riplex 0.6.3__tar.gz → 0.7.0__tar.gz

{riplex-0.6.3 → riplex-0.7.0}/.github/workflows/release.yml

@@ -164,6 +164,11 @@ jobs:
       - name: Create GitHub Release
         uses: softprops/action-gh-release@v2
         with:
+          # Auto-generate notes from commits and PRs since the previous tag,
+          # then prepend the tag annotation (if any) as a header. Using just
+          # the tag annotation produces single-line releases when the tag
+          # was created with `git tag -a -m "<subject>"` and no body.
+          generate_release_notes: true
           body: ${{ steps.tag_notes.outputs.body }}
           files: |
             release/riplex-windows.exe

{riplex-0.6.3 → riplex-0.7.0}/.gitignore

@@ -10,3 +10,6 @@ build/
 riplex.toml
 plex-planner.toml
 riplex_app.log
+
+# Local dev fixture captures (scripts/capture_fixture.py output)
+_captures/

{riplex-0.6.3 → riplex-0.7.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: riplex
-Version: 0.6.3
+Version: 0.7.0
 Summary: Automates the tedious manual work around MakeMKV: figuring out what to rip, which MKV files are actually what, and organizing everything into Plex-compatible folder structures.
 License: MIT
 Requires-Python: >=3.11

{riplex-0.6.3 → riplex-0.7.0}/docs/changelog.md

@@ -1,14 +1,36 @@
-# Documentation Changelog
+# Documentation Changelog
 
 All notable changes to the riplex documentation are recorded here.
 
 The format is based on [Keep a Changelog](https://keepachangelog.com/).
 
+## v0.7.0 — 2026-05-16
+
+Summary: organize-time match quality fixes, faster post-rip organize, and a release-picker affordance for verifying the dvdcompare film page.
+
+### Added
+
+- **GUI: "View on dvdcompare.net" link** on the disc-release screen. Shows the auto-selected film page so users can verify region/edition before committing to a long rip.
+- **GUI: manual film-id override** on the disc-release screen. Paste either a bare fid (e.g. `55540`) or a full URL (`https://www.dvdcompare.net/comparisons/film.php?fid=55540`) and riplex fetches and uses that film page instead. The chosen fid is persisted per `(title, disc_format)` so swapping discs in the same box set keeps the override, with a "Clear saved override" affordance.
+- **`riplex organize --rescan`** flag. By default `organize` now reads `_rip_manifest.json` from each disc subfolder when present (instant load, preserves rip-time classification). Pass `--rescan` to force a fresh ffprobe scan instead.
+- **GUI organize folder picker**: when every disc subfolder has a `_rip_manifest.json`, the folder loads instantly without running ffprobe. A green banner indicates the manifest load with a "Rescan with ffprobe" button to force a fresh probe.
+- **GUI organize preview**: every matched row now shows a confidence chip with the actual delta in seconds (e.g. `HIGH ±18s`, `MEDIUM ±104s`) so weak matches are easy to spot before executing.
+
+### Changed
+
+- **Tighter match tolerance for extras and episodes.** The global `_MAX_MATCH_DELTA` of 300 s is now reserved for the main-movie target; episodes and extras use a 120 s cap. This prevents short featurettes from being claimed by unrelated short clips when no good candidate exists.
+- **Classification-aware matching.** Files whose rip-time classification is `Unmatched content`, `Unknown content`, or `Very short` are no longer paired with a named extra target unless the duration delta is within ±30 s. Ambiguous shorts stay unmatched (and visible) instead of being silently assigned to the closest dvdcompare entry within the loose 300 s window.
+- **Release workflow**: GitHub Releases now include both the manually composed release notes and the auto-generated commit list, so tag annotations authored ahead of the tag push aren't lost.
+
+### Fixed
+
+- **4K disc extras classification**: 1080p extras on a 4K disc are now only skipped when a 4K counterpart actually exists on the same disc. Previously, the duplicate-detection pass could flag legitimate standalone 1080p extras as duplicates of unrelated 4K titles.
+
 ## 2026-05-13
 
 ### Changed
 
-- Bumped `dvdcompare-scraper` pin to `>=0.1.15`, which adds quoted-title disc-header parsing. Boxsets such as *Back to the Future 40th Anniversary Trilogy* that glue six physical discs into a single dvdcompare disc entry are now split correctly into one PlannedDisc per physical disc.
+- Bumped `dvdcompare-scraper` pin to `>=0.1.15`, which adds quoted-title disc-header parsing.
 
 ## 2026-05-12
 
@@ -43,7 +65,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/).
 ### Added
 
 - Troubleshooting guide: macOS-specific sections for tkinter/browse button, SSL certificate errors, Gatekeeper blocking, and tools not found despite being installed.
-- Installation guide: new "Install with pipx" section as the recommended install method for end users — provides globally available `riplex` and `riplex-ui` commands without venv activation.
+- Installation guide: new "Install with pipx" section as the recommended install method for end users — provides globally available `riplex` and `riplex-ui` commands without venv activation.
 
 ### Changed
 

{riplex-0.6.3 → riplex-0.7.0}/docs/reference/cli.md

@@ -59,6 +59,7 @@ riplex organize <folder> [options]
 | `--verbose`, `-v` | Print debug logging to stderr (log file is always written) |
 | `--no-cache` | Bypass cached dvdcompare and TMDb responses |
 | `--force` | Re-organize files even if already tagged as organized |
+| `--rescan` | Ignore `_rip_manifest.json` files and re-probe every MKV with ffprobe. By default, riplex reuses rip-time metadata from the manifest when present (faster, preserves classification). |
 | `--json` | Output as JSON |
 | `--api-key` | TMDb API key |
 | `--snapshot` | Replay from a snapshot JSON file instead of scanning live files |

{riplex-0.6.3 → riplex-0.7.0}/src/riplex/disc/analysis.py

@@ -289,22 +289,23 @@ def is_skip_title(
                 return True
 
     # On 4K discs, skip 1080p titles that match non-episode dvdcompare entries
-    # (featurettes, behind-the-scenes, etc. at lower resolution)
-    # Exception: keep 1080p featurette play-alls when no 4K version exists
+    # ONLY when a 4K physical title at the same duration also exists on the
+    # disc (i.e. the 1080p title is a true duplicate). Some studios (e.g.
+    # Universal) ship the 4K main film on a 4K disc but include most extras
+    # at 1080p only — in that case the 1080p extras are the *only* copy and
+    # must be ripped, not skipped.
     if not is_4k and dvd_entries:
         has_4k = any("3840" in (t.resolution or "") for t in all_titles if t.duration_seconds > 600)
         if has_4k:
             match = find_duration_match(dur, dvd_entries)
             if match and match[2] != "episode":
-                # Keep featurette play-alls if no 4K counterpart at same duration
-                if "play all" in match[0].lower() and _is_featurette_play_all(match[2]):
-                    has_4k_version = any(
-                        "3840" in (t.resolution or "") and abs(t.duration_seconds - dur) < 30
-                        for t in all_titles if t is not title
-                    )
-                    if not has_4k_version:
-                        return False
-                return True
+                has_4k_version = any(
+                    "3840" in (t.resolution or "")
+                    and abs(t.duration_seconds - dur) < 30
+                    for t in all_titles if t is not title
+                )
+                if has_4k_version:
+                    return True
 
     # Skip dvdcompare-based play-all if individual episodes exist at same resolution
     if total_episode_runtime > 0 and abs(dur - total_episode_runtime) < 120:

{riplex-0.6.3 → riplex-0.7.0}/src/riplex/disc/provider.py

@@ -208,6 +208,82 @@ class DiscProvider:
     # in screens / external code may still use this; prefer ``fetch_film``.
     _fetch_film_cached = fetch_film
 
+    async def fetch_film_by_id(self, film_id: int) -> FilmComparison:
+        """Fetch a specific dvdcompare film page by its film id.
+
+        Used by the UI's manual-override path when our auto-ranking picks
+        the wrong film page. Cached on success; HTTP/network failures
+        propagate as LookupError after caching a short-TTL negative entry.
+        """
+        cache_key = cache.hash_key(f"film-by-id|{film_id}")
+
+        cached = cache.cache_get(self.cache_ns, cache_key, ttl_days=self.ttl_days)
+        if cached is not None and not cached.get("_negative"):
+            log.debug("dvdcompare film-by-id cache hit for %s", film_id)
+            return _dict_to_film(cached)
+
+        url = film_url(film_id)
+        try:
+            film = await _throttled_get_film_by_url(url)
+        except httpx.HTTPStatusError as exc:
+            status = exc.response.status_code if exc.response is not None else "?"
+            log.warning("dvdcompare HTTP %s fetching fid=%s", status, film_id)
+            raise LookupError(
+                f"dvdcompare returned HTTP {status} for fid={film_id}"
+            ) from exc
+        except httpx.RequestError as exc:
+            log.warning("dvdcompare network error fetching fid=%s: %s", film_id, exc)
+            raise LookupError(
+                f"dvdcompare network error for fid={film_id}: {exc}"
+            ) from exc
+
+        if not film.releases:
+            raise LookupError(f"dvdcompare film {film_id} has no releases")
+
+        cache.cache_set(self.cache_ns, cache_key, dataclasses.asdict(film))
+        return film
+
+
+# ---------------------------------------------------------------------------
+# Public URL / fid helpers
+# ---------------------------------------------------------------------------
+
+from dvdcompare.scraper import BASE_URL as _DVDCOMPARE_BASE_URL  # noqa: E402
+
+
+def film_url(film_id: int | str) -> str:
+    """Return the public dvdcompare comparison URL for a film id."""
+    return f"{_DVDCOMPARE_BASE_URL}/comparisons/film.php?fid={film_id}"
+
+
+_FID_RE = re.compile(r"(?:fid=|film\.php\?fid=|^)(\d+)")
+
+
+def parse_film_id(value: str) -> int | None:
+    """Extract a numeric film id from a bare number or a dvdcompare URL.
+
+    Returns ``None`` if no fid can be parsed.
+
+    Accepted forms:
+      - ``"55540"``
+      - ``"fid=55540"``
+      - ``"https://www.dvdcompare.net/comparisons/film.php?fid=55540"``
+      - ``"https://www.dvdcompare.net/comparisons/film.php?fid=55540#2"``
+    """
+    if not value:
+        return None
+    s = value.strip()
+    if not s:
+        return None
+    # Pure digits
+    if s.isdigit():
+        return int(s)
+    m = _FID_RE.search(s)
+    if m:
+        return int(m.group(1))
+    return None
+
+
 
 # ---------------------------------------------------------------------------
 # Internal helpers: negative cache + throttle
@@ -284,6 +360,21 @@ async def _throttled_find_film(
             _last_request_at = time.monotonic()
 
 
+async def _throttled_get_film_by_url(url: str) -> FilmComparison:
+    """Call ``get_film_by_url`` under the same throttle as ``find_film``."""
+    global _last_request_at
+    async with _request_lock:
+        now = time.monotonic()
+        wait = _MIN_INTERVAL_S - (now - _last_request_at)
+        if wait > 0:
+            log.debug("dvdcompare throttle: waiting %.2fs before request", wait)
+            await asyncio.sleep(wait)
+        try:
+            return await get_film_by_url(url)
+        finally:
+            _last_request_at = time.monotonic()
+
+
 async def _find_film_prefer_format(
     title: str, disc_format: str, year: int | None
 ) -> FilmComparison:

{riplex-0.6.3 → riplex-0.7.0}/src/riplex/matcher.py

@@ -161,8 +161,32 @@ def format_match_report(candidates: list[MatchCandidate]) -> str:
 _FILM_DISC_MARKER = -1
 
 # Absolute maximum delta (seconds) beyond which a pairing is rejected.
-# Prevents garbage matches when no good candidate exists.
-_MAX_MATCH_DELTA = 300
+# The movie target gets a generous tolerance because main-feature
+# runtimes from TMDb are often rounded to the nearest minute and may
+# omit credits / disc-specific intros.  Extras tend to have precise
+# dvdcompare runtimes, so a much tighter cap prevents short featurettes
+# from being matched to unrelated short clips.
+_MAX_MATCH_DELTA = 300         # movie target
+_MAX_MATCH_DELTA_EXTRA = 120   # episodes + extras
+
+# Classification prefixes that signal "rip-time classifier could not
+# identify this title".  When set, the file must not be claimed by a
+# named target unless the delta is very small (< _HIGH_THRESHOLD).
+_UNIDENTIFIED_CLASSIFICATION_PREFIXES = (
+    "Unmatched content",
+    "Unknown content",
+    "Very short",
+)
+
+
+def _is_movie_target(label: str) -> bool:
+    return label.endswith("(movie)")
+
+
+def _is_unidentified(classification: str) -> bool:
+    if not classification:
+        return False
+    return classification.startswith(_UNIDENTIFIED_CLASSIFICATION_PREFIXES)
 
 _PLAY_ALL_RE = re.compile(r"\bplay\s*all\b", re.IGNORECASE)
 
@@ -437,12 +461,30 @@ def match_discs(
     for delta, fi, ti in pairings:
         if fi in claimed_files or ti in claimed_targets:
             continue
-        if delta > _MAX_MATCH_DELTA:
-            log.debug("Stopping greedy claims: delta %ds exceeds max %ds",
-                      delta, _MAX_MATCH_DELTA)
-            break
-        sf = all_scanned[fi]
         label, runtime_s, _t_disc = targets[ti]
+        is_movie = _is_movie_target(label)
+        max_delta = _MAX_MATCH_DELTA if is_movie else _MAX_MATCH_DELTA_EXTRA
+        if delta > max_delta:
+            # Skip — but don't break, since later pairings may involve a
+            # movie target with a looser cap.  Pairings are still sorted
+            # by delta, so any remaining pair this large is also too large.
+            log.debug("Skip pairing: %s -> '%s' delta=%ds exceeds cap %ds",
+                      all_scanned[fi].name, label, delta, max_delta)
+            continue
+        sf = all_scanned[fi]
+        # Honor rip-time classification: if the classifier explicitly
+        # flagged this file as unidentified/short, require a tight delta
+        # before pairing it with a named target.
+        if (
+            not is_movie
+            and _is_unidentified(sf.classification)
+            and delta > _HIGH_THRESHOLD
+        ):
+            log.debug(
+                "Reject pairing on classification: %s [%s] -> '%s' delta=%ds",
+                sf.name, sf.classification, label, delta,
+            )
+            continue
         conf = _confidence(delta)
         log.debug("Claim: %s (%ds) -> '%s' (%ds) delta=%ds [%s]",
                   sf.name, sf.duration_seconds, label, runtime_s, delta, conf)

{riplex-0.6.3 → riplex-0.7.0}/src/riplex/organizer.py

@@ -75,6 +75,7 @@ class FileMove:
     destination: str
     label: str  # what this file was matched to
     confidence: str
+    delta_seconds: int = 0  # |file_runtime - target_runtime|, for diagnostics
 
 
 @dataclass
@@ -86,6 +87,7 @@ class SplitMove:
     chapter_labels: list[str]  # one label per chapter
     confidence: str
     original_label: str = ""  # the dvdcompare match label
+    delta_seconds: int = 0
 
 
 @dataclass
@@ -396,6 +398,7 @@ def build_organize_plan(
                 destination=str(dest),
                 label=candidate.matched_label,
                 confidence=candidate.confidence,
+                delta_seconds=candidate.delta_seconds,
             )
         )
 

{riplex-0.6.3 → riplex-0.7.0}/src/riplex.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: riplex
-Version: 0.6.3
+Version: 0.7.0
 Summary: Automates the tedious manual work around MakeMKV: figuring out what to rip, which MKV files are actually what, and organizing everything into Plex-compatible folder structures.
 License: MIT
 Requires-Python: >=3.11

{riplex-0.6.3 → riplex-0.7.0}/src/riplex_app/screens/folder_picker.py

@@ -11,6 +11,7 @@ log = logging.getLogger(__name__)
 
 from riplex.config import get_rip_output
 from riplex.detect import detect_format
+from riplex.manifest import build_scanned_from_manifests
 from riplex.scanner import scan_folder
 from riplex.snapshot import load_organized_marker
 from riplex.title import infer_title_from_scanned
@@ -124,8 +125,48 @@ class FolderPickerScreen:
             return
         self.folder_field.error_text = None
 
-        self.app.state["source_folder"] = Path(path)
+        folder = Path(path)
+        self.app.state["source_folder"] = folder
 
+        # Fast path: if every disc subfolder has a _rip_manifest.json,
+        # load metadata from the manifest instead of running ffprobe.
+        if self._has_complete_manifests(folder):
+            log.info("loading scan from rip manifests in %s", folder)
+            try:
+                scanned = build_scanned_from_manifests(folder)
+            except Exception as exc:
+                log.exception("manifest load failed; falling back to ffprobe")
+                scanned = None
+            if scanned:
+                self.app.state["_scan_from_manifest"] = True
+                self.app.state["_scan_result"] = scanned
+                self.app.navigate("folder_picker")
+                return
+
+        self.app.state["_scan_from_manifest"] = False
+        self._start_ffprobe_scan(folder)
+
+    def _has_complete_manifests(self, folder: Path) -> bool:
+        subfolders_with_mkvs = [
+            c for c in folder.iterdir()
+            if c.is_dir() and any(c.glob("*.mkv"))
+        ]
+        if not subfolders_with_mkvs:
+            return False
+        return all((c / "_rip_manifest.json").exists() for c in subfolders_with_mkvs)
+
+    def _rescan_with_ffprobe(self, e):
+        """User clicked the 'Rescan with ffprobe' banner button."""
+        folder = self.app.state.get("source_folder")
+        if not folder:
+            return
+        # Clear any cached scan and force a fresh ffprobe scan.
+        self.app.state.pop("scanned", None)
+        self.app.state.pop("_scan_result", None)
+        self.app.state["_scan_from_manifest"] = False
+        self._start_ffprobe_scan(folder)
+
+    def _start_ffprobe_scan(self, folder: Path):
         # Show scanning state with progress
         self._progress_text = ft.Text("Discovering files...", size=14)
         self._progress_bar = ft.ProgressBar(width=400)
@@ -149,7 +190,7 @@ class FolderPickerScreen:
         )
         self.app.page.update()
 
-        threading.Thread(target=self._do_scan, args=(Path(path),), daemon=True).start()
+        threading.Thread(target=self._do_scan, args=(folder,), daemon=True).start()
 
     def _do_scan(self, folder: Path):
         """Run ffprobe scan in background."""
@@ -219,6 +260,31 @@ class FolderPickerScreen:
                 ),
             ]
 
+        # Banner shown when the scan came from rip manifests (instant load).
+        manifest_banner = []
+        if self.app.state.get("_scan_from_manifest"):
+            manifest_banner = [
+                ft.Container(
+                    content=ft.Row([
+                        ft.Icon(ft.Icons.BOLT, color=ft.Colors.GREEN_400, size=18),
+                        ft.Text(
+                            "Loaded instantly from rip manifests (no ffprobe needed).",
+                            size=13,
+                            color=ft.Colors.GREEN_400,
+                            expand=True,
+                        ),
+                        ft.TextButton(
+                            "Rescan with ffprobe",
+                            icon=ft.Icons.REFRESH,
+                            on_click=self._rescan_with_ffprobe,
+                        ),
+                    ], spacing=8),
+                    bgcolor=ft.Colors.with_opacity(0.08, ft.Colors.GREEN),
+                    padding=12,
+                    border_radius=8,
+                ),
+            ]
+
         # Build disc summary rows
         disc_rows = []
         for d in scanned:
@@ -252,6 +318,7 @@ class FolderPickerScreen:
                     color=ft.Colors.GREY_500,
                 ),
                 ft.Divider(height=20),
+                *manifest_banner,
                 *marker_banner,
                 ft.Text(
                     f"Scanned {len(scanned)} disc{'s' if len(scanned) != 1 else ''}, "

{riplex-0.6.3 → riplex-0.7.0}/src/riplex_app/screens/organize_preview.py

@@ -175,10 +175,25 @@ class OrganizePreviewScreen:
                 "medium": ft.Colors.YELLOW,
                 "low": ft.Colors.ORANGE,
             }.get(move.confidence, ft.Colors.GREY)
+            # Confidence chip: shows label + delta so the user can spot
+            # weak (medium/low) matches before executing.
+            delta = getattr(move, "delta_seconds", 0)
+            conf_label = (move.confidence or "?").upper()
+            if delta:
+                conf_text = f"{conf_label} \u00B1{delta}s"
+            else:
+                conf_text = conf_label
+            conf_chip = ft.Container(
+                content=ft.Text(conf_text, size=10, color=conf_color, weight=ft.FontWeight.BOLD),
+                bgcolor=ft.Colors.with_opacity(0.12, conf_color),
+                padding=ft.Padding(left=6, right=6, top=2, bottom=2),
+                border_radius=4,
+            )
             move_rows.append(
                 ft.Row([
                     ft.Icon(ft.Icons.CHECK_CIRCLE, color=conf_color, size=16),
-                    ft.Text(f"{src_name}", size=12, width=250, no_wrap=True),
+                    conf_chip,
+                    ft.Text(f"{src_name}", size=12, width=220, no_wrap=True),
                     ft.Icon(ft.Icons.ARROW_FORWARD, size=14, color=ft.Colors.GREY_500),
                     ft.Text(f"{dest_folder}/{dest_name}", size=12, expand=True, no_wrap=True),
                 ], spacing=6)