Scrapy 1.5.0:
This release brings small new features and improvements across the codebase.
Some highlights:
* Google Cloud Storage is supported in FilesPipeline and ImagesPipeline.
* Crawling with proxy servers becomes more efficient, as connections
to proxies can be reused now.
* Warnings, exception and logging messages are improved to make debugging
easier.
* scrapy parse command now allows to set custom request meta via
--meta argument.
* Compatibility with Python 3.6, PyPy and PyPy3 is improved;
PyPy and PyPy3 are now supported officially, by running tests on CI.
* Better default handling of HTTP 308, 522 and 524 status codes.
* Documentation is improved, as usual.
Backwards Incompatible Changes
* Scrapy 1.5 drops support for Python 3.3.
* Default Scrapy User-Agent now uses https link to scrapy.org.
**This is technically backwards-incompatible**; override
:setting:USER_AGENT if you relied on old value.
* Logging of settings overridden by custom_settings is fixed;
**this is technically backwards-incompatible** because the logger
changes from [scrapy.utils.log] to [scrapy.crawler]. If you're
parsing Scrapy logs, please update your log parsers.
* LinkExtractor now ignores m4v extension by default, this is change
in behavior.
* 522 and 524 status codes are added to RETRY_HTTP_CODES
New features
- Support <link> tags in Response.follow
- Support for ptpython REPL
- Google Cloud Storage support for FilesPipeline and ImagesPipeline
- New --meta option of the "scrapy parse" command allows to pass additional
request.meta
- Populate spider variable when using shell.inspect_response
- Handle HTTP 308 Permanent Redirect
- Add 522 and 524 to RETRY_HTTP_CODES
- Log versions information at startup
- scrapy.mail.MailSender now works in Python 3 (it requires Twisted 17.9.0)
- Connections to proxy servers are reused
- Add template for a downloader middleware
- Explicit message for NotImplementedError when parse callback not defined
- CrawlerProcess got an option to disable installation of root log handler
- LinkExtractor now ignores m4v extension by default
- Better log messages for responses over :setting:DOWNLOAD_WARNSIZE and
:setting:DOWNLOAD_MAXSIZE limits
- Show warning when a URL is put to Spider.allowed_domains instead of
a domain.
Bug fixes
- Fix logging of settings overridden by custom_settings;
**this is technically backwards-incompatible** because the logger
changes from [scrapy.utils.log] to [scrapy.crawler], so please
update your log parsers if needed
- Default Scrapy User-Agent now uses https link to scrapy.org.
**This is technically backwards-incompatible**; override
:setting:USER_AGENT if you relied on old value.
- Fix PyPy and PyPy3 test failures, support them officially
- Fix DNS resolver when DNSCACHE_ENABLED=False
- Add cryptography for Debian Jessie tox test env
- Add verification to check if Request callback is callable
- Port extras/qpsclient.py to Python 3
- Use getfullargspec under the scenes for Python 3 to stop DeprecationWarning
- Update deprecated test aliases
- Fix SitemapSpider support for alternate links
33769cc03b