Remove legacy ia_collection_s field, use ia_collection directly (ref #11586)

openlibrary/i18n/messages.pot

@@ -1,8 +1,8 @@
 # Translations template for Open Library.
-# Copyright (C) 2025 Internet Archive
+# Copyright (C) 2026 Internet Archive
 # This file is distributed under the same license as the Open Library
 # project.
-# FIRST AUTHOR <EMAIL@ADDRESS>, 2025.
+# FIRST AUTHOR <EMAIL@ADDRESS>, 2026.
 #
 #, fuzzy
 msgid ""

openlibrary/plugins/worksearch/code.py

@@ -584,9 +584,7 @@ def get_doc(doc: SolrDocument):
         url=f"{doc['key']}/{urlsafe(doc['title'])}",
         edition_count=doc['edition_count'],
         ia=doc.get('ia', []),
-        collections=(
-            doc['ia_collection_s'].split(';') if doc.get('ia_collection_s') else []
-        ),
+        collections=(doc.get('ia_collection') or []),
         has_fulltext=doc.get('has_fulltext', False),
         public_scan=doc.get('public_scan_b', bool(doc.get('ia'))),
         lending_edition=doc.get('lending_edition_s', None),

openlibrary/plugins/worksearch/schemes/editions.py

@@ -40,10 +40,7 @@ class EditionSearchScheme(SearchScheme):
             'publishers': 'publisher',
             'subtitle': 'alternative_subtitle',
             'title': 'alternative_title',
-            # "Private" fields
-            # This is private because we'll change it to a multi-valued field instead of a
-            # plain string at the next opportunity, which will make it much more usable.
-            '_ia_collection': 'ia_collection_s',
+            '_ia_collection': 'ia_collection',
         }
     )
     sorts = MappingProxyType(

openlibrary/plugins/worksearch/schemes/works.py

@@ -140,10 +140,7 @@ class WorkSearchScheme(SearchScheme):
             'work_subtitle': 'subtitle',
             'work_title': 'title',
             'trending': 'trending_z_score',
-            # "Private" fields
-            # This is private because we'll change it to a multi-valued field instead of a
-            # plain string at the next opportunity, which will make it much more usable.
-            '_ia_collection': 'ia_collection_s',
+            '_ia_collection': 'ia_collection',
         }
     )
     sorts = MappingProxyType(
@@ -211,7 +208,7 @@ class WorkSearchScheme(SearchScheme):
             'lending_edition_s',
             'lending_identifier_s',
             'language',
-            'ia_collection_s',
+            'ia_collection',
             # FIXME: These should be fetched from book_providers, but can't cause circular
             # dep
             'id_project_gutenberg',
@@ -278,8 +275,6 @@ def transform_user_query(
                     lcc_transform(node)
                 if node.name in ('dcc', 'dcc_sort'):
                     ddc_transform(node)
-                if node.name == 'ia_collection_s':
-                    ia_collection_s_transform(node)
 
         if not has_search_fields:
             # If there are no search fields, maybe we want just an isbn?
@@ -737,23 +732,6 @@ def isbn_transform(sf: luqum.tree.SearchField):
         logger.warning(f"Unexpected isbn SearchField value type: {type(field_val)}")
 
 
-def ia_collection_s_transform(sf: luqum.tree.SearchField):
-    """
-    Because this field is not a multi-valued field in solr, but a simple ;-separate
-    string, we have to do searches like this for now.
-    """
-    val = sf.children[0]
-    if isinstance(val, luqum.tree.Word):
-        if val.value.startswith('*'):
-            val.value = '*' + val.value
-        if val.value.endswith('*'):
-            val.value += '*'
-    else:
-        logger.warning(
-            f"Unexpected ia_collection_s SearchField value type: {type(val)}"
-        )
-
-
 def has_solr_editions_enabled():
     if 'pytest' in sys.modules:
         return True

openlibrary/plugins/worksearch/subjects.py

@@ -296,7 +296,7 @@ def get_subject(
                 "cover_edition_key",
                 "has_fulltext",
                 "subject",
-                "ia_collection_s",
+                "ia_collection",
                 "public_scan_b",
                 "lending_edition_s",
                 "lending_identifier_s",
@@ -419,7 +419,7 @@ def work_wrapper(w: dict) -> web.storage:
         These docs are weird :/ We should be using more standardized results
         across our search APIs, but that would be a big breaking change.
         """
-        ia_collection = w.get('ia_collection_s', '').split(';')
+        ia_collection = w.get('ia_collection') or []
         return web.storage(
             key=w['key'],
             title=w["title"],

openlibrary/solr/solr_types.py

@@ -129,7 +129,6 @@ class SolrDocument(TypedDict):
     public_scan_b: Optional[bool]
     printdisabled_s: Optional[str]
     lending_edition_s: Optional[str]
-    ia_collection_s: Optional[str]
     ebook_count_i: Optional[int]
 
