Preserve Discord webhook images on entry edits

.vscode/settings.json

@@ -2,13 +2,16 @@
   "cSpell.words": [
     "autoexport",
     "botuser",
+    "domcontentloaded",
     "Genshins",
     "healthcheck",
     "Hoyolab",
+    "KHTML",
     "levelname",
     "Lovinator",
     "markdownified",
     "markdownify",
+    "networkidle",
     "pipx",
     "pyproject",
     "thead",

discord_rss_bot/feeds.py

@@ -100,7 +100,6 @@ class JsonResponseLike(Protocol):
 
 
 MAX_DISCORD_UPLOAD_BYTES: int = 8 * 1024 * 1024
-JPEG_QUALITY_STEPS: tuple[int, ...] = (85, 70, 55, 40)
 SENT_WEBHOOKS_TAG: str = "sent_webhooks"
 SAVE_SENT_WEBHOOKS_TAG: str = "save_sent_webhooks"
 MESSAGE_PAYLOAD_KEYS: tuple[str, ...] = (
@@ -357,6 +356,69 @@ def hash_webhook_payload(payload: JsonObject) -> str:
     return hashlib.sha256(normalized_payload.encode()).hexdigest()
 
 
+def json_object_or_empty(value: JsonValue) -> JsonObject:
+    """Return a JSON object value or an empty object."""
+    return cast("JsonObject", value) if isinstance(value, dict) else {}
+
+
+def json_list_or_empty(value: JsonValue) -> list[JsonValue]:
+    """Return a JSON list value or an empty list."""
+    return cast("list[JsonValue]", value) if isinstance(value, list) else []
+
+
+def has_media_url(value: JsonValue) -> bool:
+    """Return whether an embed media field has a usable URL."""
+    return isinstance(value, dict) and isinstance(value.get("url"), str) and bool(value["url"])
+
+
+def preserve_previous_embed_media(payload: JsonObject, previous_payload: JsonObject) -> JsonObject:
+    """Keep existing embed image fields when an entry update cannot extract a replacement.
+
+    Returns:
+        JsonObject: Payload with previous embed media restored when the update lacks replacement media.
+    """
+    embeds: list[JsonValue] = json_list_or_empty(payload.get("embeds"))
+    previous_embeds: list[JsonValue] = json_list_or_empty(previous_payload.get("embeds"))
+    if not embeds or not previous_embeds:
+        return payload
+
+    merged_payload: JsonObject = cast("JsonObject", json.loads(json.dumps(payload, default=str)))
+    merged_embeds: list[JsonValue] = json_list_or_empty(merged_payload.get("embeds"))
+
+    for index, embed_value in enumerate(merged_embeds):
+        if index >= len(previous_embeds) or not isinstance(embed_value, dict):
+            continue
+
+        previous_embed: JsonObject = json_object_or_empty(previous_embeds[index])
+        embed: JsonObject = cast("JsonObject", embed_value)
+        for media_key in ("image", "thumbnail"):
+            previous_media: JsonValue = previous_embed.get(media_key)
+            if has_media_url(previous_media) and not has_media_url(embed.get(media_key)):
+                embed[media_key] = previous_media
+
+    return merged_payload
+
+
+def get_webhook_message_edit_payload(payload: JsonObject, record: SentWebhookRecord) -> JsonObject:
+    """Return the payload to PATCH to Discord for a saved message edit.
+
+    Returns:
+        JsonObject: Payload suitable for a Discord message edit request.
+    """
+    previous_payload: JsonObject = json_object_or_empty(record.get("payload"))
+    edit_payload: JsonObject = preserve_previous_embed_media(payload, previous_payload)
+
+    previous_embeds: list[JsonValue] = json_list_or_empty(previous_payload.get("embeds"))
+    if edit_payload.get("embeds") == [] and not previous_embeds:
+        edit_payload.pop("embeds", None)
+
+    previous_attachments: list[JsonValue] = json_list_or_empty(previous_payload.get("attachments"))
+    if edit_payload.get("attachments") == [] and not previous_attachments:
+        edit_payload.pop("attachments", None)
+
+    return edit_payload
+
+
 def json_value_to_int(value: JsonValue, default: int = 0) -> int:
     """Convert a simple JSON scalar to int.
 
@@ -661,14 +723,18 @@ def update_sent_webhook_record_for_entry(
         reader,
         use_default_message_on_empty=True,
     )
-    payload: JsonObject = get_webhook_message_payload(webhook)
+    payload: JsonObject = preserve_previous_embed_media(
+        get_webhook_message_payload(webhook),
+        json_object_or_empty(record.get("payload")),
+    )
+    edit_payload: JsonObject = get_webhook_message_edit_payload(payload, record)
     payload_hash: str = hash_webhook_payload(payload)
     if payload_hash == record.get("payload_hash"):
         return record, False, False
 
     now: str = datetime.datetime.now(tz=datetime.UTC).isoformat()
     try:
-        response = edit_sent_webhook_message(webhook_url_value, message_id_value, webhook, payload)
+        response = edit_sent_webhook_message(webhook_url_value, message_id_value, webhook, edit_payload)
     except (AssertionError, RequestException, httpx.HTTPError, OSError, ValueError) as e:
         logger.exception("Failed to edit Discord webhook message %s for entry %s", message_id_value, entry.id)
         return (
@@ -837,7 +903,7 @@ def create_screenshot_webhook(webhook_url: str, entry: Entry, reader: Reader) ->
             len(screenshot_bytes),
         )
 
-        for quality in JPEG_QUALITY_STEPS:
+        for quality in (85, 70, 55, 40):
             jpeg_bytes = capture_full_page_screenshot(
                 entry_link,
                 screenshot_layout=screenshot_layout,

tests/test_feeds.py

@@ -7,6 +7,7 @@ from datetime import UTC
 from datetime import datetime
 from pathlib import Path
 from typing import LiteralString
+from typing import cast
 from unittest.mock import MagicMock
 from unittest.mock import patch
 
@@ -1040,13 +1041,15 @@ def test_update_sent_webhooks_for_modified_entries_edits_changed_payload(
     response.json.return_value = {"id": "message-3"}
     mock_edit_sent_webhook_message.return_value = response
 
-    updated_count = feeds.update_sent_webhooks_for_modified_entries(
+    updated_count: int = feeds.update_sent_webhooks_for_modified_entries(
         reader,
         [("https://example.com/feed.xml", "entry-3")],
     )
 
     assert updated_count == 1
     mock_edit_sent_webhook_message.assert_called_once()
+    edit_payload = mock_edit_sent_webhook_message.call_args.args[3]
+    assert edit_payload == {"content": "New title"}
     records = state[feeds.SENT_WEBHOOKS_TAG]
     assert isinstance(records, list)
     assert isinstance(records[0], dict)
@@ -1058,6 +1061,74 @@ def test_update_sent_webhooks_for_modified_entries_edits_changed_payload(
     assert not records[0]["last_error"]
 
 
+@patch("discord_rss_bot.feeds.edit_sent_webhook_message")
+@patch("discord_rss_bot.feeds.create_webhook_for_entry")
+def test_update_sent_webhook_record_preserves_existing_embed_image_when_updated_entry_has_no_image(
+    mock_create_webhook_for_entry: MagicMock,
+    mock_edit_sent_webhook_message: MagicMock,
+) -> None:
+    previous_image: JsonObject = {
+        "url": "https://example.com/original-image.jpg",
+        "proxy_url": None,
+        "height": None,
+        "width": None,
+    }
+    old_payload: JsonObject = {
+        "content": "",
+        "embeds": [{"description": "Old summary", "image": previous_image, "thumbnail": None}],
+        "attachments": [],
+    }
+    record: feeds.SentWebhookRecord = {
+        "feed_url": "https://example.com/feed.xml",
+        "entry_id": "entry-4",
+        "webhook_url": "https://discord.com/api/webhooks/123/abc",
+        "message_id": "message-4",
+        "payload": old_payload,
+        "payload_hash": feeds.hash_webhook_payload(old_payload),
+        "update_count": 0,
+    }
+
+    entry = MagicMock()
+    entry.id = "entry-4"
+    entry.title = "New title"
+    entry.link = "https://example.com/entry-4"
+    entry.updated = datetime(2026, 5, 8, tzinfo=UTC)
+    entry.feed.url = "https://example.com/feed.xml"
+    entry.feed.title = "Example feed"
+
+    reader = MagicMock()
+    webhook = MagicMock()
+    webhook.json = {
+        "content": "",
+        "embeds": [{"description": "New summary", "image": None, "thumbnail": None}],
+        "attachments": [],
+    }
+    mock_create_webhook_for_entry.return_value = (webhook, "embed")
+
+    response = MagicMock()
+    response.status_code = 200
+    response.text = '{"id": "message-4"}'
+    response.json.return_value = {"id": "message-4"}
+    mock_edit_sent_webhook_message.return_value = response
+
+    updated_record, record_changed, message_was_edited = feeds.update_sent_webhook_record_for_entry(
+        reader,
+        entry,
+        record,
+    )
+
+    assert record_changed is True
+    assert message_was_edited is True
+    edit_payload = mock_edit_sent_webhook_message.call_args.args[3]
+    assert isinstance(edit_payload["embeds"], list)
+    assert edit_payload["embeds"][0]["image"] == previous_image
+    assert isinstance(updated_record["payload"], dict)
+    updated_payload = cast("JsonObject", updated_record["payload"])
+    updated_embeds = cast("list[JsonObject]", updated_payload["embeds"])
+    assert updated_embeds[0]["image"] == previous_image
+    assert updated_record["payload_hash"] == feeds.hash_webhook_payload(updated_payload)
+
+
 def test_update_feeds_and_collect_modified_entries_only_returns_modified_entries() -> None:
     class StubReader:
         def __init__(self) -> None: