Remove pdc calls from owner-sync-pagure.j2 script

Signed-off-by: Lenka Segura <lsegura@redhat.com>

roles/bodhi2/backend/templates/owner-sync-pagure.j2

@@ -22,6 +22,7 @@ from urllib.parse import urljoin
 import multiprocessing.pool
 from math import ceil
 from functools import partial
+import re
 
 import requests
 import koji
@@ -50,6 +51,7 @@ STAGING = {{ 'True' if env == 'staging' else 'False' }}
 HOSTNAME = 'bodhi{{ env_suffix }}.fedoraproject.org'
 IPA_REALM = '{{ ipa_realm }}'
 ENV_SUFFIX = '{{ env_suffix }}'
+BODHI_URL = 'https://bodhi.fedoraproject.org/'
 if STAGING:
     PAGURE_URL = 'https://src.stg.fedoraproject.org/'
     PDC_URL = 'https://pdc.stg.fedoraproject.org/rest_api/v1/'
@@ -186,8 +188,52 @@ def get_branch_and_arches(tag, version):
     return branch, arches
 
 
-def get_pdc_project_name_and_branch(session, namespace, project_name,
-                                    verbose=False):
+def get_active_releases_from_bodhi():
+    bodhi_url = '{0}releases/?exclude_archived=True'.format(BODHI_URL)
+
+    rv = requests.get(bodhi_url, timeout=60)
+
+    if not rv.ok:
+        return []
+
+    active_releases = []
+    rv_json = rv.json()
+    if rv_json['releases']:
+        for release in rv_json['releases']:
+            if re.match(r'^(f|epel)\d{1,2}$', release['branch']):
+                active_releases.append(release['branch'])
+    return list(set(active_releases))
+
+
+def get_project_branches(session, namespace, project_name):
+    """
+    Returns list of branches for the repo from Pagure dist-git.
+    :param logger: A logger object
+    :param url: a string of the URL to Pagure
+    :param namespace: a string determines a type of the repository
+    :param repo_name: a string of the repository name
+    :return: a list of branches
+    """
+    get_branches_url = '{0}api/0/{1}/{2}/git/branches'.format(PAGURE_URL, namespace, project_name)
+
+    headers = {
+        'Accept': 'application/json',
+        'Content-Type': 'application/json'
+    }
+    rv = requests.get(
+            get_branches_url, headers=headers, timeout=60)
+    rv_json = rv.json()
+
+    if not rv.ok:
+        # When specific namespace has no branches, API returns error "Project not found".
+        # Do not fail. Return "no branches found" instead.
+        return project_name, []
+
+    return rv_json.get("branches", ())
+
+
+def get_project_name_and_its_active_branches(session, namespace, active_releases,
+                                             lookaside, project_name, verbose=False):
     """
     Gets the branches on a project. This function is used for mapping.
     :param namespace: string of the namespace the project is in
@@ -196,24 +242,18 @@ def get_pdc_project_name_and_branch(session, namespace, project_name,
     :return: a tuple containing the string of the project and a list of
     branches
     """
-    project_branches_url = '{0}component-branches/'.format(PDC_URL)
-    params = dict(
-        global_component=project_name,
-        type=PDC_TYPES[namespace],
-        active=True,
-    )
     if verbose:
-        print('- Querying {0} {1}'.format(project_branches_url, params))
-    project_branches_rv = session.get(
-        project_branches_url, params=params, verify=VERIFY, timeout=60)
+        print('- Querying pagure distgit for package branches')
+    project_branches = get_project_branches(session, namespace, project_name)
 
-    # If the project's branches can't be reported, just return no branches and
-    # it will be skipped later on
-    if not project_branches_rv.ok:
-        return project_name, []
+    active_package_branches = list(set(active_releases) & set(project_branches)) + ['rawhide']
 
-    data = project_branches_rv.json()
-    return project_name, [branch['name'] for branch in data['results']]
+    # Check if a package is not retired on any of the branches
+    for branch in active_package_branches:
+        if project_name in lookaside[branch]:
+            active_package_branches.remove(branch)
+
+    return project_name, active_package_branches
 
 
 def get_pagure_project_names_from_page(session, namespace, page,
@@ -288,7 +328,7 @@ def get_pagure_project_branches(namespace, package=None, verbose=False):
     # Since we are going to multi-thread, we need to make a partial function
     # call so that all the function needs is an iterable to run
     partial_get_pdc_project_name_and_branch = partial(
-        get_pdc_project_name_and_branch, session, namespace,
+        get_project_name_and_its_active_branches, session, namespace, active_releases, lookaside,
         verbose=verbose)
     # Get a list of tuples in the form of (project, [branch...]), then convert
     # that to a dictionary
@@ -433,6 +473,22 @@ if __name__ == '__main__':
         }
         unique_namespaces.update(namespaces)
 
+    # Let's start with getting the active releases from bodhi
+    active_releases = get_active_releases_from_bodhi()
+
+    # Let's fetch the json files with retired packages per release from lookaside cache
+    # This is a bit ugly, but the idea is to have the latest release removed in favor of rawhide
+    rawhide_active_releases = active_releases[:]
+    rawhide_active_releases.remove(max(rawhide_active_releases))
+    rawhide_active_releases + ['rawhide']
+    # Let's store the json files with retired packages in lookaside
+    lookaside = {}
+    for branch in rawhide_active_releases:
+        url = "https://src.fedoraproject.org/lookaside/retired_in_{0}.json".format(branch)
+        rv = requests.get(url)  # change to session
+        lookaside[branch] = rv.json()
+
+
     # Get all the project to branch mappings for every namespace
     namespace_to_projects = {}
     for namespace in unique_namespaces:
@@ -440,7 +496,7 @@ if __name__ == '__main__':
             print('Querying for all the projects with the namespace "{0}"'
                   .format(namespace))
         namespace_to_projects[namespace] = \
-            get_pagure_project_branches(namespace, package=package, verbose=verbose)
+            get_pagure_project_branches(namespace, active_releases, lookaside, package=package, verbose=verbose)
 
     for tag, info in list(tag_info.items()):
         if verbose: