From 834cf61634fa8f3d409832a108b2aa766fc241b8 Mon Sep 17 00:00:00 2001
From: Stephan Kulow <coolo@suse.de>
Date: Sun, 6 Nov 2022 09:49:52 +0100
Subject: [PATCH] Use proper user info in commits

---
 lib/git.py          | 17 +--------------
 lib/git_exporter.py | 52 ++++++++++++++++++++++++++++-----------------
 lib/user.py         | 13 ++++++++++++
 3 files changed, 47 insertions(+), 35 deletions(-)

diff --git a/lib/git.py b/lib/git.py
index 56b1ee3..f1c20c5 100644
--- a/lib/git.py
+++ b/lib/git.py
@@ -73,10 +73,8 @@ class Git:
         committer=None,
         committer_email=None,
         committer_time=None,
-        allow_empty=False,
     ):
         """Add all the files and create a new commit in the current HEAD"""
-        assert allow_empty or self.is_dirty()
 
         if not committer:
             committer = self.committer if self.committer else self.user
@@ -85,33 +83,20 @@ class Git:
             )
             committer_time = committer_time if committer_time else user_time
 
-        try:
+        if self.is_dirty():
             self.repo.index.add_all()
-        except pygit2.GitError as e:
-            if not allow_empty:
-                raise e
 
         self.repo.index.write()
         author = pygit2.Signature(user, user_email, int(user_time.timestamp()))
         committer = pygit2.Signature(
             committer, committer_email, int(committer_time.timestamp())
         )
-        if not parents:
-            try:
-                parents = [self.repo.head.target]
-            except pygit2.GitError as e:
-                parents = []
-                if not allow_empty:
-                    raise e
 
         tree = self.repo.index.write_tree()
         return self.repo.create_commit(
             "HEAD", author, committer, message, tree, parents
         )
 
-    def merge_abort(self):
-        self.repo.state_cleanup()
-
     def last_commit(self):
         try:
             return self.repo.head.target
diff --git a/lib/git_exporter.py b/lib/git_exporter.py
index edb9b97..c13f5e4 100644
--- a/lib/git_exporter.py
+++ b/lib/git_exporter.py
@@ -7,8 +7,9 @@ from lib.binary import is_binary_or_large
 from lib.db import DB
 from lib.git import Git
 from lib.obs import OBS
-from lib.proxy_sha256 import ProxySHA256, md5
+from lib.proxy_sha256 import ProxySHA256
 from lib.tree_builder import TreeBuilder
+from lib.user import User
 
 
 class GitExporter:
@@ -31,12 +32,7 @@ class GitExporter:
     def set_gc_interval(self, gc):
         self.gc_interval = gc
 
-    def export_as_git(self):
-        db = DB()
-        tree = TreeBuilder(db).build(self.project, self.package)
-        flats = tree.as_flat_list()
-
-        branch_state = {"factory": None, "devel": None}
+    def check_repo_state(self, flats, branch_state):
         state_data = dict()
         if os.path.exists(self.state_file):
             with open(self.state_file, "r") as f:
@@ -57,19 +53,38 @@ class GitExporter:
                     found_state = True
             if not found_state:
                 left_to_commit.append(flat)
+        return left_to_commit
+
+    def export_as_git(self):
+        db = DB()
+        tree = TreeBuilder(db).build(self.project, self.package)
+        flats = tree.as_flat_list()
+
+        branch_state = {"factory": None, "devel": None}
+        left_to_commit = self.check_repo_state(flats, branch_state)
+
+        if not left_to_commit:
+            return
+
+        logging.info(f"Commiting into {self.git.path}")
+        self.run_gc()
+        users = dict()
 
-        gc_cnt = self.gc_interval
-        if len(left_to_commit) > 0:
-            logging.info(f"Commiting into {self.git.path}")
-            self.git.gc()
         for flat in left_to_commit:
-            gc_cnt -= 1
-            if gc_cnt <= 0 and self.gc_interval:
-                self.git.gc()
-                gc_cnt = self.gc_interval
+            if flat.commit.userid not in users:
+                users[flat.commit.userid] = User(db, flat.commit.userid)
+            flat.user = users[flat.commit.userid]
+            logging.debug(f"USER {flat.user}")
+            self.gc_cnt -= 1
+            if self.gc_cnt <= 0 and self.gc_interval:
+                self.run_gc()
             logging.debug(f"Committing {flat}")
             self.commit_flat(flat, branch_state)
 
+    def run_gc(self):
+        self.gc_cnt = self.gc_interval
+        self.git.gc()
+
     def commit_file(self, flat, file, size, md5):
         # have such files been detected as text mimetype before?
         is_text = self.proxy_sha256.is_text(flat.commit.package, file.name)
@@ -107,7 +122,7 @@ class GitExporter:
 
         if flat.parent1:
             if not self.branch_fits_parent1(flat, branch_state):
-                logging.info(f"Reset {flat.branch} onto {flat.parent1.short_string()}")
+                logging.debug(f"Reset {flat.branch} onto {flat.parent1.short_string()}")
                 assert flat.parent1.git_commit
                 self.git.set_branch_head(flat.branch, flat.parent1.git_commit)
                 self.git.checkout(flat.branch)
@@ -126,12 +141,11 @@ class GitExporter:
             self.commit_file(flat, file, size, md5)
 
         commit = self.git.commit(
-            f"OBS User {flat.commit.userid}",
-            "null@suse.de",
+            flat.user.realname,
+            flat.user.email,
             flat.commit.commit_time,
             # TODO: Normalize better the commit message
             f"{flat.commit.comment}\n\n{flat.commit}",
-            allow_empty=True,
             parents=parents,
         )
         flat.commit.git_commit = commit
diff --git a/lib/user.py b/lib/user.py
index 6421c39..2d2aa95 100644
--- a/lib/user.py
+++ b/lib/user.py
@@ -15,6 +15,19 @@ FAKE_ACCOUNTS = (
 
 
 class User:
+    def __init__(self, db, userid) -> None:
+        row = User.lookup(db, userid)
+        self.userid = userid
+        if row:
+            (_, _, self.email, self.realname) = row
+        else:
+            self.email = ""
+            self.realname = ""
+        if not self.email:
+            self.email = "null@suse.de"
+        if not self.realname:
+            self.realname = f"OBS User {userid}"
+
     def parse(self, xml, userid):
         self.userid = userid
         self.realname = xml.find("realname").text