Add audiofile.has_video() (#153)

* Add audiofile.has_video()

* Fix tests

* Add missing video test assest

* Fix return type

* Raise RuntimeError if file does not exist

* Only check for non SND files

audiofile/__init__.py

@@ -2,6 +2,7 @@
 from audiofile.core.info import bit_depth
 from audiofile.core.info import channels
 from audiofile.core.info import duration
+from audiofile.core.info import has_video
 from audiofile.core.info import samples
 from audiofile.core.info import sampling_rate
 from audiofile.core.io import convert_to_wav

audiofile/core/info.py

@@ -182,6 +182,47 @@ def duration(file: str, sloppy=False) -> float:
     return samples(file) / sampling_rate(file)
 
 
+def has_video(file: str) -> bool:
+    """If file contains video.
+
+    For WAV, FLAC, MP3, or OGG files
+    ``False`` is returned.
+    All other files are probed by mediainfo.
+
+    Args:
+        file: file name of input file
+
+    Returns:
+        ``True`` if file contains video data
+
+    Raises:
+        FileNotFoundError: if mediainfo binary is needed,
+            but cannot be found
+        RuntimeError: if ``file`` is missing
+            and does not end with ``"wav"``, ``"flac"``, ``"mp3"``, ``"ogg"``
+
+    Examples:
+        >>> has_video("stereo.wav")
+        False
+
+    """
+    if file_extension(file) in SNDFORMATS:
+        return False
+    else:
+        try:
+            path = audeer.path(file)
+            if not os.path.exists(path):
+                raise RuntimeError(f"'{file}' does not exist.")
+            cmd = ["mediainfo", "--Inform=Video;%Format%", path]
+            video_format = run(cmd)
+            if len(video_format) > 0:
+                return True
+            else:
+                return False
+        except FileNotFoundError:
+            raise binary_missing_error("mediainfo")
+
+
 def samples(file: str) -> int:
     """Number of samples in audio file.
 

docs/api-src/audiofile.rst

@@ -12,6 +12,7 @@ audiofile
     bit_depth
     channels
     duration
+    has_video
     samples
     sampling_rate
     convert_to_wav

tests/assets/README.md

@@ -1,4 +1,6 @@
-# Non SND format audio test files
+# Test files
+
+## Non SND format audio test files
 
 This folder contains files to test file formats
 that cannot be stored with `audiofile.write()`
@@ -10,5 +12,13 @@ They are excerpts
 from "Furious Freak"
 and "Galway",
 Kevin MacLeod (incompetech.com),
-Licensed under Creative Commons:
+licensed under Creative Commons:
+[CC-BY-3.0](http://creativecommons.org/licenses/by/3.0/).
+
+## Video test files
+
+The folder contains the video file `video.mp4`,
+which is a short excerpt from "Big Bunny"
+from Blender Foundation | www.blender.org,
+licensed under Creative Commons:
 [CC-BY-3.0](http://creativecommons.org/licenses/by/3.0/).

tests/test_audiofile.py

@@ -211,6 +211,8 @@ def test_missing_binaries(tmpdir, hide_system_path, empty_file):
         af.duration(empty_file)
     with pytest.raises(expected_error, match="mediainfo"):
         af.duration(empty_file, sloppy=True)
+    with pytest.raises(expected_error, match="mediainfo"):
+        af.has_video(empty_file)
     with pytest.raises(expected_error, match="ffmpeg"):
         af.samples(empty_file)
     with pytest.raises(expected_error, match="mediainfo"):
@@ -247,6 +249,22 @@ def test_missing_file(tmpdir, ext):
         af.convert_to_wav(missing_file, converted_file)
 
 
+@pytest.mark.parametrize(
+    "file, expected_error, expected_error_message",
+    [
+        ("missing_file.bin", RuntimeError, "'missing_file.bin' does not exist"),
+        ("missing_file.mp4", RuntimeError, "'missing_file.mp4' does not exist"),
+        ("missing_file.wav", None, None),
+    ],
+)
+def test_missing_file_has_video(file, expected_error, expected_error_message):
+    if expected_error is not None:
+        with pytest.raises(expected_error, match=expected_error_message):
+            af.has_video(file)
+    else:
+        assert af.has_video(file) is False
+
+
 @pytest.mark.parametrize(
     "non_audio_file",
     ("bin", "mp4", "wav"),
@@ -488,40 +506,40 @@ def test_file_type(tmpdir, file_type, magnitude, sampling_rate, channels):
     if file_type in ["mp3", "ogg"]:
         bit_depth = None
     assert af.bit_depth(file) == bit_depth
+    assert af.has_video(file) is False
 
 
-def test_other_formats():
-    files = [
-        "gs-16b-1c-44100hz.opus",
-        "gs-16b-1c-8000hz.amr",
-        "gs-16b-1c-44100hz.m4a",
-        "gs-16b-1c-44100hz.aac",
-    ]
-    header_durations = [  # as given by mediainfo
-        15.839,
-        15.840000,
-        15.833,
-        None,
-    ]
-    files = [os.path.join(ASSETS_DIR, f) for f in files]
-    for file, header_duration in zip(files, header_durations):
-        signal, sampling_rate = af.read(file)
-        assert af.channels(file) == _channels(signal)
-        assert af.sampling_rate(file) == sampling_rate
-        assert af.samples(file) == _samples(signal)
+@pytest.mark.parametrize(
+    "file, header_duration, audio, video",  # header duration as given by mediainfo
+    [
+        ("gs-16b-1c-44100hz.opus", 15.839, True, False),
+        ("gs-16b-1c-8000hz.amr", 15.840000, True, False),
+        ("gs-16b-1c-44100hz.m4a", 15.833, True, False),
+        ("gs-16b-1c-44100hz.aac", None, True, False),
+        ("video.mp4", None, False, True),
+    ],
+)
+def test_other_formats(file, header_duration, audio, video):
+    path = os.path.join(ASSETS_DIR, file)
+    if audio:
+        signal, sampling_rate = af.read(path)
+        assert af.channels(path) == _channels(signal)
+        assert af.sampling_rate(path) == sampling_rate
+        assert af.samples(path) == _samples(signal)
         duration = _duration(signal, sampling_rate)
-        assert af.duration(file) == duration
+        assert af.duration(path) == duration
         if header_duration is None:
             # Here we expect samplewise precision
-            assert af.duration(file, sloppy=True) == duration
+            assert af.duration(path, sloppy=True) == duration
         else:
             # Here we expect limited precision
             # as the results differ between soxi and mediainfo
             precision = 1
-            sloppy_duration = round(af.duration(file, sloppy=True), precision)
+            sloppy_duration = round(af.duration(path, sloppy=True), precision)
             header_duration = round(header_duration, precision)
             assert sloppy_duration == header_duration
-        assert af.bit_depth(file) is None
+        assert af.bit_depth(path) is None
+    assert af.has_video(path) is video
 
 
 @pytest.mark.parametrize(