From cce0ad9f8fadbd43310bb949a9d075130f277dae Mon Sep 17 00:00:00 2001
From: John Doty <john@d0ty.me>
Date: Tue, 16 Jul 2024 06:37:12 -0700
Subject: [PATCH] fix: handle permanent redirects in the database

---
 cry/cli.py      |  5 ++++-
 cry/database.py | 49 +++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 53 insertions(+), 1 deletion(-)

diff --git a/cry/cli.py b/cry/cli.py
index f04e4d0..5164e24 100644
--- a/cry/cli.py
+++ b/cry/cli.py
@@ -163,7 +163,10 @@ def refresh(url):
     results = asyncio.run(feed.fetch_many(feeds))
 
     new_count = 0
-    for d, meta in results:
+    for i, (d, meta) in enumerate(results):
+        if meta.url != feeds[i].url:
+            db.redirect_feed(feeds[i].url, meta.url)
+
         if d is None:
             # Nothing new.
             db.update_meta(meta)
diff --git a/cry/database.py b/cry/database.py
index c63b8f4..334fe7e 100644
--- a/cry/database.py
+++ b/cry/database.py
@@ -80,6 +80,10 @@ def database_path(origin: str) -> pathlib.Path:
     return pathlib.Path.home() / "Dropbox" / "cry" / f"{origin}.db"
 
 
+# TODO: Refactor into:
+# -top level: transactions
+# -bottom level: queries
+# to enable reuse
 class Database:
     db: sqlite3.Connection
     origin: str
@@ -421,3 +425,48 @@ class Database:
                 [status, int(time.time()), url],
             )
             return cursor.rowcount
+
+    def redirect_feed(self, old_url: str, new_url: str):
+        with self.db:
+            cursor = self.db.execute(
+                "SELECT COUNT(1) FROM feeds WHERE url=?", [new_url]
+            )
+            row = cursor.fetchone()
+            if row[0] == 0:
+                self.db.execute(
+                    "UPDATE feeds SET url = ? WHERE url = ?", [new_url, old_url]
+                )
+            else:
+                # Preserve the entries that were under the old url.
+                self.db.execute(
+                    """
+                    UPDATE entries
+                    SET feed_url = ?
+                    WHERE feed_url = ?
+                    ON CONFLICT DO UPDATE
+                    SET
+                      -- NOTE: This is also part of the feed merge algorithm, BUT
+                      --       we implement it here. See the comment in store_feed
+                      --       for the rationale.
+                      inserted_at=MIN(inserted_at, excluded.inserted_at),
+                      title=CASE
+                        WHEN inserted_at < excluded.inserted_at THEN title
+                        ELSE excluded.title
+                      END,
+                      link=CASE
+                        WHEN inserted_at < excluded.inserted_at THEN link
+                        ELSE excluded.link
+                      END
+                    """
+                )
+
+                # Mark the old feed dead.
+                self.db.execute(
+                    """
+                    UPDATE feeds
+                    SET status = ?,
+                        last_fetched_ts = ?
+                    WHERE url = ?
+                    """,
+                    [feed.FEED_STATUS_DEAD, int(time.time()), old_url],
+                )