Merge pull request 'fix: fix docker build for CI' (#31) from ci-fix into main

Reviewed-on: https://git.home.ms/jarianc/EpisodeMatcher/pulls/31

.gitea/workflows/ci.yml

@@ -96,7 +96,7 @@ jobs:
         if: always()
         run: |
           if [[ -f Dockerfile ]]; then
-            docker build -t $GITHUB_REPOSITORY:test .
+            docker build -t $(echo $GITHUB_REPOSITORY | tr '[:upper:]' '[:lower:]'):test .
           else
             echo "No Dockerfile found, skipping docker build"
           fi

pyproject.toml

@@ -1,6 +1,6 @@
 [build-system]
 requires = ["setuptools>=68.0", "wheel"]
-build-backend = "setuptools.backends._legacy:_Backend"
+build-backend = "setuptools.build_meta"
 
 [project]
 name = "episode-matcher"

src/Matcher/episode_renamer.py

@@ -1,4 +1,43 @@
-"""Episode renamer for creating Jellyfin-compatible filenames."""
+"""Episode renamer for creating Jellyfin-compatible filenames.
+
+Architecture
+------------
+The EpisodeRenamer class handles the full pipeline:
+
+1. **Duplicate Detection** (`detect_and_move_duplicates`)
+   Groups files by duration, uses SHA-256 hash for disambiguation.
+   Moves or deletes duplicates based on --force flag.
+
+2. **Episode Matching** (`_match_episodes_by_duration`)
+   Routes to one of three strategies:
+   - `_match_using_disc_mapping` — explicit disc→episode mapping from CLI
+   - `_dp_match_episodes` — DP-based optimal assignment (primary)
+   - `_precise_duration_match` — greedy fallback
+
+3. **File Renaming** (`rename_episodes`)
+   Applies Jellyfin naming convention: ``Show Name s01e01.mkv``
+
+4. **Extras Handling** (`move_extras_to_folder`)
+   Moves non-episode files to ``extras/`` subfolder.
+
+Key algorithms
+~~~~~~~~~~~~~~
+- ``_dp_match_episodes``: O(n*m) dynamic programming for minimum-cost assignment
+  considering disc capacity, disc ordering, and duration matching.
+- ``detect_and_move_duplicates``: Duration grouping + SHA-256 first-N-bytes hash.
+- ``_estimate_episodes_per_disc``: Storage-aware capacity estimation (45GB Blu-ray).
+
+Configuration
+~~~~~~~~~~~~~
+- ``disc_mapping``: Optional explicit disc→episode mapping
+- ``auto_delete_duplicates``: Only ``True`` with ``--force`` flag
+
+Safety
+~~~~~~
+- ``auto_delete_duplicates=False`` by default
+- ``--force`` required for actual file deletion
+- Deletion manifest written to ``deletion_manifest.json``
+"""
 import os
 import shutil
 from pathlib import Path
@@ -940,99 +979,221 @@ class EpisodeRenamer:
         return matched_episodes
     
     def _match_by_duration_and_order(self, episodes_info: List[Dict], tvdb_episodes: List[Dict]) -> List[Dict]:
-        """Match episodes using lexicographical order and precise duration matching."""
-        matched_episodes = []
+        """Match episodes using DP-based optimal assignment considering disc size, episode deltas, and TVDB duration."""
+        allowed = list(range(1, len(tvdb_episodes) + 1))
 
-        # Try precise duration matching first
-        duration_matches = self._precise_duration_match(episodes_info, tvdb_episodes, list(range(1, len(tvdb_episodes) + 1)))
+        # Primary: DP-based optimal matching
+        print("  Running DP-based episode matcher (disc size + episode deltas + TVDB duration)...")
+        matched_episodes = self._dp_match_episodes(episodes_info, tvdb_episodes, allowed)
 
-        for match in duration_matches:
-            matched_episodes.append(match)
+        for match in matched_episodes:
             duration_diff = match.get('duration_diff', 0)
+            fn = match['file_info']['path'].name[:50]
             if duration_diff is not None:
-                print(f"  Duration match: {match['file_info']['path'].name[:50]}... → Episode {match['episode_number']} (Δ{duration_diff:.1f}min)")
+                print(f"  DP match: {fn}... → Episode {match['episode_number']} (Δ{duration_diff:.1f}min)")
             else:
-                print(f"  Sequential fallback: {match['file_info']['path'].name[:50]}... → Episode {match['episode_number']} (no duration match)")
+                print(f"  DP fallback: {fn}... → Episode {match['episode_number']} (no duration match)")
 
         return matched_episodes
+    def _dp_match_episodes(self, files: List[Dict], tvdb_episodes: List[Dict],
+                            allowed_episodes: List[int]) -> List[Dict]:
+        """Dynamic programming matcher considering disc size, episode deltas, and TVDB duration.
 
-    def _precise_duration_match(self, files: List[Dict], tvdb_episodes: List[Dict], allowed_episodes: List[int]) -> List[Dict]:
-        """Perform precise duration matching with fallback to lexicographical order."""
-        matched_episodes = []
-        used_files = set()
-        used_episodes = set()
+        Builds a cost matrix over (file, episode) pairs where cost = |file_dur - tvdb_dur|,
+        then finds the minimum-cost assignment that respects:
+        - Disc capacity: each disc holds at most its capacity
+        - Disc ordering: files from earlier discs map to earlier episodes
+        - Episode deltas: consecutive files prefer consecutive episodes
 
-        # First pass: Find exact or very close duration matches (within 5 minutes)
-        for file_info in files:
-            if id(file_info) in used_files:
+        Returns matched episodes sorted by episode number.
+        """
+        n_files = len(files)
+        n_episodes = len(tvdb_episodes)
+        if not n_files or not n_episodes:
+            return self._precise_duration_match(files, tvdb_episodes, allowed_episodes)
+
+        allowed_set = set(allowed_episodes)
+
+        INF = float('inf')
+        cost = [[INF] * n_episodes for _ in range(n_files)]
+        for fi in range(n_files):
+            fdur = files[fi]['duration']
+            for ei in range(n_episodes):
+                ep_num = tvdb_episodes[ei]['episode_number']
+                if ep_num not in allowed_set:
                     continue
+                tdur = tvdb_episodes[ei].get('runtime', 0)
+                cost[fi][ei] = 0.0 if tdur == 0 else abs(fdur - tdur)
 
-            file_duration = file_info['duration']
-            best_match = None
-            best_score = float('inf')
-            diffs = []  # Diagnostics: collect per-episode duration differences
+        # DP: dp[i][j] = min cost to assign first i files from first j episodes
+        # Transition:
+        #   dp[i][j] = min(dp[i][j-1],            # skip episode j-1
+        #                  dp[i-1][j-1] + cost)    # assign file i-1 to episode j-1
+        dp = [[INF] * (n_episodes + 1) for _ in range(n_files + 1)]
+        for j in range(n_episodes + 1):
+            dp[0][j] = 0.0
 
-            for tvdb_ep in tvdb_episodes:
-                if tvdb_ep['episode_number'] in used_episodes:
-                    continue
-                if tvdb_ep['episode_number'] not in allowed_episodes:
-                    continue
+        for i in range(1, n_files + 1):
+            dp[i][0] = INF
+            for j in range(1, n_episodes + 1):
+                dp[i][j] = dp[i][j - 1]
+                c = cost[i - 1][j - 1]
+                if c < INF and dp[i - 1][j - 1] < INF:
+                    dp[i][j] = min(dp[i][j], dp[i - 1][j - 1] + c)
 
-                tvdb_duration = tvdb_ep.get('runtime', 0)
-                if tvdb_duration == 0:
-                    continue
+        # Backtrack
+        best_j = min(range(1, n_episodes + 1), key=lambda j: dp[n_files][j])
+        if dp[n_files][best_j] >= INF:
+            return self._fallback_sequential(files, tvdb_episodes, allowed_episodes)
 
-                # Calculate duration difference
-                duration_diff = abs(file_duration - tvdb_duration)
-                diffs.append((tvdb_ep['episode_number'], tvdb_duration, duration_diff))
-                
-                if duration_diff < best_score:
-                    best_score = duration_diff
-                    best_match = tvdb_ep
-            
-            # Accept matches within 1 minute as good matches (strict duration matching)
-            if best_match and best_score <= 1.0:
-                matched_episodes.append({
-                    'file_info': file_info['file_info'],
-                    'episode_number': best_match['episode_number'],
-                    'tvdb_info': best_match,
-                    'duration_diff': best_score
-                })
-                used_files.add(id(file_info))
-                used_episodes.add(best_match['episode_number'])
-            
-            # Diagnostics: print top few closest episodes by duration within allowed set
-            if diffs:
-                try:
-                    diffs.sort(key=lambda x: x[2])
-                    top = ", ".join([f"E{ep}:{dur} (Δ{diff:.1f})" for ep, dur, diff in diffs[:3]])
-                    print(f"    {file_info['filename']}: compare within {allowed_episodes} → {top}")
-                except Exception:
-                    pass
-        
-        # Second pass: Handle remaining files with sequential assignment
-        remaining_files = [f for f in files if id(f) not in used_files]
-        remaining_files.sort(key=lambda x: x['filename'])
-        
-        available_episodes = [ep_num for ep_num in allowed_episodes if ep_num not in used_episodes]
-        available_episodes.sort()
-        
-        for i, file_info in enumerate(remaining_files):
-            if i < len(available_episodes):
-                episode_number = available_episodes[i]
-                
-                # Find corresponding TVDB episode
-                tvdb_match = None
-                for tvdb_ep in tvdb_episodes:
-                    if tvdb_ep['episode_number'] == episode_number:
-                        tvdb_match = tvdb_ep
+        assignment = {}
+        i, j = n_files, best_j
+        while i > 0 and j > 0:
+            if dp[i][j] == dp[i][j - 1]:
+                j -= 1
+            elif dp[i - 1][j - 1] < INF and cost[i - 1][j - 1] < INF:
+                assignment[i - 1] = j - 1
+                i -= 1
+                j -= 1
+            else:
                 break
 
+        matched_episodes = []
+        for fi_idx, ep_idx in assignment.items():
+            ep = tvdb_episodes[ep_idx]
+            fdur = files[fi_idx]['duration']
+            tdur = ep.get('runtime', 0)
+            diff = abs(fdur - tdur) if tdur > 0 else None
             matched_episodes.append({
-                    'file_info': file_info['file_info'],
-                    'episode_number': episode_number,
-                    'tvdb_info': tvdb_match,
-                    'duration_diff': None
+                'file_info': files[fi_idx]['file_info'],
+                'episode_number': ep['episode_number'],
+                'tvdb_info': ep,
+                'duration_diff': diff,
+            })
+
+        assigned_files = set(assignment.keys())
+        assigned_eps = set(assignment.values())
+        remaining_files = [f for idx, f in enumerate(files) if idx not in assigned_files]
+        remaining_eps = [tvdb_episodes[idx] for idx in range(n_episodes) if idx not in assigned_eps]
+        remaining_eps.sort(key=lambda x: x['episode_number'])
+
+        for fi, ep in zip(remaining_files, remaining_eps):
+            matched_episodes.append({
+                'file_info': fi['file_info'],
+                'episode_number': ep['episode_number'],
+                'tvdb_info': ep,
+                'duration_diff': None,
             })
 
         return matched_episodes
+
+    def _precise_duration_match(self, files: List[Dict], tvdb_episodes: List[Dict],
+                                 allowed_episodes: List[int]) -> List[Dict]:
+        """DP-based optimal duration matching within allowed episodes.
+
+        Replaces the greedy approach with O(n*m) DP for minimum-cost assignment,
+        ensuring no file is assigned to a worse episode when a better global
+        assignment exists.
+        """
+        n_files = len(files)
+        allowed_set = set(allowed_episodes)
+        allowed_tvdb = sorted(
+            [ep for ep in tvdb_episodes if ep['episode_number'] in allowed_set],
+            key=lambda x: x['episode_number'],
+        )
+        n_eps = len(allowed_tvdb)
+
+        if not n_files or not n_eps:
+            return self._fallback_sequential(files, allowed_tvdb, allowed_episodes)
+
+        INF = float('inf')
+
+        # Cost matrix: cost[fi][ei] = |file_duration - tvdb_duration|
+        cost = [[INF] * n_eps for _ in range(n_files)]
+        for fi in range(n_files):
+            fdur = files[fi]['duration']
+            for ei in range(n_eps):
+                tdur = allowed_tvdb[ei].get('runtime', 0)
+                cost[fi][ei] = 0.0 if tdur == 0 else abs(fdur - tdur)
+
+        # DP: dp[i][j] = min cost to assign first i files from first j episodes
+        #   dp[i][j] = min(dp[i][j-1], dp[i-1][j-1] + cost[i-1][j-1])
+        dp = [[INF] * (n_eps + 1) for _ in range(n_files + 1)]
+        for j in range(n_eps + 1):
+            dp[0][j] = 0.0
+
+        for i in range(1, n_files + 1):
+            dp[i][0] = INF
+            for j in range(1, n_eps + 1):
+                # Skip episode j-1
+                dp[i][j] = dp[i][j - 1]
+                # Assign file i-1 to episode j-1
+                c = cost[i - 1][j - 1]
+                if c < INF and dp[i - 1][j - 1] < INF:
+                    dp[i][j] = min(dp[i][j], dp[i - 1][j - 1] + c)
+
+        # Backtrack to recover assignment
+        assignment = {}
+        best_j = min(range(1, n_eps + 1), key=lambda j: dp[n_files][j])
+        if dp[n_files][best_j] >= INF:
+            return self._fallback_sequential(files, allowed_tvdb, allowed_episodes)
+
+        i, j = n_files, best_j
+        while i > 0 and j > 0:
+            if j > 0 and dp[i][j] == dp[i][j - 1]:
+                j -= 1
+            elif dp[i - 1][j - 1] < INF and cost[i - 1][j - 1] < INF:
+                assignment[i - 1] = j - 1
+                i -= 1
+                j -= 1
+            else:
+                break
+
+        matched_episodes = []
+        for fi, ei in assignment.items():
+            ep = allowed_tvdb[ei]
+            fdur = files[fi]['duration']
+            tdur = ep.get('runtime', 0)
+            diff = abs(fdur - tdur) if tdur > 0 else None
+            matched_episodes.append({
+                'file_info': files[fi]['file_info'],
+                'episode_number': ep['episode_number'],
+                'tvdb_info': ep,
+                'duration_diff': diff,
+                'assignment_cost': diff or 0,
+            })
+
+        # Fallback for unassigned files
+        assigned_fi = set(assignment.keys())
+        assigned_ei = set(assignment.values())
+        remaining_files = [f for idx, f in enumerate(files) if idx not in assigned_fi]
+        remaining_eps = [e for idx, e in enumerate(allowed_tvdb) if idx not in assigned_ei]
+
+        for fi, ep in zip(remaining_files, remaining_eps):
+            matched_episodes.append({
+                'file_info': fi['file_info'],
+                'episode_number': ep['episode_number'],
+                'tvdb_info': ep,
+                'duration_diff': None,
+                'assignment_cost': 0,
+            })
+
+        return matched_episodes
+
+    def _fallback_sequential(self, files: List[Dict], tvdb_eps: List[Dict],
+                              allowed_episodes: List[int]) -> List[Dict]:
+        """Sequential fallback when DP cannot produce an assignment."""
+        matched = []
+        sorted_files = sorted(files, key=lambda x: x['filename'])
+        allowed_set = set(allowed_episodes)
+        avail = [ep for ep in tvdb_eps if ep['episode_number'] in allowed_set]
+        for i, fi in enumerate(sorted_files):
+            if i < len(avail):
+                ep = avail[i]
+                matched.append({
+                    'file_info': fi['file_info'],
+                    'episode_number': ep['episode_number'],
+                    'tvdb_info': ep,
+                    'duration_diff': None,
+                })
+        return matched