From f3278ec1db0eaae7dc6342c589555177ffeb5767 Mon Sep 17 00:00:00 2001
From: Dima Gerasimov <karlicoss@gmail.com>
Date: Tue, 9 Oct 2018 22:38:58 +0100
Subject: [PATCH] Basic implementation for commit provider

---
 commits/__init__.py | 73 +++++++++++++++++++++++++++++++++++++++++++++
 commits/__main__.py |  5 ++++
 requirements.txt    |  1 +
 3 files changed, 79 insertions(+)
 create mode 100644 commits/__init__.py
 create mode 100644 requirements.txt

diff --git a/commits/__init__.py b/commits/__init__.py
new file mode 100644
index 0000000..83d8db9
--- /dev/null
+++ b/commits/__init__.py
@@ -0,0 +1,73 @@
+from datetime import datetime
+from typing import List, NamedTuple, Optional
+from os.path import basename, islink, isdir, join
+from os import listdir
+
+import git # type: ignore
+
+# TODO do something smarter... later
+# TODO def run against bitbucket and gh backups
+SOURCES = [
+    '***REMOVED***',
+    '***REMOVED***',
+    '***REMOVED***',
+    '***REMOVED***',
+    '***REMOVED***',
+    '***REMOVED***',
+]
+
+THINGS = [
+    '***REMOVED***',
+    '***REMOVED***',
+    '***REMOVED***',
+    '***REMOVED***',
+]
+
+def by_me(actor):
+    aa = actor.email + " " + actor.name
+    if actor.email in ('***REMOVED***', '***REMOVED***@gmail.com'):
+        return True
+    if actor.name in ('***REMOVED***',):
+        return True
+    for thing in THINGS:
+        if thing in aa:
+            print("WARNING!!!", actor)
+            return True
+    return False
+
+class Commit(NamedTuple):
+    dt: datetime
+    message: str
+    repo: str
+        # TODO filter so they are authored by me
+
+def iter_commits(repo: str):
+    # TODO other branches?
+    rr = basename(repo)
+    gr = git.Repo(repo)
+    for c in gr.head.reference.log():
+        if by_me(c.actor):
+            yield Commit(
+                dt=c.time,
+                message=c.message, # TODO strip off 'commit: '? (there are also 'merge')
+                repo=rr,
+            )
+
+def is_git_repo(d: str):
+    dotgit = join(d, '.git')
+    return isdir(dotgit)
+
+# TODO eh. traverse all of filesystem?? or only specific dirs for now?
+def iter_all_commits():
+    for src in SOURCES:
+        # TODO warn if doesn't exist?
+        for d in listdir(src):
+            pr = join(src, d)
+            if is_git_repo(pr):
+                for c in iter_commits(pr):
+                    yield c
+
+
+def get_all_commits():
+    ss = set(iter_all_commits())
+    return list(sorted(ss, key=lambda c: c.dt))
diff --git a/commits/__main__.py b/commits/__main__.py
index e69de29..ee2eee1 100644
--- a/commits/__main__.py
+++ b/commits/__main__.py
@@ -0,0 +1,5 @@
+from commits import iter_commits, iter_all_commits, get_all_commits
+
+# TODO cache?
+for c in get_all_commits(): # ('***REMOVED***'):
+    print(c)
diff --git a/requirements.txt b/requirements.txt
new file mode 100644
index 0000000..59348f9
--- /dev/null
+++ b/requirements.txt
@@ -0,0 +1 @@
+gitpython