdefault("collection_name", None)
+        kwargs.setdefault("language", None)
         order = kwargs.setdefault("order", None)
 
         if order in (None, self.ORDER_BY_RELEVANCE):
@@ -2935,6 +2970,7 @@ def update_size(self, _db, search_engine=None):
                 order=FacetConstants.ORDER_WORK_ID,
                 distributor=FacetConstants.DISTRIBUTOR_ALL,
                 collection_name=FacetConstants.COLLECTION_NAME_ALL,
+                language=FacetConstants.LANGUAGE_ALL,
                 entrypoint=entrypoint,
             )
             filter = self.filter(_db, facets)

tests/api/controller/test_work.py

@@ -23,6 +23,7 @@
 from core.classifier import Classifier
 from core.entrypoint import AudiobooksEntryPoint
 from core.external_search import SortKeyPagination, mock_search_index
+from core.facets import FacetConstants
 from core.feed.acquisition import OPDSAcquisitionFeed
 from core.feed.annotator.circulation import LibraryAnnotator
 from core.feed.types import WorkEntry
@@ -139,7 +140,7 @@ def test_contributor(self, work_fixture: WorkFixture):
         facet_links = [
             link for link in links if link["rel"] == "http://opds-spec.org/facet"
         ]
-        assert 10 == len(facet_links)
+        assert 10 + len(FacetConstants.LANGUAGE_FACETS) == len(facet_links)
 
         # At this point we don't want to generate real feeds anymore.
         # We can't do a real end-to-end test without setting up a real
@@ -905,7 +906,7 @@ def test_series(self, work_fixture: WorkFixture):
         facet_links = [
             link for link in links if link["rel"] == "http://opds-spec.org/facet"
         ]
-        assert 11 == len(facet_links)
+        assert 11 + len(FacetConstants.LANGUAGE_FACETS) == len(facet_links)
 
         # The facet link we care most about is the default sort order,
         # put into place by SeriesFacets.

tests/api/feed/test_opds_acquisition_feed.py

@@ -1151,7 +1151,7 @@ def run(wl=None, facets=None, pagination=None):
         # The make_link function that was passed in calls
         # TestAnnotator.feed_url() when passed an EntryPoint. The
         # Facets object's other facet groups are propagated in this URL.
-        first_page_url = "http://wl/?available=all&collection=full&collectionName=All&distributor=All&entrypoint=Book&order=author"
+        first_page_url = "http://wl/?available=all&collection=full&collectionName=All&distributor=All&entrypoint=Book&language=all&order=author"
         assert first_page_url == make_link(EbooksEntryPoint)
 
         # Pagination information is not propagated through entry point links

tests/api/test_lanes.py

@@ -918,6 +918,7 @@ def test_default(self, db: DatabaseTransactionFixture):
         assert facets.collection == CrawlableFacets.COLLECTION_FULL
         assert facets.availability == CrawlableFacets.AVAILABLE_ALL
         assert facets.order == CrawlableFacets.ORDER_LAST_UPDATE
+        assert facets.language == CrawlableFacets.LANGUAGE_ALL
         assert facets.order_ascending is False
 
         [
@@ -926,10 +927,11 @@ def test_default(self, db: DatabaseTransactionFixture):
             collection,
             distributor,
             collectionName,
+            language,
         ] = facets.enabled_facets
 
         # The default facets are the only ones enabled.
-        for facet in [order, availability, collection]:
+        for facet in [order, availability, collection, language]:
             assert len(facet) == 1
 
         # Except for distributor and collectionName, which have the default

tests/core/models/test_library.py

@@ -187,6 +187,8 @@ def test_explain(self, db: DatabaseTransactionFixture):
 facets_default_available='all'
 facets_enabled_collection='['full', 'featured']'
 facets_default_collection='full'
+facets_enabled_language='['all', 'fin', 'swe', 'eng', 'others']'
+facets_default_language='all'
 help_web='http://library.com/support'
 default_notification_email_address='noreply@thepalaceproject.org'
 color_scheme='blue'