Updated setup.py and code formatting for packaging

Ismael Balafrej · Ismael Balafrej · commit 67edc1c23fbb · 2021-05-21T15:57:06.000-04:00
diff --git a/LICENSE b/LICENSE
@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2021 Ismael Balafrej
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
diff --git a/MANIFEST.in b/MANIFEST.in
diff --git a/README.md b/README.md
@@ -6,9 +6,9 @@ An event based dataset loader under one common python (>=3.5) API built on top o
 # Supported datasets
 
 1. Neuromorphic Mnist dataset from
-Orchard, G.; Cohen, G.; Jayawant, A.; and Thakor, N.
-“Converting Static Image Datasets to Spiking Neuromorphic Datasets Using Saccades",
-Frontiers in Neuroscience, vol.9, no.437, Oct. 2015. Available for download at https://www.garrickorchard.com/datasets/n-mnist
+    Orchard, G.; Cohen, G.; Jayawant, A.; and Thakor, N.
+    “Converting Static Image Datasets to Spiking Neuromorphic Datasets Using Saccades",
+    Frontiers in Neuroscience, vol.9, no.437, Oct. 2015. Available for download at https://www.garrickorchard.com/datasets/n-mnist
 
 2. NCaltech101 dataset from
     Orchard, G.; Cohen, G.; Jayawant, A.; and Thakor, N.
@@ -42,7 +42,7 @@ Frontiers in Neuroscience, vol.9, no.437, Oct. 2015. Available for download at h
 # Installation
 You can install the latest version of this package with:
 ```bash
-pip install git+https://github.com/tihbe/python-ebdataset.git
+pip install ebdataset
 ```
 
 # Getting started
@@ -72,3 +72,7 @@ python -m ebdataset.visualization.spike_train_to_vid NMnist path
 ```
 
 ![](images/nmnist-2.gif) ![](images/nmnist-9.gif)
+
+# Contributing
+
+Feel free to create a pull request if you're interested in this project. 
diff --git a/ebdataset/audio/ntidigits.py b/ebdataset/audio/ntidigits.py
@@ -12,9 +12,7 @@ class NTidigits(data.Dataset):
     Available for download at https://docs.google.com/document/d/1Uxe7GsKKXcy6SlDUX4hoJVAC0-UkH-8kr5UXp0Ndi1M
     """
 
-    def __init__(
-        self, path: str, is_train=True, transforms=None, only_single_digits=False
-    ):
+    def __init__(self, path: str, is_train=True, transforms=None, only_single_digits=False):
         assert os.path.exists(path)
         self.prename = "train" if is_train else "test"
         self.path = path
@@ -24,11 +22,7 @@ def __init__(
             self.samples = f[self.prename + "_labels"][()]
 
         if only_single_digits:
-            self.samples = list(
-                filter(
-                    lambda s: len(NTidigits._get_label_for_sample(s)) == 1, self.samples
-                )
-            )
+            self.samples = list(filter(lambda s: len(NTidigits._get_label_for_sample(s)) == 1, self.samples))
 
     @staticmethod
     def _get_label_for_sample(sample_id):
@@ -43,9 +37,7 @@ def __getitem__(self, index):
             addresses = f[self.prename + "_addresses"][sample_id][()]
             ts = f[self.prename + "_timestamps"][sample_id][()]
 
-        sparse_spike_train = np.recarray(
-            shape=len(ts), dtype=[("addr", addresses.dtype), ("ts", ts.dtype)]
-        )
+        sparse_spike_train = np.recarray(shape=len(ts), dtype=[("addr", addresses.dtype), ("ts", ts.dtype)])
         sparse_spike_train.addr = addresses
         sparse_spike_train.ts = ts
 
diff --git a/ebdataset/utils/__init__.py b/ebdataset/utils/__init__.py
@@ -28,7 +28,11 @@ def unzip(zip_file_path, output_directory, verbose=True, desc="Extracting"):
     with ZipFile(zip_file_path, "r") as zf:
         size = sum((f.file_size for f in zf.infolist()))
         with tqdm(
