Remove unused combine_subtitles functionality (#69)

* chore: Remove unused combine_subtitles functionality

- Remove src/sub_tools/subtitles/combiner.py
- Remove paths_with_offsets() from system/directory.py
- Remove TestPathsWithOffsets class from test_directory.py
- Remove unused 're' import from directory.py

Fixes #67

* chore: Remove unused python-ffmpeg dependency

The codebase uses ffmpeg via subprocess, not the python-ffmpeg package.

Author: dohyeondk

pyproject.toml

@@ -11,9 +11,8 @@ dependencies = [
     "pycountry>=24.6.1",
     "pysrt>=1.1.2",
     "pytest>=9.0.1",
-    "pytest-asyncio>=1.2.0",
-    "python-ffmpeg>=2.0.12",
-    "rich>=14.2.0",
+    "pytest-asyncio>=0.23.5",
+    "rich>=13.9.4",
     "whisperx>=3.1.1",
 ]
 

src/sub_tools/subtitles/combiner.py

@@ -4,7 +4,6 @@
 
 import hashlib
 import os
-import re
 import tempfile
 
 
@@ -86,20 +85,3 @@ def get_cached_file_path(
     """
     temp_dir = get_temp_directory(url, subfolder)
     return os.path.join(temp_dir, filename)
-
-
-def paths_with_offsets(
-    prefix: str, file_format: str, directory: str = "."
-) -> list[tuple[str, int]]:
-    """
-    Returns a list of paths with offsets, sorted numerically by offset.
-    """
-    pattern = rf"{prefix}_(\d+)\.{file_format}"
-    paths = [
-        (path, int(match.group(1)))
-        for path in os.listdir(directory)
-        for match in [re.match(pattern, path)]
-        if match and match.group(1)
-    ]
-    # Sort by numeric offset, not alphabetically
-    return sorted(paths, key=lambda x: x[1])

src/sub_tools/system/directory.py

@@ -14,7 +14,6 @@
     get_cached_file_path,
     get_temp_directory,
     get_url_hash,
-    paths_with_offsets,
 )
 
 
@@ -287,119 +286,3 @@ def test_works_with_none_url(self):
         temp_dir = os.path.dirname(file_path)
         if os.path.exists(temp_dir):
             os.rmdir(temp_dir)
-
-
-class TestPathsWithOffsets:
-    """Tests for paths_with_offsets function."""
-
-    def test_returns_empty_list_for_empty_directory(self, tmp_path):
-        """Test that returns empty list when no matching files."""
-        result = paths_with_offsets("prefix", "mp3", str(tmp_path))
-        assert result == []
-
-    def test_finds_files_with_offsets(self, tmp_path):
-        """Test that finds files with numeric offsets."""
-        # Create test files
-        (tmp_path / "audio_0.mp3").touch()
-        (tmp_path / "audio_5000.mp3").touch()
-        (tmp_path / "audio_10000.mp3").touch()
-
-        result = paths_with_offsets("audio", "mp3", str(tmp_path))
-
-        assert len(result) == 3
-        assert ("audio_0.mp3", 0) in result
-        assert ("audio_5000.mp3", 5000) in result
-        assert ("audio_10000.mp3", 10000) in result
-
-    def test_returns_sorted_list(self, tmp_path):
-        """Test that returns files in sorted order by numeric offset."""
-        # Create files in random order
-        (tmp_path / "seg_10000.wav").touch()
-        (tmp_path / "seg_0.wav").touch()
-        (tmp_path / "seg_5000.wav").touch()
-
-        result = paths_with_offsets("seg", "wav", str(tmp_path))
-
-        # Should be sorted numerically by offset, not alphabetically
-        assert result[0] == ("seg_0.wav", 0)
-        assert result[1] == ("seg_5000.wav", 5000)
-        assert result[2] == ("seg_10000.wav", 10000)
-
-    def test_ignores_non_matching_files(self, tmp_path):
-        """Test that ignores files that don't match pattern."""
-        (tmp_path / "audio_0.mp3").touch()
-        (tmp_path / "audio_5000.mp3").touch()
-        (tmp_path / "other_file.mp3").touch()
-        (tmp_path / "audio.mp3").touch()
-        (tmp_path / "audio_abc.mp3").touch()
-
-        result = paths_with_offsets("audio", "mp3", str(tmp_path))
-
-        assert len(result) == 2
-        assert all("audio_" in path for path, _ in result)
-
-    def test_matches_different_prefixes(self, tmp_path):
-        """Test matching with different prefixes."""
-        (tmp_path / "video_0.mp4").touch()
-        (tmp_path / "audio_0.mp3").touch()
-
-        video_result = paths_with_offsets("video", "mp4", str(tmp_path))
-        audio_result = paths_with_offsets("audio", "mp3", str(tmp_path))
-
-        assert len(video_result) == 1
-        assert len(audio_result) == 1
-        assert video_result[0][0] == "video_0.mp4"
-        assert audio_result[0][0] == "audio_0.mp3"
-
-    def test_matches_different_formats(self, tmp_path):
-        """Test matching with different file formats."""
-        (tmp_path / "seg_0.mp3").touch()
-        (tmp_path / "seg_0.wav").touch()
-        (tmp_path / "seg_0.srt").touch()
-
-        mp3_result = paths_with_offsets("seg", "mp3", str(tmp_path))
-        wav_result = paths_with_offsets("seg", "wav", str(tmp_path))
-        srt_result = paths_with_offsets("seg", "srt", str(tmp_path))
-
-        assert len(mp3_result) == 1
-        assert len(wav_result) == 1
-        assert len(srt_result) == 1
-
-    def test_handles_large_offsets(self, tmp_path):
-        """Test handling of large offset numbers."""
-        (tmp_path / "seg_999999999.mp3").touch()
-
-        result = paths_with_offsets("seg", "mp3", str(tmp_path))
-
-        assert len(result) == 1
-        assert result[0][1] == 999999999
-
-    def test_uses_current_directory_by_default(self):
-        """Test that uses current directory when no directory specified."""
-        original_dir = os.getcwd()
-        temp_dir = tempfile.mkdtemp()
-
-        try:
-            os.chdir(temp_dir)
-            # Create test file
-            with open("test_0.txt", "w") as f:
-                f.write("test")
-
-            result = paths_with_offsets("test", "txt")
-            assert len(result) == 1
-            assert result[0][0] == "test_0.txt"
-
-        finally:
-            os.chdir(original_dir)
-            shutil.rmtree(temp_dir)
-
-    def test_offset_as_int(self, tmp_path):
-        """Test that offset is returned as int for proper numeric sorting."""
-        (tmp_path / "audio_12345.mp3").touch()
-
-        result = paths_with_offsets("audio", "mp3", str(tmp_path))
-
-        assert len(result) == 1
-        path, offset = result[0]
-        assert isinstance(offset, int)
-        assert offset == 12345