initial commit

This commit is contained in:
dave 2018-07-15 18:17:51 -07:00
commit 4a98f2316b
8 changed files with 753 additions and 0 deletions

57
dl_projects.py Normal file
View File

@ -0,0 +1,57 @@
from mirror import SSHAgent
from subprocess import check_call
import os
import json
import gitlab
from concurrent.futures import ThreadPoolExecutor
class DlGitlabProjects(object):
def __init__(self, agent, gl):
self.agent = agent
self.gl = gl
def run(self):
projects = self.get_gitlab_projects()
futures = []
with ThreadPoolExecutor(max_workers=8) as pool:
for name, project in projects.items():
# import ipdb ; ipdb.set_trace()
if not os.path.exists(os.path.join("tmp", project.name)):
futures.append(pool.submit(check_call, ["git", "clone", project.ssh_url_to_repo, project.name], cwd="./tmp/"))
else:
with open(os.path.join("tmp", project.name + ".txt"), "w") as f:
f.write(project.description)
for item in futures:
e = item.exception()
if e:
raise e
def get_gitlab_projects(self):
page = 0
all_projects = {}
while True:
projects = self.gl.projects.owned(per_page=10, page=page)
page += 1
if not projects:
break
for item in projects:
if item.visibility_level == gitlab.VISIBILITY_PUBLIC: # or "githubmirror" not in item.tag_list:
continue
all_projects[item.name] = item
return all_projects
def main():
with open("creds.json") as f:
creds = json.load(f)
gitlab_api = gitlab.Gitlab(creds["gitlab"]["url"], creds["gitlab"]["token"], api_version=3)
with SSHAgent() as agent:
check_call(["ssh-add", creds["identity"]], env={"SSH_AUTH_SOCK": agent})
DlGitlabProjects(agent, gitlab_api).run()
if __name__ == '__main__':
main()

51
dl_webhooks.py Normal file
View File

@ -0,0 +1,51 @@
from mirror import SSHAgent
from subprocess import check_call
from collections import defaultdict
import json
import gitlab
class DlGitlabProjects(object):
def __init__(self, agent, gl):
self.agent = agent
self.gl = gl
def run(self):
projects = self.get_gitlab_projects()
hooks = defaultdict(list)
for name, project in projects.items():
# import ipdb ; ipdb.set_trace()
for hook in self.gl.project_hooks.list(project_id=project.id):
hooks[name].append(hook.url)
with open("./hooks.json", "w") as f:
json.dump(dict(hooks), f, sort_keys=True, indent=4)
def get_gitlab_projects(self):
page = 0
all_projects = {}
while True:
projects = self.gl.projects.list(per_page=10, page=page, owned=True)
page += 1
if not projects:
break
for item in projects:
# if item.visibility == 'public': # or "githubmirror" not in item.tag_list:
# continue
all_projects[item.name] = item
return all_projects
def main():
with open("creds.json") as f:
creds = json.load(f)
gitlab_api = gitlab.Gitlab(creds["gitlab"]["url"], creds["gitlab"]["token"], api_version=4)
with SSHAgent() as agent:
check_call(["ssh-add", creds["identity"]], env={"SSH_AUTH_SOCK": agent})
DlGitlabProjects(agent, gitlab_api).run()
if __name__ == '__main__':
main()