-            total=size, unit="B", unit_scale=True, desc=desc, disable=not verbose,
+            total=size,
+            unit="B",
+            unit_scale=True,
+            desc=desc,
+            disable=not verbose,
         ) as pbar:
             for file in zf.infolist():
                 if file.is_dir():
diff --git a/ebdataset/vision/ibm_gesture.py b/ebdataset/vision/ibm_gesture.py
@@ -61,15 +61,11 @@ def __init__(self, path: str, shuffle: bool = True):
 
         # Read train trials file
         with open(os.path.join(path, self._TRAIN_TRIALS_FILE), "r") as f:
-            self._TRAIN_FILES = map(
-                lambda d: os.path.join(path, d.rstrip()), f.readlines()
-            )
+            self._TRAIN_FILES = map(lambda d: os.path.join(path, d.rstrip()), f.readlines())
 
         # Read test trials file
         with open(os.path.join(path, self._TEST_TRIALS_FILE), "r") as f:
-            self._TEST_FILES = map(
-                lambda d: os.path.join(path, d.rstrip()), f.readlines()
-            )
+            self._TEST_FILES = map(lambda d: os.path.join(path, d.rstrip()), f.readlines())
 
         self._TRAIN_FILES = list(filter(lambda f: os.path.isfile(f), self._TRAIN_FILES))
         self._TEST_FILES = list(filter(lambda f: os.path.isfile(f), self._TEST_FILES))
@@ -80,9 +76,7 @@ def __init__(self, path: str, shuffle: bool = True):
 
     def _read_labels(self, file: str) -> np.array:
         assert os.path.exists(file), "File %s doesn't exist" % file
-        return np.genfromtxt(
-            file, delimiter=",", skip_header=1, dtype=self._LABELS_DTYPE
-        )
+        return np.genfromtxt(file, delimiter=",", skip_header=1, dtype=self._LABELS_DTYPE)
 
     def _parse_filename(self, file: str) -> Tuple[str, str, str]:
         trial = re.search(r"^user([0-9]+)_(.+)\.(aedat|csv)$", file, re.IGNORECASE)
@@ -108,13 +102,9 @@ def _create_generator(self, files: List[str]):
             labels = self._read_labels(file.replace(".aedat", "_labels.csv"))
             multilabel_spike_train = readAEDATv3(file)
             for (label_id, start_time, end_time) in labels:
-                event_mask = (multilabel_spike_train.ts >= start_time) & (
-                    multilabel_spike_train.ts < end_time
-                )
+                event_mask = (multilabel_spike_train.ts >= start_time) & (multilabel_spike_train.ts < end_time)
                 ts = multilabel_spike_train.ts[event_mask] - start_time
-                spike_train = DVSSpikeTrain(
-                    ts.size, width=128, height=128, duration=end_time - start_time + 1
-                )
+                spike_train = DVSSpikeTrain(ts.size, width=128, height=128, duration=end_time - start_time + 1)
                 spike_train.ts = ts
                 spike_train.x = multilabel_spike_train.x[event_mask]
                 spike_train.y = multilabel_spike_train.y[event_mask]
@@ -165,9 +155,7 @@ def __init__(self, path: str, is_train: bool = True):
         """
         _, file_extension = os.path.splitext(path)
         if file_extension != ".h5":
-            raise Exception(
-                "The dvs gesture must first be converted to a .h5 file. Please call H5DvsGesture.Convert"
-            )
+            raise Exception("The dvs gesture must first be converted to a .h5 file. Please call H5DvsGesture.Convert")
 
         self.indx = 0 if is_train else 1
         self.file_path = path
@@ -189,9 +177,7 @@ def convert(dvs_folder_path: str, h5_output_path: str, verbose=True):
         position_type = h5py.vlen_dtype(np.dtype("uint16"))
         time_type = h5py.vlen_dtype(np.dtype("uint32"))
 
