New parser request: git log --numstat

[avi-cenna]

It would be nice to have a parser for git log --numstat. This command shows the number of additions and deletions for each file in a commit.

[kellyjonbrazil]

Thanks for the parser request! Seems like a good addition. Might be a new parser or a possible addition to the existing --git-log parsers.

[pwnetrationguru]

I was about to file a similar issue. Before digging in, I'd like to verify @kellyjonbrazil that you'd be interested in merging the change in.

Proposal would be to change output of --git-log from:

  {
    "commit": "0e49db35a8f39fa8cdcdbe18fc02efbe6b38bf3a",
    "author": "debo",
    "author_email": "[email protected]",
    "date": "Sun Mar 21 13:01:59 2021 -0700",
    "commit_by": "debo",
    "commit_by_email": "[email protected]",
    "commit_by_date": "Sun Mar 21 13:01:59 2021 -0700",
    "stats": {
      "files_changed": 5,
      "insertions": 248,
      "deletions": 7,
      "files": [
        ".gitignore",
        "Makefile",
        "README.md",
        "go.mod",
        "go.sum"
      ]
    }

To output of:

{
  "commit": "0e49db35a8f39fa8cdcdbe18fc02efbe6b38bf3a",
  "author": "debo",
  "author_email": "[email protected]",
  "date": "Sun Mar 21 13:01:59 2021 -0700",
  "commit_by": "debo",
  "commit_by_email": "[email protected]",
  "commit_by_date": "Sun Mar 21 13:01:59 2021 -0700",
  "stats": {
    "files_changed": 5,
    "insertions": 248,
    "deletions": 7,
    "files": [
      {"file": ".gitignore", "insertions": 169, "deletions": 0},
      {"file": "Makefile", "insertions": 26, "deletions": 3},
      {"file": "README.md", "insertions": 13, "deletions": 4},
      {"file": "go.mod", "insertions": 8, "deletions": 0},
      {"file": "go.sum", "insertions": 32, "deletions": 0}
    ]
  }
}

Without digging in more, I have some concern around complexity of the change as we'd need to use --format=fuller in combination with --numstat.

A few questions:

• @kellyjonbrazil would you be interested in a change like this?
• @kellyjonbrazil if so, do you see and meaningful complexity that I might be missing?

[kellyjonbrazil]

With respect to the schema change, could we add a new field called file_stats or similar that would contain the list of file objects? This way the schema is backwards compatible and the new field would come up when more detailed data is available. I’ll definitely take a look at a PR if someone wants to take this on. If the code changes are too crazy we can also do a check to see what “type” of git-log output we are dealing with and dispatch to the correct parse function for that type. That can keep the code cleaner and we have done that in the past for netstat and others.