forked from zaclys/searxng
[mod] google-images: slightly improvements of the engine
Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
This commit is contained in:
parent
df5f8d0e8e
commit
cf7ee67f71
|
@ -1,22 +1,16 @@
|
|||
# SPDX-License-Identifier: AGPL-3.0-or-later
|
||||
# lint: pylint
|
||||
"""This is the implementation of the google images engine using the google internal API used the Google Go Android app.
|
||||
"""This is the implementation of the google images engine using the google
|
||||
internal API used the Google Go Android app.
|
||||
|
||||
This internal API offer results in
|
||||
|
||||
- JSON (_fmt:json)
|
||||
- Protobuf (_fmt:pb)
|
||||
- Protobuf compressed? (_fmt:pc)
|
||||
- HTML (_fmt:html)
|
||||
- Protobuf encoded in JSON (_fmt:jspb).
|
||||
|
||||
.. admonition:: Content-Security-Policy (CSP)
|
||||
|
||||
This engine needs to allow images from the `data URLs`_ (prefixed with the
|
||||
``data:`` scheme)::
|
||||
|
||||
Header set Content-Security-Policy "img-src 'self' data: ;"
|
||||
|
||||
.. _data URLs:
|
||||
https://developer.mozilla.org/en-US/docs/Web/HTTP/Basics_of_HTTP/Data_URIs
|
||||
"""
|
||||
|
||||
from urllib.parse import urlencode
|
||||
|
@ -95,21 +89,34 @@ def response(resp):
|
|||
|
||||
detect_google_sorry(resp)
|
||||
|
||||
response_2nd_line = resp.text.split("\n", 1)[1]
|
||||
json_data = loads(response_2nd_line)["ischj"]
|
||||
json_start = resp.text.find('{"ischj":')
|
||||
json_data = loads(resp.text[json_start:])
|
||||
|
||||
for item in json_data["metadata"]:
|
||||
results.append(
|
||||
{
|
||||
'url': item["result"]["referrer_url"],
|
||||
'title': item["result"]["page_title"],
|
||||
'content': item["text_in_grid"]["snippet"],
|
||||
'source': item["result"]["site_title"],
|
||||
'format': f'{item["original_image"]["width"]} x item["original_image"]["height"]',
|
||||
'img_src': item["original_image"]["url"],
|
||||
'thumbnail_src': item["thumbnail"]["url"],
|
||||
'template': 'images.html',
|
||||
}
|
||||
)
|
||||
for item in json_data["ischj"]["metadata"]:
|
||||
|
||||
result_item = {
|
||||
'url': item["result"]["referrer_url"],
|
||||
'title': item["result"]["page_title"],
|
||||
'content': item["text_in_grid"]["snippet"],
|
||||
'source': item["result"]["site_title"],
|
||||
'img_format': f'{item["original_image"]["width"]} x {item["original_image"]["height"]}',
|
||||
'img_src': item["original_image"]["url"],
|
||||
'thumbnail_src': item["thumbnail"]["url"],
|
||||
'template': 'images.html',
|
||||
}
|
||||
|
||||
author = item["result"].get('iptc', {}).get('creator')
|
||||
if author:
|
||||
result_item['author'] = ', '.join(author)
|
||||
|
||||
copyright_notice = item["result"].get('iptc', {}).get('copyright_notice')
|
||||
if copyright_notice:
|
||||
result_item['source'] += ' / ' + copyright_notice
|
||||
|
||||
file_size = item.get('gsa', {}).get('file_size')
|
||||
if file_size:
|
||||
result_item['source'] += ' (%s)' % file_size
|
||||
|
||||
results.append(result_item)
|
||||
|
||||
return results
|
||||
|
|
Loading…
Reference in New Issue