-        step_counter = tqdm(
-            total=sum(H5IBMGesture._nb_of_samples), disable=(not verbose)
-        )
+        step_counter = tqdm(total=sum(H5IBMGesture._nb_of_samples), disable=(not verbose))
 
         with h5py.File(h5_output_path, "w-") as f:
             for (name, gen, length) in zip(
@@ -223,9 +209,7 @@ def __getitem__(self, index):
             tos = file_hndl[name + "_tos"][index]
             label = file_hndl[name + "_label"][index]
 
-        spike_train = DVSSpikeTrain(
-            tos.size, width=128, height=128, duration=tos.max() + 1
-        )
+        spike_train = DVSSpikeTrain(tos.size, width=128, height=128, duration=tos.max() + 1)
         spike_train.x = pos[0]
         spike_train.y = pos[1]
         spike_train.p = pos[2]
diff --git a/ebdataset/vision/ini_roshambo.py b/ebdataset/vision/ini_roshambo.py
@@ -30,9 +30,7 @@ def __init__(self, path: str, with_backgrounds=False, transforms=None):
 
         if os.path.isdir(path):  # AEDat v2 directory
             self.backend = "aedat"
-            self.samples = filter(
-                lambda f: os.path.splitext(f)[1] == ".aedat", os.listdir(path)
-            )
+            self.samples = filter(lambda f: os.path.splitext(f)[1] == ".aedat", os.listdir(path))
         elif os.path.splitext(path)[1] == ".h5":
             self.backend = "h5"
             with File(path, "r", libver="latest") as f_hndl:
@@ -54,27 +52,19 @@ def convert(self, out_path, verbose=False):
         :return: New Roshambo object with h5 file as backend
         """
 
-        if (
-            self.backend == "h5"
-        ):  # Send back object if we're already using an h5 backend
+        if self.backend == "h5":  # Send back object if we're already using an h5 backend
             return self
 
         if not (".h5" in out_path):
             out_path += ".h5"
 
         with File(out_path, "w-", libver="latest") as f_hndl:
             for sample_id in tqdm(self.samples, disable=not verbose):
-                sparse_spike_train = readAEDATv2_davies(
-                    os.path.join(self.path, sample_id)
-                )
-                sparse_spike_train.ts = sparse_spike_train.ts - np.min(
-                    sparse_spike_train.ts
-                )  # Start the sample at t=0
+                sparse_spike_train = readAEDATv2_davies(os.path.join(self.path, sample_id))
+                sparse_spike_train.ts = sparse_spike_train.ts - np.min(sparse_spike_train.ts)  # Start the sample at t=0
                 f_hndl[sample_id] = sparse_spike_train
 
-        return INIRoshambo(
-            out_path, with_backgrounds=self.with_backgrounds, transforms=self.transforms
-        )
+        return INIRoshambo(out_path, with_backgrounds=self.with_backgrounds, transforms=self.transforms)
 
     def split_to_subsamples(self, out_path, duration_per_sample, verbose=False):
         if not (".h5" in out_path):
@@ -96,15 +86,11 @@ def split_to_subsamples(self, out_path, duration_per_sample, verbose=False):
                 ):
                     if start_time + duration_per_sample > total_duration:  # End
                         break
-                    sub_mask = (sample.ts >= start_time) & (
-                        sample.ts < start_time + duration_per_sample
-                    )
+                    sub_mask = (sample.ts >= start_time) & (sample.ts < start_time + duration_per_sample)
                     nb_of_spikes = np.sum(sub_mask)
                     if nb_of_spikes <= 10:
                         continue
-                    sub_sample = DVSSpikeTrain(
