cffbots / filtering Goto Github PK

Almost every function splits the URL into owner and repo.
It is probably easier to just store only the owner and repo directly.

If the number of resulting repos is too large, sorting may be necessary. This is a multicriteria problem, so we need to handle it with care, instead of simply piping sorting functions.

Criteria:

• Number of stars
• Number of contributors
• Number of updates to CITATION.cff (#10)

Maybe things to include are:

• who is making the issue / where we are coming from?
• what to do next?
• what if they don't feel like following up / accepting our suggestion?
• where can they leave feedback?
• include notification/mention of other team members. How can we get a clean list of notifications, only the ones that have something to do with these PRs/issues? (Maybe by including @nlescgt?)

Might be that certain repos prefer that issues are created in a certain way, using certain formats. (How) do we try to stick to those rules?

(Spawned from #1)

Abel

• citation-file-format/cff-converter-python citation-file-format/cffconvert#270
• citation-file-format/cff-initializer-javascript citation-file-format/cff-initializer-javascript#450
• citation-file-format/citation-file-format citation-file-format/citation-file-format#368
• CLARIAH/grlc - CLARIAH/grlc#368

Faruk

• eWaterCycle/ewatercycle
  • eWaterCycle/ewatercycle#293
• fair-workflows/nanopub
  • fair-workflows/nanopub#146
• NLeSC/mcfly
  • NLeSC/mcfly#279
• phenology/cgc
  • phenology/cgc#77

Jurriaan

• research-software-directory/research-software-directory
  • research-software-directory/research-software-directory#818
• wadpac/GGIR
  • wadpac/GGIR#506
• xenon-middleware/xenon
  • There are no workflows on main, but there is already an issue on adding cffconvert: xenon-middleware/xenon#686
  • xenon-middleware/xenon#687

add a simple package.json to install the dependencies. Currently the only dependency is octokit.

• some of the filters would need to talk to the github API, so probably something from this list: https://docs.github.com/en/rest/overview/libraries would be convenient
• Python is probably convenient for combining with cffconvert validator
• node would be good (but maybe not necesssary per se) if we want to run this in a scheduled GitHub Action workflow, for example if we're interested in collecting statistics over time.

(Spawned from #1)

(Spawned from #1)

Make the condition parameterizable like the other filters.

(Spawned from #1)

Should be urls instead of whitelist.

(Spawned from #1)

For example,

• the filtering steps (can they all be automated?)
• forking of any qualifying repositories to this organization
• submitting issues in repositories that qualify

(Spawned from #1)

For example, intersect the list of all repos-that-have-a-CITATION.cff (1) with repos that occur in research-software.nl/api/software (2)

1. https://github.com/search?l=&q=filename%3ACITATION.cff&type=code
2. curl https://research-software.nl/api/software_cache | jq '.[] | .repositoryURLs.github[]' --raw-output > urls.txt

Possibly not for this sprint.

Changing the data structure from an array of URLs to something more complicated that kept additional information about the repos would be useful in a few places:

• #18
• #23
• #36

We used sourcegraph to list the repositories including CITATION.cff file.
https://sourcegraph.com/search?q=context:global+file:CITATION.cff+count:100&patternType=literal

Cherry picked repositories with high number of stars and contributors

Need to create pull request

Abel:

• https://github.com/allenai/allennlp (:fork_and_knife: Our fork) allenai/allennlp#5561
  Commit fixing cffversion was created.
• https://github.com/automl/auto-sklearn (:fork_and_knife: Our fork) automl/auto-sklearn#1388 automl/auto-sklearn#1390
  Commit fixing family-names key was created.
• https://github.com/JuliaLang/julia (:fork_and_knife: Our fork) JuliaLang/julia#44062
  Commit fixing publisher key was created.
• https://github.com/lutzroeder/netron (:fork_and_knife: Our fork) lutzroeder/netron#878
  Commit fixing missing message was created.

