Recommend feeds when creating new feed if broken

discord_rss_bot/feeds.py

@@ -79,6 +79,10 @@ type SentWebhookRecord = dict[str, JsonValue]
 type UpdateCallback = Callable[[], UpdatedFeed | None]
 
 
+class FeedUpdateError(HTTPException):
+    """Raised when the initial update for a newly added feed fails."""
+
+
 class JsonResponseLike(Protocol):
     """Response interface needed for Discord webhook JSON parsing."""
 
@@ -1899,6 +1903,7 @@ def create_feed(reader: Reader, feed_url: str, webhook_dropdown: str) -> None:
         webhook_dropdown: The webhook we should send entries to.
 
     Raises:
+        FeedUpdateError: If the initial feed update fails.
         HTTPException: If webhook_dropdown does not equal a webhook or default_custom_message not found.
     """
     clean_feed_url: str = feed_url.strip()
@@ -1929,7 +1934,7 @@ def create_feed(reader: Reader, feed_url: str, webhook_dropdown: str) -> None:
     try:
         reader.update_feed(clean_feed_url)
     except ReaderError as e:
-        raise HTTPException(status_code=404, detail=f"Error updating feed: {e}") from e
+        raise FeedUpdateError(status_code=404, detail=f"Error updating feed: {e}") from e
 
     # Mark every entry as read, so we don't send all the old entries to Discord.
     entries: Iterable[Entry] = reader.get_entries(feed=clean_feed_url, read=False)

discord_rss_bot/main.py

@@ -52,6 +52,7 @@ from discord_rss_bot.custom_message import get_embed
 from discord_rss_bot.custom_message import get_first_image
 from discord_rss_bot.custom_message import replace_tags_in_text_message
 from discord_rss_bot.custom_message import save_embed
+from discord_rss_bot.feeds import FeedUpdateError
 from discord_rss_bot.feeds import SentWebhookRecord
 from discord_rss_bot.feeds import coerce_media_gallery_image_limit
 from discord_rss_bot.feeds import create_feed
@@ -122,6 +123,12 @@ class FilterPreviewContext(TypedDict):
     preview_helper_text: str
 
 
+class AutodiscoverLink(TypedDict):
+    href: str
+    type: str | None
+    title: str | None
+
+
 LOGGING_CONFIG = {
     "version": 1,
     "disable_existing_loggers": False,
@@ -262,6 +269,57 @@ templates.env.globals["get_backup_path"] = get_backup_path  # pyright: ignore[re
 templates.env.globals["has_webhooks"] = has_webhooks  # pyright: ignore[reportArgumentType]
 
 
+def get_global_delivery_mode(reader: Reader) -> str:
+    """Return the normalized default delivery mode for new feeds.
+
+    Args:
+        reader: The Reader instance.
+
+    Returns:
+        The configured delivery mode, falling back to embed.
+    """
+    global_delivery_mode: str = str(reader.get_tag((), "delivery_mode", "embed")).strip().lower()
+    return global_delivery_mode if global_delivery_mode in {"embed", "text"} else "embed"
+
+
+def get_autodiscover_links(reader: Reader, feed_url: str) -> list[AutodiscoverLink]:
+    """Return valid autodiscovered links stored for a failed feed update.
+
+    Args:
+        reader: The Reader instance.
+        feed_url: The URL that failed to parse as a feed.
+
+    Returns:
+        Valid discovered feed link dictionaries.
+    """
+    try:
+        stored_links = reader.get_tag(feed_url, ".reader.autodiscover", [])
+    except ReaderError:
+        return []
+
+    if not isinstance(stored_links, list):
+        return []
+
+    links: list[AutodiscoverLink] = []
+    for stored_link in stored_links:
+        if not isinstance(stored_link, dict):
+            continue
+
+        href = stored_link.get("href")
+        if not isinstance(href, str) or not href:
+            continue
+
+        link_type = stored_link.get("type")
+        title = stored_link.get("title")
+        links.append({
+            "href": href,
+            "type": link_type if isinstance(link_type, str) else None,
+            "title": title if isinstance(title, str) else None,
+        })
+
+    return links
+
+
 @app.post("/add_webhook")
 async def post_add_webhook(
     webhook_name: Annotated[str, Form()],
@@ -357,24 +415,51 @@ async def post_delete_webhook(
     return RedirectResponse(url="/", status_code=303)
 
 
-@app.post("/add")
+@app.post("/add", response_model=None)
 async def post_create_feed(
+    request: Request,
     feed_url: Annotated[str, Form()],
     webhook_dropdown: Annotated[str, Form()],
     reader: Annotated[Reader, Depends(get_reader_dependency)],
-) -> RedirectResponse:
+) -> RedirectResponse | HTMLResponse:
     """Add a feed to the database.
 
     Args:
+        request: The request object.
         feed_url: The feed to add.
         webhook_dropdown: The webhook to use.
         reader: The Reader instance.
 
     Returns:
         RedirectResponse: Redirect to the feed page.
+
+    Raises:
+        FeedUpdateError: If updating the feed fails without discovered feed links.
     """
     clean_feed_url: str = feed_url.strip()
-    create_feed(reader, feed_url, webhook_dropdown)
+    try:
+        create_feed(reader, feed_url, webhook_dropdown)
+    except FeedUpdateError as exception:
+        autodiscover_links = get_autodiscover_links(reader, clean_feed_url)
+        if not autodiscover_links:
+            raise
+
+        context = {
+            "request": request,
+            "webhooks": reader.get_tag((), "webhooks", []),
+            "global_delivery_mode": get_global_delivery_mode(reader),
+            "feed_url": clean_feed_url,
+            "selected_webhook": webhook_dropdown,
+            "messages": exception.detail,
+            "autodiscover_links": autodiscover_links,
+        }
+        return templates.TemplateResponse(
+            request=request,
+            name="add.html",
+            context=context,
+            status_code=exception.status_code,
+        )
+
     commit_state_change(reader, f"Add feed {clean_feed_url}")
     return RedirectResponse(url=f"/feed?feed_url={urllib.parse.quote(clean_feed_url)}", status_code=303)
 
@@ -1609,14 +1694,10 @@ def get_add(
     Returns:
         HTMLResponse: The add feed page.
     """
-    global_delivery_mode: str = str(reader.get_tag((), "delivery_mode", "embed")).strip().lower()
-    if global_delivery_mode not in {"embed", "text"}:
-        global_delivery_mode = "embed"
-
     context = {
         "request": request,
         "webhooks": reader.get_tag((), "webhooks", []),
-        "global_delivery_mode": global_delivery_mode,
+        "global_delivery_mode": get_global_delivery_mode(reader),
     }
     return templates.TemplateResponse(request=request, name="add.html", context=context)
 

discord_rss_bot/settings.py

@@ -3,6 +3,7 @@ from __future__ import annotations
 import os
 import typing
 from functools import lru_cache
+from importlib.util import find_spec
 from pathlib import Path
 
 from platformdirs import user_data_dir
@@ -31,6 +32,31 @@ default_custom_embed: dict[str, str] = {
 }
 
 
+def has_plugin(plugin_name: str) -> bool:
+    """Return whether the installed reader version provides a built-in plugin.
+
+    We started using .autodiscover, but that is from Reader version 3.25.
+    """
+    try:
+        return find_spec(f"reader.plugins.{plugin_name.removeprefix('.')}") is not None
+    except ModuleNotFoundError:
+        return False
+
+
+def make_app_reader(db_location: Path) -> Reader:
+    """Create a reader with plugins supported by the installed reader version.
+
+    Returns:
+        The configured reader.
+    """
+    plugins_we_want = (".ua_fallback", ".autodiscover")
+    plugins: list[str] = [name for name in plugins_we_want if has_plugin(name)]
+
+    if plugins:
+        return make_reader(url=str(db_location), plugins=plugins)
+    return make_reader(url=str(db_location))
+
+
 @lru_cache(maxsize=1)
 def get_reader(custom_location: Path | None = None) -> Reader:
     """Get the reader.
@@ -42,7 +68,7 @@ def get_reader(custom_location: Path | None = None) -> Reader:
         The reader.
     """
     db_location: Path = custom_location or Path(data_dir) / "db.sqlite"
-    reader: Reader = make_reader(url=str(db_location))
+    reader: Reader = make_app_reader(db_location)
 
     # https://reader.readthedocs.io/en/latest/api.html#reader.types.UpdateConfig
     # Set the default update interval to 15 minutes if not already configured

discord_rss_bot/templates/add.html

@@ -8,11 +8,6 @@
 {% block content %}
     <div class="p-2 border border-dark">
         <form action="/add" method="post">
-            <div class="mb-3 text-muted">
-                New feeds currently default to
-                <strong>{{ global_delivery_mode }}</strong>
-                delivery mode.
-            </div>
             <!-- Feed URL -->
             <div class="row pb-2">
                 <label for="feed_url" class="col-sm-2 col-form-label">Feed URL</label>
@@ -20,7 +15,8 @@
                     <input name="feed_url"
                            type="text"
                            class="form-control bg-dark border-dark text-muted"
-                           id="feed_url" />
+                           id="feed_url"
+                           value="{{ feed_url }}" />
                 </div>
             </div>
             <!-- Webhook dropdown -->
@@ -30,8 +26,11 @@
                     <select class="col-auto form-select bg-dark border-dark text-muted"
                             id="webhook_dropdown"
                             name="webhook_dropdown">
-                        <option selected>Choose webhook...</option>
-                        {% for hook in webhooks %}<option value="{{ hook.name }}">{{- hook.name -}}</option>{% endfor %}
+                        <option {% if not selected_webhook %}selected{% endif %}>Choose webhook...</option>
+                        {% for hook in webhooks %}
+                            <option value="{{ hook.name }}"
+                                    {% if hook.name == selected_webhook %}selected{% endif %}>{{ hook.name }}</option>
+                        {% endfor %}
                     </select>
                 </div>
             </div>
@@ -41,4 +40,28 @@
             </div>
         </form>
     </div>
+    {% if autodiscover_links %}
+        <section class="card border border-dark shadow-sm text-light rounded-0 mt-3">
+            <div class="card-body p-3 p-md-4">
+                <h2 class="h6 text-uppercase text-muted mb-2">Discovered feed links</h2>
+                <p class="text-muted mb-3">The submitted URL is not a feed. Choose one of the feeds advertised by that page.</p>
+                <div class="d-flex flex-column gap-2">
+                    {% for link in autodiscover_links %}
+                        <form action="/add"
+                              method="post"
+                              class="d-flex flex-column flex-md-row align-items-md-center gap-3 p-3 border border-dark rounded-0">
+                            <input type="hidden" name="feed_url" value="{{ link.href }}" />
+                            <input type="hidden" name="webhook_dropdown" value="{{ selected_webhook }}" />
+                            <div class="flex-grow-1 feed-page__content">
+                                {% if link.title %}<div class="text-light fw-semibold">{{ link.title }}</div>{% endif %}
+                                <code class="d-block text-muted feed-page__wrap">{{ link.href }}</code>
+                                {% if link.type %}<div class="small text-muted mt-1">{{ link.type }}</div>{% endif %}
+                            </div>
+                            <button class="btn btn-outline-light btn-sm ms-md-auto" type="submit">Add feed</button>
+                        </form>
+                    {% endfor %}
+                </div>
+            </div>
+        </section>
+    {% endif %}
 {% endblock content %}