Merge pull request #9 from RadarML/dev/bugfix

Miscellaneous Bugfixes and Improvements

Author: thetianshuhuang

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "nrdk"
-version = "0.1.2"
+version = "0.1.3"
 authors = [
   { name="Tianshu Huang", email="tianshu2@andrew.cmu.edu" },
 ]

src/nrdk/framework/result.py

@@ -33,12 +33,41 @@ class Result:
         └── events.out.tfevents...
         ```
 
+    !!! danger
+
+        Since `.hydra` is technically a hidden folder, many file operations
+        (e.g., `mv experiment_result/*`) will skip or hide this folder by
+        default.
+
     Args:
         path: path to results directory.
+        validate: check that the path exists, and that it matches the expected
+            structure.
     """
 
-    def __init__(self, path: str) -> None:
+    def __init__(self, path: str, validate: bool = True) -> None:
         self.path = path
+        if validate:
+            self.validate(path)
+
+    @staticmethod
+    def validate(path: str) -> None:
+        """Validate that the given path is a valid results directory.
+
+        Raises:
+            ValueError: if the path does not exist, or does not have the
+                expected structure.
+        """
+        if not os.path.exists(path):
+            raise ValueError(f"Result {path} does not exist.")
+        if not os.path.exists(os.path.join(path, ".hydra", "config.yaml")):
+            raise ValueError(
+                f"Result {path} exists, but does not have an associated "
+                f"hydra configuration {path}/.hydra/config.yaml.")
+        if not os.path.exists(os.path.join(path, "checkpoints.yaml")):
+            raise ValueError(
+                f"Result {path} exists, does not have a checkpoint index "
+                f"file {path}/checkpoints.yaml.")
 
     @overload
     def config(self, omegaconf: Literal[True] = True) -> DictConfig: ...

src/nrdk/roverd/dataloader.py

@@ -29,7 +29,7 @@ def datamodule(
     traces: Mapping[str, Sequence[str]],
     transforms: spec.Pipeline,
     batch_size: int = 32, samples: int | Sequence[int] = 0,
-    num_workers: int = 32, prefetch_factor: int = 2,
+    num_workers: int = 32, prefetch_factor: int | None = 2,
     subsample: Mapping[str, int | float | None] = {},
     ptrain: float = 0.8, pval: float = 0.2
 ) -> ADLDataModule:

src/nrdk/roverd/lidar.py

@@ -12,6 +12,17 @@
 from .transforms import SpectrumData
 
 
+def _n_range(
+    data: types.XWRRadarIQ[np.ndarray] | SpectrumData[np.ndarray]
+) -> int:
+    if isinstance(data, SpectrumData):
+        _batch, _t, _doppler, _el, _az, n_rng, _ch = data.spectrum.shape
+        return n_rng
+    else:  # XWRRadarIQ
+        _batch, _t, _n_slow, _tx, _rx, n_fast = data.iq.shape
+        return n_fast // 2
+
+
 @dataclass
 class Occupancy3DData(Generic[TArray]):
     """3D occupancy data.
@@ -62,7 +73,7 @@ def __init__(
 
     def __call__(
         self, lidar: types.OSDepth[np.ndarray],
-        radar: SpectrumData[np.ndarray],
+        radar: SpectrumData[np.ndarray] | types.XWRRadarIQ[np.ndarray],
         aug: Mapping[str, Any] = {}
     ) -> Occupancy3DData[np.ndarray]:
         """Create 3D occupancy map from Lidar depth data.
@@ -87,7 +98,7 @@ def __call__(
         if aug.get("azimuth_flip", False):
             rng = np.flip(rng, axis=3)
 
