work on fsspec

dascore/io/ap_sensing/utils.py

@@ -4,7 +4,8 @@
 
 import dascore as dc
 from dascore.core import get_coord, get_coord_manager
-from dascore.utils.misc import _maybe_unpack, get_path, unbyte
+from dascore.utils.misc import _maybe_unpack, unbyte
+from dascore.utils.fs import get_uri
 
 
 def _get_version_string(resource):
@@ -83,7 +84,7 @@ def _get_attrs_dict(resource, format_name):
         instrumet_id=unbyte(_maybe_unpack(daq["SerialNumber"])),
         gauge_length=_maybe_unpack(pserver["GaugeLength"]),
         radians_to_nano_strain=_maybe_unpack(pserver["RadiansToNanoStrain"]),
-        path=get_path(resource),
+        path=get_uri(resource),
         format_name=format_name,
         format_version=version,
     )

dascore/io/asn/utils.py

@@ -6,7 +6,9 @@
 import dascore.core
 from dascore.core.coords import get_coord
 from dascore.utils.hdf5 import unpack_scalar_h5_dataset
-from dascore.utils.misc import get_path, unbyte
+from dascore.utils.misc import unbyte
+from dascore.utils.fs import get_uri
+
 
 # --- Getting format/version
 
@@ -78,7 +80,7 @@ def _get_attr_dict(header, path, format_name, format_version):
 def _get_opto_das_coords_attrs(fi, format_name) -> tuple[dc.CoordManager, dict]:
     """Scan a OptoDAS file, return metadata."""
     cm = _get_coord_manager(fi)
-    path = get_path(fi)
+    path = get_uri(fi)
     version = _get_opto_das_version_str(fi)
     attrs = _get_attr_dict(fi["header"], path, format_name, version)
     return cm, attrs

dascore/io/core.py

@@ -38,7 +38,8 @@
 )
 from dascore.utils.io import IOResourceManager, get_handle_from_resource
 from dascore.utils.mapping import FrozenDict
-from dascore.utils.misc import _iter_filesystem, cached_method, iterate, warn_or_raise
+from dascore.utils.misc import cached_method, iterate, warn_or_raise
+from dascore.utils.fs import _iter_filesystem
 from dascore.utils.models import (
     CommaSeparatedStr,
     DascoreBaseModel,

dascore/io/dasdae/core.py

@@ -9,7 +9,8 @@
     H5Reader,
     H5Writer,
 )
-from dascore.utils.misc import get_path, unbyte
+from dascore.utils.misc import unbyte
+from ...utils.fs import get_uri
 from dascore.utils.patch import get_patch_names
 
 from .utils import (
@@ -88,7 +89,7 @@ def get_format(self, resource: H5Reader, **kwargs) -> tuple[str, str] | bool:
     def read(self, resource: H5Reader, **kwargs) -> SpoolType:
         """Read a DASDAE file."""
         patches = []
-        path = get_path(resource)
+        path = get_uri(resource)
         format_version = unbyte(resource.attrs["__DASDAE_version__"])
         format_name = self.name
         try:

dascore/io/dasdae/utils.py

@@ -9,7 +9,8 @@
 from dascore.core.coordmanager import get_coord_manager
 from dascore.core.coords import get_coord
 from dascore.utils.hdf5 import Empty
-from dascore.utils.misc import get_path, suppress_warnings, unbyte
+from dascore.utils.misc import suppress_warnings, unbyte
+from dascore.utils.fs import get_uri
 from dascore.utils.time import to_int
 
 # --- Functions for writing DASDAE format
@@ -189,7 +190,7 @@ def _read_patch(patch_group, path, format_name, format_version, **kwargs):
 
 def _get_summary_from_patch_groups(h5, format_name="DASDAE"):
     """Get the contents from each patch group."""
-    path = get_path(h5)
+    path = get_uri(h5)
     format_version = h5.attrs["__DASDAE_version__"]
     out = []
     for name, group in h5[("/waveforms")].items():

dascore/io/dashdf5/core.py

@@ -6,7 +6,7 @@
 from dascore.constants import opt_timeable_types
 from dascore.io import FiberIO
 from dascore.utils.hdf5 import H5Reader
-from dascore.utils.misc import get_path
+from ...utils.fs import get_uri
 
 from .utils import _get_cf_attrs, _get_cf_coords, _get_cf_version_str
 
@@ -21,7 +21,7 @@ class DASHDF5(FiberIO):
     def _get_attr(self, resource: H5Reader):
         """Get the attrs dict with path and such populated."""
         attrs = _get_cf_attrs(resource)
-        attrs["path"] = get_path(resource)
+        attrs["path"] = get_uri(resource)
         attrs["format_name"] = self.name
         attrs["format_version"] = self.version
         return dc.PatchAttrs.model_validate(attrs)

dascore/io/febus/core.py

@@ -10,7 +10,7 @@
 from dascore.constants import opt_timeable_types
 from dascore.io import FiberIO
 from dascore.utils.hdf5 import H5Reader
-from dascore.utils.misc import get_path
+from ...utils.fs import get_uri
 from dascore.utils.models import UTF8Str
 
