[PYTHON] Download the raw link of GitLab private repository with CLI

Download gitlab private repository's raw link using CLI

Starting with GitLab 11.5.1, you can no longer download raw links using the PRIVATE-TOKEN header (cf https://about.gitlab.com/releases/2018/11/28/security-release-gitlab-11 -dot-5-dot-1-released / Improper Enforcement of Token Scope). When the raw link was passed, I parsed and converted the URL so that I could download the file itself as well.

#!/usr/bin/python
#gitlab_download

import os
import sys
import json
import shutil
from contextlib import closing

if sys.version_info[0]>=3:
    import http.client as httplib
    from urllib.request import urlparse
    binstdout = sys.stdout.buffer
else:
    import httplib
    from urlparse import urlparse
    binstdout = sys.stdout

url = urlparse(sys.argv[1])
a=url.path[1:].split('/')
projname='/'.join([a[0],a[1]])
revision=a[3]
filepath='%2F'.join(a[4:])

token = os.environ.get('gitlab_token_'+url.hostname.replace('.','_'),'')

with closing(httplib.HTTPSConnection(url.hostname,port=url.port)) as https:
    https.request('GET','/api/v4/projects?search_namespaces=true&search='+projname,None,{'PRIVATE-TOKEN':token})
    resp = https.getresponse()
    jso = json.loads(resp.read())
    proj = next(e for e in jso if e['path_with_namespace']==projname)
    resolvedPath = '/api/v4/projects/'+str(proj['id'])+'/repository/files/'+filepath.replace('/','%2F')+'/raw?ref='+revision
    sys.stderr.write('resolved: https://%s%s\n'%(url.netloc,resolvedPath))
    https.request('GET',resolvedPath,None,{'PRIVATE-TOKEN':token})
    resp = https.getresponse()
    #print(resp.getheaders())
    shutil.copyfileobj(resp,binstdout)

200602

Changed to give search_namespaces = true when converting from project name to project ID.