Fix runtime error

[SanghyunKo]

Fix trivial cmsRun runtime error

[SanghyunKo]

Sorry, nevermind, I see this already in #42

[kdlong]

Thanks! I still need a bit to review dylan's PR before merging, we could just make this change in the meantime?

[kdlong]

@SanghyunKo can you list some samples affected by your latest fix?

[SanghyunKo]

@kdlong Oh gosh... I didn't notice that you re-opened the PR and I added another commit... Forgive me :(

If you allow it, then let me explain here about the additional commit:
Some samples occasionally have extra < or > after the last </weightgroup>, and this crashes entire XML parsing.
Unfortunately, this seems to happen quite often, even in the some of the DY samples e.g. /DYJetsToLL_M-50_TuneCP5_13TeV-amcatnloFXFX-pythia8/RunIIFall17MiniAODv2-PU2017_12Apr2018_94X_mc2017_realistic_v14-v1/MINIAODSIM.

    <weightgroup combine="none" name="PDF_variation">
      <weight id="2079"> PDF=  262000 NNPDF30_lo_as_0118 </weight>
    </weightgroup>
    <weightgroup combine="none" name="PDF_variation">
      <weight id="2080"> PDF=  263000 NNPDF30_lo_as_0130 </weight>
    </weightgroup>
  <

Since LHEEventProduct themselves are OK, this can be simply fixed by deleting extra strings after the last </weightgroup> in the LHERunInfoProduct (indeed any kind of extra strings after the very last ket wouldn't make any sense for XML format)

[kdlong]

No worries! IF you want you can close and open a new one. This looks like a good idea to me. Otherwise the samples fail, right?

Can you collect this together with the other conditions that depend on failIfInvalidXML_ and GetError() != 0 and make the logic a bit cleaner? Right now there are several if statements that could be collected in a more clear/logical order. It might also be worth making these lines you've added into a short function named removeTrailingCharsFromHeader, for example

[SanghyunKo]

Done. Although this may not be the most compact way, it should prevent repeating if (failIfInvalidXML_) over and over again.

[kdlong]

Thanks Sanghyun, this looks much better to me. One quick thing, can you run scram b code-checks and scram b code-format?

[SanghyunKo]

Hi @kdlong, indeed there was a couple of warnings from code format check. Fixed it!

[kdlong]

Thanks!