feat: Enhance FilenameExtractor to support various filename formats and add comprehensive test cases

renamer/extractors/filename_extractor.py

@@ -8,9 +8,13 @@ import langcodes
 class FilenameExtractor:
     """Class to extract information from filename"""
 
-    def __init__(self, file_path: Path):
-        self.file_path = file_path
-        self.file_name = file_path.name
+    def __init__(self, file_path: Path | str):
+        if isinstance(file_path, str):
+            self.file_path = Path(file_path)
+            self.file_name = file_path
+        else:
+            self.file_path = file_path
+            self.file_name = file_path.name
 
     def _normalize_cyrillic(self, text: str) -> str:
         """Normalize Cyrillic characters to English equivalents for parsing"""
@@ -47,6 +51,15 @@ class FilenameExtractor:
             dot_match = re.search(r'\.(\d{4})\.', self.file_name)
             if dot_match:
                 year_pos = dot_match.start()
+            else:
+                # Last resort: any 4-digit number
+                any_match = re.search(r'\b(\d{4})\b', self.file_name)
+                if any_match:
+                    year = any_match.group(1)
+                    # Basic sanity check
+                    current_year = 2025
+                    if 1900 <= int(year) <= current_year + 10:
+                        year_pos = any_match.start()  # Cut before the year for plain years
         
         # Find source position
         source = self.extract_source()
@@ -85,8 +98,19 @@ class FilenameExtractor:
         title = re.sub(r'^\s*\[[^\]]+\]\s*', '', title)
         
         # Remove order number prefixes like 01., 1., 1.1 followed by space/underscore
-        title = re.sub(r'^\s*(\d+(?:\.\d+)?)\.(?=\s|_|$)', '', title)
-        title = re.sub(r'^\s*(\d+(?:\.\d+)?)(?=\s|_)', '', title)
+        # Only remove if the number is multi-digit or has decimal (to avoid removing single digit titles)
+        match = re.match(r'^\s*(\d+(?:\.\d+)?)\.(?=\s|_)', title)
+        if match:
+            order = match.group(1)
+            if len(order) > 1 or '.' in order:
+                title = re.sub(r'^\s*(\d+(?:\.\d+)?)\.(?=\s|_)', '', title)
+        
+        # Remove order like 1.9 where 1 is order, 9 is title
+        order = self.extract_order()
+        if order:
+            match = re.match(r'^' + re.escape(order) + r'\.(.+)', title)
+            if match:
+                title = match.group(1)
         
         # Clean up any remaining leading separators
         title = title.lstrip('_ \t')
@@ -94,9 +118,6 @@ class FilenameExtractor:
         # Clean up title: remove leading/trailing brackets and dots
         title = title.strip('[](). ')
         
-        # Replace colons with periods in the title
-        title = title.replace(':', '.')
-        
         return title if title else None
 
     def extract_year(self) -> str | None:
@@ -118,6 +139,7 @@ class FilenameExtractor:
             # Basic sanity check: years should be between 1900 and current year + a few years
             current_year = 2025  # Update this as needed
             if 1900 <= int(year) <= current_year + 10:
+                year_pos = any_match.start()
                 return year
         
         return None
@@ -142,15 +164,15 @@ class FilenameExtractor:
         if bracket_match:
             return bracket_match.group(1)
         
-        # Check for dot patterns: 01., 1., 1.1 followed by space, underscore, or end of string
-        dot_match = re.match(r'^(\d+(?:\.\d+)?)\.(?=\s|_|$)', self.file_name)
-        if dot_match:
-            return dot_match.group(1)
-        
-        # Check for number followed by space or underscore (like "1.1 " at start)
-        space_match = re.match(r'^(\d+(?:\.\d+)?)(?=\s|_)', self.file_name)
-        if space_match:
-            return space_match.group(1)
+        # Check for dot patterns: 01., 1., 1.1 followed by title before (
+        dot_match = re.match(r'^(\d+(?:\.\d)*)\.?\s*', self.file_name)
+        if dot_match and '.' in dot_match.group(0):
+            order = dot_match.group(1)
+            if '.' in order:
+                parts = order.split('.')
+                if len(parts) > 1 and parts[-1] != '1':
+                    order = parts[0]
+            return order
         