335
gitea.py Normal file
View File

@ -0,0 +1,335 @@
import json
import requests
class ApiException(Exception):
pass
class Gitea(object):
def __init__(self, url, token):
self.url = url
self.session = requests.session()
self.session.headers = {"Authorization": "token " + token}
def get_url(self, endpoint):
url = self.url + endpoint
print(url)
return url
def parse_result(self, result):
print(result.status_code)
if (result.text and len(result.text) > 3):
return json.loads(result.text)
return {}
def requests_get(self, endpoint):
return self.parse_result(self.session.get(self.get_url(endpoint)))
def requests_post(self, endpoint, data):
return self.parse_result(self.session.post(self.get_url(endpoint), json=data))
def get_version(self):
path = '/version'
return self.requests_get(path)
def create_repo(self, **kwargs):
path = '/user/repos'
return self.requests_post(path, kwargs)
def create_hook(self, owner, repo, url):
path = '/repos/{owner}/{repo}/hooks'.format(owner=owner, repo=repo)
return self.requests_post(path,
{"active": True,
"config": {
"url": url,
"content_type": "json",
"secret": ""
},
"events": ["push"],
"type": "gitea"})
# def get_repos_search(self, q, uid, limit):
# path = '/repos/search'
# return self.requests_get(path)
# def get_users_gpg_keys(self, username):
# path = '/users/' + username + '/gpg_keys'
# return self.requests_get(path)
# def get_orgs_repos(self, orgname):
# path = '/orgs/' + orgname + '/repos'
# return self.requests_get(path)
# def put_user_starred(self, username, reponame):
# path = '/user/starred/' + username + '/' + reponame
# return self.requests.put(path)
# def delete_user_starred(self, username, reponame):
# path = '/user/starred/' + username + '/' + reponame
# return self.requests.delete(path)
# def get_user_starred(self, username, reponame):
# path = '/user/starred/' + username + '/' + reponame
# return self.requests_get(path)
# def get_users_search(self, ):
# path = '/users/search'
# return self.requests_get(path)
# def delete_repos(self, username, reponame):
# path = '/repos/' + username + '/' + reponame
# return self.requests.delete(path)
# def get_repos(self, username, reponame):
# path = '/repos/' + username + '/' + reponame
# return self.requests_get(path)
# def get_users_repos(self, username):
# path = '/users/' + username + '/repos'
# return self.requests_get(path)
# def post_repos__mirror_sync(self, username, reponame):
# path = '/repos/' + username + '/' + reponame + '/mirror-sync'
# return self.requests_post(path, data={})
# def post_admin_users_orgs(self, username, full_name, description, website, location):
# path = '/admin/users/' + username + '/orgs'
# return self.requests_post(path, data={'username': username, 'full_name': full_name, 'description': description,
# 'website': website, 'location': location})
# def post_user_gpg_keys(self, armored_public_key):
# path = '/user/gpg_keys'
# return self.requests_post(path, data={'armored_public_key': armored_public_key})
# def get_user_gpg_keys_all(self):
# path = '/user/gpg_keys'
# return self.requests_get(path)
# def get_user_subscriptions(self, ):
# path = '/user/subscriptions'
# return self.requests_get(path)
# def post_markdown(self, Text, Mode, Context, Wiki):
# path = '/markdown'
# return self.requests_post(path, data={'Text': Text, 'Mode': Mode, 'Context': Context, 'Wiki': Wiki})
# def patch_repos__hooks(self, config, events, active, username, reponame, id):
# path = '/repos/' + username + '/' + reponame + '/hooks/' + id
# return self.requests.patch(path)
# def delete_repos_hooks(self, username, reponame, id):
# path = '/repos/' + username + '/' + reponame + '/hooks/' + id
# return self.requests.delete(path)
# def get_users_keys(self, username):
# path = '/users/' + username + '/keys'
# return self.requests_get(path)
# def put_user_following(self, username):
# path = '/user/following/' + username
# return self.requests.put(path)
# def delete_user_following(self, username):
# path = '/user/following/' + username
# return self.requests.delete(path)
# def get_user_following(self, username):
# path = '/user/following/' + username
# return self.requests_get(path)
# def post_markdown_raw(self, ):
# path = '/markdown/raw'
# return self.requests_post(path, data={})
# def get_users_subscriptions(self, username):
# path = '/users/' + username + '/subscriptions'
# return self.requests_get(path)
# def delete_user_keys(self, id):
# path = '/user/keys/' + id
# return self.requests.delete(path)
# def get_user_keys(self, id):
# path = '/user/keys/' + id
# return self.requests_get(path)
# def post_orgs_hooks(self, type, config, events, active, orgname):
# path = '/orgs/' + orgname + '/hooks/'
# return self.requests_post(path, data={'type': type, 'config': config, 'events': events, 'active': active})
# def get_user(self, ):
# path = '/user'
# return self.requests_get(path)
# def get_orgs_members_all(self, orgname):
# path = '/orgs/' + orgname + '/members'
# return self.requests_get(path)
# def patch_admin_users(self, source_id, login_name, full_name, email, password, website, location, active, admin,
# allow_git_hook, allow_import_local, max_repo_creation, username):
# path = '/admin/users/' + username
# return self.requests.patch(path)
# def delete_admin_users(self, username):
# path = '/admin/users/' + username
# return self.requests.delete(path)
# def post_admin_users_keys(self, title, key, username):
# path = '/admin/users/' + username + '/keys'
# return self.requests_post(path, data={'title': title, 'key': key})
# def post_user_keys(self, title, key):
# path = '/user/keys'
# return self.requests_post(path, data={'title': title, 'key': key})
# def get_user_keys_all(self):
# path = '/user/keys'
# return self.requests_get(path)
# def get_users_tokens(self, username):
# path = '/users/' + username + '/tokens'
# return self.requests_get(path)
# def get_orgs_hooks_all(self, orgname):
# path = '/orgs/' + orgname + '/hooks'
# return self.requests_get(path)
# def get_users(self, username):
# path = '/users/' + username
# return self.requests_get(path)
# def get_orgs_public_members_all(self, orgname):
# path = '/orgs/' + orgname + '/public_members'
# return self.requests_get(path)
# def post_repos__forks(self, organization, repo, owner):
# path = '/repos/' + owner + '/' + repo + '/forks'
# return self.requests_post(path, data={'organization': organization})
# def get_repos_forks(self, repo, owner):
# path = '/repos/' + owner + '/' + repo + '/forks'
# return self.requests_get(path)
# def post_admin_users(self, source_id, login_name, username, full_name, email, password, send_notify):
# path = '/admin/users'
# return self.requests_post(path, data={'source_id': source_id, 'login_name': login_name, 'username': username,
# 'full_name': full_name, 'email': email, 'password': password,
# 'send_notify': send_notify})
# def get_user_starred_all(self):
# path = '/user/starred'
# return self.requests_get(path)
# def get_users_followers(self, username):
# path = '/users/' + username + '/followers'
# return self.requests_get(path)
# def get_repositories(self, id):
# path = '/repositories/' + id
# return self.requests_get(path)
# def put_repos__subscription(self, username, reponame):
# path = '/repos/' + username + '/' + reponame + '/subscription'
# return self.requests.put(path)
# def delete_repos_subscription(self, username, reponame):
# path = '/repos/' + username + '/' + reponame + '/subscription'
# return self.requests.delete(path)
# def get_repos_subscription(self, username, reponame):
# path = '/repos/' + username + '/' + reponame + '/subscription'
# return self.requests_get(path)
# def get_users_following(self, username):
# path = '/users/' + username + '/following'
# return self.requests_get(path)
# def get_users_starred(self, username):
# path = '/users/' + username + '/starred'
# return self.requests_get(path)
# def put_orgs_public_members(self, username, orgname):
# path = '/orgs/' + orgname + '/public_members/' + username
# return self.requests.put(path)
# def delete_orgs_public_members(self, username, orgname):
# path = '/orgs/' + orgname + '/public_members/' + username
# return self.requests.delete(path)
# def get_orgs_public_members(self, username, orgname):
# path = '/orgs/' + orgname + '/public_members/' + username
# return self.requests_get(path)
# def get_user_following_all(self, ):
# path = '/user/following'
# return self.requests_get(path)
# def get_user_followers(self, ):
# path = '/user/followers'
# return self.requests_get(path)
# def post_admin_users_repos(self, name, description, private, auto_init, gitignores, license, readme, username):
# path = '/admin/users/' + username + '/repos'
# return self.requests_post(path, data={'name': name, 'description': description, 'private': private,
# 'auto_init': auto_init, 'gitignores': gitignores, 'license': license,
# 'readme': readme})
# def patch_orgs_hooks(self, config, events, active, orgname, id):
# path = '/orgs/' + orgname + '/hooks/' + id
# return self.requests.patch(path)
# def delete_orgs_hooks(self, orgname, id):
# path = '/orgs/' + orgname + '/hooks/' + id
# return self.requests.delete(path)
# def get_orgs_hooks(self, orgname, id):
# path = '/orgs/' + orgname + '/hooks/' + id
# return self.requests_get(path)
# def post_org_repos(self, name, description, private, auto_init, gitignores, license, readme, org):
# path = '/org/' + org + '/repos'
# return self.requests_post(path, data={'name': name, 'description': description, 'private': private,
# 'auto_init': auto_init, 'gitignores': gitignores, 'license': license,
# 'readme': readme})
# def delete_orgs_members(self, orgname, username):
# path = '/orgs/' + orgname + '/members/' + username
# return self.requests.delete(path)
# def get_orgs_members(self, orgname, username):
# path = '/orgs/' + orgname + '/members/' + username
# return self.requests_get(path)
# def post_repos__hooks(self, type, config, events, active, reponame, username):
# path = '/repos/' + username + '/' + reponame + '/hooks'
# return self.requests_post(path, data={'type': type, 'config': config, 'events': events, 'active': active})
# def get_repos_hooks(self, reponame, username):
# path = '/repos/' + username + '/' + reponame + '/hooks'
# return self.requests_get(path)
# def post_repos_migrate(self, clone_addr, auth_username, auth_password, uid, repo_name, mirror, private,
# description):
# path = '/repos/migrate'
# return self.requests_post(path, data={'clone_addr': clone_addr, 'auth_username': auth_username,
# 'auth_password': auth_password, 'uid': uid, 'repo_name': repo_name,
# 'mirror': mirror, 'private': private, 'description': description})
# def post_user_repos(self, name, description, private, auto_init, gitignores, license, readme):
# path = '/user/repos'
# return self.requests_post(path, data={'name': name, 'description': description, 'private': private,
# 'auto_init': auto_init, 'gitignores': gitignores, 'license': license,
# 'readme': readme})
def get_user_repos(self, ):
path = '/user/repos'
return self.requests_get(path)
# def delete_user_gpg_keys(self, id):
# path = '/user/gpg_keys/' + id
# return self.requests.delete(path)
# def get_user_gpg_keys(self, id):
# path = '/user/gpg_keys/' + id
# return self.requests_get(path)

112
gitea_to_github.py Normal file
View File

@ -0,0 +1,112 @@
#!/usr/bin/env python3
import json
import gitea
import github
import os
from shutil import rmtree
import re
from subprocess import check_call, check_output
from time import time
import signal
class SSHAgent(object):
sockre = re.compile(r'SSH_AUTH_SOCK=(.*?);')
pidre = re.compile(r'SSH_AGENT_PID=(\d*);', )
def __init__(self, cluster=None):
self.addargs = []
if cluster:
self.addargs.extend(['-a', os.path.join(cluster, 'agent.sock.{}'.format(int(time())))])
def __enter__(self):
output = check_output(['ssh-agent'] + self.addargs).decode("UTF-8")
self.sock = self.sockre.search(output).group(1)
self.pid = int(self.pidre.search(output).group(1))
return self.sock
def __exit__(self, errtype, value, traceback):
os.kill(self.pid, signal.SIGHUP)
class GiteaToGithubMirror(object):
def __init__(self, creds, agent):
self.creds = creds
self.agent = agent
self.ge = gitea.Gitea(self.creds["gitea"]["url"], self.creds["gitea"]["token"])
self.gh = github.Github(self.creds["github"]["username"], self.creds["github"]["password"])
self.ghu = self.gh.get_user()
if not os.path.exists("./tmp"):
os.mkdir("./tmp")
def run(self):
self.github_projects = self.get_github_projects()
self.gitea_projects = self.get_gitea_projects()
print("{} projects to check".format(len(self.gitea_projects)))
visited = []
for project_name, project in self.gitea_projects.items():
if project_name in visited:
continue
visited.append(project_name)
print("({}/{}) Mirroring {}".format(len(visited), len(self.gitea_projects), project_name))
self.mirror_to_gh(project_name)
def get_gitea_projects(self):
projects = {}
for project in filter(lambda x: x["private"] == False and x["empty"] == False, self.ge.get_user_repos()):
projects[project['name']] = project
return projects
def get_github_projects(self):
all_projects = {}
for repo in self.ghu.get_repos():
all_projects[repo.name] = repo
return all_projects
def mirror_to_gh(self, project_name):
assert project_name
source = self.gitea_projects[project_name]
dest = None
if project_name in self.github_projects:
dest = self.github_projects[project_name]
else:
print("{} not found in github, creating".format(project_name))
dest = self.ghu.create_repo(name=source['name'],
description=source['description'],
homepage=source['html_url']) # Gitea project url
repo_dir = os.path.join("tmp", source['name'])
if not os.path.exists(repo_dir):
try:
check_call(["git", "clone", source['ssh_url'], source['name']], cwd="./tmp/")
except:
rmtree(repo_dir)
raise
try:
check_call(["git", "remote", "add", "github", dest.ssh_url], cwd=repo_dir)
except: # Remote already exists
pass
else:
check_call(["git", "fetch"], cwd=repo_dir)
check_call(["git", "pull", "origin", "master"], cwd=repo_dir)
check_call(["git", "push", "-f", "-u", "github", "master"], cwd=repo_dir, env={"SSH_AUTH_SOCK": self.agent})
print("Completed {}\n".format(project_name))
def main():
with open("creds.json") as f:
creds = json.load(f)
with SSHAgent() as agent:
check_call(["ssh-add", creds["identity"]], env={"SSH_AUTH_SOCK": agent})
GiteaToGithubMirror(creds, agent).run()
if __name__ == '__main__':
main()

