Fix issue with shortcut recompression

[MicahGale]

Description

Bug fix #499

This fixes the bug from #499. This bug came down to a bug when two repeat shortcuts are touching. The second repeat assumes that the repeat node to its left is the value this one is repeating. To solve this the logic is updated to see if these edge values match.

Test Improvement #497

In the process I implemented testing where all valid input files are readin, "written" out and then re-read in, and the original and new files are compared. This found various padding bugs.

Bug fix: #489

For the bug: #489, interpolations (and all shortcuts) in geometry definitions were not recompressed. This was due to the fact that the GeometryTree had no idea that a shortcut was involved. This will be fixed by:

1. Making it possible to have an interpolation of integer values only.
2. Labeling GeometryTree accordingly when it comes from a shortcut
3. Creating a method to recompress a shortcut in a GeoemtryTree.

Bug Fix #526

This is actually #352 in disguise.

Bug fix #352

This was due to the trailing comments not being found properly due to a few reasons. In #352 the issue was due to the default trees for cell modifiers being added. This makes it look like there were no trailing comments.

This case was fixed by flagging default tree items in the parameters, and ignoring them for transferring comments.

The other case found was not being transferred from a tally comment. In this case the padding was attached to the classifier. The get_trailing_comment logic had to be updated to allow falling through to that very first node in the tree. This issue was in part causing #526 due to the new line additions due to the comment being in the wrong object.

• Fixes Shortcuts recompressed incorrectly #499
• Fixes Add support for reading from non-file steams #519
• Fixes Test writing all valid MCNP input files out #497
• fixes Botched Integer Interpolation Expansion #489
• fixes Extra blank lines are sometimes added on write_problem #526
• Fixes Trailing comments aren't being properly transferred #352.

Checklist

• I have performed a self-review of my own code
• I have made corresponding changes to the documentation (if applicable)
• I have added tests that prove my fix is effective or that my feature works (if applicable)
• Update unit testing to be more thurough
• Cover logic case of Jump Repeat
• Implement more unit tests for geometry shortcuts.

[MicahGale]

Not actually replicating #499 yet.

[tjlaboss]

The first expansion is ok. It seems to occur on the $N^\text{th}$ expansion on a line where $N>1$.

[MicahGale]

Awww. I see. Thanks.

[MicahGale]

This looks more correctly failing:

>               assert new_line == gold_line.rstrip().expandtabs(8)
E               AssertionError: assert 'imp:e   0 2r 2r' == 'imp:e   0 2r 1 r'
E
E                 - imp:e   0 2r 1 r
E                 ?              ^^
E                 + imp:e   0 2r 2r
E                 ?              ^

[MicahGale]

I accidentally stumbled upon #489 in this testing:

E               AssertionError: assert '10214   0    (1  2.03.04)' == '10214   0    (1  2i  4)'
E
E                 - 10214   0    (1  2i  4)
E                 ?                   ^^^
E                 + 10214   0    (1  2.03.04)
E                 ?                   ^^^^^

(this is for tests/inputs/test_interp_edge.imcnp).

[tjlaboss]

Try reading and rewriting this input: #520 (comment)

I'm using commit 843db18

[MicahGale]

Ohh I see it now, I focused on the wrong problem there.

[MicahGale]

Ok your bug with regards to Importance seems to be solved.

This had to be manually addressed because importance is a special snow-flake and has its own system for handling syntax trees.

[MicahGale]

And... it broke again.