From be1c0b097bda697227aac1674a37cbcfd8b51ac3 Mon Sep 17 00:00:00 2001
From: Nick Sweeting <git@sweeting.me>
Date: Wed, 12 Aug 2026 22:31:46 -0700
Subject: [PATCH] Preserve snapshot completion from 0.8 betas

---
 .../core/migrations/0023_upgrade_to_0_9_0.py  | 34 +++++++++++--------
 archivebox/tests/test_migrations_08_to_09.py  |  8 +++--
 2 files changed, 26 insertions(+), 16 deletions(-)

diff --git a/archivebox/core/migrations/0023_upgrade_to_0_9_0.py b/archivebox/core/migrations/0023_upgrade_to_0_9_0.py
index 0bcd6b19..2bc9101a 100644
--- a/archivebox/core/migrations/0023_upgrade_to_0_9_0.py
+++ b/archivebox/core/migrations/0023_upgrade_to_0_9_0.py
@@ -318,6 +318,22 @@ def upgrade_core_tables(apps, schema_editor):
             snapshot_cols = get_table_columns("core_snapshot")
             has_added = "added" in snapshot_cols
             has_bookmarked_at = "bookmarked_at" in snapshot_cols
+            inferred_status_sql = """
+                CASE
+                    WHEN EXISTS (
+                        SELECT 1 FROM core_archiveresult
+                        WHERE core_archiveresult.snapshot_id = core_snapshot.id
+                        AND core_archiveresult.status IN ('queued', 'started', 'backoff')
+                    )
+                    THEN 'queued'
+                    WHEN EXISTS (
+                        SELECT 1 FROM core_archiveresult
+                        WHERE core_archiveresult.snapshot_id = core_snapshot.id
+                    )
+                    THEN 'sealed'
+                    ELSE 'queued'
+                END
+            """
 
             if has_added and not has_bookmarked_at:
                 # Migrating from v0.7.2 (has added/updated fields)
@@ -343,20 +359,7 @@ def upgrade_core_tables(apps, schema_editor):
                         COALESCE(added, CURRENT_TIMESTAMP) as created_at,
                         COALESCE(updated, added, CURRENT_TIMESTAMP) as modified_at,
                         updated as downloaded_at,
-                        CASE
-                            WHEN EXISTS (
-                                SELECT 1 FROM core_archiveresult
-                                WHERE core_archiveresult.snapshot_id = core_snapshot.id
-                                AND core_archiveresult.status IN ('queued', 'started', 'backoff')
-                            )
-                            THEN 'queued'
-                            WHEN EXISTS (
-                                SELECT 1 FROM core_archiveresult
-                                WHERE core_archiveresult.snapshot_id = core_snapshot.id
-                            )
-                            THEN 'sealed'
-                            ELSE 'queued'
-                        END as status
+                        {inferred_status_sql} as status
                     FROM core_snapshot;
                 """)
                 print(f"      copied {cursor.rowcount} Snapshots")
@@ -398,6 +401,9 @@ def upgrade_core_tables(apps, schema_editor):
                         END
                         """,
                     )
+                else:
+                    insert_cols.append("status")
+                    select_cols.append(inferred_status_sql)
                 if has_retry_at:
                     insert_cols.append("retry_at")
                     select_cols.append("retry_at")
diff --git a/archivebox/tests/test_migrations_08_to_09.py b/archivebox/tests/test_migrations_08_to_09.py
index 375e99b3..02bcbd0d 100644
--- a/archivebox/tests/test_migrations_08_to_09.py
+++ b/archivebox/tests/test_migrations_08_to_09.py
@@ -408,7 +408,8 @@ def test_migration_preserves_foreign_keys(migration_08_data):
     assert ok, msg
 
 
-def test_migration_preserves_08_timestamp_meanings(migration_08_data):
+@pytest.mark.parametrize("has_snapshot_status", (True, False), ids=("with-status", "without-status"))
+def test_migration_preserves_08_timestamp_meanings(migration_08_data, has_snapshot_status):
     """0.8.x already has separated timestamp/bookmarked_at/created_at/downloaded_at fields."""
     work_dir, db_path, original_data = migration_08_data
     snapshot = original_data["snapshots"][0]
@@ -420,6 +421,8 @@ def test_migration_preserves_08_timestamp_meanings(migration_08_data):
 
     conn = sqlite3.connect(str(db_path))
     cursor = conn.cursor()
+    if not has_snapshot_status:
+        cursor.execute("ALTER TABLE core_snapshot DROP COLUMN status")
     cursor.execute(
         """
         UPDATE core_snapshot
@@ -437,7 +440,7 @@ def test_migration_preserves_08_timestamp_meanings(migration_08_data):
     conn = sqlite3.connect(str(db_path))
     cursor = conn.cursor()
     cursor.execute(
-        "SELECT timestamp, bookmarked_at, created_at, modified_at, downloaded_at FROM core_snapshot WHERE id = ?",
+        "SELECT timestamp, bookmarked_at, created_at, modified_at, downloaded_at, status FROM core_snapshot WHERE id = ?",
         (snapshot["id"],),
     )
     migrated = cursor.fetchone()
@@ -448,6 +451,7 @@ def test_migration_preserves_08_timestamp_meanings(migration_08_data):
     assert migrated[2].startswith("2024-08-28"), migrated[2]
     assert migrated[3].startswith("2024-08-29"), migrated[3]
     assert migrated[4].startswith("2024-08-30"), migrated[4]
+    assert migrated[5] == "sealed"
 
 
 def test_hyphenated_crawl_ids_are_normalized_before_snapshot_saves(migration_08_data):