MEGAINFRA-WEBAPP/scripts/site-model/extract_munyaka_village.py

"""Village network of Munyaka (C3175-VIL-SDP-001 Rev C, issued for information) as plan linework.

The SDP is a vector PDF at 1:1000 with no coordinates, so it is aligned to the CAD model by
matching its manhole symbols to the issued SR1-SR12 sewer and SW stormwater structures.
Result is plan position only: no levels, sizes or quantities are read from it.
"""
import itertools
import math
import warnings

import numpy as np
from shapely.geometry import LineString

SCALE = 1000*0.0254/72   # metres per PDF point at 1:1000
# Legend colours of the SDP (see its LEGEND block)
ROUTES = {'stormwater': ((0.0, 0.72, 0.0), 1.08), 'sewer': ((1.0, 0.0, 0.0), 1.68), 'water': ((0.0, 0.25, 1.0), 1.08)}
POND = ((0.0, 1.0, 0.5), 0.72)
BOUNDARY = ((0.0, 0.0, 0.0), 0.72)
LAYOUT = ((0.0, 0.0, 0.0), 0.24)


def _open(pdf_path):
    import pymupdf
    warnings.filterwarnings('ignore')
    page = pymupdf.open(str(pdf_path))[0]
    return pymupdf, page, page.rotation_matrix


def _symbols(page, rot, colour, lo, hi):
    out = []
    for it in page.get_drawings():
        r = it['rect']
        if (it['fill'] or it['color']) is None or not lo <= round(r.width, 1) <= hi or r.height >= 8:
            continue
        c = tuple(round(x, 2) for x in (it['fill'] or it['color']))
        if c == colour:
            q = (r.tl + r.br)*0.5*rot
            out.append((q.x*SCALE, -q.y*SCALE))
    return np.array(out)


def _rigid(P, Q):
    pc, qc = P.mean(0), Q.mean(0)
    U, _, Vt = np.linalg.svd((P-pc).T@(Q-qc))
    R = Vt.T@U.T
    if np.linalg.det(R) < 0:
        Vt[-1] *= -1
        R = Vt.T@U.T
    return R, qc-pc@R.T


def align(page, rot, sewer_mh, storm):
    """Rigid transform (no scale, no mirroring) from PDF metres to local model metres."""
    P = _symbols(page, rot, (1.0, 0.0, 0.0), 2.3, 3.4)
    Pw = _symbols(page, rot, (0.0, 0.72, 0.0), 2.8, 3.4)
    Q = np.array(sewer_mh)
    best = (0,)
    dP = np.linalg.norm(P[:, None]-P[None], axis=2)
    for i, j in itertools.combinations(range(len(Q)), 2):
        d = np.linalg.norm(Q[i]-Q[j])
        for a, b in zip(*np.where(np.abs(dP-d) < .6)):
            vp, vq = P[b]-P[a], Q[j]-Q[i]
            th = math.atan2(vq[1], vq[0])-math.atan2(vp[1], vp[0])
            R = np.array([[math.cos(th), -math.sin(th)], [math.sin(th), math.cos(th)]])
            T = Q[i]-P[a]@R.T
            hits = (np.linalg.norm((P@R.T+T)[:, None]-Q[None], axis=2).min(0) < 1.5).sum()
            if hits > best[0]:
                best = (hits, R, T)
    if best[0] < 8:
        raise ValueError(f'Village PDF alignment failed: only {best[0]} sewer manholes matched')
    _, R, T = best
    both = np.vstack([Q, np.array(storm)])
    for _ in range(6):
        pairs = []
        for cand, targets in ((P, Q), (Pw, np.array(storm))):
            X = cand@R.T+T
            for q in targets:
                d = np.linalg.norm(X-q, axis=1)
                if d.min() < 3:
                    pairs.append((cand[d.argmin()], q))
        R, T = _rigid(np.array([a for a, _ in pairs]), np.array([b for _, b in pairs]))
    res = np.linalg.norm(np.array([a for a, _ in pairs])@R.T+T-np.array([b for _, b in pairs]), axis=1)
    if len(pairs) < 15 or np.median(res) > .6:
        raise ValueError(f'Village PDF alignment is weak: {len(pairs)} pairs, median residual {np.median(res):.2f} m')
    return R, T, {'matchedPoints': len(pairs), 'medianResidualM': round(float(np.median(res)), 3), 'maxResidualM': round(float(res.max()), 3),
                  'rotationDeg': round(math.degrees(math.atan2(R[1, 0], R[0, 0])), 2), 'of': len(both)}


def village_model(pdf_path, sewer_mh, storm, simplify=.25):
    pymupdf, page, rot = _open(pdf_path)
    R, T, fit = align(page, rot, sewer_mh, storm)
    tf = lambda q: (np.array([(q*rot).x*SCALE, -(q*rot).y*SCALE])@R.T+T)

    def points(it):
        out = []
        for seg in it['items']:
            if seg[0] == 'l':
                out += [seg[1], seg[2]]
            elif seg[0] == 'c':
                a, b, c, d = (np.array(v) for v in seg[1:5])
                out += [pymupdf.Point(*(((1-t)**3)*a+3*((1-t)**2)*t*b+3*(1-t)*t*t*c+t**3*d)) for t in np.linspace(0, 1, 6)]
        return out

    buckets = {k: [] for k in ('stormwater', 'sewer', 'water', 'pond', 'boundary', 'layout')}
    styles = {**{k: v for k, v in ROUTES.items()}, 'pond': POND, 'boundary': BOUNDARY, 'layout': LAYOUT}
    for it in page.get_drawings():
        if it['type'] != 's' or not it['color'] or (it['rect']*rot).x0 > 2000:   # right-hand strip is the title block/legend
            continue
        key = (tuple(round(c, 2) for c in it['color']), round(it['width'] or 0, 2))
        for name, style in styles.items():
            if key == style:
                pts = points(it)
                if len(pts) >= 2:
                    line = LineString([tf(q) for q in pts])
                    if name == 'layout' and line.length < 10:
                        continue
                    if name == 'pond' and line.length < 25:
                        continue
                    buckets[name].append(line)

    def rows(lines):
        out = []
        for ln in lines:
            s = ln.simplify(simplify) if simplify else ln
            if s.length > .5:
                out.append({'pts': [[round(x, 2), round(y, 2)] for x, y in s.coords], 'length': round(s.length, 2)})
        return out
    routes = [{**r, 'service': svc} for svc in ('stormwater', 'sewer', 'water') for r in rows(buckets[svc])]
    hydrants = []
    for w in page.get_text('words'):
        if w[4] == 'FH':
            q = tf(pymupdf.Point((w[0]+w[2])/2, (w[1]+w[3])/2))
            hydrants.append([round(float(q[0]), 2), round(float(q[1]), 2)])
    return {'source': 'C3175-VIL-SDP-001 Rev C (22 Oct 2026), issued for information. Plan position only, aligned to the CAD model; no levels, sizes or quantities.',
            'alignment': fit, 'routes': routes, 'ponds': rows(buckets['pond']), 'boundary': rows(buckets['boundary']),
            'layout': rows(buckets['layout']), 'hydrants': hydrants}