rework wikidata and add primary source helpers

mex/common/primary_source/helpers.py

@@ -0,0 +1,43 @@
+from functools import cache
+
+from mex.common.models import (
+    ExtractedPrimarySource,
+)
+from mex.common.primary_source.extract import extract_seed_primary_sources
+from mex.common.primary_source.transform import (
+    transform_seed_primary_sources_to_extracted_primary_sources,
+)
+
+
+@cache
+def get_all_extracted_primary_sources() -> list[ExtractedPrimarySource]:
+    """Extract and transform all primary sources.
+
+    Extract the primary sources from the raw-data JSON file and transform them into
+    a list of ExtractedPrimarySources.
+
+    Returns:
+        List of all ExtractedPrimarySources
+    """
+    seed_primary_sources = extract_seed_primary_sources()
+    return list(
+        transform_seed_primary_sources_to_extracted_primary_sources(
+            seed_primary_sources
+        )
+    )
+
+
+@cache
+def get_extracted_primary_source_by_name(name: str) -> ExtractedPrimarySource | None:
+    """Pick the extracted primary source with the given name and return it.
+
+    Args:
+        name: Name (`identifierInPrimarySource`) of the primary source
+
+    Returns:
+        Extracted primary source if it was found, else None
+    """
+    primary_sources_by_name = {
+        p.identifierInPrimarySource: p for p in get_all_extracted_primary_sources()
+    }
+    return primary_sources_by_name.get(name)

mex/common/primary_source/transform.py

@@ -28,6 +28,7 @@ def transform_seed_primary_sources_to_extracted_primary_sources(
         )
 
 
+# TODO: Remove this in MX-1698
 def get_primary_sources_by_name(
     extracted_primary_sources: Iterable[ExtractedPrimarySource], *names: str
 ) -> tuple[ExtractedPrimarySource, ...]:

mex/common/wikidata/helpers.py

@@ -0,0 +1,40 @@
+from functools import cache
+
+from mex.common.models import ExtractedOrganization, ExtractedPrimarySource
+from mex.common.wikidata.extract import search_organization_by_label
+from mex.common.wikidata.transform import (
+    transform_wikidata_organization_to_extracted_organization,
+)
+
+
+@cache
+def get_extracted_organization_from_wikidata(
+    query_string: str,
+    wikidata_primary_source: ExtractedPrimarySource,
+) -> ExtractedOrganization | None:
+    """Get extracted organization matching the query string.
+
+    Search wikidata for organization and transform it into an ExtractedOrganization.
+
+    Args:
+        query_string: query string to search in wikidata
+        wikidata_primary_source: wikidata primary source
+
+    Returns:
+        ExtractedOrganization if one matching organization is found in
+           Wikidata lookup.
+        None if multiple matches / no organization is found.
+    """
+    found_organization = search_organization_by_label(query_string)
+
+    if found_organization is None:
+        return None
+
+    extracted_organization = transform_wikidata_organization_to_extracted_organization(
+        found_organization, wikidata_primary_source
+    )
+
+    if extracted_organization is None:
+        return None
+
+    return extracted_organization