Readd city term processing and don't add duplicate xref edges

nanglo123 · Oct 1, 2024 · 7346375 · 7346375
1 parent 30875a7
commit 7346375
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 3 deletions.
diff --git a/mira/dkg/construct.py b/mira/dkg/construct.py
@@ -1112,8 +1112,7 @@ def _get_edge_name(curie_: str, strict: bool = False) -> str:
  if xref.value.prefix in obograph.PROVENANCE_PREFIXES:
  # Don't add provenance information as xrefs
  continue
- edges.append(
- (
+ xref_edge_info = (
  node.curie,
  xref.value.curie,
  "xref",
@@ -1122,7 +1121,8 @@ def _get_edge_name(curie_: str, strict: bool = False) -> str:
  graph_id,
  version or "",
  )
- )
+ if xref_edge_info not in edges:
+ edges.append(xref_edge_info)
  if xref.value.curie not in nodes:
  node_sources[node.replaced_by].add(prefix)
  nodes[xref.value.curie] = NodeInfo(

diff --git a/mira/dkg/resources/geonames.py b/mira/dkg/resources/geonames.py
@@ -122,6 +122,7 @@ def get_cities(code_to_country, code_to_admin1, code_to_admin2, *, minimum_popul
  cols = ["geonames_id", "name", "synonyms", "country_code", "admin1",
  "admin2", "population"]
  for identifier, name, synonyms, country, admin1, admin2, population in (cities_df[cols].values):
+ terms[identifier] = term = Term.from_triple("geonames", identifier,name)
  if synonyms and not isinstance(synonyms, float):
  for synoynm in synonyms:
  term.append_synonym(synoynm)