Keep the errors as we are seeing them and print or send a report at the end

While the script runs, we're keeping in memory a list of all the errors
we have encountered and at the end of the run we send to the admins a
report with all of them, categorized in a way that will hopefully make
it easier to fix for them.

We're also adding an option to preserve the function to send the report
to the admins but disable sending the notifications to the
users/packagers.
This would be useful to start get an idea of the number of people who
would have issues with the script.

Signed-off-by: Pierre-Yves Chibon <pingou@pingoured.fr>

distgit_bugzilla_sync/script.py

@@ -30,6 +30,7 @@ into bugzilla.
 '''
 
 import argparse
+import collections
 import datetime
 from email.message import EmailMessage
 import itertools
@@ -126,6 +127,7 @@ class BugzillaProxy:
         self.product_cache = {}
         self.user_cache = {}
         self.inverted_user_cache = {}
+        self.errors = []
 
         # Connect to the fedora account system
         self.fas = AccountSystem(
@@ -286,8 +288,13 @@ class BugzillaProxy:
                 initial_cc_emails.append(bz_email)
                 initial_cc_fasnames.append(watcher)
             else:
-                print(f"** {watcher} has no bugzilla_email or mailing_list set "
-                      f"({collection}/{package}) **")
+                self.errors.append(
+                    f"{watcher} has no bugzilla_email or mailing_list set on "
+                    f"({collection}/{package})"
+                )
+                if self.config["verbose"]:
+                    print(f"** {watcher} has no bugzilla_email or mailing_list set "
+                          f"({collection}/{package}) **")
 
         # Add owner to the cclist so comaintainers taking over a bug don't
         # have to do this manually
@@ -499,6 +506,7 @@ class DistgitBugzillaSync:
     _namespace_to_product = None
     _product_to_branch_regex = None
     _branch_regex_to_product = None
+    errors = collections.defaultdict(list)
 
     def send_email(self, from_address, to_address, subject, message, cc_address=None):
         '''Send an email if there's an error.
@@ -615,6 +623,11 @@ class DistgitBugzillaSync:
         parser.add_argument(
             '--print-no-change', action='store_true', default=False,
             help="Print elements that are not being changed as they are checked")
+        parser.add_argument(
+            '--no-user-notifications', dest="user_notifications", action='store_false',
+            default=True,
+            help="Do not notify every packager whose account is wrongly set-up, but do send the "
+            "overall report to the admins")
 
         self.args = parser.parse_args()
 
@@ -723,19 +736,18 @@ class DistgitBugzillaSync:
             if project['namespace'] not in self.env['pdc_types']:
                 project['branches'] = []
                 project['products'] = []
-                if self.env["verbose"]:
-                    print(
-                        f'! Namespace {project["namespace"]} not found in the pdc_type '
-                        f'configuration key, project {project["namespace"]}/{project["name"]} '
-                        'ignored'
-                    )
+                self.errors["configuration"].append(
+                    f'Namespace `{project["namespace"]}` not found in the pdc_type '
+                    f'configuration key, project {project["namespace"]}/{project["name"]} '
+                    'ignored'
+                )
                 continue
 
             pdc_type = self.env['pdc_types'][project['namespace']]
             project['branches'] = pdc_branches.get(pdc_type, {}).get(project['name'], [])
             if not project['branches']:
-                if self.env["verbose"]:
-                    print(f"! No PDC branch found for {project['namespace']}/{project['name']}")
+                self.errors["PDC"].append(
+                    f"No PDC branch found for {project['namespace']}/{project['name']}")
 
             # Products
             products = set()
@@ -782,10 +794,18 @@ class DistgitBugzillaSync:
         if self.env["verbose"]:
             print('Querying {0}'.format(pagure_override_url))
         override_rv = session.get(pagure_override_url, timeout=30)
+        output = {}
         if override_rv.status_code == 200:
-            override_yaml = yaml.safe_load(override_rv.text)
-            return override_yaml.get('bugzilla_contact', {})
-        return {}
+            try:
+                override_yaml = yaml.safe_load(override_rv.text)
+                output = override_yaml.get('bugzilla_contact', {})
+            except yaml.YAMLError:
+                self.errors["SCM overrides"].append(
+                    f"Failed to load yaml file at: {pagure_override_url}")
+            except AttributeError:
+                self.errors["SCM overrides"].append(
+                    f"Invalid yaml file at: {pagure_override_url}")
+        return output
 
     @classmethod
     def main(cls):
@@ -895,33 +915,64 @@ class DistgitBugzillaSync:
                     )
                 except ValueError as e:
                     # A username didn't have a bugzilla address
-                    errors.append(str(e.args))
+                    self.errors["bugzilla_raw"].append(str(e.args))
+                    self.errors["bugzilla"].append(
+                        f"Failed to update: `{product}/{project['name']}`:"
+                        f"\n   {e}"
+                        f"\n   {e.args}"
+                    )
                 except DataChangedError as e:
                     # A Package or Collection was returned via xmlrpc but wasn't
                     # present when we tried to change it
-                    errors.append(str(e.args))
+                    self.errors["bugzilla_raw"].append(str(e.args))
+                    self.errors["bugzilla"].append(
+                        f"Failed to update: `{product}/{project['name']}`: "
+                        f"\n   {e}"
+                        f"\n   {e.args}"
+                    )
                 except xmlrpc.client.ProtocolError as e:
                     # Unrecoverable and likely means that nothing is going to
                     # succeed.
-                    errors.append(str(e.args))
+                    self.errors["bugzilla_raw"].append(str(e.args))
+                    self.errors["bugzilla"].append(
+                        f"Failed to update: `{product}/{project['name']}`: "
+                        f"\n   {e}"
+                        f"\n   {e.args}"
+                    )
                     break
                 except xmlrpc.client.Error as e:
                     # An error occurred in the xmlrpc call.  Shouldn't happen but
                     # we better see what it is
-                    errors.append('%s -- %s' % (project["name"], e.args[-1]))
+                    self.errors["bugzilla_raw"].append('%s -- %s' % (project["name"], e.args[-1]))
+                    self.errors["bugzilla"].append(
+                        f"Failed to update: `{product}/{project['name']}`: "
+                        f"\n   {e}"
+                        f"\n   {e.args}"
+                    )
 
         # Send notification of errors
-        if errors:
+        if self.errors:
+            if not self.env["dryrun"] and self.args.user_notifications:
+                self.notify_users(self.errors["bugzilla"])
+
+            # Build the report for the admins
+            report = ["ERROR REPORT"]
+            for key in ["configuration", "PDC", "SCM overrides", "bugzilla"]:
+                if self.errors[key]:
+                    report.append(key)
+                    report.append('  - {}'.format("\n  - ".join(self.errors[key])))
+                    report.append('')
+
             if self.env["verbose"] or self.env["dryrun"]:
-                print('[DEBUG]', '\n'.join(errors))
+                print("*" * 80)
+                print('\n'.join(report))
             else:
-                self.notify_users(errors)
                 self.send_email(
                     self.env['email']['from'],
                     self.env['email']['notify_admins'],
                     'Errors while syncing bugzilla with the PackageDB',
                     self.env['email']['templates']['admin_notification'].format(
-                        errors='\n'.join(errors)
+                        errors='\n'.join(report)
                     )
                 )
         else: