Spaces:
Sleeping
Sleeping
Update appStore/prep_data.py
Browse files- appStore/prep_data.py +7 -1
appStore/prep_data.py
CHANGED
@@ -88,7 +88,13 @@ def process_giz_worldwide():
|
|
88 |
|
89 |
# Create an empty 'url' column as the new dataset has an empty URL
|
90 |
giz_df['url'] = ''
|
91 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
92 |
# Compute text_size based on merged_text and assign full text to the 'chunks' column
|
93 |
giz_df['text_size'] = giz_df['merged_text'].apply(lambda text: len(text.split()) if isinstance(text, str) else 0)
|
94 |
|
|
|
88 |
|
89 |
# Create an empty 'url' column as the new dataset has an empty URL
|
90 |
giz_df['url'] = ''
|
91 |
+
|
92 |
+
# Convert the CRS value to integer (if the column exists)
|
93 |
+
if 'crs_value' in giz_df.columns:
|
94 |
+
giz_df['crs_value'] = giz_df['crs_value'].apply(
|
95 |
+
lambda x: int(float(x)) if pd.notnull(x) and str(x).strip() != "" else x
|
96 |
+
)
|
97 |
+
|
98 |
# Compute text_size based on merged_text and assign full text to the 'chunks' column
|
99 |
giz_df['text_size'] = giz_df['merged_text'].apply(lambda text: len(text.split()) if isinstance(text, str) else 0)
|
100 |
|