fixed .gpkg pipeline

Author: amyfromandi

py-modules/map-integration/macrostrat/map_integration/commands/ingest.py

@@ -75,11 +75,15 @@ def preprocess_dataframe(
         meta_df = P.read_excel(meta_path)
         # TODO xls pipeline for if feature_suffix == "polygons", "lines" OR "points"
     elif pipeline == ".gpkg":
-        meta_df = map_t_b_standard(poly_line_pt_df, "epoch", "period")
         ingest_pipeline = ".gpkg pipeline"
         state = "needs review"
         comments = ""
-        return meta_df, ingest_pipeline, comments, state
+        if feature_suffix == "polygons":
+            meta_df = map_t_b_standard(poly_line_pt_df, "epoch", "period")
+            return meta_df, ingest_pipeline, comments, state
+        else:
+            return poly_line_pt_df, ingest_pipeline, comments, state
+
     elif pipeline == ".gdb":
         if feature_suffix == "polygons":
             join_col = "mapunit"

py-modules/map-integration/macrostrat/map_integration/utils/ingestion_utils.py

@@ -147,6 +147,20 @@ def map_t_b_standard(meta_df: G.GeoDataFrame, col_one: str, col_two: str) -> G.G
             meta_df.loc[needs_fill, "b_interval"] = mapped_col_two
             meta_df.loc[needs_fill, "t_interval"] = mapped_col_two
 
+    if "era" in meta_df.columns:
+        needs_fill = meta_df["b_interval"].isna() | meta_df["t_interval"].isna()
+        if needs_fill.any():
+            mapped_col_three = (
+                meta_df.loc[needs_fill, "era"]
+                .astype("string")
+                .str.strip()
+                .str.lower()
+                .replace("", pd.NA)
+                .map(interval_lookup)
+            )
+            meta_df.loc[needs_fill, "b_interval"] = mapped_col_three
+            meta_df.loc[needs_fill, "t_interval"] = mapped_col_three
+
     return meta_df