annikwag commited on
Commit
b5da228
·
verified ·
1 Parent(s): cebff59

Update appStore/prep_data.py

Browse files
Files changed (1) hide show
  1. appStore/prep_data.py +3 -4
appStore/prep_data.py CHANGED
@@ -100,10 +100,9 @@ def process_giz_worldwide():
100
  giz_df['url'] = ''
101
 
102
  # Convert CRS value to numeric then to integer if possible.
103
- if 'crs_key' in giz_df.columns:
104
- # Convert to numeric, coercing errors to NaN, then fill NaN with 0 and cast to int.
105
- giz_df['crs_key'] = pd.to_numeric(giz_df['crs_key'], errors='coerce').fillna(0).astype(int)
106
-
107
  # Compute text_size based on merged_text and assign full text to the 'chunks' column
108
  giz_df['text_size'] = giz_df['merged_text'].apply(lambda text: len(text.split()) if isinstance(text, str) else 0)
109
 
 
100
  giz_df['url'] = ''
101
 
102
  # Convert CRS value to numeric then to integer if possible.
103
+ if 'crs_key' in df.columns:
104
+ df['crs_key'] = df['crs_key'].apply(lambda x: str(int(float(x))) if pd.notnull(x) and str(x).strip() != "" else x)
105
+
 
106
  # Compute text_size based on merged_text and assign full text to the 'chunks' column
107
  giz_df['text_size'] = giz_df['merged_text'].apply(lambda text: len(text.split()) if isinstance(text, str) else 0)
108