Spaces:
Sleeping
Sleeping
Update appStore/prep_data.py
Browse files- appStore/prep_data.py +10 -3
appStore/prep_data.py
CHANGED
@@ -102,9 +102,16 @@ def process_giz_worldwide():
|
|
102 |
# Convert CRS value to numeric then to integer if possible.
|
103 |
# After reading the JSON file into giz_df, convert the crs_key column:
|
104 |
if 'crs_key' in giz_df.columns:
|
105 |
-
|
106 |
-
|
107 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
108 |
|
109 |
|
110 |
# Compute text_size based on merged_text and assign full text to the 'chunks' column
|
|
|
102 |
# Convert CRS value to numeric then to integer if possible.
|
103 |
# After reading the JSON file into giz_df, convert the crs_key column:
|
104 |
if 'crs_key' in giz_df.columns:
|
105 |
+
def clean_crs_key(x):
|
106 |
+
if pd.isnull(x):
|
107 |
+
return ""
|
108 |
+
try:
|
109 |
+
# Try converting to float, then to int, then to string.
|
110 |
+
return str(int(float(x)))
|
111 |
+
except Exception:
|
112 |
+
# Fallback: remove trailing ".0" if present.
|
113 |
+
return re.sub(r'\.0$', '', str(x))
|
114 |
+
giz_df['crs_key'] = giz_df['crs_key'].apply(clean_crs_key)
|
115 |
|
116 |
|
117 |
# Compute text_size based on merged_text and assign full text to the 'chunks' column
|