def download(data_home=None,
partial_download=None,
force_overwrite=False,
cleanup=True):
"""Download GuitarSet.
Args:
data_home (str):
Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
force_overwrite (bool):
Whether to overwrite the existing downloaded data
partial_download (list):
List indicating what to partially download. The list can include any of:
* `'annotations'` the annotation files
* `'audio_hex_original'` original 6 channel wave file from hexaphonic pickup
* `'audio_hex_debleeded'` hex wave files with interference removal applied
* `'audio_mic'` monophonic recording from reference microphone
* `'audio_mix'` monophonic mixture of original 6 channel file
If `None`, all data is downloaded.
cleanup (bool):
Whether to delete the zip/tar file after extracting.
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
download_utils.downloader(
data_home,
remotes=REMOTES,
partial_download=partial_download,
info_message=None,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
def download(data_home=None, force_overwrite=False):
"""DALI is not available for downloading directly.
This function prints a helper message to download DALI
through zenodo.org.
Args:
data_home (str): Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
info_message = """
To download this dataset, visit:
https://zenodo.org/record/2577915 and request access.
Once downloaded, unzip the file DALI_v1.0.zip
and place the result in:
{save_path}
Use the function dali_code.get_audio you can find at:
https://github.com/gabolsgabs/DALI for getting the audio and place them at:
{audio_path}
""".format(
save_path=os.path.join(data_home, 'annotations'),
force_overwrite=force_overwrite,
audio_path=os.path.join(data_home, 'audio'),
)
download_utils.downloader(data_home,
remotes=REMOTES,
info_message=info_message)
def download(
data_home=None, partial_download=None, force_overwrite=False, cleanup=True
):
"""Download TinySOL.
Args:
data_home (str):
Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
force_overwrite (bool):
Whether to overwrite the existing downloaded data
partial_download (list):
List indicating what to partially download. The list can include any of:
* `'annotations'` the annotation files
* `'audio'` the audio files
If `None`, all data is downloaded.
cleanup (bool):
Whether to delete the zip/tar file after extracting.
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
download_utils.downloader(
data_home,
remotes=REMOTES,
partial_download=partial_download,
info_message=None,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
def download(self,
partial_download=None,
force_overwrite=False,
cleanup=False):
"""Download the dataset
Args:
partial_download (list or None):
A list of keys of remotes to partially download.
If None, all data is downloaded
force_overwrite (bool):
If True, existing files are overwritten by the downloaded files.
cleanup (bool):
Whether to delete any zip/tar files after extracting.
Raises:
ValueError: if invalid keys are passed to partial_download
IOError: if a downloaded file's checksum is different from expected
"""
download_utils.downloader(
self.data_home,
remotes=self.remotes,
partial_download=partial_download,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
self._find_replace(os.path.join(self.data_home, "meta"), ": nan",
": null", "*.json")
def download(self, partial_download=None, force_overwrite=False, cleanup=False):
"""Download the dataset
Args:
partial_download (list or None):
A list of keys of remotes to partially download.
If None, all data is downloaded
force_overwrite (bool):
If True, existing files are overwritten by the downloaded files.
cleanup (bool):
Whether to delete any zip/tar files after extracting.
Raises:
ValueError: if invalid keys are passed to partial_download
IOError: if a downloaded file's checksum is different from expected
"""
# in MAESTRO "metadata" is contained in "midi" is contained in "all"
if partial_download is None or "all" in partial_download:
partial_download = ["all"]
elif "midi" in partial_download:
partial_download = ["midi"]
download_utils.downloader(
self.data_home,
remotes=self.remotes,
partial_download=partial_download,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
def download(data_home=None, force_overwrite=False):
"""Download the Beatles Dataset (annotations).
The audio files are not provided due to the copyright.
Args:
data_home (str): Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
force_overwrite (bool): Whether to overwrite the existing downloaded data
"""
# use the default location: ~/mir_datasets/Beatles
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
download_message = """
Unfortunately the audio files of the Beatles dataset are not available
for download. If you have the Beatles dataset, place the contents into
a folder called Beatles with the following structure:
> Beatles/
> annotations/
> audio/
and copy the Beatles folder to {}
""".format(
data_home
)
download_utils.downloader(
data_home,
tar_downloads=[ANNOTATIONS_REMOTE],
info_message=download_message,
force_overwrite=force_overwrite,
)
def _download(save_dir, remotes, partial_download, info_message,
force_overwrite, cleanup):
"""Download the dataset.
Args:
save_dir (str):
The directory to download the data
remotes (dict or None):
A dictionary of RemoteFileMetadata tuples of data in zip format.
If None, there is no data to download
partial_download (list or None):
A list of keys to partially download the remote objects of the download dict.
If None, all data is downloaded
info_message (str or None):
A string of info to print when this function is called.
If None, no string is printed.
force_overwrite (bool):
If True, existing files are overwritten by the downloaded files.
cleanup (bool):
Whether to delete the zip/tar file after extracting.
"""
download_utils.downloader(
save_dir,
remotes=remotes,
partial_download=partial_download,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
# removing nans from JSON files
find_replace(os.path.join(save_dir, "meta"), ": nan", ": null", "*.json")
def download(data_home=None, force_overwrite=False):
"""Download ORCHSET Dataset.
Args:
data_home (str): Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
force_overwrite (bool): Whether to overwrite the existing downloaded data
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
download_utils.downloader(data_home,
zip_downloads=[REMOTE],
force_overwrite=force_overwrite)
# files get downloaded to a folder called Orchset - move everything up a level
duplicated_orchset_dir = os.path.join(data_home, 'Orchset')
orchset_files = glob.glob(os.path.join(duplicated_orchset_dir, '*'))
for fpath in orchset_files:
shutil.move(fpath, data_home)
if os.path.exists(duplicated_orchset_dir):
os.removedirs(duplicated_orchset_dir)
def download(data_home=None, force_overwrite=False):
"""Download iKala Dataset. However, iKala dataset is not available for
download anymore. This function prints a helper message to organize
pre-downloaded iKala dataset.
Args:
data_home (str): Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
force_overwrite (bool): If True, existing files are overwritten by the
downloaded files.
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
download_message = """
Unfortunately the iKala dataset is not available for download.
If you have the iKala dataset, place the contents into a folder called
{ikala_dir} with the following structure:
> {ikala_dir}/
> Lyrics/
> PitchLabel/
> Wavfile/
and copy the {ikala_dir} folder to {save_path}
""".format(ikala_dir=DATASET_DIR, save_path=data_home)
download_utils.downloader(
data_home,
file_downloads=[ID_MAPPING_REMOTE],
info_message=download_message,
force_overwrite=force_overwrite,
)
def download(data_home=None):
"""MedleyDB is not available for downloading directly.
This function prints a helper message to download MedleyDB
through zenodo.org.
Args:
data_home (str):
Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
info_message = """
To download this dataset, visit:
https://zenodo.org/record/2628782#.XKZdABNKh24
and request access.
Once downloaded, unzip the file MedleyDB-Melody.zip
and copy the result to:
{data_home}
""".format(data_home=data_home)
download_utils.downloader(data_home, info_message=info_message)
def download(self,
partial_download=None,
force_overwrite=False,
cleanup=False):
"""Download data to `save_dir` and optionally print a message.
Args:
partial_download (list or None):
A list of keys of remotes to partially download.
If None, all data is downloaded
force_overwrite (bool):
If True, existing files are overwritten by the downloaded files.
cleanup (bool):
Whether to delete any zip/tar files after extracting.
Raises:
ValueError: if invalid keys are passed to partial_download
IOError: if a downloaded file's checksum is different from expected
"""
download_utils.downloader(
self.data_home,
remotes=self.remotes,
partial_download=partial_download,
info_message=self._download_info,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
def download(data_home=None, force_overwrite=False, cleanup=True):
"""Download SALAMI Dataset (annotations).
The audio files are not provided.
Args:
data_home (str):
Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
force_overwrite (bool):
Whether to overwrite the existing downloaded data
cleanup (bool):
Whether to delete the zip/tar file after extracting.
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
info_message = """
Unfortunately the audio files of the Salami dataset are not available
for download. If you have the Salami dataset, place the contents into a
folder called Salami with the following structure:
> Salami/
> salami-data-public-hierarchy-corrections/
> audio/
and copy the Salami folder to {}
""".format(data_home)
download_utils.downloader(
data_home,
remotes=REMOTES,
info_message=info_message,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
def download(data_home=None, force_overwrite=False):
"""Download SALAMI Dataset (annotations).
The audio files are not provided.
Args:
data_home (str): Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
force_overwrite (bool): whether to overwrite the existing downloaded data
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
info_message = """
Unfortunately the audio files of the Salami dataset are not available
for download. If you have the Salami dataset, place the contents into a
folder called Salami with the following structure:
> Salami/
> salami-data-public-master/
> audio/
and copy the Salami folder to {}
""".format(data_home)
download_utils.downloader(
data_home,
zip_downloads=[ANNOTATIONS_REMOTE],
info_message=info_message,
force_overwrite=force_overwrite,
)
def download(data_home=None,
force_overwrite=False,
cleanup=True,
partial_download=None):
"""Download the giantsteps_key Dataset (annotations).
The audio files are not provided due to copyright issues.
Args:
data_home (str):
Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
force_overwrite (bool):
Whether to overwrite the existing downloaded data
cleanup (bool):
Whether to delete the zip/tar file after extracting.
partial_download(list of str)
arguments can be 'audio' 'metadata' or/and 'keys'
"""
# use the default location: ~/mir_datasets/giantsteps_key
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
download_message = ""
download_utils.downloader(
data_home,
remotes=REMOTES,
partial_download=partial_download,
info_message=download_message,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
def download(data_home=None, force_overwrite=False, cleanup=True):
"""Download Groove MIDI.
Args:
data_home (str):
Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
force_overwrite (bool):
Whether to overwrite the existing downloaded data
cleanup (bool):
Whether to delete the zip/tar file after extracting.
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
download_utils.downloader(
data_home,
remotes=REMOTES,
info_message=None,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
# files get downloaded to a folder called groove - move everything up a level
groove_dir = os.path.join(data_home, 'groove')
groove_files = glob.glob(os.path.join(groove_dir, '*'))
for fpath in groove_files:
shutil.move(fpath, data_home)
if os.path.exists(groove_dir):
shutil.rmtree(groove_dir)
def download(data_home=None, force_overwrite=False):
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
info_message = """
Unfortunately the audio files of the RWC-Popular dataset are not available
for download. If you have the RWC-Popular dataset, place the contents into a
folder called RWC-Popular with the following structure:
> RWC-Popular/
> annotations/
> audio/rwc-p-m0i with i in [1 .. 7]
> metadata-master/
and copy the RWC-Popular folder to {}
""".format(data_home)
download_utils.downloader(
data_home,
zip_downloads=[
METADATA_REMOTE,
ANNOTATIONS_REMOTE_1,
ANNOTATIONS_REMOTE_2,
ANNOTATIONS_REMOTE_3,
ANNOTATIONS_REMOTE_4,
],
info_message=info_message,
force_overwrite=force_overwrite,
)
def download(self,
partial_download=None,
force_overwrite=False,
cleanup=False):
"""Download the dataset
Args:
partial_download (list or None):
A list of keys of remotes to partially download.
If None, all data is downloaded
force_overwrite (bool):
If True, existing files are overwritten by the downloaded files.
cleanup (bool):
Whether to delete any zip/tar files after extracting.
Raises:
ValueError: if invalid keys are passed to partial_download
IOError: if a downloaded file's checksum is different from expected
"""
# in MAESTRO "metadata" is contained in "midi" is contained in "all"
if partial_download is None or "all" in partial_download:
partial_download = ["all"]
elif "midi" in partial_download:
partial_download = ["midi"]
download_utils.downloader(
self.data_home,
remotes=self.remotes,
partial_download=partial_download,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
# files get downloaded to a folder called maestro-v2.0.0
# move everything up a level
maestro_dir = os.path.join(self.data_home, "maestro-v2.0.0")
if not os.path.exists(maestro_dir):
logging.info(
"Maestro data not downloaded, because it probably already exists on your computer. "
+
"Run .validate() to check, or rerun with force_overwrite=True to delete any "
+ "existing files and download from scratch")
return
maestro_files = glob.glob(os.path.join(maestro_dir, "*"))
for fpath in maestro_files:
target_path = os.path.join(self.data_home, os.path.basename(fpath))
if os.path.exists(target_path):
logging.info(
"{} already exists. Run with force_overwrite=True to download from scratch"
.format(target_path))
continue
shutil.move(fpath, self.data_home)
if os.path.exists(maestro_dir):
shutil.rmtree(maestro_dir)
def index(self):
if self.remote_index is not None:
working_dir = os.path.dirname(os.path.realpath(__file__))
path_index_file = os.path.join(working_dir, "datasets/indexes",
self.index_file)
if not os.path.isfile(path_index_file):
path_indexes = os.path.join(working_dir, "datasets/indexes")
download_utils.downloader(path_indexes,
remotes=self.remote_index)
return load_json_index(self.index_file)
def download(data_home=None):
"""Download the GTZAN-Genre dataset.
Args:
data_home (str): Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
download_utils.downloader(data_home, tar_downloads=[DATASET_REMOTE])
def download(self,
partial_download=None,
force_overwrite=False,
cleanup=False):
"""Download the dataset
Args:
partial_download (list or None):
A list of keys of remotes to partially download.
If None, all data is downloaded
force_overwrite (bool):
If True, existing files are overwritten by the downloaded files.
cleanup (bool):
Whether to delete any zip/tar files after extracting.
Raises:
ValueError: if invalid keys are passed to partial_download
IOError: if a downloaded file's checksum is different from expected
"""
download_utils.downloader(
self.data_home,
partial_download=partial_download,
remotes=self.remotes,
info_message=None,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
# files get downloaded to a folder called groove - move everything up a level
groove_dir = os.path.join(self.data_home, "groove")
if not os.path.exists(groove_dir):
logging.info(
"Groove MIDI data not downloaded, because it probably already exists on your computer. "
+
"Run .validate() to check, or rerun with force_overwrite=True to delete any "
+ "existing files and download from scratch")
return
groove_files = glob.glob(os.path.join(groove_dir, "*"))
for fpath in groove_files:
target_path = os.path.join(self.data_home, os.path.basename(fpath))
if os.path.exists(target_path):
logging.info(
"{} already exists. Run with force_overwrite=True to download from scratch"
.format(target_path))
continue
shutil.move(fpath, self.data_home)
if os.path.exists(groove_dir):
shutil.rmtree(groove_dir)
def download(
data_home=None, partial_download=None, force_overwrite=False, cleanup=True
):
"""Download the dataset.
Args:
data_home (str):
Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
force_overwrite (bool):
Whether to overwrite the existing downloaded data
partial_download (list):
List indicating what to partially download. The list can include any of:
* 'all': audio, midi and metadata
* 'midi': midi and metadata only
* 'metadata': metadata only
If `None`, all data is downloaded.
cleanup (bool):
Whether to delete the zip/tar file after extracting.
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
# in MAESTRO "metadata" is contained in "midi" is contained in "all"
if partial_download is None or 'all' in partial_download:
partial_download = ['all']
elif 'midi' in partial_download:
partial_download = ['midi']
download_utils.downloader(
data_home,
remotes=REMOTES,
partial_download=partial_download,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
# files get downloaded to a folder called maestro-v2.0.0
# move everything up a level
maestro_dir = os.path.join(data_home, 'maestro-v2.0.0')
maestro_files = glob.glob(os.path.join(maestro_dir, '*'))
for fpath in maestro_files:
shutil.move(fpath, data_home)
if os.path.exists(maestro_dir):
shutil.rmtree(maestro_dir)
def _download(save_dir, remotes, partial_download, info_message,
force_overwrite, cleanup):
"""Download the dataset.
Args:
save_dir (str):
The directory to download the data
remotes (dict or None):
A dictionary of RemoteFileMetadata tuples of data in zip format.
If None, there is no data to download
partial_download (list or None):
List indicating what to partially download. The list can include any of:
* 'all': audio, midi and metadata
* 'midi': midi and metadata only
* 'metadata': metadata only
If None, all data is downloaded
info_message (str or None):
A string of info to print when this function is called.
If None, no string is printed.
force_overwrite (bool):
If True, existing files are overwritten by the downloaded files.
cleanup (bool):
Whether to delete the zip/tar file after extracting.
"""
# in MAESTRO "metadata" is contained in "midi" is contained in "all"
if partial_download is None or "all" in partial_download:
partial_download = ["all"]
elif "midi" in partial_download:
partial_download = ["midi"]
download_utils.downloader(
save_dir,
remotes=remotes,
partial_download=partial_download,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
# files get downloaded to a folder called maestro-v2.0.0
# move everything up a level
maestro_dir = os.path.join(save_dir, "maestro-v2.0.0")
maestro_files = glob.glob(os.path.join(maestro_dir, "*"))
for fpath in maestro_files:
shutil.move(fpath, save_dir)
if os.path.exists(maestro_dir):
shutil.rmtree(maestro_dir)
def download(data_home=None,
partial_download=None,
force_overwrite=False,
cleanup=True):
"""Download the RWC Popular (annotations and metadata).
The audio files are not provided due to copyright issues.
Args:
data_home (str):
Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
force_overwrite (bool):
Whether to overwrite the existing downloaded data
partial_download (list):
List indicating what to partially download. The list can include any of:
* `'annotations_beat'` the beat annotation files
* `'annotations_sections'` the sections annotation files
* `'annotations_chords'` the chords annotation files
* `'annotations_vocal_act'` the vocal activity annotation files
* `'metadata'` the metadata files
If `None`, all data is downloaded.
cleanup (bool):
Whether to delete the zip/tar file after extracting.
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
info_message = """
Unfortunately the audio files of the RWC-Popular dataset are not available
for download. If you have the RWC-Popular dataset, place the contents into a
folder called RWC-Popular with the following structure:
> RWC-Popular/
> annotations/
> audio/rwc-p-m0i with i in [1 .. 7]
> metadata-master/
and copy the RWC-Popular folder to {}
""".format(data_home)
download_utils.downloader(
data_home,
remotes=REMOTES,
partial_download=partial_download,
info_message=info_message,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
def download(data_home=None):
"""Download Groove MIDI.
Args:
data_home (str): Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
download_utils.downloader(data_home,
zip_downloads=[AUDIO_MIDI_REMOTE],
cleanup=True)
os.rename(os.path.join(data_home, 'groove'),
os.path.join(data_home, 'Groove MIDI'))
def download(data_home=None):
"""Download TinySOL.
Args:
data_home (str): Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
download_utils.downloader(
data_home,
tar_downloads=[AUDIO_REMOTE],
file_downloads=[ANNOTATION_REMOTE],
cleanup=True,
)
def download(data_home=None):
"""Download GuitarSet.
Args:
data_home (str): Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
download_utils.downloader(
data_home,
zip_downloads=[
ANNOTATION_REMOTE,
AUDIO_HEX_CLN_REMOTE,
AUDIO_HEX_REMOTE,
AUDIO_MIC_REMOTE,
AUDIO_MIX_REMOTE,
],
cleanup=True,
)
def download(data_home=None, force_overwrite=False, cleanup=True):
"""Download the GTZAN-Genre dataset.
Args:
data_home (str):
Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
force_overwrite (bool):
Whether to overwrite the existing downloaded data
cleanup (bool):
Whether to delete the zip/tar file after extracting.
"""
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
download_utils.downloader(
data_home,
remotes=REMOTES,
info_message=None,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
def _download(save_dir, remotes, partial_download, info_message,
force_overwrite, cleanup):
"""Download the dataset.
Args:
save_dir (str):
The directory to download the data
remotes (dict or None):
A dictionary of RemoteFileMetadata tuples of data in zip format.
