command to add blog

src/memory/common/celery_app.py

@@ -41,6 +41,7 @@ UPDATE_METADATA_FOR_ITEM = f"{MAINTENANCE_ROOT}.update_metadata_for_item"
 SYNC_WEBPAGE = f"{BLOGS_ROOT}.sync_webpage"
 SYNC_ARTICLE_FEED = f"{BLOGS_ROOT}.sync_article_feed"
 SYNC_ALL_ARTICLE_FEEDS = f"{BLOGS_ROOT}.sync_all_article_feeds"
+ADD_ARTICLE_FEED = f"{BLOGS_ROOT}.add_article_feed"
 SYNC_WEBSITE_ARCHIVE = f"{BLOGS_ROOT}.sync_website_archive"
 
 

src/memory/common/settings.py

@@ -106,6 +106,7 @@ ARTICLE_FEED_SYNC_INTERVAL = int(os.getenv("ARTICLE_FEED_SYNC_INTERVAL", 30 * 60
 CLEAN_COLLECTION_INTERVAL = int(os.getenv("CLEAN_COLLECTION_INTERVAL", 24 * 60 * 60))
 CHUNK_REINGEST_INTERVAL = int(os.getenv("CHUNK_REINGEST_INTERVAL", 60 * 60))
 NOTES_SYNC_INTERVAL = int(os.getenv("NOTES_SYNC_INTERVAL", 15 * 60))
+LESSWRONG_SYNC_INTERVAL = int(os.getenv("LESSWRONG_SYNC_INTERVAL", 60 * 60 * 24))
 
 CHUNK_REINGEST_SINCE_MINUTES = int(os.getenv("CHUNK_REINGEST_SINCE_MINUTES", 60 * 24))
 

src/memory/parsers/archives.py

@@ -295,4 +295,7 @@ feeds = [
     "https://karlin.blog/",
     "https://slatestarcodex.com/",
     "https://www.astralcodexten.com/",
+    "https://nayafia.substack.com",
+    "https://homosabiens.substack.com",
+    "https://usefulfictions.substack.com",
 ]

src/memory/parsers/blogs.py

@@ -609,6 +609,9 @@ PARSER_REGISTRY = {
     r"rachelbythebay\.com": RachelByTheBayParser,
     r"nadia\.xyz": NadiaXyzParser,
     r"slatestarcodex\.com": SlateStarCodexParser,
+    r"nayafia\.substack\.com": SubstackParser,
+    r"homosabiens\.substack\.com": SubstackParser,
+    r"usefulfictions\.substack\.com": SubstackParser,
 }
 
 

src/memory/parsers/lesswrong.py

@@ -237,6 +237,7 @@ def fetch_lesswrong(
 
 def fetch_lesswrong_posts(
     since: datetime | None = None,
+    until: datetime | None = None,
     min_karma: int = 10,
     limit: int = 50,
     cooldown: float = 0.5,
@@ -280,6 +281,9 @@ def fetch_lesswrong_posts(
             break
 
         for post in page_posts:
+            published_at = post.get("published_at")
+            if published_at and until and published_at > until:
+                break
             yield post
 
         last_item = page_posts[-1]

src/memory/workers/ingest.py

@@ -8,6 +8,7 @@ from memory.common.celery_app import (
     SYNC_ALL_COMICS,
     SYNC_ALL_ARTICLE_FEEDS,
     TRACK_GIT_CHANGES,
+    SYNC_LESSWRONG,
 )
 
 logger = logging.getLogger(__name__)
@@ -38,4 +39,8 @@ app.conf.beat_schedule = {
         "task": TRACK_GIT_CHANGES,
         "schedule": settings.NOTES_SYNC_INTERVAL,
     },
+    "sync-lesswrong": {
+        "task": SYNC_LESSWRONG,
+        "schedule": settings.LESSWRONG_SYNC_INTERVAL,
+    },
 }

src/memory/workers/tasks/blogs.py

@@ -12,6 +12,7 @@ from memory.common.celery_app import (
     SYNC_WEBPAGE,
     SYNC_ARTICLE_FEED,
     SYNC_ALL_ARTICLE_FEEDS,
+    ADD_ARTICLE_FEED,
     SYNC_WEBSITE_ARCHIVE,
 )
 from memory.workers.tasks.content_processing import (
@@ -169,10 +170,52 @@ def sync_all_article_feeds() -> list[dict]:
         return results
 
 
+@app.task(name=ADD_ARTICLE_FEED)
+def add_article_feed(
+    url: str,
+    title: str | None = None,
+    description: str | None = None,
+    tags: Iterable[str] = [],
+    active: bool = True,
+    check_interval: int = 60 * 24,  # 24 hours
+) -> dict:
+    """
+    Add a new ArticleFeed.
+
+    Args:
+        url: URL of the feed
+        title: Title of the feed (optional)
+        description: Description of the feed (optional)
+        tags: Tags to apply to the feed
+        active: Whether the feed is active
+        check_interval: Interval in minutes to check the feed
+
+    Returns:
+        dict: Summary of the added feed
+    """
+    with make_session() as session:
+        feed = session.query(ArticleFeed).filter(ArticleFeed.url == url).first()
+        if feed:
+            logger.info(f"Feed already exists: {url}")
+            return {"status": "error", "error": "Feed already exists"}
+
+        feed = ArticleFeed(
+            url=url,
+            title=title or url,
+            description=description,
+            active=active,
+            check_interval=check_interval,
+            tags=tags,
+        )
+        session.add(feed)
+        session.commit()
+        return {"status": "success", "feed_id": feed.id}
+
+
 @app.task(name=SYNC_WEBSITE_ARCHIVE)
 @safe_task_execution
 def sync_website_archive(
-    url: str, tags: Iterable[str] = [], max_pages: int = 100
+    url: str, tags: Iterable[str] = [], max_pages: int = 100, add_feed: bool = True
 ) -> dict:
     """
     Synchronize all articles from a website's archive.
@@ -187,6 +230,16 @@ def sync_website_archive(
     """
     logger.info(f"Starting archive sync for: {url}")
 
+    if add_feed:
+        with make_session() as session:
+            feed = session.query(ArticleFeed).filter(ArticleFeed.url == url).first()
+            if not feed:
+                feed = ArticleFeed(
+                    url=url,
+                    title=url,
+                    active=True,
+                )
+
     # Get archive fetcher for the website
     fetcher = get_archive_fetcher(url)
     if not fetcher:
@@ -200,10 +253,10 @@ def sync_website_archive(
     new_articles = 0
     task_ids = []
 
-    for feed_item in fetcher.fetch_all_items():
+    with make_session() as session:
-        articles_found += 1
+        for feed_item in fetcher.fetch_all_items():
+            articles_found += 1
 
-        with make_session() as session:
             existing = check_content_exists(session, BlogPost, url=feed_item.url)
             if existing:
                 continue

src/memory/workers/tasks/forums.py

@@ -49,7 +49,8 @@ def sync_lesswrong_post(
 @app.task(name=SYNC_LESSWRONG)
 @safe_task_execution
 def sync_lesswrong(
-    since: str = (datetime.now() - timedelta(days=30)).isoformat(),
+    since: str | None = None,
+    until: str | None = None,
     min_karma: int = 10,
     limit: int = 50,
     cooldown: float = 0.5,
@@ -57,9 +58,27 @@ def sync_lesswrong(
     af: bool = False,
     tags: list[str] = [],
 ):
+    if until:
+        end_date = datetime.fromisoformat(until)
+    else:
+        end_date = datetime.now() - timedelta(hours=8)
+
     logger.info(f"Syncing LessWrong posts since {since}")
-    start_date = datetime.fromisoformat(since)
+
-    posts = fetch_lesswrong_posts(start_date, min_karma, limit, cooldown, max_items, af)
+    if since:
+        start_date = datetime.fromisoformat(since)
+    else:
+        start_date = end_date - timedelta(days=30)
+
+    posts = fetch_lesswrong_posts(
+        since=start_date,
+        until=end_date,
+        min_karma=min_karma,
+        limit=limit,
+        cooldown=cooldown,
+        max_items=max_items,
+        af=af,
+    )
 
     posts_num, new_posts = 0, 0
     with make_session() as session:

tests/memory/workers/tasks/test_forums_tasks.py

@@ -1,5 +1,5 @@
 import pytest
-from datetime import datetime, timedelta, timezone
+from datetime import datetime, timezone
 from unittest.mock import Mock, patch
 
 from memory.common.db.models import ForumPost
@@ -228,17 +228,19 @@ def test_sync_lesswrong_success(mock_fetch, mock_lesswrong_post, db_session):
         assert result["since"] == "2024-01-01T00:00:00"
         assert result["min_karma"] == 10
         assert result["max_items"] == 100
-        assert result["af"] == False
+        assert not result["af"]
 
-        # Verify fetch_lesswrong_posts was called with correct arguments
+        # Verify fetch_lesswrong_posts was called with correct arguments (kwargs)
-        mock_fetch.assert_called_once_with(
+        mock_fetch.assert_called_once()
-            datetime.fromisoformat("2024-01-01T00:00:00"),
+        kwargs = mock_fetch.call_args.kwargs
-            10,  # min_karma
+        assert kwargs["since"] == datetime.fromisoformat("2024-01-01T00:00:00")
-            50,  # limit
+        assert kwargs["min_karma"] == 10
-            0.1,  # cooldown
+        assert kwargs["limit"] == 50
-            100,  # max_items
+        assert kwargs["cooldown"] == 0.1
-            False,  # af
+        assert kwargs["max_items"] == 100
-        )
+        assert kwargs["af"] is False
+        assert "until" in kwargs
+        assert isinstance(kwargs["until"], datetime)
 
         # Verify sync_lesswrong_post was called for the new post
         mock_sync_post.delay.assert_called_once_with(mock_lesswrong_post, ["test"])
@@ -343,11 +345,14 @@ def test_sync_lesswrong_since_parameter(mock_fetch, db_session):
     forums.sync_lesswrong(since="2024-01-01T00:00:00")
     expected_since = datetime.fromisoformat("2024-01-01T00:00:00")
 
-    # Verify fetch was called with correct since date
+    # Verify fetch was called with correct since date (kwargs)
-    call_args = mock_fetch.call_args[0]
+    kwargs = mock_fetch.call_args.kwargs
-    actual_since = call_args[0]
+    actual_since = kwargs["since"]
 
     assert actual_since == expected_since
+    assert "until" in kwargs
+    assert isinstance(kwargs["until"], datetime)
+    assert kwargs["until"] >= actual_since
 
 
 @pytest.mark.parametrize(
@@ -373,14 +378,14 @@ def test_sync_lesswrong_parameters(
         max_items=500,
     )
 
-    # Verify fetch was called with correct parameters
+    # Verify fetch was called with correct parameters (kwargs)
-    call_args = mock_fetch.call_args[0]
+    kwargs = mock_fetch.call_args.kwargs
 
-    assert call_args[1] == min_karma  # min_karma
+    assert kwargs["min_karma"] == min_karma
-    assert call_args[2] == limit  # limit
+    assert kwargs["limit"] == limit
-    assert call_args[3] == cooldown  # cooldown
+    assert kwargs["cooldown"] == cooldown
-    assert call_args[4] == 500  # max_items
+    assert kwargs["max_items"] == 500
-    assert call_args[5] == af_value  # af
+    assert kwargs["af"] == af_value
 
     assert result["min_karma"] == min_karma
     assert result["af"] == af_value

tools/run_celery_task.py

@@ -27,6 +27,7 @@ from memory.common import settings
 from memory.common.celery_app import (
     SYNC_ALL_ARTICLE_FEEDS,
     SYNC_ARTICLE_FEED,
+    ADD_ARTICLE_FEED,
     SYNC_WEBPAGE,
     SYNC_WEBSITE_ARCHIVE,
     SYNC_ALL_COMICS,
@@ -49,6 +50,7 @@ from memory.common.celery_app import (
     UPDATE_METADATA_FOR_ITEM,
     UPDATE_METADATA_FOR_SOURCE_ITEMS,
     SETUP_GIT_NOTES,
+    TRACK_GIT_CHANGES,
     app,
 )
 
@@ -78,6 +80,7 @@ TASK_MAPPINGS = {
         "sync_article_feed": SYNC_ARTICLE_FEED,
         "sync_all_article_feeds": SYNC_ALL_ARTICLE_FEEDS,
         "sync_website_archive": SYNC_WEBSITE_ARCHIVE,
+        "add_article_feed": ADD_ARTICLE_FEED,
     },
     "comic": {
         "sync_all_comics": SYNC_ALL_COMICS,
@@ -92,6 +95,7 @@ TASK_MAPPINGS = {
     },
     "notes": {
         "setup_git_notes": SETUP_GIT_NOTES,
+        "track_git_changes": TRACK_GIT_CHANGES,
     },
 }
 QUEUE_MAPPINGS = {
@@ -249,6 +253,13 @@ def notes_setup_git_notes(ctx, origin, email, name):
     execute_task(ctx, "notes", "setup_git_notes", origin=origin, email=email, name=name)
 
 
+@notes.command("track-git-changes")
+@click.pass_context
+def notes_track_git_changes(ctx):
+    """Track git changes."""
+    execute_task(ctx, "notes", "track_git_changes")
+
+
 @cli.group()
 @click.pass_context
 def maintenance(ctx):
@@ -376,6 +387,34 @@ def blogs_sync_website_archive(ctx, url):
     execute_task(ctx, "blogs", "sync_website_archive", url=url)
 
 
+@blogs.command("add-article-feed")
+@click.option("--url", required=True, help="URL of the feed")
+@click.option("--title", help="Title of the feed")
+@click.option("--description", help="Description of the feed")
+@click.option("--tags", help="Comma-separated tags to apply to the feed", default="")
+@click.option("--active", is_flag=True, help="Whether the feed is active")
+@click.option(
+    "--check-interval",
+    type=int,
+    help="Interval in minutes to check the feed",
+    default=60 * 24,  # 24 hours
+)
+@click.pass_context
+def blogs_add_article_feed(ctx, url, title, description, tags, active, check_interval):
+    """Add a new article feed."""
+    execute_task(
+        ctx,
+        "blogs",
+        "add_article_feed",
+        url=url,
+        title=title,
+        description=description,
+        tags=tags.split(","),
+        active=active,
+        check_interval=check_interval,
+    )
+
+
 @cli.group()
 @click.pass_context
 def comic(ctx):