-                        nb_of_spikes, duration=duration_per_sample
-                    )
+                    sub_sample = DVSSpikeTrain(nb_of_spikes, duration=duration_per_sample)
                     sub_sample.ts = sample.ts[sub_mask]
                     sub_sample.ts = sub_sample.ts - np.min(sub_sample.ts)  # Start at 0
                     sub_sample.x = sample.x[sub_mask]
@@ -121,15 +107,11 @@ def __getitem__(self, index):
         if self.backend == "aedat":
             filename = os.path.join(self.path, sample_id)
             sparse_spike_train = readAEDATv2_davies(filename)
-            sparse_spike_train.ts = sparse_spike_train.ts - np.min(
-                sparse_spike_train.ts
-            )  # Start the sample at t=0
+            sparse_spike_train.ts = sparse_spike_train.ts - np.min(sparse_spike_train.ts)  # Start the sample at t=0
         elif self.backend == "h5":
             with File(self.path, "r", libver="latest") as f_hndl:
                 sparse_spike_train = f_hndl[sample_id][()]
-            sparse_spike_train = np.rec.array(
-                sparse_spike_train, dtype=sparse_spike_train.dtype
-            ).view(DVSSpikeTrain)
+            sparse_spike_train = np.rec.array(sparse_spike_train, dtype=sparse_spike_train.dtype).view(DVSSpikeTrain)
 
         sparse_spike_train.width = 240
         sparse_spike_train.height = 180
diff --git a/ebdataset/vision/nmnist.py b/ebdataset/vision/nmnist.py
@@ -56,10 +56,12 @@ def _download_and_unzip(self, output_directory):
         test_url = "https://www.dropbox.com/sh/tg2ljlbmtzygrag/AADSKgJ2CjaBWh75HnTNZyhca/Test.zip?dl=1"
         train_loc = os.path.join(output_directory, "Train%i.zip" % time.time())
         test_loc = os.path.join(output_directory, "Test%i.zip" % time.time())
-        success = download(train_url, train_loc, desc="Downloading training files") and \
-            unzip(train_loc, output_directory, desc="Extracting training files") and \
-            download(test_url, test_loc, desc="Downloading test files") and \
-            unzip(test_loc, output_directory, desc="Extracting test files")
+        success = (
+            download(train_url, train_loc, desc="Downloading training files")
+            and unzip(train_loc, output_directory, desc="Extracting training files")
+            and download(test_url, test_loc, desc="Downloading test files")
+            and unzip(test_loc, output_directory, desc="Extracting test files")
+        )
 
         if success:
             os.remove(train_loc)
diff --git a/ebdataset/vision/transforms.py b/ebdataset/vision/transforms.py
@@ -8,8 +8,7 @@
 
 
 class ScaleDown(object):
-    """Scale down a 2d sparse spike train by factor (both in x and y)
-    """
+    """Scale down a 2d sparse spike train by factor (both in x and y)"""
 
     def __init__(self, width, height, factor):
         self.authorized_x = list(range(0, width, factor))
@@ -65,10 +64,7 @@ def __call__(self, sparse_spike_train):
         ts = (sparse_spike_train.ts * time_scale).astype(int)
 
         dense_spike_train[
-            sparse_spike_train.x.astype(int),
-            sparse_spike_train.y.astype(int),
-            sparse_spike_train.p.astype(int),
-            ts
+            sparse_spike_train.x.astype(int), sparse_spike_train.y.astype(int), sparse_spike_train.p.astype(int), ts
         ] = 1
 
         return dense_spike_train
diff --git a/ebdataset/vision/type.py b/ebdataset/vision/type.py
@@ -1,28 +1,15 @@
 import numpy as np
 
-_dtype = np.dtype(
-    [("x", np.uint16), ("y", np.uint16), ("p", np.bool_), ("ts", np.uint64)]