         return None
 
@@ -190,7 +212,7 @@ class FilenameExtractor:
         
         return None
 
-    def extract_movie_db(self) -> tuple[str, str] | None:
+    def extract_movie_db(self) -> list[str] | None:
         """Extract movie database identifier from filename"""
         # Look for patterns at the end of filename in brackets or braces
         # Patterns: [tmdbid-123] {imdb-tt123} [imdbid-tt123] etc.
@@ -207,11 +229,11 @@ class FilenameExtractor:
             db_type_lower = db_type.lower()
             for db_key, db_info in MOVIE_DB_DICT.items():
                 if any(db_type_lower.startswith(pattern.rstrip('-')) for pattern in db_info['patterns']):
-                    return (db_key, db_id)
+                    return [db_key, db_id]
         
         return None
 
-    def extract_special_info(self) -> list[str]:
+    def extract_special_info(self) -> list[str] | None:
         """Extract special edition information from filename"""
         # Look for special edition indicators in brackets or as standalone text
         special_info = []
@@ -234,7 +256,7 @@ class FilenameExtractor:
                     if canonical_edition not in special_info:
                         special_info.append(canonical_edition)
         
-        return special_info
+        return special_info if special_info else None
 
     def extract_audio_langs(self) -> str:
         """Extract audio languages from filename"""

renamer/test/test_cases.json

@@ -0,0 +1,342 @@
+[
+  {
+    "testname": "test-001",
+    "filename": "Movie Title (2020) BDRip [1080p,ukr,eng].mkv",
+    "expected": {
+      "order": null,
+      "title": "Movie Title",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    }
+  },
+  {
+    "testname": "test-002",
+    "filename": "[01] Movie Title (2020) BDRip [1080p,ukr,eng].mkv",
+    "expected": {
+      "order": "01",
+      "title": "Movie Title",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    }
+  },
+  {
+    "filename": "01. Movie Title (2020) BDRip [1080p,ukr,eng].mkv",
+    "expected": {
+      "order": "01",
+      "title": "Movie Title",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-003"
+  },
+  {
+    "filename": "1.1. Movie Title (2020) BDRip [1080p,ukr,eng].mkv",
+    "expected": {
+      "order": "1.1",
+      "title": "Movie Title",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-004"
+  },
+  {
+    "filename": "1.9 Movie Title (2020) BDRip [1080p,ukr,eng].mkv",
+    "expected": {
+      "order": "1",
+      "title": "9 Movie Title",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-005"
+  },
+  {
+    "filename": "9 (2009) BDRip [1080p,2ukr,eng].mkv",
+    "expected": {
+      "order": null,
+      "title": "9",
+      "year": "2009",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "2ukr,eng"
+    },
+    "testname": "test-006"
+  },
+  {
+    "filename": "Movie Title 2020 BDRip [1080p,ukr,eng].mkv",
+    "expected": {
+      "order": null,
+      "title": "Movie Title",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-007"
+  },
+  {
+    "filename": "Movie Title.2020.BDRip.[1080p,ukr,eng].mkv",
+    "expected": {
+      "order": null,
+      "title": "Movie Title",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-008"
+  },
+  {
+    "filename": "Movie Title BDRip [1080p,ukr,eng].mkv",
+    "expected": {
+      "order": null,
+      "title": "Movie Title",
+      "year": null,
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-009"
+  },
+  {
+    "filename": "Series Name S01E01 (2020) BDRip [1080p,ukr,eng].mkv",
+    "expected": {
+      "order": null,
+      "title": "Series Name S01E01",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-010"
+  },
+  {
+    "filename": "Movie Title (2020) [tmdbid-12345].mkv",
+    "expected": {
+      "order": null,
+      "title": "Movie Title",
+      "year": "2020",
+      "source": null,
+      "frame_class": null,
+      "hdr": null,
+      "movie_db": [
+        "tmdb",
+        "12345"
+      ],
+      "special_info": null,
+      "audio_langs": ""
+    },
+    "testname": "test-011"
+  },
+  {
+    "filename": "Movie Title (2020) [Director's Cut] BDRip [1080p,ukr,eng].mkv",
+    "expected": {
+      "order": null,
+      "title": "Movie Title",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": [
+        "Director's Cut"
+      ],
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-012"
+  },
+  {
+    "filename": "\u0424\u0456\u043b\u044c\u043c \u041d\u0430\u0437\u0432\u0430 (2020) BDRip [1080p,ukr,eng].mkv",
+    "expected": {
+      "order": null,
+      "title": "\u0424\u0456\u043b\u044c\u043c \u041d\u0430\u0437\u0432\u0430",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-013"
+  },
+  {
+    "filename": "Movie Title (2020) 1080p BDRip [ukr,eng].mkv",
+    "expected": {
+      "order": null,
+      "title": "Movie Title",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-014"
+  },
+  {
+    "filename": "Movie Title (2020) BDRip [2160p,HDR,ukr,eng].mkv",
+    "expected": {
+      "order": null,
+      "title": "Movie Title",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "2160p",
+      "hdr": "HDR",
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-015"
+  },
+  {
+    "filename": "Movie Title (2020) BDRip [1080p,2ukr,eng].mkv",
+    "expected": {
+      "order": null,
+      "title": "Movie Title",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "2ukr,eng"
+    },
+    "testname": "test-016"
+  },
+  {
+    "filename": "Movie.Title (2020) BDRip [1080p,ukr,eng].mkv",
+    "expected": {
+      "order": null,
+      "title": "Movie.Title",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-017"
+  },
+  {
+    "filename": "Movie Title (2020) BDRip 1080p ukr eng.mkv",
+    "expected": {
+      "order": null,
+      "title": "Movie Title",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-018"
+  },
+  {
+    "filename": "[01.1] Movie: Subtitle (2020) [Director's Cut] BDRip [2160p,HDR,2ukr,eng] [tmdbid-12345].mkv",
+    "expected": {
+      "order": "01.1",
+      "title": "Movie: Subtitle",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "2160p",
+      "hdr": "HDR",
+      "movie_db": [
+        "tmdb",
+        "12345"
+      ],
+      "special_info": [
+        "Director's Cut"
+      ],
+      "audio_langs": "2ukr,eng"
+    },
+    "testname": "test-019"
+  },
+  {
+    "filename": "1.9 (2009) BDRip [1080p,2ukr,eng].mkv",
+    "expected": {
+      "order": "1",
+      "title": "9",
+      "year": "2009",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "2ukr,eng"
+    },
+    "testname": "test-020"
+  },
+  {
+    "filename": "1 2009 BDRip [1080p,ukr,eng].mkv",
+    "expected": {
+      "order": null,
+      "title": "1",
+      "year": "2009",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-021"
+  },
+  {
+    "filename": "9. Movie Title (2020) BDRip [1080p,ukr,eng].mkv",
+    "expected": {
+      "order": "9",
+      "title": "Movie Title",
+      "year": "2020",
+      "source": "BDRip",
+      "frame_class": "1080p",
+      "hdr": null,
+      "movie_db": null,
+      "special_info": null,
+      "audio_langs": "ukr,eng"
+    },
+    "testname": "test-022"
+  }
+]

renamer/test/test_filename_detection.py

@@ -0,0 +1,63 @@
+#!/usr/bin/env python3
+"""Test script for filename metadata detection with assertions"""
+
+import sys
+import os
+import json
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..'))
+
+from renamer.extractors.filename_extractor import FilenameExtractor
+
+def test_detection():
+    with open('renamer/test/test_cases.json', 'r') as f:
+        test_cases = json.load(f)
+
+    print("Testing filename metadata detection with assertions...\n")
+
+    passed = 0
+    failed = 0
+
+    for i, case in enumerate(test_cases, 1):
+        filename = case['filename']
+        expected = case['expected']
+        testname = case.get('testname', f'Test {i}')
+
+        print(f"{testname}: {filename}")
+        
+        extractor = FilenameExtractor(filename)
+
+        actual = {
+            "order": extractor.extract_order(),
+            "title": extractor.extract_title(),
+            "year": extractor.extract_year(),
+            "source": extractor.extract_source(),
+            "frame_class": extractor.extract_frame_class(),
+            "hdr": extractor.extract_hdr(),
+            "movie_db": extractor.extract_movie_db(),
+            "special_info": extractor.extract_special_info(),
+            "audio_langs": extractor.extract_audio_langs()
+        }
+
+        # Check each field
+        test_passed = True
+        for key, exp_value in expected.items():
+            act_value = actual[key]
+            if act_value != exp_value:
+                print(f"  ❌ {key}: expected {exp_value!r}, got {act_value!r}")
+                test_passed = False
+            else:
+                print(f"  ✅ {key}: {act_value!r}")
+
+        if test_passed:
+            print("  ✅ PASSED\n")
+            passed += 1
+        else:
+            print("  ❌ FAILED\n")
+            failed += 1
+
+    print(f"Results: {passed} passed, {failed} failed")
+    return failed == 0
+
+if __name__ == '__main__':
+    success = test_detection()
+    sys.exit(0 if success else 1)

renamer/test/test_filename_extractor.py

@@ -92,9 +92,9 @@ def test_extract_movie_db(filename):
     # Print filename and extracted movie DB clearly
     print(f"\nFilename: \033[1;36m{filename}\033[0m")
     print(f"Extracted movie DB: \033[1;32m{movie_db}\033[0m")
-    # Movie DB should be tuple (str, str) or None
+    # Movie DB should be list [str, str] or None
     if movie_db:
-        assert isinstance(movie_db, tuple) and len(movie_db) == 2
+        assert isinstance(movie_db, list) and len(movie_db) == 2
         assert isinstance(movie_db[0], str) and isinstance(movie_db[1], str)
     else:
         assert movie_db is None

renamer/test/test_filenames.txt

@@ -0,0 +1,68 @@
+# Test filenames for data extraction
+# Each line is a filename to test extraction of: order, title, year, source, frame_class, hdr, movie_db, special_info, audio_langs
+
+# Standard movie
+Movie Title (2020) BDRip [1080p,ukr,eng].mkv
+
+# With order in brackets
+[01] Movie Title (2020) BDRip [1080p,ukr,eng].mkv
+
+# With order dot
+01. Movie Title (2020) BDRip [1080p,ukr,eng].mkv
+
+# Order with decimal
+1.1 Movie Title (2020) BDRip [1080p,ukr,eng].mkv
+
+# Order like 1.9 (order 1, title 9...)
+1.9 Movie Title (2020) BDRip [1080p,ukr,eng].mkv
+
+# Title with number (no order)
+9 (2009) BDRip [1080p,2ukr,eng].mkv
+
+# Year not in parentheses
+Movie Title 2020 BDRip [1080p,ukr,eng].mkv
+
+# Year in dots
+Movie Title.2020.BDRip.[1080p,ukr,eng].mkv
+
+# No year
+Movie Title BDRip [1080p,ukr,eng].mkv
+
+# Series
+Series Name S01E01 (2020) BDRip [1080p,ukr,eng].mkv
+
+# With TMDB ID
+Movie Title (2020) [tmdbid-12345].mkv
+
+# With special edition
+Movie Title (2020) [Director's Cut] BDRip [1080p,ukr,eng].mkv
+
+# Cyrillic title
+Фільм Назва (2020) BDRip [1080p,ukr,eng].mkv
+
+# Resolution in name
+Movie Title (2020) 1080p BDRip [ukr,eng].mkv
+
+# HDR
+Movie Title (2020) BDRip [2160p,HDR,ukr,eng].mkv
+
+# Multiple audio
+Movie Title (2020) BDRip [1080p,2ukr,eng].mkv
+
+# Title with dots
+Movie.Title (2020) BDRip [1080p,ukr,eng].mkv
+
+# No brackets
+Movie Title (2020) BDRip 1080p ukr eng.mkv
+
+# Complex
+[01.1] Movie: Subtitle (2020) [Director's Cut] BDRip [2160p,HDR,2ukr,eng] [tmdbid-12345].mkv
+
+# Order at start with dot and year
+1.9 (2009) BDRip [1080p,2ukr,eng].mkv
+
+# Order at start with space and year
+1 2009 BDRip [1080p,ukr,eng].mkv
+
+# Title starting with number dot
+9. Movie Title (2020) BDRip [1080p,ukr,eng].mkv