From bc06de8e5c5002f621571ed144851992d3965d9d Mon Sep 17 00:00:00 2001
From: msramalho <19508417+msramalho@users.noreply.github.com>
Date: Mon, 27 Apr 2026 12:34:47 +0100
Subject: [PATCH] fixes incomplete yt-dlp parts download

---
 src/auto_archiver/core/metadata.py |  4 +++
 src/auto_archiver/utils/misc.py    |  3 ++
 tests/test_metadata.py             | 16 +++++++++++
 tests/utils/test_misc.py           | 45 ++++++++++++++++++++++++++++++
 4 files changed, 68 insertions(+)

diff --git a/src/auto_archiver/core/metadata.py b/src/auto_archiver/core/metadata.py
index 05f7d9c..f8fb5f5 100644
--- a/src/auto_archiver/core/metadata.py
+++ b/src/auto_archiver/core/metadata.py
@@ -11,6 +11,7 @@ Key Functionalities:
 
 from __future__ import annotations
 import hashlib
+import os
 from typing import Any, List, Union, Dict
 from dataclasses import dataclass, field
 from dataclasses_json import dataclass_json
@@ -186,6 +187,9 @@ class Metadata:
                 continue
             h = m.get("hash")
             if not h:
+                if not os.path.exists(m.filename):
+                    logger.warning(f"Skipping missing media file: {m.filename}")
+                    continue
                 h = calculate_hash_in_chunks(hashlib.sha256(), int(1.6e7), m.filename)
             if len(h) and h in media_hashes:
                 continue
diff --git a/src/auto_archiver/utils/misc.py b/src/auto_archiver/utils/misc.py
index 747d57e..c16ed5c 100644
--- a/src/auto_archiver/utils/misc.py
+++ b/src/auto_archiver/utils/misc.py
@@ -120,6 +120,9 @@ def ydl_entry_to_filename(ydl, entry: dict) -> str:
     directory = os.path.dirname(base_filename)  # '/get/path/to'
     basename = os.path.basename(base_filename)  # 'file'
     for f in os.listdir(directory):
+        # skip incomplete downloads left behind by yt-dlp
+        if f.endswith(".part"):
+            continue
         if (
             f.startswith(basename)
             or (entry_url and os.path.splitext(f)[0] in entry_url)
diff --git a/tests/test_metadata.py b/tests/test_metadata.py
index e838979..9908d92 100644
--- a/tests/test_metadata.py
+++ b/tests/test_metadata.py
@@ -86,6 +86,22 @@ def test_media_management(basic_metadata, media_file):
     assert basic_metadata.get_media_by_id("m1") == media1
 
 
+def test_remove_duplicate_skips_missing_files(basic_metadata, media_file, tmp_path):
+    """Missing files should be dropped instead of crashing with FileNotFoundError."""
+    real_file = tmp_path / "exists.txt"
+    real_file.write_text("content")
+    valid = media_file(filename=str(real_file), hash_value="abc")
+    missing = media_file(filename="/nonexistent/path/gone.mp4")
+
+    basic_metadata.add_media(valid, "valid")
+    basic_metadata.add_media(missing, "missing")
+
+    assert len(basic_metadata.media) == 2
+    basic_metadata.remove_duplicate_media_by_hash()
+    assert len(basic_metadata.media) == 1
+    assert basic_metadata.get_media_by_id("valid") == valid
+
+
 def test_success():
     m = Metadata()
     assert not m.is_success()
diff --git a/tests/utils/test_misc.py b/tests/utils/test_misc.py
index a3902c7..05dce22 100644
--- a/tests/utils/test_misc.py
+++ b/tests/utils/test_misc.py
@@ -14,6 +14,7 @@ from auto_archiver.utils.misc import (
     calculate_file_hash,
     random_str,
     get_timestamp,
+    ydl_entry_to_filename,
 )
 
 
@@ -139,3 +140,47 @@ class TestMiscUtils:
 
     def test_invalid_timestamp_returns_none(self):
         assert get_timestamp("invalid-date") is None
+
+
+class TestYdlEntryToFilename:
+    """Tests for ydl_entry_to_filename, especially .part file filtering."""
+
+    def _make_mock_ydl(self, prepared_filename):
+        class MockYDL:
+            def prepare_filename(self, entry):
+                return prepared_filename
+
+        return MockYDL()
+
+    def test_returns_exact_file_if_exists(self, tmp_path):
+        video = tmp_path / "video.mp4"
+        video.write_bytes(b"data")
+        ydl = self._make_mock_ydl(str(video))
+        assert ydl_entry_to_filename(ydl, {}) == str(video)
+
+    def test_skips_part_file_returns_complete(self, tmp_path):
+        """Simulates yt-dlp leaving a .part file from a failed format
+        while a complete .webm exists."""
+        (tmp_path / "f5U3IKfoSYs.f399.mp4.part").write_bytes(b"incomplete")
+        webm = tmp_path / "f5U3IKfoSYs.webm"
+        webm.write_bytes(b"complete video")
+
+        # ydl.prepare_filename returns the expected .mp4 which doesn't exist
+        ydl = self._make_mock_ydl(str(tmp_path / "f5U3IKfoSYs.mp4"))
+        result = ydl_entry_to_filename(ydl, {})
+
+        assert result == str(webm)
+        assert not result.endswith(".part")
+
+    def test_skips_part_file_returns_false_if_no_other_match(self, tmp_path):
+        """Only a .part file exists — should return False."""
+        (tmp_path / "video.f399.mp4.part").write_bytes(b"incomplete")
+
+        ydl = self._make_mock_ydl(str(tmp_path / "video.mp4"))
+        assert ydl_entry_to_filename(ydl, {}) is False
+
+    def test_returns_false_when_no_files_match(self, tmp_path):
+        (tmp_path / "unrelated.txt").write_bytes(b"data")
+
+        ydl = self._make_mock_ydl(str(tmp_path / "video.mp4"))
+        assert ydl_entry_to_filename(ydl, {}) is False