src/cad_parser.py

from __future__ import annotations

from collections import Counter
from pathlib import Path
from typing import Iterable

from .geometry import bbox_local, centroid_local, polygon_area_local, polygon_perimeter_local
from .models import BoundaryCandidate, SiteSelection

UNIT_CODES = {
    "0": "unitless",
    "4": "millimeters",
    "6": "meters",
}

BOUNDARY_HINTS = ("site boundary", "site boundry", "boundary", "boundry", "plot", "property")
CONTEXT_HINTS = {
    "road": "roads/access",
    "surround": "surroundings",
    "building": "existing building",
    "built": "built-up",
    "veget": "vegetation",
    "tree": "vegetation",
    "sea": "water/sea",
    "water": "water/sea",
    "contour": "contours",
}


def parse_dxf(path: str | Path) -> dict[str, object]:
    source = Path(path)
    pairs = _read_pairs(source)
    unit = _detect_unit(pairs)
    scale = 0.001 if unit == "millimeters" else 1.0
    layers = _extract_layers(pairs)
    polylines = _extract_lwpolylines(pairs)
    boundary_candidates: list[BoundaryCandidate] = []
    layer_counts: Counter[str] = Counter()
    context_layers: Counter[str] = Counter()

    for layer, closed, raw_points in polylines:
        layer_counts[layer] += 1
        category = _context_category(layer)
        if category:
            context_layers[category] += 1
        points_m = [(x * scale, y * scale) for x, y in raw_points]
        if _is_boundary_layer(layer) and len(points_m) >= 3:
            closed_points = points_m if points_m[0] == points_m[-1] else points_m + [points_m[0]]
            area = polygon_area_local(closed_points)
            perimeter = polygon_perimeter_local(closed_points)
            if area <= 0:
                continue
            boundary_candidates.append(
                BoundaryCandidate(
                    id=f"{source.stem}:{len(boundary_candidates) + 1}",
                    source_file=source.name,
                    layer_name=layer,
                    unit=unit,
                    area_sqm=area,
                    perimeter_m=perimeter,
                    vertex_count=len(points_m),
                    is_closed=closed or points_m[0] == points_m[-1],
                    bbox=bbox_local(points_m),
                    confidence_reason="Layer name matched a boundary/site hint.",
                    points_m=points_m,
                )
            )
    boundary_candidates.sort(key=lambda c: c.area_sqm, reverse=True)
    return {
        "source_file": source.name,
        "unit": unit,
        "layers": sorted(layers),
        "layer_counts": dict(layer_counts),
        "context_layers": dict(context_layers),
        "boundary_candidates": boundary_candidates,
    }


def candidate_to_selection(candidate: BoundaryCandidate, anchor_lat: float, anchor_lon: float) -> SiteSelection:
    return SiteSelection(
        id=f"S-{candidate.id}",
        selection_type="dxf_boundary",
        coordinate_mode="anchored_local_cad",
        geometry_geojson=None,
        local_geometry=candidate.points_m,
        anchor_lat=anchor_lat,
        anchor_lon=anchor_lon,
        radius_m=None,
        area_sqm=candidate.area_sqm,
        perimeter_m=candidate.perimeter_m,
        centroid=centroid_local(candidate.points_m),
        bbox=candidate.bbox,
        unit_source=f"DXF {candidate.unit}",
        accuracy_label="CAD local boundary with public-data anchor",
        source_files=[candidate.source_file],
        selected_boundary_id=candidate.id,
        limitations=[
            "CAD geometry uses local drawing coordinates unless georeferenced.",
            "Public climate, sun, and OSM data are based on the anchor point, not exact CAD georeferencing.",
            "CAD boundary is only as reliable as the uploaded drawing.",
        ],
    )


def _read_pairs(path: Path) -> list[tuple[str, str]]:
    lines = path.read_text(errors="ignore").splitlines()
    return [
        (lines[index].strip(), lines[index + 1].strip())
        for index in range(0, len(lines) - 1, 2)
    ]


def _detect_unit(pairs: list[tuple[str, str]]) -> str:
    for idx, (code, value) in enumerate(pairs):
        if code == "9" and value == "$INSUNITS":
            for c2, v2 in pairs[idx + 1 : idx + 8]:
                if c2 == "70":
                    return UNIT_CODES.get(v2.strip(), f"unit-code-{v2.strip()}")
    return "unknown"


def _extract_layers(pairs: list[tuple[str, str]]) -> set[str]:
    layers = set()
    for idx, (code, value) in enumerate(pairs):
        if code == "0" and value == "LAYER":
            for c2, v2 in pairs[idx : idx + 30]:
                if c2 == "2":
                    layers.add(v2)
                    break
    return layers


def _extract_lwpolylines(pairs: list[tuple[str, str]]) -> Iterable[tuple[str, bool, list[tuple[float, float]]]]:
    index = 0
    while index < len(pairs):
        code, value = pairs[index]
        if code == "0" and value == "LWPOLYLINE":
            layer = "0"
            closed = False
            xs: list[float] = []
            ys: list[float] = []
            index += 1
            while index < len(pairs) and pairs[index][0] != "0":
                c2, v2 = pairs[index]
                if c2 == "8":
                    layer = v2
                elif c2 == "70":
                    try:
                        closed = bool(int(v2.strip()) & 1)
                    except ValueError:
                        closed = False
                elif c2 == "10":
                    try:
                        xs.append(float(v2))
                    except ValueError:
                        pass
                elif c2 == "20":
                    try:
                        ys.append(float(v2))
                    except ValueError:
                        pass
                index += 1
            yield layer, closed, list(zip(xs, ys))
            continue
        index += 1


def _is_boundary_layer(layer_name: str) -> bool:
    lower = layer_name.lower()
    return any(hint in lower for hint in BOUNDARY_HINTS)


def _context_category(layer_name: str) -> str | None:
    lower = layer_name.lower()
    for hint, category in CONTEXT_HINTS.items():
        if hint in lower:
            return category
    return None