Fix broken links

[pfheatwole]

Fixes three broken links in the docs (one is just a broken anchor on Wikipedia, but still)

Cheers, and thanks for the tool!

[nicoddemus]

Many thanks!

[coveralls]

Coverage remained the same at 92.816% when pulling ef90346 on gogoengie:broken-links into 78f0388 on pytest-dev:master.

[The-Compiler]

How did you find out about those? Is this something we could integrate in our CI?

[pfheatwole]

I encountered thetox broken link while reading "good practices". Once I'd opened the branch I figured I'd look around for other (subjectively) important links. I'm short on time, so I triaged with this:

$ cd pytest/doc/en
$ make linkcheck
$ grep broken _build/linkcheck/output.txt | cut -d: -f1 | uniq

In #1722 you discussed this method, but decided you'd like to avoid all those requests for every commit. (Reasonable, but I wasn't worried for my one-off run.) If you're looking for a compromise, you could check only those files affected by the commit set. For example, suppose something like:

$ cd pytest/doc/en
$ git diff --name-only HEAD~   # Suppose I deliberately broke links in these files
doc/en/goodpractices.rst
doc/en/assert.rst
doc/en/links.inc
$ git diff --name-only HEAD~ | cut -d/ -f3   # Use these as input to sphinx-build
goodpractices.rst
assert.rst
links.inc
$ sphinx-build -b linkcheck -d _build/doctrees -D latex_paper_size=letter . _build/linkcheck $(git diff --name-only HEAD~ | cut -d/ -f3)
Running Sphinx v1.5
loading pickled environment... done
building [mo]: targets for 0 po files that are specified
building [linkcheck]: 3 source files given on command line
updating environment: 0 added, 7 changed, 0 removed
reading sources... [100%] usage                                                 
links.inc:17: WARNING: Duplicate explicit target name: "virtualenv".
/home/peter/scratch/pytest/doc/en/goodpractices.rst:142: ERROR: Duplicate target name, cannot be used as a unique reference: "virtualenv".
looking for now-outdated files... none found
pickling environment... done
checking consistency... done
preparing documents... done
writing output... [ 33%] assert                                                 
(line  310) broken    http://example.com/blargh - 404 Client Error: Not Found for url: http://example.com/blargh
writing output... [ 66%] contents                                               
(line    6) ok        https://media.readthedocs.org/pdf/pytest/latest/pytest.pdf
writing output... [100%] goodpractices                                          
(line  157) ok        https://wiki.jenkins-ci.org/display/JENKINS/xUnit+Plugin
(line  147) broken    https://example.com/blargh - 404 Client Error: Not Found for url: https://example.com/blargh
(line  157) redirect  http://jenkins-ci.org/ - permanently to https://jenkins.io/index.html
(line  166) ok        https://pypi.python.org/pypi/pytest-runner
(line  147) redirect  http://testrun.org/tox - with Found to https://tox.readthedocs.io/en/latest/
(line  142) redirect  http://pypi.python.org/pypi/pip - permanently to https://pypi.python.org/pypi/pip

build finished with problems, 2 warnings.

$ grep broken _build/linkcheck/output.txt | cut -d: -f1 | uniq
assert.rst
goodpractices.rst
$  # Notice, it missed breakage in `links.inc`

Note, this example is intentionally incomplete: the last commit changed doc/en/links.inc, but sphinx-build seems to ignore non-output-targets in the list of filenames. Still, it's an idea. :-)

Hope that helps!