 from .utils import (
@@ -71,7 +71,7 @@ def scan(self, resource: H5Reader, **kwargs) -> list[dc.PatchSummary]:
         """Scan a febus file, return summary information about the file's contents."""
         return _scan_febus(
             resource,
-            path=get_path(resource),
+            path=get_uri(resource),
             format_name=self.name,
             format_version=self.version,
             attr_cls=FebusPatchAttrs,
@@ -87,7 +87,7 @@ def read(
         """Read a febus spool of patches."""
         patches = _read_febus(
             resource,
-            path=get_path(resource),
+            path=get_uri(resource),
             format_name=self.name,
             format_version=self.version,
             time=time,

dascore/io/gdr/core.py

@@ -17,7 +17,7 @@
     _maybe_trim_data,
 )
 from dascore.utils.hdf5 import H5Reader
-from dascore.utils.misc import get_path
+from dascore.utils.fs import get_uri
 
 
 class GDRPatchAttrs(dc.PatchAttrs):
@@ -40,7 +40,7 @@ class GDR_V1(FiberIO):  # noqa
     def _get_attr_coord_data(self, resource, snap=True):
         """Get the attributes, coordinates, and h5 dataset."""
         attr_dict, cm, data = _get_attrs_coords_and_data(resource, snap=snap)
-        attr_dict["path"] = get_path(resource)
+        attr_dict["path"] = get_uri(resource)
         attr_dict["format_name"] = self.name
         attr_dict["version"] = self.version
         attr = GDRPatchAttrs(**attr_dict)

dascore/io/indexer.py

@@ -18,7 +18,6 @@
 import dascore as dc
 from dascore.constants import ONE_SECOND_IN_NS, PROGRESS_LEVELS
 from dascore.exceptions import InvalidIndexVersionError
-from dascore.utils.hdf5 import HDFPatchIndexManager
 from dascore.utils.misc import iterate
 from dascore.utils.pd import filter_df
 from dascore.utils.time import get_max_min_times, to_timedelta64

dascore/io/neubrex/core.py

@@ -12,7 +12,7 @@
 from dascore.constants import SpoolType
 from dascore.io import FiberIO
 from dascore.utils.hdf5 import H5Reader
-from dascore.utils.misc import get_path
+from dascore.utils.fs import get_uri
 
 
 class NeubrexRFSPatchAttrs(dc.PatchAttrs):
@@ -53,7 +53,7 @@ class NeubrexRFSV1(FiberIO):
     def _get_attrs(self, resource) -> NeubrexRFSPatchAttrs:
         """Get the patch attributes."""
         attr = rfs_utils._get_attr_dict(resource)
-        attr["path"] = get_path(resource)
+        attr["path"] = get_uri(resource)
         attr["format_name"] = self.name
         attr["format_version"] = self.version
         return NeubrexRFSPatchAttrs(**attr)
@@ -106,7 +106,7 @@ class NeubrexDASV1(FiberIO):
     def _get_attr(self, resource) -> NeubrexDASPatchAttrs:
         """Get the attrs for from the file."""
         attr = das_utils._get_attr_dict(resource["Acoustic"])
-        attr["path"] = get_path(resource)
+        attr["path"] = get_uri(resource)
         attr["format_name"] = self.name
         attr["format_version"] = self.version
         return NeubrexDASPatchAttrs(**attr)

dascore/io/pickle/core.py

@@ -6,7 +6,7 @@
 
 import dascore as dc
 from dascore.io import BinaryReader, BinaryWriter, FiberIO
-from dascore.utils.misc import get_path
+from dascore.utils.fs import get_uri
 
 
 class PickleIO(FiberIO):
@@ -53,7 +53,7 @@ def read(self, resource: BinaryReader, **kwargs):
         patch: dc.Patch = pickle.load(resource)
         # Add the relevant file info.
         out = patch.update_attrs(
-            path=get_path(resource),
+            path=get_uri(resource),
             format_name=self.name,
             format_version=self.version,
         )

dascore/io/prodml/utils.py

@@ -6,7 +6,8 @@
 from dascore.constants import VALID_DATA_TYPES
 from dascore.core.coordmanager import get_coord_manager
 from dascore.core.coords import get_coord
-from dascore.utils.misc import get_path, iterate, maybe_get_items, unbyte
+from dascore.utils.misc import iterate, maybe_get_items, unbyte
+from dascore.utils.fs import get_uri
 
 # --- Getting format/version
 
@@ -100,7 +101,7 @@ def _get_prodml_attrs(fi, format_name, format_version) -> list[dict]:
     acq = fi["Acquisition"]
     base_info = maybe_get_items(acq.attrs, _root_attrs)
     raw_nodes = _get_raw_node_dict(acq)
-    path = get_path(fi)
+    path = get_uri(fi)
 
     # Iterate each raw data node. I have only ever seen 1 in a file but since
     # it is indexed like Raw[0] there might be more.

