pysits 2.0.0.dev8__tar.gz → 2.0.0.dev9__tar.gz

{pysits-2.0.0.dev8 → pysits-2.0.0.dev9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: pysits
-Version: 2.0.0.dev8
+Version: 2.0.0.dev9
 Summary: Python wrapper for the sits R package
 Project-URL: Homepage, https://github.com/e-sensing/pysits
 Project-URL: Issue Tracker, https://github.com/e-sensing/pysits/issues

{pysits-2.0.0.dev8 → pysits-2.0.0.dev9}/pyproject.toml

@@ -17,7 +17,7 @@
 
 [project]
 name = "pysits"
-version = "2.0.0.dev8"
+version = "2.0.0.dev9"
 description = "Python wrapper for the sits R package"
 readme = "README.md"
 requires-python = ">=3.10,<4"

{pysits-2.0.0.dev8 → pysits-2.0.0.dev9}/pysits/__init__.py

@@ -87,8 +87,10 @@ from .sits.data import sits_summary as summary
 from .sits.exporters import (
     sits_as_geopandas,
     sits_as_xarray,
+    sits_from_parquet,
     sits_timeseries_to_csv,
     sits_to_csv,
+    sits_to_parquet,
     sits_to_xlsx,
 )
 from .sits.impute import (
@@ -276,6 +278,8 @@ __all__ = (
     "sits_as_geopandas",
     "sits_to_xlsx",
     "sits_timeseries_to_csv",
+    "sits_to_parquet",
+    "sits_from_parquet",
     # Tuning
     "sits_tuning_hparams",
     "sits_tuning",

pysits-2.0.0.dev9/pysits/docs/content/sits_from_parquet.md

@@ -0,0 +1,26 @@
+Read sits time series from the Parquet format
+
+Reads a file written by `sits_to_parquet` and rebuilds the time series,
+with its class, column order and nested series.
+
+Files written by other programs are also read. Their class is then
+inferred from the columns, which must include "longitude", "latitude",
+"start_date", "end_date" and "label".
+
+Args:
+    file (str | pathlib.Path): Full path or HTTP(S) URL of the file to
+        read (valid file name with extension ".parquet").
+
+Returns:
+    SITSTimeSeriesModel: Time series.
+
+Raises:
+    httpx2.HTTPError: If a remote file cannot be downloaded.
+
+Examples:
+    import tempfile
+    from pysits import *
+
+    parquet_file = tempfile.gettempdir() + "/samples.parquet"
+    sits_to_parquet(samples_modis_ndvi, file=parquet_file)
+    samples = sits_from_parquet(parquet_file)

pysits-2.0.0.dev9/pysits/docs/content/sits_to_parquet.md

@@ -0,0 +1,23 @@
+Export sits time series to the Parquet format
+
+Writes a set of time series to a Parquet file. The layout is long: one
+row per sample and date, with the time series as plain columns. The
+file metadata keeps what is needed to rebuild the data with
+`sits_from_parquet` (class, column order and bands). It also follows
+GeoParquet 1.1, so tools like GDAL and QGIS read the file as points.
+
+Args:
+    data (SITSTimeSeriesModel): Time series.
+
+    file (str | pathlib.Path): Full path of the exported file (valid
+        file name with extension ".parquet").
+
+Returns:
+    None
+
+Examples:
+    import tempfile
+    from pysits import *
+
+    parquet_file = tempfile.gettempdir() + "/samples.parquet"
+    sits_to_parquet(samples_modis_ndvi, file=parquet_file)

{pysits-2.0.0.dev8 → pysits-2.0.0.dev9}/pysits/sits/exporters/__init__.py

@@ -17,7 +17,13 @@
 
 """Exporters module."""
 
-from .files import sits_timeseries_to_csv, sits_to_csv, sits_to_xlsx
+from .files import (
+    sits_from_parquet,
+    sits_timeseries_to_csv,
+    sits_to_csv,
+    sits_to_parquet,
+    sits_to_xlsx,
+)
 from .sf import sits_as_geopandas
 from .xarray import sits_as_xarray
 
@@ -27,4 +33,6 @@ __all__ = (
     "sits_to_csv",
     "sits_to_xlsx",
     "sits_timeseries_to_csv",
+    "sits_to_parquet",
+    "sits_from_parquet",
 )

{pysits-2.0.0.dev8 → pysits-2.0.0.dev9}/pysits/sits/exporters/files.py

@@ -17,10 +17,15 @@
 
 """Files exporter."""
 
