GitDiff Python

git_diff.py

import git
from unidiff import PatchSet
import os

from io import StringIO


class GitDiff:
    """
    Class to get the diff between two branches
    """

    def __init__(
        self,
        repo_path=os.getcwd(),
        base_branch=os.environ.get("BASE_BRANCH", "main"),
        current_branch=os.environ.get("CURRENT_BRANCH", "HEAD"),
    ):
        """
        Init function
        :param repo_path: path to the git repo, default is current working directory
        :param base_branch: base branch to compare with, default is main
        :param current_branch: current branch to compare, default is HEAD
        """
        self.base_branch = base_branch
        self.current_branch = current_branch
        self.repository = git.Repo(repo_path)

    def json(self):
        """
        Returns a git diff between two branches in json format
        :return: Array of dicts
        """
        uni_diff_text = self.repository.git.diff(
            self.base_branch,
            self.current_branch,
            ignore_blank_lines=True,
            ignore_space_at_eol=True,
        )

        patch_set = PatchSet(StringIO(uni_diff_text))

        change_list = []

        for patched_file in patch_set:

            file_path = patched_file.path

            added_lines = []
            for hunk in patched_file:
                for line in hunk:
                    if line.is_added and line.value.strip() != "":
                        added_lines.append(
                            {"line_number": line.target_line_no, "line": line.value}
                        )

            deleted_lines = []
            for hunk in patched_file:
                for line in hunk:
                    if line.is_removed and line.value.strip() != "":
                        deleted_lines.append(
                            {"line_number": line.source_line_no, "line": line.value}
                        )

            # get modifed lines by comparing added and deleted lines
            added_to_remove = []
            del_to_remove = []
            modified_lines = []
            for added_line in added_lines:
                for deleted_line in deleted_lines:
                    if added_line["line_number"] == deleted_line["line_number"]:
                        modified_lines.append(
                            {
                                "line_number": added_line["line_number"],
                                "line": added_line["line"],
                            }
                        )

                        added_to_remove.append(added_line)
                        del_to_remove.append(deleted_line)

            # remove modified lines from added and deleted lines
            for line in added_to_remove:
                added_lines.remove(line)

            for line in del_to_remove:
                deleted_lines.remove(line)

            change_list.append(
                {
                    "file_path": file_path,
                    "added": added_lines,
                    "deleted": deleted_lines,
                    "modified": modified_lines,
                }
            )

        return change_list

Usage 🔨

Set your env vars:

export BASE_BRANCH=main
export CURRENT_BRANCH=my-cool-feature

Use the class:

import json
git_diff = GitDiff().json()

print(json.dumps(git_diff, indent=4))