-)
+_dtype = np.dtype([("x", np.uint16), ("y", np.uint16), ("p", np.bool_), ("ts", np.uint64)])
 
 
 class DVSSpikeTrain(np.recarray):
     """Common type for event based vision datasets"""
 
     __name__ = "SparseVisionSpikeTrain"
 
-    def __new__(
-        cls,
-        nb_of_spikes,
-        *args,
-        width=-1,
-        height=-1,
-        duration=-1,
-        time_scale=1e-6,
-        **nargs
-    ):
-        obj = super(DVSSpikeTrain, cls).__new__(
-            cls, nb_of_spikes, dtype=_dtype, *args, **nargs
-        )
+    def __new__(cls, nb_of_spikes, *args, width=-1, height=-1, duration=-1, time_scale=1e-6, **nargs):
+        obj = super(DVSSpikeTrain, cls).__new__(cls, nb_of_spikes, dtype=_dtype, *args, **nargs)
         obj.width = width
         obj.height = height
         obj.duration = duration
diff --git a/ebdataset/visualization/__init__.py b/ebdataset/visualization/__init__.py
diff --git a/ebdataset/visualization/spike_train_to_vid.py b/ebdataset/visualization/spike_train_to_vid.py
@@ -28,29 +28,19 @@
     NMnist,
     PropheseeNCars,
 ]
-dataset_map = dict(
-    zip([dataset.__name__ for dataset in available_datasets], available_datasets)
-)
+dataset_map = dict(zip([dataset.__name__ for dataset in available_datasets], available_datasets))
 
-parser.add_argument(
-    "dataset", help="Dataset - One of [%s]" % " | ".join(dataset_map.keys())
-)
-parser.add_argument(
-    "path", help="Path of the data directory or file for the chosen dataset"
-)
+parser.add_argument("dataset", help="Dataset - One of [%s]" % " | ".join(dataset_map.keys()))
+parser.add_argument("path", help="Path of the data directory or file for the chosen dataset")
 parser.add_argument(
     "-n",
     "--num_samples",
     help="Number of video samples to generate",
     type=int,
     default=10,
 )
-parser.add_argument(
-    "-d", "--dilatation", help="Time dilatation scale", type=float, default=1.0
-)  # Default Real time
-parser.add_argument(
-    "-s", "--scale", help="Spatial scaling", type=float, default=1.0
-)  # Default Real size
+parser.add_argument("-d", "--dilatation", help="Time dilatation scale", type=float, default=1.0)  # Default Real time
+parser.add_argument("-s", "--scale", help="Spatial scaling", type=float, default=1.0)  # Default Real size
 
 args = parser.parse_args()
 dataset = dataset_map[args.dataset]
@@ -70,17 +60,13 @@
         int(spike_train.height * spatial_scale),
     )
     out_duration = spike_train.duration * spike_train.time_scale * time_scale
-    out = cv2.VideoWriter(
-        filename, cv2.VideoWriter_fourcc(*"MP42"), 60.0, (out_width, out_height)
-    )
+    out = cv2.VideoWriter(filename, cv2.VideoWriter_fourcc(*"MP42"), 60.0, (out_width, out_height))
     for frame_start in np.arange(0.0, out_duration, 1 / 60.0):
         frame_end = frame_start + 1 / 60.0
         ts = spike_train.ts * spike_train.time_scale * time_scale
         mask = (ts >= frame_start) & (ts < frame_end)
         frame = np.zeros((out_width, out_height, 3), dtype=np.uint8)
-        for x, y, p in zip(
-            spike_train.x[mask], spike_train.y[mask], spike_train.p[mask]
-        ):
+        for x, y, p in zip(spike_train.x[mask], spike_train.y[mask], spike_train.p[mask]):
             frame[
                 int(x * spatial_scale) : int((x + 1) * spatial_scale),
                 int(y * spatial_scale) : int((y + 1) * spatial_scale),
diff --git a/ebdataset/visualization/time_binning.py b/ebdataset/visualization/time_binning.py
diff --git a/pyproject.toml b/pyproject.toml
diff --git a/setup.py b/setup.py
