From b6463f7cc5dec603c3abe55c9edc7c3f4c337a70 Mon Sep 17 00:00:00 2001 From: Kovid Goyal Date: Sat, 16 Jul 2022 17:58:50 +0530 Subject: [PATCH] Also add consent cookie when searching google directly --- .../ebooks/metadata/sources/search_engines.py | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) diff --git a/src/calibre/ebooks/metadata/sources/search_engines.py b/src/calibre/ebooks/metadata/sources/search_engines.py index 22d91f8a5a..9ced08ba85 100644 --- a/src/calibre/ebooks/metadata/sources/search_engines.py +++ b/src/calibre/ebooks/metadata/sources/search_engines.py @@ -20,7 +20,7 @@ from calibre import browser as _browser, prints, random_user_agent from calibre.utils.monotonic import monotonic from calibre.utils.random_ua import accept_header_for_ua -current_version = (1, 0, 10) +current_version = (1, 0, 11) minimum_calibre_version = (2, 80, 0) @@ -284,11 +284,24 @@ def google_search(terms, site=None, br=None, log=prints, safe_search=False, dump url = 'https://www.google.com/search?q={q}'.format(q=q) log('Making google query: ' + url) br = br or browser() + br.set_simple_cookie('CONSENT', 'YES+', '.google.com', path='/') r = [] root = query(br, url, 'google', dump_raw, timeout=timeout, save_raw=r.append) return google_parse_results(root, r[0], log=log), url +def gt(): + url = 'https://www.google.com/search?q=%289780525509998+or+978-0525509998%29+site%3Awww.amazon.com' + raw = browser().open_novisit(url).read() + with open('/t/raw.html', 'wb') as f: + f.write(raw) + root = parse_html(raw) + for result in google_parse_results(root, raw): + print(result.title) + print(' ', result.url) + print(' ', result.cached_url) + + def google_develop(search_terms='1423146786', raw_from=''): if raw_from: with open(raw_from, 'rb') as f: