youpdated 0.3.0__tar.gz → 0.3.1__tar.gz

{youpdated-0.3.0 → youpdated-0.3.1}/CHANGELOG.md

@@ -4,6 +4,39 @@ All notable changes to this project are documented here. The format follows
 [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), and this project adheres to
 [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.3.1] — 2026-10-10
+
+### Fixed
+
+- **itch builds no longer re-report on unchanged pages.** itch.io shuffles HTML
+  attribute order, and the upload list was only read in one
+  order, so some fetches saw no files and fingerprinted a "new build". Successive
+  runs each reported a different handful of games until every variant had been
+  seen. Upload and timestamp parsing no longer depend on attribute order.
+- **YouTube Shorts no longer re-report when the feed falls back.** The official
+  feed links Shorts as `/shorts/<id>`, which was not recognised, so Shorts were
+  recorded as `yt:video:yt:video:<id>` there and `yt:video:<id>` through Invidious
+  or Data API. Each switch reported Shorts again. Existing
+  history is migrated on first open, so nothing already seen comes back.
+- **`--no-save` really leaves the state alone.** It still stored each feed's
+  ETag/Last-Modified, so the next real run was answered 304 and silently skipped
+  whatever the preview had shown. (Whoops)
+- **A target that fails partway no longer hides its new items.** If one of a
+  target's fetches failed (e.g. GitHub `watch: [releases, commits]` with commits
+  down), the others' validators were still stored while their items were
+  discarded, so the next run got a 304 and never reported them. Validators are
+  now stored only alongside the items they cover.
+- **Feeds listed oldest-first report new entries.** Entries were cut to the
+  limit in document order before sorting, so on an oldest-first feed longer than
+  the limit the newest posts never made it in. Affects `feed`, Steam, itch
+  devlogs, and GitHub atom feeds.
+- **Chrome no longer errors** when a version is first listed without a rollout
+  start time and later with.
+- **Removing an `ignore:` rule shows what it hid.** Ignored items
+  are not recorded, but the feed's validators were, so hidden items stayed hidden until it next changed.
+  Each target's rules are now recorded, and a target whose rules changed is fetched fully once.
+  Every target is fetched in full once after upgrading.
+
 ## [0.3.0] — 2026-10-07
 
 ### Added
@@ -237,6 +270,7 @@ First release.
 - Firefox publishes current versions, so it reports one item per channel.
 - Edge exposes release notes only for the stable and beta channels. (But like, it's Edge, why do you want to know when it updates?)
 
+[0.3.1]: https://github.com/Void1-1/youpdated/releases/tag/v0.3.1
 [0.3.0]: https://github.com/Void1-1/youpdated/releases/tag/v0.3.0
 [0.2.1]: https://github.com/Void1-1/youpdated/releases/tag/v0.2.1
 [0.2.0]: https://github.com/Void1-1/youpdated/releases/tag/v0.2.0

{youpdated-0.3.0 → youpdated-0.3.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: youpdated
-Version: 0.3.0
+Version: 0.3.1
 Summary: Simple update tracker for games, apps, and packages
 Author-email: Void1-1 <161782542+Void1-1@users.noreply.github.com>
 License-Expression: MIT

{youpdated-0.3.0 → youpdated-0.3.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "youpdated"
-version = "0.3.0"
+version = "0.3.1"
 description = "Simple update tracker for games, apps, and packages"
 readme = "README.md"
 requires-python = ">=3.11"

{youpdated-0.3.0 → youpdated-0.3.1}/youpdated/__init__.py

@@ -1,3 +1,3 @@
 """Youpdated, a simple update tracker."""
 
-__version__ = "0.3.0"
+__version__ = "0.3.1"

{youpdated-0.3.0 → youpdated-0.3.1}/youpdated/http.py

@@ -29,6 +29,9 @@ except ImportError:  # pragma: no cover
 
 NETWORK_ERRORS = (httpx.HTTPError, SOCKSError)
 
+#: ``(url, etag, last_modified)`` from a 200, see :meth:`Client.deferred_validators`
+Validator = tuple[str, str | None, str | None]
+
 # Small pool of current desktop UAs. Rotating inside a plausible set
 # it blends in better, but can be changed here
 USER_AGENTS: tuple[str, ...] = (
@@ -220,6 +223,17 @@ class Client:
         finally:
             self._local.unconditional = previous
 
+    @contextmanager
+    def deferred_validators(self) -> "Iterator[list[Validator]]":
+        """Collect validators this thread receives instead of storing"""
+        previous = getattr(self._local, "pending", None)
+        pending: list[Validator] = []
+        self._local.pending = pending
+        try:
+            yield pending
+        finally:
+            self._local.pending = previous
+
     # internals
 
     def _user_agent(self) -> str:
@@ -339,11 +353,16 @@ class Client:
                     if self._run_active:
                         self._run_bodies[cache_key] = result
                 if conditional and self.state is not None:
-                    self.state.remember_validators(
+                    validator = (
                         url,
                         response.headers.get("etag"),
                         response.headers.get("last-modified"),
                     )
+                    pending = getattr(self._local, "pending", None)
+                    if pending is not None:
+                        pending.append(validator)
+                    else:
+                        self.state.remember_validators(*validator)
                 return result
 
             if response.status_code in soft:

{youpdated-0.3.0 → youpdated-0.3.1}/youpdated/runner.py

@@ -9,7 +9,7 @@ from datetime import datetime, timedelta, timezone
 from typing import Any, Protocol, runtime_checkable
 
 from .config import Config, ConfigError, parse_duration, parse_ignore_list
-from .http import Client
+from .http import Client, Validator
 from .models import RunError, Target, Update
 from .registry import all_sources
 from .state import State
@@ -118,16 +118,26 @@ def run(
     sources = all_sources()
     fetched: list[Update] = []
     succeeded: dict[tuple[str, str], Target] = {}
+    # Stored only for targets whose items are recorded, see Client.deferred_validators
+    validators: list[Validator] = []
 
-    def work(target: Target) -> tuple[Target, list[Update] | Exception, int]:
+    def work(
+        target: Target,
+    ) -> tuple[Target, list[Update] | Exception, int, list[Validator]]:
         if progress is not None:
             progress.target_started(target)
         outcome: list[Update] | Exception
         dropped = 0
-        # A first fetch must see the whole document
-        first = not state.is_baselined(target.source, target.key)
+        # A first fetch must see the whole document, same with one thats ignore rules changed
+        full = (
+            not state.is_baselined(target.source, target.key)
+            or state.ignore_rules_changed(target.source, target.key, target.ignore)
+        )
         try:
-            with client.unconditional() if first else nullcontext():
+            with (
+                client.deferred_validators() as pending,
+                client.unconditional() if full else nullcontext(),
+            ):
                 items = list(sources[target.source].fetch(target, client))
             kept = [u for u in items if not target.ignores(u.tags)]
             # Ignored items are not recorded as seen
@@ -139,7 +149,7 @@ def run(
             progress.target_finished(
                 target, outcome if isinstance(outcome, Exception) else None
             )
-        return target, outcome, dropped
+        return target, outcome, dropped, pending
 
     if progress is not None:
         progress.run_started(len(targets))
@@ -147,7 +157,7 @@ def run(
     workers = max(1, min(config.privacy.concurrency, len(targets)))
     # run_scope: targets that share an upstream document fetch it once between them
     with client.run_scope(), ThreadPoolExecutor(max_workers=workers) as pool:
-        for target, outcome, dropped in pool.map(work, targets):
+        for target, outcome, dropped, pending in pool.map(work, targets):
             result.ignored += dropped
             if isinstance(outcome, Exception):
                 result.errors.append(
@@ -159,6 +169,7 @@ def run(
                 )
             else:
                 fetched.extend(outcome)
+                validators.extend(pending)
                 succeeded.setdefault((target.source, target.key), target)
 
     result.total_fetched = len(fetched)
@@ -181,7 +192,10 @@ def run(
 
     if save:
         state.mark_seen(fetched)
+        for validator in validators:
+            state.remember_validators(*validator)
         state.mark_baselined(succeeded)
+        state.mark_ignore_rules((s, k, t.ignore) for (s, k), t in succeeded.items())
         state.set_last_run()
         # Prune only after marking. Targets not fetched cleanly (failed, or left out by --source) keep history
         result.pruned = state.prune(

{youpdated-0.3.0 → youpdated-0.3.1}/youpdated/sources/browser.py

@@ -164,13 +164,15 @@ class BrowserSource:
             return []
 
         # One version can appear several times as its rollout fraction grows: keeps the earliest start time per version.
-        starts: dict[str, datetime] = {}
+        starts: dict[str, datetime | None] = {}
         for release in fetched.json().get("releases", []):
             version = release.get("version")
             started = _parse_iso((release.get("serving") or {}).get("startTime"))
             if not version:
                 continue
-            if version not in starts or (started and started < starts[version]):
+            if version not in starts:
+                starts[version] = started
+            elif started and (starts[version] is None or started < starts[version]):
                 starts[version] = started
 
         ordered = sorted(

{youpdated-0.3.0 → youpdated-0.3.1}/youpdated/sources/feed.py

@@ -101,8 +101,12 @@ def parse_entries(
 ) -> list[Update]:
     """Turn an already-parsed feed into Updates, newest first."""
     updates: list[Update] = []
+    # Sort before truncating: some feeds list oldest first. Undated entries sort last
+    entries = sorted(
+        parsed.entries, key=lambda e: entry_datetime(e) or _UNDATED, reverse=True
+    )
 
-    for entry in parsed.entries[: limit or None]:
+    for entry in entries[: limit or None]:
         link = entry.get("link") or ""
         title = (entry.get("title") or "").strip() or "(untitled)"
         uid = entry.get("id") or link or f"{title}|{entry.get('published', '')}"
@@ -120,8 +124,6 @@ def parse_entries(
             )
         )
 
-    # Undated entries sort last
-    updates.sort(key=lambda u: u.published or _UNDATED, reverse=True)
     return updates
 
 

{youpdated-0.3.0 → youpdated-0.3.1}/youpdated/sources/itch.py

@@ -27,8 +27,11 @@ GAME_URL_RE = re.compile(
 
 # info panel: <tr><td>Updated</td><td><abbr title="17 August 2026 @ 10:34 UTC">…
 INFO_ROW_RE = re.compile(r"<td>([^<]{1,30})</td><td>(.{0,400}?)</td>", re.S)
-ABBR_TITLE_RE = re.compile(r'<abbr title="([^"]+)"')
-UPLOAD_NAME_RE = re.compile(r'<strong title="([^"]*)" class="name">')
+# itch shuffles attribute order between responses, so never assume any
+ABBR_TITLE_RE = re.compile(r'<abbr\b[^>]*?\stitle="([^"]+)"')
+STRONG_TAG_RE = re.compile(r"<strong\b([^>]*)>")
+NAME_CLASS_RE = re.compile(r'(?:^|\s)class="name"')
+TITLE_ATTR_RE = re.compile(r'(?:^|\s)title="([^"]*)"')
 UPLOAD_SIZE_RE = re.compile(r'class="file_size"><span>([^<]*)</span>')
 ITCH_DATE_RE = re.compile(
     r"(\d{1,2})\s+([A-Za-z]+)\s+(\d{4})(?:\s*@\s*(\d{1,2}):(\d{2}))?", re.I
@@ -188,10 +191,19 @@ class ItchSource:
         return rows
 
     def _uploads(self, html: str) -> list[tuple[str, str]]:
-        names = UPLOAD_NAME_RE.findall(html)
-        sizes = UPLOAD_SIZE_RE.findall(html)
-        # Sizes can be missing for some entries
-        return [(n, sizes[i] if i < len(sizes) else "") for i, n in enumerate(names)]
+        names: list[tuple[str, int]] = []
+        for tag in STRONG_TAG_RE.finditer(html):
+            attrs = tag.group(1)
+            title = TITLE_ATTR_RE.search(attrs)
+            if title and NAME_CLASS_RE.search(attrs):
+                names.append((title.group(1), tag.end()))
+        uploads = []
+        for i, (name, start) in enumerate(names):
+            # Sizes can be missing for some entries; only look up to the next upload
+            end = names[i + 1][1] if i + 1 < len(names) else len(html)
+            size = UPLOAD_SIZE_RE.search(html, start, end)
+            uploads.append((name, size.group(1) if size else ""))
+        return uploads
 
 
 def _parse_itch_date(value: str | None) -> datetime | None:

{youpdated-0.3.0 → youpdated-0.3.1}/youpdated/sources/youtube.py

@@ -263,7 +263,8 @@ class YouTubeSource:
         # Rewrite uids to the video id so the three paths dedupe
         normalized = []
         for update in updates:
-            vid = _video_id(update.url) or update.uid
+            # Entry ids are already ``yt:video:<id>``
+            vid = _video_id(update.url) or update.uid.removeprefix("yt:video:")
             normalized.append(
                 Update(
                     source=update.source,
@@ -326,7 +327,8 @@ def _extract_playlist_id(raw: str) -> str:
 
 
 def _video_id(url: str) -> str | None:
-    match = re.search(r"[?&]v=([A-Za-z0-9_-]{11})", url or "")
+    # Shorts are linked as /shorts/<id> rather than watch?v=<id>
+    match = re.search(r"(?:[?&]v=|/shorts/)([A-Za-z0-9_-]{11})", url or "")
     return match.group(1) if match else None
 
 

{youpdated-0.3.0 → youpdated-0.3.1}/youpdated/state.py

@@ -45,6 +45,10 @@ def _ago(age: timedelta) -> str:
     return (datetime.now(timezone.utc) - age).isoformat()
 
 
+def _rules_value(ignore: frozenset[str]) -> str:
+    return ",".join(sorted(ignore))
+
+
 class State:
     """Thread-safe wrapper over SQLite file."""
 
@@ -90,6 +94,17 @@ class State:
             self._conn.execute("ALTER TABLE seen ADD COLUMN last_seen TEXT")
             self._conn.execute("UPDATE seen SET last_seen = first_seen")
             self._dirty = True
+        # <= 0.3.0 stored YouTube Shorts from the official feed as yt:video:yt:video:<id>
+        fixed = self._conn.execute(
+            "UPDATE OR IGNORE seen SET uid = substr(uid, 10) "
+            "WHERE source = 'youtube' AND uid LIKE 'yt:video:yt:video:%'"
+        ).rowcount
+        # Rows left over already had the correct id recorded by a fallback path
+        fixed += self._conn.execute(
+            "DELETE FROM seen WHERE source = 'youtube' AND uid LIKE 'yt:video:yt:video:%'"
+        ).rowcount
+        if fixed:
+            self._dirty = True
 
     def _read_encrypted(self) -> bytes | None:
         """Decrypt the state file into a database image, or None if there is no file yet."""
@@ -247,8 +262,13 @@ class State:
                 "DELETE FROM kv WHERE namespace='baseline' AND key=?",
                 [(key,) for key in stale_marks],
             )
+            # Rules are recorded with the baseline mark, and go with it
+            removed_rules = self._conn.execute(
+                "DELETE FROM kv WHERE namespace='ignore' AND key NOT IN "
+                "(SELECT key FROM kv WHERE namespace='baseline')"
+            ).rowcount
             self._conn.commit()
-            if removed or removed_validators or stale_marks:
+            if removed or removed_validators or stale_marks or removed_rules:
                 # reclaim empty so the file shrinks
                 self._conn.execute("VACUUM")
                 self._dirty = True
@@ -301,6 +321,28 @@ class State:
             self._conn.commit()
             self._dirty = True
 
+    # per target ignore rules
+
+    def ignore_rules_changed(self, source: str, target: str, ignore: frozenset[str]) -> bool:
+        """If ``ignore`` is different from the rules the target's last saved fetch ran under"""
+        return self.cache_get("ignore", f"{source}:{target}") != _rules_value(ignore)
+
+    def mark_ignore_rules(self, targets: Iterable[tuple[str, str, frozenset[str]]]) -> None:
+        rows = [
+            ("ignore", f"{source}:{target}", _rules_value(ignore))
+            for source, target, ignore in targets
+        ]
+        if not rows:
+            return
+        with self._lock:
+            self._conn.executemany(
+                "INSERT INTO kv (namespace, key, value) VALUES (?,?,?) "
+                "ON CONFLICT(namespace, key) DO UPDATE SET value=excluded.value",
+                rows,
+            )
+            self._conn.commit()
+            self._dirty = True
+
     # upkeeping
 
     def last_run(self) -> datetime | None:

{youpdated-0.3.0 → youpdated-0.3.1}/youpdated.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: youpdated
-Version: 0.3.0
+Version: 0.3.1
 Summary: Simple update tracker for games, apps, and packages
 Author-email: Void1-1 <161782542+Void1-1@users.noreply.github.com>
 License-Expression: MIT