Do not drop .pdf urls

[GlazerMann]

Includes &uid= on end

Fixes deleting open acces copies

[ms609]

We should include a test case here with an unambiguously necessary URL, so that this does not get removed in future. (I'm still keen to see an example that cannot simply be reached through the DOI link.)

[codecov]

Codecov Report

Merging #704 into master will decrease coverage by 1.42%.
The diff coverage is 100%.

@@             Coverage Diff              @@
##             master     #704      +/-   ##
============================================
- Coverage     69.79%   68.37%   -1.43%     
- Complexity     1624     1672      +48     
============================================
  Files            12       12              
  Lines          3364     3377      +13     
============================================
- Hits           2348     2309      -39     
- Misses         1016     1068      +52

Impacted Files	Coverage Δ	Complexity Δ
Template.php	73.63% <100%> (-2.37%)	1490 <81> (+48)
DOItools.php	93.75% <0%> (+0.09%)	0% <0%> (ø)	⬇️
expandFns.php	90.97% <0%> (+0.69%)	0% <0%> (ø)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 77785b8...90aca29. Read the comment docs.

[ms609]

Do we need to use preg_match here? Would we be better using a stripos and a constant in a suitable file?

[GlazerMann]

I probably should use real urls

[GlazerMann]

I made up the urls since I have yet to find a journal PDF url that ends in .pdf. Thus, the second test does not occur. The real url must: have the doi, end in .pdf, and be from a publisher.

[GlazerMann]

Todo: better URLs

[ms609]

Could it be that a .pdf ending is enough to signify that a link is not from a publisher, after all?

[GlazerMann]

I have improved the user output. I agree that the publisher list is overkill, but we do know that publishers do have pdf files

[GlazerMann]

@ms609 I think this is ready to go.

[ms609]

I still think that without real URLs to use as tests, there's no proven value behind this additional code complexity.

[GlazerMann]

complexity removed @ms609

[ms609]

Could it be that a .pdf ending is enough to signify that a link is not from a publisher, after all?

Against this assertion, here's a real life publisher link ending in .pdf (but not containing a doi...) http://jcs.biologists.org/content/joces/s3-100/49/89.full.pdf

[GlazerMann]

I guess we just accept this pull as it is, and wait for an example of a URL that:

1. contains DOI
2. ends in PDF
3. is a publisher's website
   Once we find one of those, start forgetting those urls

Unneeded code removed

[ms609]

Do we need to use process_citation in the tests, or will prepare_citation do the trick?

Does the free-floating URL duplicate an existing test? Can we in fact do without the first of the three tests, whcih duplicates the second?

[GlazerMann]

prepare_citation() did the trick