Remove flatdict dependency by replacing fiboa-cli with lightweight parquet utilities

ftw_tools/inference/utils.py

@@ -3,7 +3,8 @@
 import geopandas as gpd
 import numpy as np
 import shapely
-from fiboa_cli.parquet import create_parquet
+
+from ftw_tools.parquet_utils import create_parquet
 
 
 def merge_polygons(

ftw_tools/parquet_utils.py

@@ -0,0 +1,78 @@
+"""Utility functions for writing GeoParquet files with fiboa metadata.
+
+This module provides lightweight replacements for fiboa-cli's create_parquet
+and features_to_dataframe functions to avoid the flatdict dependency.
+"""
+
+import json
+
+import geopandas as gpd
+import pyarrow.parquet as pq
+import shapely.geometry
+
+
+def features_to_dataframe(features, columns):
+    """Convert GeoJSON-like features to a GeoDataFrame.
+
+    Args:
+        features: List of GeoJSON-like feature dictionaries with geometry and properties.
+        columns: List of column names to include in the GeoDataFrame.
+
+    Returns:
+        GeoDataFrame with the features converted to rows, using EPSG:4326 CRS.
+
+    Note:
+        All input features are assumed to be in WGS84 (EPSG:4326) coordinates.
+    """
+    rows = []
+    for feature in features:
+        feature_id = feature.get("id")
+        geometry = (
+            shapely.geometry.shape(feature["geometry"])
+            if "geometry" in feature
+            else None
+        )
+        row = {
+            "id": feature_id,
+            "geometry": geometry,
+        }
+        properties = feature.get("properties", {})
+        row.update(properties)
+        rows.append(row)
+
+    return gpd.GeoDataFrame(rows, columns=columns, geometry="geometry", crs="EPSG:4326")
+
+
+def create_parquet(data, columns, collection, output_file, config=None, compression=None):
+    """Write a GeoDataFrame to a Parquet file with fiboa metadata.
+
+    Args:
+        data: GeoDataFrame to write.
+        columns: List of column names to include in the output.
+        collection: Dictionary with fiboa collection metadata.
+        output_file: Path to the output Parquet file.
+        config: Configuration dictionary (optional, retained for API compatibility).
+        compression: Compression algorithm to use (default: 'zstd').
+
+    Note:
+        The config parameter is retained for API compatibility with the original
+        fiboa-cli interface but is not used in this implementation.
+    """
+    if compression is None:
+        compression = "zstd"
+
+    # Write to Parquet using geopandas built-in method
+    # which handles geometry serialization properly
+    data[columns].to_parquet(
+        output_file,
+        compression=compression,
+        index=False,
+    )
+
+    # Now reopen and add the fiboa metadata
+    # This is needed to add custom metadata to the parquet file
+    parquet_file = pq.read_table(output_file)
+    existing_metadata = parquet_file.schema.metadata or {}
+    existing_metadata[b"fiboa"] = json.dumps(collection).encode("utf-8")
+    parquet_file = parquet_file.replace_schema_metadata(existing_metadata)
+    pq.write_table(parquet_file, output_file, compression=compression)

ftw_tools/postprocess/polygonize.py

@@ -10,13 +10,13 @@
 import rasterio.features
 import shapely.geometry
 from affine import Affine
-from fiboa_cli.parquet import create_parquet, features_to_dataframe
 from pyproj import CRS, Transformer
 from rtree import index
 from shapely.ops import transform, unary_union
 from skimage.morphology import dilation, erosion
 from tqdm import tqdm
 
+from ftw_tools.parquet_utils import create_parquet, features_to_dataframe
 from ftw_tools.settings import SUPPORTED_POLY_FORMATS_TXT
 
 

pyproject.toml

@@ -30,7 +30,6 @@ classifiers = [
 dependencies = [
   "click>=8.2.1,<9",
   "dask[distributed]>=2025.5.1",
-  "fiboa-cli==0.7",
   "fiona>=1.9,<2",
   "geopandas>=0.14,<2",
   "kornia>=0.7,<1",