feat: new script seem to be super efficient
This commit is contained in:
46
new_write.py
Normal file
46
new_write.py
Normal file
@@ -0,0 +1,46 @@
|
|||||||
|
import json
|
||||||
|
import pandas as pd
|
||||||
|
|
||||||
|
# Step 1: Load the JSON data
|
||||||
|
with open('CIK0000320193.json') as f:
|
||||||
|
data = json.load(f)
|
||||||
|
|
||||||
|
# Step 2: Extract the relevant fields
|
||||||
|
cik = data.get('cik')
|
||||||
|
entity_name = data.get('entityName')
|
||||||
|
facts = data.get('facts')
|
||||||
|
|
||||||
|
# Prepare a list to hold the rows
|
||||||
|
rows = []
|
||||||
|
|
||||||
|
# Traverse through the facts dictionary
|
||||||
|
for taxonomy, fact_items in facts.items():
|
||||||
|
for fact_id, fact_data in fact_items.items():
|
||||||
|
label = fact_data.get('label')
|
||||||
|
description = fact_data.get('description')
|
||||||
|
units = fact_data.get('units')
|
||||||
|
|
||||||
|
# For each unit and its details, add a new row
|
||||||
|
for unit, details_list in units.items():
|
||||||
|
for details in details_list:
|
||||||
|
row = {
|
||||||
|
'entity_cik': cik,
|
||||||
|
'entity_name': entity_name,
|
||||||
|
'fact_id': fact_id,
|
||||||
|
'fact_taxonomy': taxonomy,
|
||||||
|
'fact_label': label,
|
||||||
|
'fact_description': description,
|
||||||
|
'fact_unit': unit
|
||||||
|
}
|
||||||
|
|
||||||
|
# Include the additional details in the row
|
||||||
|
row.update(details)
|
||||||
|
|
||||||
|
# Append the row to the rows list
|
||||||
|
rows.append(row)
|
||||||
|
|
||||||
|
# Step 3: Create the DataFrame
|
||||||
|
df = pd.DataFrame(rows)
|
||||||
|
|
||||||
|
# Step 4: Output the head of the DataFrame
|
||||||
|
print(df.head())
|
||||||
Reference in New Issue
Block a user