Faruk:

• https://github.com/microsoft/onnxruntime (:fork_and_knife: Our fork)
  microsoft/onnxruntime#10478
  Commit fixing duplicate title and wrong license was created.
• https://github.com/mxgmn/WaveFunctionCollapse (:fork_and_knife: Our fork)
  mxgmn/WaveFunctionCollapse#72
  Commit fixing cff-version was created.
• https://github.com/nextflow-io/nextflow (:fork_and_knife: Our fork)
  nextflow-io/nextflow#2630
  Commit fixing DOI key was created.
• https://github.com/nlohmann/json (:fork_and_knife: Our fork)
  nlohmann/json#3320
  Commit fixing cff-version to 1.2.0 was created.
  Commit fixing date-released was created.
• https://github.com/nltk/nltk (:fork_and_knife: Our fork)
  nltk/nltk#2944
  Was filtered out before. No changes required in CITATION.cff.

Jurrriaan:

• https://github.com/open-mmlab/mmdetection (:fork_and_knife: Our fork) open-mmlab/mmdetection#7125
  Was filtered out before. No changes required in CITATION.cff.
• https://github.com/pydata/xarray (:fork_and_knife: Our fork) pydata/xarray#6252
  Was filtered out before. No changes required in CITATION.cff.
• https://github.com/RaRe-Technologies/gensim (:fork_and_knife: Our fork) piskvorky/gensim#3287
  Was filtered out before. No changes required in CITATION.cff.
• [ ] https://github.com/sagemath/sage (:fork_and_knife: Our fork) Does not use GitHub. Repo is only a mirror.
  Was filtered out before. No changes required in CITATION.cff.
• https://github.com/tensorflow/tensorflow (:fork_and_knife: Our fork)
  tensorflow/tensorflow#54291
  Was filtered out before. No changes required in CITATION.cff.

Present in the filtered list, already attributed.

https://github.com/duckdb/duckdb
https://github.com/effector/effector
https://github.com/ManimCommunity/manim
https://github.com/microsoft/nni
https://github.com/PyTorchLightning/pytorch-lightning
https://github.com/Zettlr/Zettlr

Won't fix

• https://github.com/automl/Auto-PyTorch
  Only uses preferred-citation.
• https://github.com/explosion/spaCy
  Only uses preferred-citation.
• https://github.com/github/linguist
  Missing CITATION.cff.
• https://github.com/huggingface/transformers
  Wrong date-released. Candidate for future CITATION.cff dissemination.
• https://github.com/microsoft/FLAML
  Only uses preferred-citation.
• https://github.com/spack/spack
  Only uses preferred-citation.
• https://github.com/tensorflow/federated
  Not using GitHub actions.

Consider repos:

1. #2
2. #3
3. #4
4. #5
5. #6
6. #9
7. #8
8. #7
9. #10
10. #11)
11. (maybe) LIMIT 30 or something

The result of these filters should be a list of repositories that qualify.

(Spawned from #1)

Use the GitHub search API to get this list as an array of repo names.

like https://github.com/UnofficialJuliaMirrorSnapshots/TensorToolbox.jl-9c690861-8ade-587a-897e-15364bc6f718/blob/master/CITATION.cff

Plan:

• Fork
• Change cffversion to 1.2.0
• Create cffconvert.yml
• Run validator
• If validation work, make a PR with updated CITATION.cff and cffconvert.yml
• If validation does not work, assess the error and decide on a case-by-case basis

Summary of repos aggregated by @fdiblen:

