Works around connection reset errors in the test suite by disabling
persistent connections.
20) Error:
Sources::PixivTest#test_: in all cases fetching source data for a new manga image should get the tags. :
Net::HTTP::Persistent::Error: too many connection resets (due to closed stream - IOError) after 0 requests on 47071328584700, last used 1.842702476 seconds ago
app/logical/pixiv_web_agent.rb:46:in `build'
app/logical/sources/strategies/pixiv.rb:104:in `agent'
app/logical/sources/strategies/pixiv.rb:72:in `get'
app/logical/sources/site.rb:6:in `get'
test/unit/sources/pixiv_test.rb:7:in `get_source'
test/unit/sources/pixiv_test.rb:64:in `block (3 levels) in <class:PixivTest>'
ref: github.com/sparklemotion/mechanize/issues/123
ref: http://www.rubydoc.info/gems/mechanize/Mechanize#retry_change_requests%3D-instance_method
e.g. If you were on an html work page on pixiv, clicked a link to a
different html work page on pixiv, and then clicked the bookmarklet,
then it used to fetch the source from the FIRST work you were on instead
of the second.
* Support rewriting source when user uploads from a thumbnail url or
html page url
* Fix bug where site did not log in correctly
* Fix bug where the image url couldn't be extracted from the page if the
image was rated as adults only on seiga
* Normalize direct image url to html page url so tags, etc., can be
extracted
