feat: read stac-geoparquet

Cargo.toml

@@ -7,7 +7,7 @@ members = [
     "stac-async",
     "stac-cli",
     "stac-server",
-    "stac-validate",
+    "stac-validate", "stac-arrow",
 ]
 default-members = [
     "stac",

scripts/create_geoparquet.py

@@ -0,0 +1,29 @@
+import stac_geoparquet
+import json
+from pathlib import Path
+from pyarrow import Table
+from pystac_client import Client
+
+root = Path(__file__).parents[1]
+examples = root / "stac-arrow" / "examples"
+if not examples.is_dir():
+    examples.mkdir(parents=True)
+
+client = Client.open("https://planetarycomputer.microsoft.com/api/stac/v1")
+item_search = client.search(
+    max_items=10,
+    collections=["sentinel-2-l2a"],
+    intersects={
+        "type": "Point",
+        "coordinates": [-105.10, 40.17],
+        "sortby": "-properties.datetime",
+    },
+)
+items = list(item_search.items_as_dicts())
+batches = stac_geoparquet.arrow.parse_stac_items_to_arrow(items)
+table = Table.from_batches(batches)
+for version in ["1.0.0", "1.1.0"]:
+    path = examples / f"sentinel-2-l2a-{version}.parquet"
+    stac_geoparquet.arrow.to_parquet(table, path, schema_version=version)
+with open(examples / "sentinel-2-l2a.json", "w") as f:
+    json.dump(items, f)

scripts/requirements.in

@@ -1 +1,4 @@
+pyarrow
+pystac-client
 stac-api-validator
+git+https://github.com/stac-utils/stac-geoparquet@e13f237

scripts/requirements.txt

@@ -6,17 +6,25 @@ attrs==23.2.0
     #   referencing
 certifi==2024.6.2
     # via
+    #   pyogrio
+    #   pyproj
     #   requests
     #   stac-api-validator
 charset-normalizer==3.3.2
     # via requests
+ciso8601==2.3.1
+    # via stac-geoparquet
 click==8.1.7
     # via
     #   stac-api-validator
     #   stac-check
     #   stac-validator
 deepdiff==6.7.1
     # via stac-api-validator
+deltalake==0.18.1
+    # via stac-geoparquet
+geopandas==1.0.0
+    # via stac-geoparquet
 idna==3.7
     # via requests
 jsonschema==4.22.0
@@ -30,23 +38,55 @@ jsonschema-specifications==2023.12.1
 more-itertools==8.14.0
     # via stac-api-validator
 numpy==2.0.0
-    # via shapely
+    # via
+    #   geopandas
+    #   pandas
+    #   pyarrow
+    #   pyogrio
+    #   shapely
 ordered-set==4.1.0
     # via deepdiff
 orjson==3.10.5
-    # via pystac
+    # via
+    #   pystac
+    #   stac-geoparquet
+packaging==24.1
+    # via
+    #   geopandas
+    #   pyogrio
+    #   stac-geoparquet
+pandas==2.2.2
+    # via
+    #   geopandas
+    #   stac-geoparquet
+pyarrow==16.1.0
+    # via
+    #   deltalake
+    #   stac-geoparquet
+pyarrow-hotfix==0.6
+    # via deltalake
+pyogrio==0.9.0
+    # via geopandas
+pyproj==3.6.1
+    # via
+    #   geopandas
+    #   stac-geoparquet
 pystac==1.10.1
     # via
     #   pystac-client
     #   stac-api-validator
+    #   stac-geoparquet
 pystac-client==0.7.7
     # via stac-api-validator
 python-dateutil==2.9.0.post0
     # via
+    #   pandas
     #   pystac
     #   pystac-client
 python-dotenv==1.0.1
     # via stac-check
+pytz==2024.1
+    # via pandas
 pyyaml==6.0.1
     # via
     #   stac-api-validator
@@ -66,16 +106,23 @@ rpds-py==0.18.1
     #   jsonschema
     #   referencing
 shapely==2.0.4
-    # via stac-api-validator
+    # via
+    #   geopandas
+    #   stac-api-validator
+    #   stac-geoparquet
 six==1.16.0
     # via python-dateutil
 stac-api-validator==0.6.2
     # via -r scripts/requirements.in
 stac-check==1.3.3
     # via stac-api-validator
+stac-geoparquet @ git+https://github.com/stac-utils/stac-geoparquet@e13f237be2aa341ceb21abb341fc47d8fe3efeda
+    # via -r scripts/requirements.in
 stac-validator==3.3.2
     # via
     #   stac-api-validator
     #   stac-check
+tzdata==2024.1
+    # via pandas
 urllib3==2.2.2
     # via requests

stac-arrow/Cargo.toml

@@ -0,0 +1,17 @@
+[package]
+name = "stac-arrow"
+version = "0.1.0"
+edition = "2021"
+
+[features]
+parquet = ["geoarrow/parquet"]
+parquet-compression = ["geoarrow/parquet_compression"]
+
+[dependencies]
+arrow = "52"
+arrow-cast = "52"
+geoarrow = { git = "https://github.com/geoarrow/geoarrow-rs", rev = "6fd07d5555f03b113a5cd36c7bf71c1a7cec8579" }
+geojson = "0.24"
+serde_json = "1"
+stac = { version = "0.7", path = "../stac" }
+thiserror = "1"

stac-arrow/src/error.rs

@@ -0,0 +1,33 @@
+use thiserror::Error;
+
+/// Crate-specific error enum.
+#[derive(Debug, Error)]
+pub enum Error {
+    /// [arrow::error::ArrowError]
+    #[error(transparent)]
+    Arrow(#[from] arrow::error::ArrowError),
+
+    /// The bounding box is not a mapping.
+    #[error("bbox is not a map: {0:?}")]
+    BBoxIsNotAMap(serde_json::Value),
+
+    /// [geoarrow::error::GeoArrowError]
+    #[error(transparent)]
+    GeoArrow(#[from] geoarrow::error::GeoArrowError),
+
+    /// An invalid bbox mapping.
+    #[error("invalid bbox mapping: {0:?}")]
+    InvalidBBoxMap(serde_json::Map<String, serde_json::Value>),
+
+    /// [std::io::Error]
+    #[error(transparent)]
+    Io(#[from] std::io::Error),
+
+    /// [serde_json::Error]
+    #[error(transparent)]
+    SerdeJson(#[from] serde_json::Error),
+
+    /// [stac::Error]
+    #[error(transparent)]
+    Stac(#[from] stac::Error),
+}