PR Feedback

smart-on-fhir · Jan 14, 2025 · 137e952 · 137e952
1 parent 470ddb8
commit 137e952
Show file tree

Hide file tree

Showing 4 changed files with 15 additions and 14 deletions.
diff --git a/cumulus_library_umls/.DS_Store b/cumulus_library_umls/.DS_Store
diff --git a/cumulus_library_umls/ancilary_tables.sql b/cumulus_library_umls/ancilary_tables.sql
@@ -269,7 +269,7 @@ SELECT
     str,
     5 AS depth
 FROM code_5
-UNION
+UNION ALL
 SELECT
     rui,
     cui1,
@@ -280,7 +280,7 @@ SELECT
     str,
     6 AS depth
 FROM code_6
-UNION
+UNION ALL
 SELECT
     rui,
     cui1,
@@ -291,7 +291,7 @@ SELECT
     str,
     7 AS depth
 FROM code_7
-UNION
+UNION ALL
 SELECT
     rui,
     cui1,
@@ -314,7 +314,7 @@ SELECT
     str,
     2 AS depth
 FROM umls__icd10_chapter
-UNION
+UNION ALL
 SELECT
     rui,
     cui1,
@@ -325,7 +325,7 @@ SELECT
     str,
     3 AS depth
 FROM umls__icd10_block
-UNION
+UNION ALL
 SELECT
     rui,
     cui1,
@@ -336,7 +336,7 @@ SELECT
     str,
     4 AS depth
 FROM umls__icd10_category
-UNION
+UNION ALL
 SELECT
     rui,
     cui1,

diff --git a/cumulus_library_umls/manifest.toml b/cumulus_library_umls/manifest.toml
@@ -4,7 +4,7 @@ study_prefix = "umls"
 file_names = [
     "umls_builder.py",
     "static_builder.py",
-    "ancilary_tables.sql"
+    "ancilary_tables.sql",
 ]
 
 [advanced_options]

diff --git a/cumulus_library_umls/umls_builder.py b/cumulus_library_umls/umls_builder.py
@@ -127,8 +127,9 @@ def create_parquet(
         :param force_upload: if true, upload to a remote source regardless of what
             already exists there
         """
+        parquet_path = parquet_path / rrf_path.stem
         if not force_upload:
-            if (parquet_path / f"{rrf_path.stem}/{rrf_path.stem}.parquet").exists():
+            if (parquet_path / f"{rrf_path.stem}.parquet").exists():
                 return
         df = pandas.read_csv(
             rrf_path,
@@ -137,8 +138,8 @@ def create_parquet(
             dtype=table["dtype"],
             index_col=False,
         )
-        (parquet_path / f"{rrf_path.stem}").mkdir(parents=True, exist_ok=True)
-        df.to_parquet(parquet_path / f"{rrf_path.stem}/{rrf_path.stem}.parquet")
+        parquet_path.mkdir(parents=True, exist_ok=True)
+        df.to_parquet(parquet_path / f"{rrf_path.stem}.parquet")
 
     def prepare_queries(
         self,
@@ -151,10 +152,10 @@ def prepare_queries(
         download_path.mkdir(exist_ok=True, parents=True)
         parquet_path = pathlib.Path(__file__).resolve().parent / "generated_parquet"
         parquet_path.mkdir(exist_ok=True, parents=True)
-        files, new_version, folder = self.get_umls_data(
+        files, new_version, umls_version = self.get_umls_data(
             download_path, parquet_path, config.force_upload, config.umls_key
         )
-        parquet_path = parquet_path / folder
+        parquet_path = parquet_path / umls_version
         parquet_path.mkdir(exist_ok=True, parents=True)
 
         with base_utils.get_progress_bar() as progress:
@@ -166,7 +167,7 @@ def prepare_queries(
                 with open(file) as f:
                     datasource, table = self.parse_ctl_file(f.readlines())
                     progress.update(task, description=f"Compressing {datasource}...")
-                    rrf_path = download_path / f"./{folder}/META/{datasource}"
+                    rrf_path = download_path / f"./{umls_version}/META/{datasource}"
                     self.create_parquet(
                         rrf_path, parquet_path, table, force_upload=config.force_upload
                     )
@@ -192,5 +193,5 @@ def prepare_queries(
                     )
                     progress.advance(task)
             log_utils.log_transaction(
-                config=config, manifest=manifest, message=f"UMLS version: {folder}"
+                config=config, manifest=manifest, message=f"UMLS version: {umls_version}"
             )