+from pathlib import Path
+
 from pysits.backend.pkgs import r_pkg_sits
 from pysits.conversions.decorators import function_call
 from pysits.docs import attach_doc
 from pysits.models.data.frame import SITSFrame
+from pysits.models.data.ts import SITSTimeSeriesModel
+from pysits.models.resolver import resolve_and_invoke_content_class
+from pysits.network import fetch_to_tempfile, is_remote
 
 
 @function_call(r_pkg_sits.sits_to_csv, SITSFrame)
@@ -41,3 +46,28 @@ def sits_to_xlsx(*args, **kwargs) -> None:
 @attach_doc("sits_timeseries_to_csv")
 def sits_timeseries_to_csv(*args, **kwargs) -> None:
     """Export a full sits timeseries to CSV format."""
+
+
+@function_call(r_pkg_sits.sits_to_parquet, lambda x: None)
+@attach_doc("sits_to_parquet")
+def sits_to_parquet(*args, **kwargs) -> None:
+    """Export sits time series to the Parquet format."""
+
+
+@function_call(r_pkg_sits.sits_from_parquet, resolve_and_invoke_content_class)
+def _sits_from_parquet(*args, **kwargs) -> SITSTimeSeriesModel:
+    """Read sits time series from a local Parquet file."""
+
+
+@attach_doc("sits_from_parquet")
+def sits_from_parquet(file: str | Path) -> SITSTimeSeriesModel:
+    """Read sits time series from the Parquet format."""
+    # If file is remote
+    if is_remote(file):
+        # Download it to a temporary file
+        with fetch_to_tempfile(str(file), suffix=".parquet") as local_file:
+            # Read it
+            return _sits_from_parquet(local_file)
+
+    # Otherwise, read it directly
+    return _sits_from_parquet(file)

{pysits-2.0.0.dev8 → pysits-2.0.0.dev9}/tests/test_exporters.py

@@ -25,8 +25,15 @@ import pytest
 from pysits.backend.pkgs import r_pkg_sits
 from pysits.models.data.frame import SITSFrame
 from pysits.models.data.matrix import SITSConfusionMatrix
+from pysits.models.data.ts import SITSTimeSeriesModel
 from pysits.sits.context import cerrado_2classes
-from pysits.sits.exporters import sits_timeseries_to_csv, sits_to_csv, sits_to_xlsx
+from pysits.sits.exporters import (
+    sits_from_parquet,
+    sits_timeseries_to_csv,
+    sits_to_csv,
+    sits_to_parquet,
+    sits_to_xlsx,
+)
 from pysits.sits.ml import sits_rfor
 from pysits.sits.ts import sits_sample, sits_validate
 
@@ -94,3 +101,46 @@ def test_sits_to_xlsx(tmp_path: Path, accuracy):
     r_pkg_sits.sits_to_xlsx(accuracy._instance, file=str(r_file))
 
     assert read_xlsx(py_file) == read_xlsx(r_file)
+
+
+def test_sits_parquet(tmp_path: Path):
+    """Test time series export / import using parquet."""
+    # local file
+    file = tmp_path / "samples.parquet"
+
+    # export samples as parquet
+    sits_to_parquet(cerrado_2classes, file=file)
+
+    # import samples from parquet
+    samples = sits_from_parquet(file)
+
+    # samples must be a time series model
+    assert isinstance(samples, SITSTimeSeriesModel)
+
+    # metadata and nested series must be equal to the original samples
+    metadata = samples.drop(columns="time_series")
+    expected = cerrado_2classes.drop(columns="time_series")
+
+    assert metadata.equals(expected)
+    assert all(
+        ts.equals(expected_ts)
+        for ts, expected_ts in zip(
+            samples["time_series"], cerrado_2classes["time_series"], strict=True
+        )
+    )
+
+
+def test_sits_from_parquet_remote(http_server):
+    """Test time series import using remote parquet file."""
+    # remote file
+    directory, base_url = http_server
+
+    # export samples as parquet
+    sits_to_parquet(cerrado_2classes, file=directory / "samples.parquet")
+
+    # import samples from parquet
+    samples = sits_from_parquet(f"{base_url}/samples.parquet")
+
+    # samples must be a time series model
+    assert isinstance(samples, SITSTimeSeriesModel)
+    assert len(samples) == len(cerrado_2classes)

{pysits-2.0.0.dev8 → pysits-2.0.0.dev9}/uv.lock

@@ -2328,7 +2328,7 @@ wheels = [
 
 [[package]]
 name = "pysits"
-version = "2.0.0.dev8"
+version = "2.0.0.dev9"
 source = { editable = "." }
 dependencies = [
     { name = "geopandas" },