Add option to create Sourcehut tickets for missing Github issues.

1 files changed, 107 insertions, 18 deletions

diff --git a/import_issues.py b/import_issues.py
index 245cb55..1c27f80 100755
--- a/import_issues.py
+++ b/import_issues.py
@@ -27,16 +27,17 @@
 # Tickets are created in sr.ht via SMTP, so a working mail setup is required.
 # Surely using Sourcehut's API would be better.
 #
-# There are a few of big caveats:
+# There are a few caveats:
 #
 # 1. If all issue IDs from 1 to the max ID are available in your export, and the
 # tracker you import into is a new tracker, then your Gitlab and Sourcehut issue
-# IDs will match up one-to-one.  If not, then all bets are off.  The ticket
-# descriptions in Sourcehut will indicate the original Gitlab ID, but no
-# rewriting of text is done, so issue references will be incorrect.  By default,
-# the script tries to avoid this by checking that all IDs from 1 to the number
-# of issues are in the export, and aborting if not.  To override this, pass
-# --allow-missing-issues.
+# IDs will match up one-to-one, and mentions of one ticket from another will
+# work.  If not, you need to decide how you want to handle this.  You can choose
+# to create empty Sourcehut tickets for the missing Gitlab issues so that IDs
+# still match, by passing --create-missing-issues.  Blank issues will be created
+# then closed.  Alternatively, you can pass --skip-missing-issues to not create
+# any extra Sourcehut tickets, but IDs will not line up.  If one of these issues
+# is needed, this program will tell you.
 #
 # 2. Because emails are used to create tickets, we also assume that emails are
 # processed in the order that they are sent, so that tickets don't get created
@@ -253,6 +254,45 @@ def open_ticket(
     return issue_count
 
 
+def file_missing_ticket(
+        *,
+        smtp,
+        smtp_delay: float,
+        mode: str,
+        srht_owner: str,
+        srht_tracker: str,
+        frm: str,
+        issue_id: int,
+) -> int:
+    global issue_count
+
+    do_mail(
+        smtp=smtp,
+        smtp_delay=smtp_delay,
+        mode=mode,
+        frm=frm,
+        to=f"~{srht_owner}/{srht_tracker}@todo.sr.ht",
+        subject="Missing issue",
+        body=f"Issue {issue_id} is not known.",
+    )
+
+    issue_count += 1
+
+    # TODO Send these emails at the end, so that there isn't such a need for the
+    # previous issue to be processed promptly.
+    close_ticket(
+        smtp=smtp,
+        smtp_delay=smtp_delay,
+        mode=mode,
+        srht_owner=srht_owner,
+        srht_tracker=srht_tracker,
+        frm=frm,
+        issue_id=issue_count,
+        closed_at=None,
+        is_closed=False,  # Save one line of text.
+    )
+
+
 def send_comment(
         *,
         smtp,
@@ -336,7 +376,8 @@ def run(
         frm: str,
         export_dir_path: Path,
         gitlab_project_url: str,
-        allow_missing_issues: bool,
+        skip_missing_issues: bool,
+        create_missing_issues: bool,
 ):
     milestone_jsons = []
     with open(export_dir_path / 'milestones.ndjson') as milestones_file:
@@ -353,10 +394,21 @@ def run(
             issue_jsons.append(json.loads(line))
 
     issue_jsons.sort(key=lambda x: x['iid'])
-    if not allow_missing_issues:
-        assert [x['iid'] for x in issue_jsons] == list(range(1, len(issue_jsons) + 1)), \
-            f"Don't have all issues from 1 to {len(issue_jsons)}, cannot proceed."
 
+    max_issue_id = max(x['iid'] for x in issue_jsons)
+    present_issue_id_set = {x['iid'] for x in issue_jsons}
+    missing_issue_ids = set(range(1, max_issue_id + 1)) - present_issue_id_set
+    if missing_issue_ids and not (skip_missing_issues or create_missing_issues):
+        raise RuntimeError(
+            f"Don't have all issues from 1 to {max_issue_id}, please pass "
+            f"--create-missing-issues or --skip-missing-issues to proceed."
+        )
+
+    issues_by_id = {}
+    for issue_json in issue_jsons:
+        issues_by_id[issue_json['iid']] = issue_json
+
+    # Need to sort notes by date, they seem to come unsorted.
     for issue_json in issue_jsons:
         issue_json['notes'].sort(key=lambda x: x['created_at'])
 
@@ -364,8 +416,31 @@ def run(
 
     issue_id_map: Dict[int, int] = {}
 
-    for issue_json in issue_jsons:
-        gitlab_issue_id = issue_json['iid']
+    # While we're creating tickets, we can't just loop over the sorted
+    # issue_jsons.  We have to loop over potential issue IDs and handle any that
+    # are missing as well.
+    for gitlab_issue_id in range(1, max_issue_id + 1):
+        if gitlab_issue_id not in issues_by_id:
+            if create_missing_issues:
+                file_missing_ticket(
+                    smtp=smtp,
+                    smtp_delay=smtp_delay,
+                    mode=mode,
+                    srht_owner=srht_owner,
+                    srht_tracker=srht_tracker,
+                    frm=frm,
+                    issue_id=gitlab_issue_id,
+                )
+            elif not skip_missing_issues:
+                raise RuntimeError(
+                    f"Internal error, don't know what to do with missing "
+                    f"issue ID {gitlab_issue_id}."
+                )
+
+            continue
+
+        issue_json = issues_by_id[gitlab_issue_id]
+
         author_id = issue_json['author_id']
         created_by: Optional[str]
         if USERS is None:
@@ -393,10 +468,12 @@ def run(
             gitlab_ticket_url=f"{gitlab_project_url}/-/issues/{gitlab_issue_id}",
         )
 
-        if not allow_missing_issues:
+        if not skip_missing_issues:
             assert srht_issue_id == gitlab_issue_id, \
                 f"Internal error, srht_issue_id {srht_issue_id} != " \
-                f"gitlab_issue_id {gitlab_issue_id}."
+                f"gitlab_issue_id {gitlab_issue_id} " \
+                f"(skip_missing_issues={skip_missing_issues}, " \
+                f"create_missing_issues={create_missing_issues})."
 
         issue_id_map[gitlab_issue_id] = srht_issue_id
 
@@ -530,9 +607,15 @@ def main():
     )
 
     parser.add_argument(
-        '--allow-missing-issues',
+        '--skip-missing-issues',
         action='store_true',
-        help="Don't abort if there are missing issue IDs in the export.",
+        help="Skip missing Gitlab issue IDs; GL and sr.ht IDs will not match.",
+    )
+
+    parser.add_argument(
+        '--create-missing-issues',
+        action='store_true',
+        help="Create missing GL issues in sr.ht to make issue IDs match.",
     )
 
     parser.add_argument(
@@ -551,6 +634,11 @@ def main():
     mode = args['mode']
     frm = args['from']
 
+    skip_missing_issues = args['skip_missing_issues']
+    create_missing_issues = args['create_missing_issues']
+    assert not (skip_missing_issues and create_missing_issues), \
+        f"Can accept at most one of --skip-missing-issues and --create-missing-issues."
+
     if mode == 'print':
         smtp = None
     elif mode == 'send':
@@ -584,7 +672,8 @@ def main():
         frm=frm,
         export_dir_path=export_dir_path,
         gitlab_project_url=args['gitlab_project_url'].rstrip('/'),
-        allow_missing_issues=args['allow_missing_issues'],
+        skip_missing_issues=args['skip_missing_issues'],
+        create_missing_issues=args['create_missing_issues'],
     )
 
     if mode == 'send':