 # fmt: on

openlibrary/solr/types_generator.py

@@ -8,7 +8,6 @@
     'public_scan_b': 'Optional[bool]',
     'printdisabled_s': 'Optional[str]',
     'lending_edition_s': 'Optional[str]',
-    'ia_collection_s': 'Optional[str]',
     'ebook_count_i': 'Optional[int]',
 }
 

openlibrary/solr/updater/work.py

@@ -500,10 +500,6 @@ def ia(self) -> list[str]:
     def ia_collection(self) -> list[str]:
         return sorted(uniq(c for e in self._solr_editions for c in e.ia_collection))
 
-    @property
-    def ia_collection_s(self) -> str:
-        return ';'.join(self.ia_collection)
-
     @cached_property
     def _ia_editions(self) -> list[EditionSolrBuilder]:
         def get_ia_sorting_key(ed: EditionSolrBuilder) -> tuple[int, str]:

openlibrary/tests/solr/updater/test_work.py

@@ -200,7 +200,7 @@ def test_with_one_lending_edition(self):
         assert d.printdisabled_s is None
         assert d.lending_edition_s == 'OL1M'
         assert d.ia == ['foo00bar']
-        assert sss(d.ia_collection_s) == sss("americana;inlibrary")
+        assert sorted(d.ia_collection) == sorted(["americana", "inlibrary"])
         assert d.edition_count == 1
         assert d.ebook_count_i == 1
 
@@ -222,7 +222,9 @@ def test_with_two_lending_editions(self):
         assert d.printdisabled_s is None
         assert d.lending_edition_s == 'OL1M'
         assert sorted(d.ia) == ['foo01bar', 'foo02bar']
-        assert sss(d.ia_collection_s) == sss("inlibrary;americana;internetarchivebooks")
+        assert sorted(d.ia_collection) == sorted(
+            ["inlibrary", "americana", "internetarchivebooks"]
+        )
         assert d.edition_count == 2
         assert d.ebook_count_i == 2
 
@@ -238,7 +240,7 @@ def test_with_one_inlibrary_edition(self):
         assert d.printdisabled_s == 'OL1M'
         assert d.lending_edition_s == 'OL1M'
         assert d.ia == ['foo00bar']
-        assert sss(d.ia_collection_s) == sss("printdisabled;inlibrary")
+        assert sorted(d.ia_collection) == sorted(["printdisabled", "inlibrary"])
         assert d.edition_count == 1
         assert d.ebook_count_i == 1
 
@@ -254,7 +256,7 @@ def test_with_one_printdisabled_edition(self):
         assert d.printdisabled_s == 'OL1M'
         assert d.lending_edition_s is None
         assert d.ia == ['foo00bar']
-        assert sss(d.ia_collection_s) == sss("printdisabled;americana")
+        assert sorted(d.ia_collection) == sorted(["printdisabled", "americana"])
         assert d.edition_count == 1
         assert d.ebook_count_i == 1
 
@@ -295,7 +297,9 @@ def test_with_multiple_editions(self):
         assert d.printdisabled_s == 'OL4M'
         assert d.lending_edition_s == 'OL2M'
         assert sorted(d.ia) == ['foo00bar', 'foo01bar', 'foo02bar']
-        assert sss(d.ia_collection_s) == sss("americana;inlibrary;printdisabled")
+        assert sorted(d.ia_collection) == sorted(
+            ["americana", "inlibrary", "printdisabled"]
+        )
 
         assert d.edition_count == 4
         assert d.ebook_count_i == 3
@@ -544,4 +548,4 @@ def test_excludes_fav_ia_collections(self):
             },
         )
 
-        assert wsb.ia_collection_s == "americanlibraries;blah"
+        assert sorted(wsb.ia_collection) == sorted(["americanlibraries", "blah"])