dd-l2-bangkok-multicrs-audit

Rerun

import csv
import fiona

path = "/work/bangkok_contractor_delivery.gpkg"
layers = fiona.listlayers(path)

rows = []

for layer_name in layers:
    with fiona.open(path, layer=layer_name) as src:
        crs = src.crs
        schema = src.schema
        geom_type = schema["geometry"]
        feature_count = len(src)

        # Build the EPSG string from the CRS
        crs_str = str(crs)
        if crs_str.upper().startswith("EPSG"):
            declared_crs = crs_str.upper()
        elif crs_str.startswith("+"):  # proj string
            # Try to extract EPSG from fiona CRS object
            epsg = crs.to_epsg()
            if epsg:
                declared_crs = f"EPSG:{epsg}"
            else:
                declared_crs = crs_str
        else:
            declared_crs = crs_str

        # Get a sample coordinate
        sample_x = None
        sample_y = None
        for feat in src:
            geom = feat["geometry"]
            if geom_type in ("Point", "MultiPoint"):
                if geom_type == "Point":
                    sample_x, sample_y = geom["coordinates"]
                else:
                    sample_x, sample_y = geom["coordinates"][0]
            elif geom_type in ("LineString", "MultiLineString"):
                if geom_type == "LineString":
                    sample_x, sample_y = geom["coordinates"][0]
                else:
                    sample_x, sample_y = geom["coordinates"][0][0]
            elif geom_type in ("Polygon", "MultiPolygon"):
                if geom_type == "Polygon":
                    sample_x, sample_y = geom["coordinates"][0][0]
                else:
                    sample_x, sample_y = geom["coordinates"][0][0][0]
            else:
                sample_x, sample_y = 0, 0
            break

        # Detect encoding by inspecting text attributes
        # Reset iterator
        src = fiona.open(path, layer=layer_name)
        encoding_detected = "utf-8"
        for feat in src:
            props = feat["properties"]
            for key, val in props.items():
                if val is None:
                    continue
                if not isinstance(val, str):
                    continue
                # Check for classic latin1-mojibake patterns:
                # Byte sequences like à¹, à¸, etc. that occur when
                # UTF-8 bytes are decoded as Latin-1 then re-encoded as UTF-8
                # The presence of these high-latin-1-range chars suggests mojibake
                for ch in val:
                    cp = ord(ch)
                    if 0x80 <= cp <= 0x9F or 0xA0 <= cp <= 0xBF:
                        # Characters in Latin-1 Supplement control/printable range
                        # that commonly appear in mojibake of Thai text
                        encoding_detected = "latin1-mojibake"
                        break
                if encoding_detected == "latin1-mojibake":
                    break
            if encoding_detected == "latin1-mojibake":
                break

        rows.append({
            "layer_name": layer_name,
            "declared_crs": declared_crs,
            "geometry_type": geom_type,
            "feature_count": feature_count,
            "sample_x": sample_x,
            "sample_y": sample_y,
            "encoding_detected": encoding_detected,
        })

with open("/work/crs_audit.csv", "w", newline="") as f:
    w = csv.DictWriter(f, fieldnames=[
        "layer_name", "declared_crs", "geometry_type",
        "feature_count", "sample_x", "sample_y", "encoding_detected"
    ])
    w.writeheader()
    w.writerows(rows)

print("Done. Rows written:")
for r in rows:
    print(r)