Abel

• https://github.com/benvanwerkhoven/kernel_tuner KernelTuner/kernel_tuner#132
• https://github.com/amusecode/amuse amusecode/amuse#818
• https://github.com/ADAH-EviDENce/evidence ADAH-EviDENce/evidence#126
• https://github.com/3D-e-Chem/knime-kripodb 3D-e-Chem/knime-kripodb#33
• https://github.com/dianna-ai/dianna dianna-ai/dianna#192
• https://github.com/DeepRank/iScore DeepRank/iScore#32
• https://github.com/DeepRank/pdb2sql DeepRank/pdb2sql#70
• https://github.com/entangled/entangled entangled/entangled#93
• https://github.com/eWaterCycle/grpc4bmi eWaterCycle/grpc4bmi#109
• https://github.com/ESMValGroup/ESMValCore ESMValGroup/ESMValCore#1459 ESMValGroup/ESMValCore#1467
• https://github.com/ewatercycle/era5cli eWaterCycle/era5cli#109

Faruk

• https://github.com/eucp-project/atlas eucp-project/atlas#82
• https://github.com/ESMValGroup/ESMValTool ESMValGroup/ESMValTool#2517
• https://github.com/FAIR-data-for-CAPACITY/FHIR-to-CAPACITY - Fix is not trivial
• https://github.com/eucp-project/storyboards eucp-project/storyboards#62
• https://github.com/iomega/spec2vec iomega/spec2vec#74
• https://github.com/fair-workflows/NanopubJL fair-workflows/NanopubJL#10
• https://github.com/iomega/paired-data-form iomega/paired-data-form#197
• https://github.com/Filter-Bubble/e2e-Dutch Filter-Bubble/e2e-Dutch#32
• https://github.com/fair-software/howfairis fair-software/howfairis#335
• https://github.com/fair-workflows/fairworkflows fair-workflows/fairworkflows#198
• https://github.com/LINNAE-project/SFB-Annotator - Fix is not trivial
• https://github.com/matchms/matchms - Fix is not trivial
• https://github.com/matchms/ms2deepscore matchms/ms2deepscore#96

Jurriaan

• https://github.com/nlesc-nano/flamingo nlesc-nano/flamingo#67
• https://github.com/NLESC-JCER/EigenCuda NLESC-JCER/EigenCuda#27
• https://github.com/nlesc-nano/ceiba nlesc-nano/ceiba#32
• https://github.com/multiscale/muscle3 multiscale/muscle3#90
• https://github.com/nlesc-nano/swan nlesc-nano/swan#96
• https://github.com/NLeSC/spot NLeSC/spot#228
• https://github.com/NLeSC/litstudy/ NLeSC/litstudy#10
• https://github.com/SCM-NV/qmflows - Fix is not trivial
• https://github.com/NLeSC/fairdatapoint fair-data/fairdatapoint#67
• https://github.com/yatiml/yatiml yatiml/yatiml#38
• https://github.com/ci-for-science/self-hosted-runners ci-for-research/self-hosted-runners#143
• https://github.com/nlesc-nano/ceiba-cli nlesc-nano/ceiba-cli#31

(Spawned from #1)

Some repos have not seen activity in a while (over 3 years, for instance).
Should we remove them?

(Spawned from #1)

Choosing cffconvert as the validator probably affects #12. Alternatively, you could pick a language that is able to run docker commands or shell commands.

like run.sh

e.g. the total number of stars, followers, forks, separated into two groups: within eScienceCenter repos, and outside repos.

sorting might be better, but that requires changes to the data model I think

Spawned from standup.

Each one creates a plan (answering the issues) and we discuss approaches in the meeting.

(Spawned from #1)

Both for speed and because each script evaluation is using 767 of my quota of 5000 requests.

As of 31 January, cffconvert GitHub Action has 6 stars, 1 fork, and 4 people are watching the repository.

1. describe how to install Node, which version
2. describe how to get token, how to use it, only needs repo permissions
3. how to run the script
4. editorconfig settings, plugin vscode
5. installing docker
6. running docker without admin rights
7. docker pull citationcff/cffconvert:2.0.0
8. docker run cffconvert
9. documentation:
   1. octokit
   2. mdn
   3. github api
10. etc

What will the filters of #1 pass to each other, how will they do it?