120
gitlab_to_github.py Executable file
View File

@ -0,0 +1,120 @@
#!/usr/bin/env python3
import json
import gitlab
import github
import os
from shutil import rmtree
import re
from subprocess import check_call, check_output
from time import time
import signal
class SSHAgent(object):
sockre = re.compile(r'SSH_AUTH_SOCK=(.*?);')
pidre = re.compile(r'SSH_AGENT_PID=(\d*);', )
def __init__(self, cluster=None):
self.addargs = []
if cluster:
self.addargs.extend(['-a', os.path.join(cluster, 'agent.sock.{}'.format(int(time())))])
def __enter__(self):
output = check_output(['ssh-agent'] + self.addargs).decode("UTF-8")
self.sock = self.sockre.search(output).group(1)
self.pid = int(self.pidre.search(output).group(1))
return self.sock
def __exit__(self, errtype, value, traceback):
os.kill(self.pid, signal.SIGHUP)
class GitlabToGithubMirror(object):
def __init__(self, creds, agent):
self.creds = creds
self.agent = agent
self.gl = gitlab.Gitlab(self.creds["gitlab"]["url"], self.creds["gitlab"]["token"], api_version=3)
self.gh = github.Github(self.creds["github"]["username"], self.creds["github"]["password"])
self.ghu = self.gh.get_user()
if not os.path.exists("./tmp"):
os.mkdir("./tmp")
def get_gitlab_projects(self):
page = 0
all_projects = {}
while True:
projects = self.gl.projects.owned(per_page=10, page=page)
page += 1
if not projects:
break
for item in projects:
if item.visibility_level != gitlab.VISIBILITY_PUBLIC or "githubmirror" not in item.tag_list:
continue
all_projects[item.name] = item
return all_projects
def get_github_projects(self):
all_projects = {}
for repo in self.ghu.get_repos():
all_projects[repo.name] = repo
return all_projects
def run(self):
self.github_projects = self.get_github_projects()
self.gitlab_projects = self.get_gitlab_projects()
print("{} projects to check".format(len(self.gitlab_projects)))
visited = []
for project_name, project in self.gitlab_projects.items():
if project_name in visited:
continue
visited.append(project_name)
print("({}/{}) Mirroring {}".format(len(visited), len(self.gitlab_projects), project_name))
self.mirror_to_gh(project_name)
def mirror_to_gh(self, project_name):
assert project_name
source = self.gitlab_projects[project_name]
dest = None
if project_name in self.github_projects:
dest = self.github_projects[project_name]
else:
print("{} not found in github, creating".format(project_name))
dest = self.ghu.create_repo(name=source.name,
description=source.description,
homepage=source.web_url) # Gitlab url
repo_dir = os.path.join("tmp", source.name)
if not os.path.exists(repo_dir):
try:
check_call(["git", "clone", source.ssh_url_to_repo, source.name], cwd="./tmp/")
except:
rmtree(repo_dir)
raise
try:
check_call(["git", "remote", "add", "github", dest.ssh_url], cwd=repo_dir)
except: # Remote already exists
pass
else:
check_call(["git", "fetch"], cwd=repo_dir)
check_call(["git", "pull", "origin", "master"], cwd=repo_dir)
check_call(["git", "push", "-f", "-u", "github", "master"], cwd=repo_dir, env={"SSH_AUTH_SOCK": self.agent})
print("Completed {}\n".format(project_name))
def main():
with open("creds.json") as f:
creds = json.load(f)
with SSHAgent() as agent:
check_call(["ssh-add", creds["identity"]], env={"SSH_AUTH_SOCK": agent})
GitlabToGithubMirror(creds, agent).run()
if __name__ == '__main__':
main()

50
proj2gitea.py Normal file
View File

@ -0,0 +1,50 @@
import gitea
from pprint import pprint
import os
from concurrent.futures import ThreadPoolExecutor
from subprocess import check_call
import json
def main():
with open("creds.json") as f:
creds = json.load(f)
client = gitea.Gitea(creds["gitea"]["url"], creds["gitea"]["token"])
projects = [i.name for i in os.scandir("tmp") if i.is_dir()]
# import ipdb ; ipdb.set_trace()
futures = []
with ThreadPoolExecutor(max_workers=1) as pool:
for project in projects:
projdir = os.path.join("tmp", project)
with open(projdir + ".txt") as f:
description = f.read()
new_repo = client.create_repo(**{"auto_init": False,
"description": description,
# "gitignores": "string",
# "license": "string",
"name": project,
"private": True,
# "readme": "string"
})
if "ssh_url" in new_repo:
check_call(["git", "remote", "add", "gitea", new_repo["ssh_url"]], cwd=projdir)
futures.append(pool.submit(check_call,
["git", "push", "gitea"],
cwd=projdir))
errors = []
for item in futures:
e = item.exception()
if e:
errors.append(e)
print("Errors: ", len(errors))
pprint(errors)
if __name__ == '__main__':
main()

9
requirements.txt Normal file
View File

@ -0,0 +1,9 @@
certifi==2018.4.16
chardet==3.0.4
idna==2.7
PyGithub==1.40
PyJWT==1.6.4
python-gitlab==1.5.1
requests==2.19.1
six==1.11.0
urllib3==1.23

19
set_hooks.py Normal file
View File

@ -0,0 +1,19 @@
import gitea
import json
def main():
with open("creds.json") as f:
creds = json.load(f)
client = gitea.Gitea(creds["gitea"]["url"], creds["gitea"]["token"])
with open("hooks.json") as f:
hooks = json.load(f)
for repo, urls in hooks.items():
print(repo)
for url in urls:
print(client.create_hook("dave", repo, url))
if __name__ == '__main__':
main()