dascore/io/segy/core.py

@@ -5,7 +5,8 @@
 import dascore as dc
 from dascore.io.core import FiberIO
 from dascore.utils.io import BinaryReader
-from dascore.utils.misc import get_path, optional_import
+from dascore.utils.misc import optional_import
+from ...utils.fs import get_uri
 
 from .utils import (
     _get_coords,
@@ -31,7 +32,7 @@ class SegyV1_0(FiberIO):  # noqa
     def _get_attrs(self, resource):
         """Get the basic attributes for a segy file."""
         info = dict(
-            path=get_path(resource),
+            path=get_uri(resource),
             format_name=self.name,
             format_version=self.version,
         )

dascore/io/sentek/core.py

@@ -7,7 +7,7 @@
 import dascore as dc
 from dascore.io import BinaryReader
 from dascore.io.core import FiberIO
-from dascore.utils.misc import get_path
+from ...utils.fs import get_uri
 from dascore.utils.models import ArraySummary
 
 from .utils import _get_patch_attrs, _get_version
@@ -24,7 +24,7 @@ def _get_attrs_coords_offsets(self, resource):
         """Get attributes, coordinates, and data offsets from file."""
         attrs_dict, coords, offsets = _get_patch_attrs(
             resource,
-            path=get_path(resource),
+            path=get_uri(resource),
             format_name=self.name,
             format_version=self.version,
         )

dascore/io/sentek/utils.py

@@ -8,12 +8,12 @@
 
 import dascore as dc
 from dascore.core import get_coord, get_coord_manager
-from dascore.utils.misc import get_path
+from dascore.utils.fs import get_uri
 
 
 def _get_version(fid):
     """Determine if Sentek file."""
-    path = get_path(fid)
+    path = get_uri(fid)
     # Sentek files cannot change the extension, or file name.
     sw_data = path.endswith(".das")
     fid.seek(0)

dascore/io/silixah5/core.py

@@ -11,7 +11,7 @@
 from dascore.constants import opt_timeable_types
 from dascore.io import FiberIO
 from dascore.utils.hdf5 import H5Reader
-from dascore.utils.misc import get_path
+from dascore.utils.fs import get_uri
 
 
 class SilixaPatchAttrs(dc.PatchAttrs):
@@ -33,7 +33,7 @@ class SilixaH5V1(FiberIO):
     def _get_attr_coords(self, resource):
         """Get attributes and coordinates of patch in file."""
         info, coords = util._get_attr_dict(resource)
-        info["path"] = get_path(resource)
+        info["path"] = get_uri(resource)
         info["format_name"] = self.name
         info["format_version"] = self.version
         return SilixaPatchAttrs(**info), coords

dascore/io/tdms/core.py

@@ -6,7 +6,7 @@
 from dascore.constants import timeable_types
 from dascore.core import Patch
 from dascore.io import BinaryReader, FiberIO
-from dascore.utils.misc import get_path
+from ...utils.fs import get_uri
 
 from .utils import _get_attrs_coords, _get_data, _get_version_str
 
@@ -22,7 +22,7 @@ class TDMSFormatterV4713(FiberIO):
     def _get_attr_coords(self, resource):
         """Get a PatchAttrs for the file."""
         out, coords, _ = _get_attrs_coords(resource)
-        out["path"] = get_path(resource)
+        out["path"] = get_uri(resource)
         out["file_format"] = self.name
         out["file_version"] = self.version
         return dc.PatchAttrs(**out), coords

dascore/io/terra15/core.py

@@ -8,7 +8,7 @@
 from dascore.constants import timeable_types
 from dascore.io import FiberIO
 from dascore.utils.hdf5 import H5Reader
-from dascore.utils.misc import get_path
+from ...utils.fs import get_uri
 
 from .utils import (
     _get_default_attrs,
@@ -39,7 +39,7 @@ def _get_attrs_coords_data_node(self, resource):
         """Get attributes, coords, and datanode for this file."""
         version, data_node = _get_version_data_node(resource)
         attrs = _get_default_attrs(resource)
-        attrs["path"] = get_path(resource)
+        attrs["path"] = get_uri(resource)
         attrs["format_name"] = self.name
         attrs["format_version"] = version
         coords_dict = {

dascore/io/xml_binary/core.py

@@ -11,6 +11,7 @@
 import dascore as dc
 from dascore.io import FiberIO
 from dascore.utils.models import UTF8Str
+from ...utils.fs import get_uri
 
 from .utils import _load_patches, _paths_to_attrs, _read_xml_metadata
 
@@ -36,9 +37,9 @@ class XMLBinaryV1(FiberIO):
     # File extension for data files.
     _data_extension = ".raw"
 
-    def scan(self, resource, timestamp=None, **kwargs) -> list[dc.PatchAttrs]:
+    def scan(self, resource, timestamp=None, **kwargs) -> list[dc.PatchSummary]:
         """Scan the contents of the directory."""
-        path = Path(resource)
+        path = get_uri(resource)
         metadata = _read_xml_metadata(path / self._metadata_name)
         data_files = list(path.glob(f"*{self._data_extension}"))
         extra_attrs = {