add manifest.json

This commit is contained in:
Abel Luck 2026-03-29 13:40:03 +02:00
parent b5a66a2665
commit 81bb8afc41

View file

@ -1,7 +1,31 @@
"""Pygea main entry point"""
import hashlib
import json
import os
from pygea.pangeafeed import PangeaFeed
from pygea.pexception import PangeaServiceException
from pygea import utilities
OUTPUT_TO_FILE = utilities.get_configuration_variable('results', 'output_to_file_p')
OUTPUT_FILE_NAME = utilities.get_configuration_variable('results', 'output_file_name')
OUTPUT_DIRECTORY = utilities.get_configuration_variable('results', 'output_directory')
def write_manifest(categories):
"""Write the category manifest beside the generated feed output."""
if OUTPUT_TO_FILE is not True:
return
output_directory = os.path.normpath(OUTPUT_DIRECTORY)
if not os.path.exists(output_directory):
os.makedirs(output_directory)
manifest_path = os.path.join(output_directory, 'manifest.json')
with open(manifest_path, 'w', encoding='utf-8') as mfile:
json.dump({'categories': categories}, mfile, indent=2, ensure_ascii=False)
mfile.write('\n')
def main():
# Feeds are generated for a single, specified, domain
@ -34,6 +58,7 @@ def main():
# 2. Generate different feeds for each defined category
try:
manifest_categories = []
for cat_tuple in args['categories']:
# form new args for each category/query
newargs = {
@ -46,7 +71,13 @@ def main():
# put each feed into a different sub-directory
feed_subdir = hashlib.md5(cat_tuple[0].encode('utf-8')).hexdigest()[:7]
pf.disgorge(feed_subdir)
manifest_categories.append({
'name': cat_tuple[0],
'short-hash': feed_subdir,
'local-path': os.path.join(feed_subdir, OUTPUT_FILE_NAME).replace(os.sep, '/')
})
print("feed for {} output to sub-directory {}".format(cat_tuple[0], feed_subdir))
write_manifest(manifest_categories)
except PangeaServiceException as error:
print(error)