amber TI parser (#10)

[orbeckst]

This PR supercedes PR #32: rebased and cleaned up history.

NOTE: This PR must be merged (not rebased, otherwise author information is lost)

[orbeckst]

I will look at this later in the afternoon. Hopefully the tests will still pass.

[orbeckst]

@shuail please review when you have time.

[codecov-io]

Codecov Report

Merging #36 into master will increase coverage by 2.47%.
The diff coverage is 98.24%.

@@            Coverage Diff             @@
##           master      #36      +/-   ##
==========================================
+ Coverage   95.12%   97.59%   +2.47%     
==========================================
  Files           8        9       +1     
  Lines         205      374     +169     
  Branches       28       59      +31     
==========================================
+ Hits          195      365     +170     
+ Misses          5        4       -1     
  Partials        5        5

Impacted Files	Coverage Δ
src/alchemlyb/parsing/gmx.py	96.25% <ø> (+2.34%)	⬆️
src/alchemlyb/parsing/amber.py	98.24% <98.24%> (ø)
src/alchemlyb/parsing/util.py	100% <0%> (+11.11%)	⬆️

---

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 e018fdb...234edc0. Read the comment docs.

[shuail]

@orbeckst thanks a lot! I have to run into a flight, will review this later today.

[dotsdl]

I pulled the amber tests for TI into the existing test module for less code duplication. The effect is identical as before. Thanks for all your work on this, @shuail and @orbeckst!

[davidlmobley]

Just spectating here. Nice job everyone -- great to see the process working. :)

[orbeckst]

I am trying out pragma: no cover to get the 95% target. The justification is to be explicit about what you exclude from test coverage. I haven't done this that aggressively before but it seems a good idea, and with @dotsdl 's great initial coverage we can be very clear about which code paths are tested and which ones are not.

(Of course, the tests themselves still have to make sense but once you read the source and see that

                # FIXME: assumes fields are only integers or floats
                if '*' in value:            # Fortran format overflow
                    result.append(float('Inf') )

is not covered (and not excluded from coverage) then you get an idea where potential pain points are.)

[orbeckst]

I cleaned up the history and force-pushed, sorry for any inconvenience.

When you git pull you should see a forced update. If not, make sure that you have no changes, then git reset --hard master; git pull.

EDIT: Just to be clear: git reset --hard master will wipe out any changes in your branch. If you have uncommitted changes, create a new branch for them or use git stash.

[orbeckst]

@dotsdl feel free to review but please don't merge until @shuail got a chance to review. I want to make sure that his code goes in properly.

@shuail if you and @dotsdl both approve you can merge the PR as is. (Do not rebase or squash because (1) this commit history has meaningful smaller commits and (2) rebasing is going to mess up the author attributions on any commit that has already gone through a rebase.) Thanks!

Excellent work!

[shuail]

All looks good to me! Thanks @orbeckst for cleaning this up and merging all the commits here, I like this version better.

[orbeckst]

Feel free to merge ("Merge pull request"). You can then delete this branch.