NickKalar/LMS-DB-ETL
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

cleaned up book entry declaration

Browse Source
This commit is contained in:
Nick Kalar
2025-08-09 21:23:28 -04:00
parent 24345c8b4e
commit 425e27e189
1 changed files with 36 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
src/transform.py
View File

@@ -56,33 +56,45 @@ def combine_raw_jsons(google_json, ol_json):
if 'categories' in google_json['book_data'][index]['volumeInfo']: if 'categories' in google_json['book_data'][index]['volumeInfo']:
categories = ', '.join(google_json['book_data'][index]['volumeInfo']['categories']) categories = ', '.join(google_json['book_data'][index]['volumeInfo']['categories'])
else: else:
categories = None categories = ''
if 'publisher' in google_json['book_data'][index]['volumeInfo']: if 'publisher' in google_json['book_data'][index]['volumeInfo']:
publisher = str(google_json['book_data'][index]['volumeInfo']['publisher']).translate(replace_quote) publisher = str(google_json['book_data'][index]['volumeInfo']['publisher']).translate(replace_quote)
else: else:
publisher = None publisher = ''
if 'publishedDate' in google_json['book_data'][index]['volumeInfo']: if 'publishedDate' in google_json['book_data'][index]['volumeInfo']:
published_date = google_json['book_data'][index]['volumeInfo']['publishedDate'] published_date = google_json['book_data'][index]['volumeInfo']['publishedDate']
if len(published_date) == 4:
published_date += '-12-31'
elif len(published_date) < 10:
published_date = published_date[0:3] + '-12-31'
else: else:
published_date = None published_date = '9999-12-31'
lost_date = '9999-12-31'
if 'printType' in google_json['book_data'][index]['volumeInfo']: if 'printType' in google_json['book_data'][index]['volumeInfo']:
print_type = google_json['book_data'][index]['volumeInfo']['printType'] print_type = google_json['book_data'][index]['volumeInfo']['printType']
else: else:
print_type = None print_type = ''
if 'language' in google_json['book_data'][index]['volumeInfo']: if 'language' in google_json['book_data'][index]['volumeInfo']:
language = google_json['book_data'][index]['volumeInfo']['language'] language = google_json['book_data'][index]['volumeInfo']['language']
else: else:
language = None language = ''
if 'pageCount' in google_json['book_data'][index]['volumeInfo']: if 'pageCount' in google_json['book_data'][index]['volumeInfo']:
pageCount = google_json['book_data'][index]['volumeInfo']['pageCount'] pageCount = google_json['book_data'][index]['volumeInfo']['pageCount']
else: else:
pageCount = 0 pageCount = 0
loc_number = ol_json['book_data'][index]['loc_number']
dewey_decimal_number = ol_json['book_data'][index]['dewey_decimal_number']
description = ol_json['book_data'][index]['description']
price_in_cents = ol_json['book_data'][index]['price_in_cents']
cover_image_uri = ol_json['book_data'][index]['cover_image_uri']
transformed_dictionary_entry = { transformed_dictionary_entry = {
'title': title, 'title': title,
'author': author, 'author': author,
@@ -97,6 +109,12 @@ def combine_raw_jsons(google_json, ol_json):
'is_checked_in': True, 'is_checked_in': True,
'is_archived': False, 'is_archived': False,
'is_lost': False, 'is_lost': False,
'lost_date': lost_date,
'loc_number': loc_number,
'dewey_decimal_number': dewey_decimal_number,
'description': description,
'price_in_cents': price_in_cents,
'cover_image_uri': cover_image_uri,
} }
transformed_dictionary['books'].append(transformed_dictionary_entry) transformed_dictionary['books'].append(transformed_dictionary_entry)