Merge pull request internetarchive#11180 from cdrini/11144/feature/solr-provider-field

Add ebook_provider field to solr

Author: mekarpeles

conf/solr/conf/managed-schema.xml

@@ -160,6 +160,7 @@
     <field name="oclc" type="text_en_splitting" multiValued="true"/>
     <field name="isbn" type="string" multiValued="true"/>
     <field name="ebook_access" type="ebookAccessLevel" multiValued="false"/>
+    <field name="ebook_provider" type="string" multiValued="true"/>
 
     <!-- Reading Levels -->
     <field name="lexile" type="pint" multiValued="true"/>

openlibrary/book_providers.py

@@ -1,4 +1,5 @@
 import logging
+import typing
 from collections.abc import Callable, Iterator
 from dataclasses import dataclass
 from typing import Literal, TypedDict, TypeVar, cast, override
@@ -157,6 +158,10 @@ class AbstractBookProvider[TProviderMetadata]:
     """
     identifier_key: str | None
 
+    @property
+    def provider_name(self) -> str:
+        return self.identifier_key or self.short_name
+
     def get_olids(self, identifier: str) -> list[str]:
         return web.ctx.site.things(
             {"type": "/type/edition", self.db_selector: identifier}
@@ -272,6 +277,11 @@ class InternetArchiveProvider(AbstractBookProvider[IALiteMetadata]):
     long_name = 'Internet Archive'
     identifier_key = 'ocaid'
 
+    @property
+    @typing.override
+    def provider_name(self) -> str:
+        return 'ia'
+
     @property
     def db_selector(self) -> str:
         return self.identifier_key
@@ -648,8 +658,14 @@ def is_non_ia_ocaid(ocaid: str) -> bool:
     return any(provider.is_own_ocaid(ocaid) for provider in providers)
 
 
-def get_book_provider_by_name(short_name: str) -> AbstractBookProvider | None:
-    return next((p for p in PROVIDER_ORDER if p.short_name == short_name), None)
+def get_book_provider_by_name(name: str) -> AbstractBookProvider | None:
+    """
+    :param name: Either the provider's short_name (legacy) or its provider_name
+    """
+    return next(
+        (p for p in PROVIDER_ORDER if name in (p.provider_name, p.short_name)),
+        None,
+    )
 
 
 ia_provider = cast(InternetArchiveProvider, get_book_provider_by_name('ia'))

openlibrary/plugins/worksearch/schemes/works.py

@@ -51,6 +51,7 @@ class WorkSearchScheme(SearchScheme):
             "alternative_subtitle",
             "cover_i",
             "ebook_access",
+            "ebook_provider",
             "edition_count",
             "edition_key",
             "format",

openlibrary/solr/solr_types.py

@@ -35,6 +35,7 @@ class SolrDocument(TypedDict):
     oclc: Optional[list[str]]
     isbn: Optional[list[str]]
     ebook_access: Optional[Literal['no_ebook', 'unclassified', 'printdisabled', 'borrowable', 'public']]
+    ebook_provider: Optional[list[str]]
     lexile: Optional[list[int]]
     lcc: Optional[list[str]]
     lcc_sort: Optional[str]

openlibrary/solr/updater/edition.py

@@ -104,7 +104,8 @@ def __init__(
         self._edition = edition
         self._solr_work = solr_work
         self._ia_metadata = ia_metadata
-        self._provider = bp.get_book_provider(edition)
+        self._providers = list(bp.get_book_providers(edition))
+        self._best_provider = self._providers[0] if self._providers else None
 
     @property
     def key(self):
@@ -273,12 +274,16 @@ def identifiers(self) -> dict:
 
     @cached_property
     def ebook_access(self) -> bp.EbookAccess:
-        if not self._provider:
+        if not self._best_provider:
             return bp.EbookAccess.NO_EBOOK
-        elif isinstance(self._provider, bp.InternetArchiveProvider):
-            return self._provider.get_access(self._edition, self._ia_metadata)
+        elif isinstance(self._best_provider, bp.InternetArchiveProvider):
+            return self._best_provider.get_access(self._edition, self._ia_metadata)
         else:
-            return self._provider.get_access(self._edition)
+            return self._best_provider.get_access(self._edition)
+
+    @property
+    def ebook_provider(self) -> list[str]:
+        return [provider.provider_name for provider in (self._providers or [])]
 
     @property
     def has_fulltext(self) -> bool:
@@ -335,6 +340,7 @@ def build(self) -> SolrDocument:
                 'ia_box_id': self.ia_box_id,
                 # Ebook access
                 'ebook_access': self.ebook_access.to_solr_str(),
+                'ebook_provider': self.ebook_provider,
                 'has_fulltext': self.has_fulltext,
                 'public_scan_b': self.public_scan_b,
             },

openlibrary/solr/updater/work.py

@@ -474,6 +474,12 @@ def ebook_access(self) -> bp.EbookAccess:
             default=bp.EbookAccess.NO_EBOOK,
         )
 
+    @property
+    def ebook_provider(self) -> list[str]:
+        return uniq(
+            provider for e in self._solr_editions for provider in e.ebook_provider
+        )
+
     @property
     def has_fulltext(self) -> bool:
         return any(e.has_fulltext for e in self._solr_editions)