-        _batch, _t, _doppler, _el, _az, n_rng, _ch = radar.spectrum.shape
+        n_rng = _n_range(radar)
         n_bins = n_rng // self.d_rng
         bin = (rng // (radar.range_resolution * self.d_rng)).astype(np.uint16)
         bin[bin >= n_bins] = 0
@@ -177,7 +188,7 @@ def __init__(
 
     def __call__(
         self, lidar: types.OSDepth[np.ndarray],
-        radar: SpectrumData[np.ndarray],
+        radar: SpectrumData[np.ndarray] | types.XWRRadarIQ[np.ndarray],
         aug: Mapping[str, Any] = {}
     ) -> Occupancy2DData[np.ndarray]:
         """Create 2D occupancy map from Lidar depth data.
@@ -216,7 +227,7 @@ def __call__(
         r[(z > self.z_max) | (z < self.z_min)] = 0
 
         # Create map
-        _batch, _t, _doppler, _el, _az, n_rng, _ch = radar.spectrum.shape
+        n_rng = _n_range(radar)
         bin: UInt16[np.ndarray, "T El Az"] = (
             r // radar.range_resolution).astype(np.uint16)
         bin[bin >= n_rng] = 0

src/nrdk/tss/_cli.py

@@ -10,19 +10,23 @@
 
 def _cli(
     path: str, /,
-    pattern: str =  r"^(?P<experiment>(.*)).npz$",
-    key: str = "loss", timestamps: str | None = None,
-    experiments: list[str] | None = None,
-    baseline: str | None = None,
-    follow_symlinks: bool = False,
-    cut: float | None = None,
-    t_max: int | None = None,
+    pattern: str | None = None, key: str | None = None,
+    timestamps: str | None = None, experiments: list[str] | None = None,
+    baseline: str | None = None, follow_symlinks: bool = False,
+    cut: float | None = None, t_max: int | None = None,
     config: str | None = None,
 ) -> int:
     """Calculate statistics for time series metrics.
 
-    - pipe `tss ... > results.csv` to save the results to a file
-    - use `--config config.yaml` to avoid having to specify all these arguments
+    - pipe `tss ... > results.csv` to save the results to a file.
+    - use `--config config.yaml` to avoid having to specify all these
+        arguments; any arguments which are explicitly provided will override
+        the values in the config file.
+
+    !!! warning
+
+        `path` (and `--follow_symlinks`, if specified) are required to be
+        passed via the command line, and cannot be specified via the config.
 
     Args:
         path: directory to find evaluations in.
@@ -43,18 +47,23 @@ def _cli(
     if config is not None:
         with open(config) as f:
             cfg = yaml.safe_load(f)
-            return _cli(
-                path,
-                pattern=cfg.get("pattern",  r"^(?P<experiment>(.*)).npz$"),
-                experiments=cfg.get("experiments", None),
-                key=cfg.get("key", "loss"),
-                timestamps=cfg.get("timestamps", None),
-                baseline=cfg.get("baseline", None),
-                cut=cfg.get("cut", None),
-                t_max=cfg.get("t_max", None),
-                follow_symlinks=follow_symlinks)
-
-    index = api.index(path, pattern=pattern, follow_symlinks=follow_symlinks)
+    else:
+        cfg = {}
+
+    def setdefault(value, param, default):
+        if value is None:
+            value = cfg.get(param, default)
+        return value
+
+    pattern = setdefault(pattern, "pattern", r"^(?P<experiment>(.*)).npz$")
+    key = setdefault(key, "key", "loss")
+    timestamps = setdefault(timestamps, "timestamps", None)
+    baseline = setdefault(baseline, "baseline", None)
+    cut = setdefault(cut, "cut", None)
+    t_max = setdefault(t_max, "t_max", None)
+
+    index = api.index(
+        path, pattern=pattern, follow_symlinks=follow_symlinks)  # type: ignore
 
     if len(index) == 0:
         print("No result files found!")
@@ -64,7 +73,7 @@ def _cli(
         return -1
 
     df = api.dataframe_from_index(
-        index, key=key, baseline=baseline,
+        index, key=key, baseline=baseline,  # type: ignore
         experiments=experiments, cut=cut, t_max=t_max, timestamps=timestamps)
 
     buf = StringIO()