From 9da566fd73de02d6ef79dedb7e887176a24ab93b Mon Sep 17 00:00:00 2001
From: Joris Roovers <joris.roovers@gmail.com>
Date: Tue, 6 Sep 2022 09:37:41 +0000
Subject: [PATCH] WIP: File level diff stats

Relates to #314
---
 gitlint-core/gitlint/git.py | 33 ++++++++++++++++++++++++++++-----
 1 file changed, 28 insertions(+), 5 deletions(-)

diff --git a/gitlint-core/gitlint/git.py b/gitlint-core/gitlint/git.py
index f27b7ee9..134b0696 100644
--- a/gitlint-core/gitlint/git.py
+++ b/gitlint-core/gitlint/git.py
@@ -152,6 +152,7 @@ def __init__(
         author_email=None,
         parents=None,
         changed_files=None,
+        changed_files_stats=None,
         branches=None,
     ):
         self.context = context
@@ -162,6 +163,7 @@ def __init__(
         self.author_email = author_email
         self.parents = parents or []  # parent commit hashes
         self.changed_files = changed_files or []
+        self.changed_files_stats = changed_files_stats or {}
         self.branches = branches or []
 
     @property
@@ -198,6 +200,7 @@ def __str__(self):
             f"is-revert-commit: {self.is_revert_commit}\n"
             f"Branches: {self.branches}\n"
             f"Changed Files: {self.changed_files}\n"
+            f"Changed Files Stats: {self.changed_files_stats}\n"
             "-----------------------"
         )
 
@@ -217,6 +220,7 @@ def __eq__(self, other):
             and self.is_squash_commit == other.is_squash_commit
             and self.is_revert_commit == other.is_revert_commit
             and self.changed_files == other.changed_files
+            and self.changed_files_stats == other.changed_files_stats
             and self.branches == other.branches
         )
 
@@ -307,18 +311,31 @@ def is_merge_commit(self):
 
     @property
     def changed_files(self):
-        def cache_changed_files():
-            self._cache["changed_files"] = _git(
+        return self._try_cache("changed_files", self.changed_files_stats)
+
+    @property
+    def changed_files_stats(self):
+        def cache_changed_files_stats():
+            changed_files_stats_raw = _git(
                 "diff-tree",
                 "--no-commit-id",
-                "--name-only",
+                "--numstat",
                 "-r",
                 "--root",
                 self.sha,
                 _cwd=self.context.repository_path,
-            ).split()
+            ).split("\n")
+
+            changed_files_stats = {}
+            for line in changed_files_stats_raw:
+                line_stats = line.split()
+                # TODO(jroovers): create a ChangedFileStats object?
+                changed_files_stats[line_stats[2]] = {"added": int(line_stats[0]), "deleted": int(line_stats[1])}
 
-        return self._try_cache("changed_files", cache_changed_files)
+            self._cache["changed_files_stats"] = changed_files_stats
+            self._cache["changed_files"] = changed_files_stats.keys()
+
+        return self._try_cache("changed_files_stats", cache_changed_files_stats)
 
 
 class StagedLocalGitCommit(GitCommit, PropertyCache):
@@ -371,6 +388,12 @@ def branches(self):
     def changed_files(self):
         return _git("diff", "--staged", "--name-only", "-r", _cwd=self.context.repository_path).split()
 
+    @property
+    def changed_files_stats(self):
+        raise NotImplementedError()
+        # TODO(jroovers): implement this!
+        # return _git("diff", "--staged", "--name-only", "-r", _cwd=self.context.repository_path).split()
+
 
 class GitContext(PropertyCache):
     """Class representing the git context in which gitlint is operating: a data object storing information about