feat: implemented git-show-coedited

home/.bin/git-show-coedited

@@ -0,0 +1,104 @@
+#!/usr/bin/env python
+"""
+Usage: python gitanalyse.py [repo_path] [file1] [file2] ...
+"""
+
+# TODO: fix bug where dirs are not counted correctly
+"""
+Bug example:
+
+$ git-show-coedited ~/dotfiles/ install-packages.sh
+Path Frequency
+----------------------------- -----------
+home/.vimrc 0.782609
+home 0.695652 <--- should be at least 0.782609
+home/.config 0.608696
+"""
+
+import itertools
+import sys
+from collections import Counter, defaultdict
+
+from git import Repo
+from tabulate import tabulate
+from tqdm import tqdm
+
+
+def main():
+ # Path to your local Git repository
+ if len(sys.argv) >= 3:
+ repo_path = sys.argv[1]
+ files = sys.argv[2:]
+ else:
+ print(__doc__.strip())
+ sys.exit(1)
+
+ coedit_frequency = analyze_coedits(repo_path)
+
+ for file in files:
+ print_file_stats(coedit_frequency, file)
+
+
+def print_file_stats(coedit_frequency, file):
+ print(f"Co-editing frequency for {file}")
+ print(
+ tabulate(
+ sorted(coedit_frequency[file].items(), key=lambda t: t[1], reverse=True)[
+ :10
+ ],
+ headers=["Path", "Frequency"],
+ )
+ )
+
+
+def analyze_coedits(repo_path):
+ print(f"Analyzing co-edits in {repo_path}")
+ repo = Repo(repo_path)
+ coedit_count: dict[str, dict[str, int]] = defaultdict(Counter)
+ total_commits = 0
+ file_commits: dict[str, int] = defaultdict(int)
+
+ # Traverse the commit history
+ for commit in tqdm(list(repo.iter_commits())[:1000]):
+ files_in_commit = set(commit.stats.files.keys())
+ if len(files_in_commit) > 100:
+ # Skip commits with more than 100 files
+ print(f"Skipping commit {commit.hexsha} with {len(files_in_commit)} files")
+ continue
+
+ total_commits += 1
+
+ # Collect directories for each file and count commits per file and directory
+ dirs_in_commit = set()
+ for file in files_in_commit:
+ file_commits[file] += 1
+ sub_path = ""
+ for part in file.split("/"):
+ sub_path = "/".join([sub_path, part]) if sub_path else part
+ dirs_in_commit.add(sub_path)
+
+ # Update commit counts for directories
+ for directory in dirs_in_commit:
+ file_commits[directory] += 1
+
+ # Count pairs of files and directories that are edited together
+ all_items_in_commit = list(files_in_commit) + list(dirs_in_commit)
+ for item_pair in itertools.combinations(all_items_in_commit, 2):
+ coedit_count[item_pair[0]][item_pair[1]] += 1
+ coedit_count[item_pair[1]][item_pair[0]] += 1
+
+ # Calculate co-editing frequency
+ coedit_frequency = {
+ file: {
+ coedited_file: count / file_commits[file]
+ for coedited_file, count in counters.items()
+ if file != coedited_file
+ }
+ for file, counters in coedit_count.items()
+ }
+
+ return coedit_frequency
+
+
+if __name__ == "__main__":
+ main()