From 37d2f77964b0c717c0d52ab2e0d32f8e5b57f918 Mon Sep 17 00:00:00 2001
From: Ralph Bean <rbean@redhat.com>
Date: Nov 09 2017 02:20:25 +0000
Subject: A script for finding which packages have EPEL BZ components, that shouldn't.


Signed-off-by: Ralph Bean <rbean@redhat.com>

---

diff --git a/scripts/bugzilla/list-bad-epel.py b/scripts/bugzilla/list-bad-epel.py
new file mode 100644
index 0000000..348ce95
--- /dev/null
+++ b/scripts/bugzilla/list-bad-epel.py
@@ -0,0 +1,63 @@
+""" A script for listing which components have bz epel entries that shouldn't.
+
+https://pagure.io/releng/issue/7148
+"""
+
+import bugzilla
+import dogpile.cache
+import requests
+
+bz = bugzilla.Bugzilla('https://bugzilla.redhat.com')
+cache = dogpile.cache.make_region().configure(
+    'dogpile.cache.dbm',
+    arguments=dict(
+        filename='/var/tmp/bugzilla-component-cache.dbm',
+        expiration_time=-1,
+    )
+)
+
+
+@cache.cache_on_arguments()
+def get_pdc_data(url):
+    response = requests.get(url)
+    if not response:
+        raise ValueError("%r failed %r" % (url, response))
+    return response.json()
+
+
+@cache.cache_on_arguments()
+def in_bugzilla(component):
+    matches = bz._proxy.Component.get(dict(names=[
+        dict(product='Fedora EPEL', component=component)]))
+    return bool(matches['components'])
+
+
+def get_all_branches_from_pdc():
+    next = 'https://pdc.fedoraproject.org/rest_api/v1/component-branches' + \
+        '?type=rpm&page_size=100&page=1'
+    while next:
+        print("Getting %r" % next)
+        data = get_pdc_data(next)
+        for entry in data['results']:
+            yield entry
+        next = data['next']
+
+
+all_packages = set()
+in_epel_in_pdc = set()
+in_epel_in_bz = set()
+for branch in get_all_branches_from_pdc():
+    all_packages.add(branch['global_component'])
+    if branch['name'].startswith('epel') or branch['name'].startswith('el'):
+        in_epel_in_pdc.add(branch['global_component'])
+    if in_bugzilla(branch['global_component']):
+        in_epel_in_bz.add(branch['global_component'])
+    print("All: %r, EPEL(PDC): %r, EPEL(BZ): %r, "
+          "In BZ and shouldn't be: %r, "
+          "Not in BZ and should be: %r" % (
+              len(all_packages),
+              len(in_epel_in_pdc),
+              len(in_epel_in_bz),
+              len(in_epel_in_bz - in_epel_in_pdc),
+              len(in_epel_in_pdc - in_epel_in_bz),
+          ))
\ No newline at end of file