If None, there is no data to download
partial_download (list or None):
A list of keys to partially download the remote objects of the download dict.
If None, all data is downloaded
info_message (str or None):
A string of info to print when this function is called.
If None, no string is printed.
force_overwrite (bool):
If True, existing files are overwritten by the downloaded files.
cleanup (bool):
Whether to delete the zip/tar file after extracting.
"""
download_utils.downloader(
save_dir,
remotes=remotes,
info_message=None,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
# files get downloaded to a folder called groove - move everything up a level
groove_dir = os.path.join(save_dir, "groove")
groove_files = glob.glob(os.path.join(groove_dir, "*"))
for fpath in groove_files:
shutil.move(fpath, save_dir)
if os.path.exists(groove_dir):
shutil.rmtree(groove_dir)
def download(data_home=None, force_overwrite=False, cleanup=True):
"""Download the giantsteps_tempo Dataset (annotations).
The audio files are not provided due to copyright issues.
Args:
data_home (str):
Local path where the dataset is stored.
If `None`, looks for the data in the default directory, `~/mir_datasets`
force_overwrite (bool):
Whether to overwrite the existing downloaded data
cleanup (bool):
Whether to delete the zip/tar file after extracting.
partial_download(list of str)
arguments can be 'audio' 'metadata' or/and 'tempos'
"""
# use the default location: ~/mir_datasets/giantsteps_tempo
if data_home is None:
data_home = utils.get_default_dataset_path(DATASET_DIR)
download_message = """
Unfortunately the audio files of the Giant Steps Tempo dataset are not available
for download. If you have the Giant Steps audio dataset, place the contents into
a folder called GiantSteps_tempo with the following structure:
> GiantSteps_tempo/
> giantsteps-tempo-dataset-0b7d47ba8cae59d3535a02e3db69e2cf6d0af5bb/
> audio/
and copy the folder to {}
""".format(data_home)
download_utils.downloader(
data_home,
remotes=REMOTES,
info_message=download_message,
force_overwrite=force_overwrite,
cleanup=cleanup,
)
def download(self,
partial_download=None,
force_overwrite=False,
cleanup=False):
"""Download the dataset
Args:
partial_download (list or None):
A list of keys of remotes to partially download.
If None, all data is downloaded
force_overwrite (bool):
If True, existing files are overwritten by the downloaded files.
By default False.
cleanup (bool):
Whether to delete any zip/tar files after extracting.
Raises:
ValueError: if invalid keys are passed to partial_download
IOError: if a downloaded file's checksum is different from expected
"""
if not os.path.exists(self.data_home):
os.makedirs(self.data_home)
# Create these directories if doesn't exist
train = "acousticbrainz-mediaeval-train"
train_dir = os.path.join(self.data_home, train)
if not os.path.isdir(train_dir):
os.mkdir(train_dir)
validate = "acousticbrainz-mediaeval-validation"
validate_dir = os.path.join(self.data_home, validate)
if not os.path.isdir(validate_dir):
os.mkdir(validate_dir)
# start to download
for key, remote in self.remotes.items():
# check overwrite
file_downloaded = False
if not force_overwrite:
fold, first_dir = key.split("-")
first_dir_path = os.path.join(
train_dir if fold == "train" else validate_dir, first_dir)
if os.path.isdir(first_dir_path):
file_downloaded = True
logging.info(
"File " + remote.filename +
" downloaded. Skip download (force_overwrite=False).")
if not file_downloaded:
# if this typical error happend it repeat download
download_utils.downloader(
self.data_home,
remotes={key: remote},
partial_download=None,
info_message=None,
force_overwrite=True,
cleanup=cleanup,
)
# move from a temporary directory to final one
source_dir = os.path.join(self.data_home, "temp",
train if "train" in key else validate)
target_dir = train_dir if "train" in key else validate_dir
dir_names = os.listdir(source_dir)
for dir_name in dir_names:
shutil.move(
os.path.join(source_dir, dir_name),
os.path.join(target_dir, dir_name),
)
