#!/usr/bin/env python3
"""Explore the concrete-lining structure drawings (04-004 .. 04-009).

Read-only on the source RAG folder. For each DXF prints layers with entity
counts, INSERT block names, dimension measurements and every text string with
its position, and writes a JSON dump for later use.
Usage: explore_lining.py <RAG_04_DIR> <OUT_JSON> [max_texts_per_file]
"""
import collections
import glob
import json
import os
import sys

import ezdxf
from ezdxf import recover

RAG_04 = sys.argv[1]
OUT_JSON = sys.argv[2]
MAX_TEXT = int(sys.argv[3]) if len(sys.argv) > 3 else 400


def load(path):
    try:
        return ezdxf.readfile(path), None
    except Exception as exc:
        try:
            doc, _ = recover.readfile(path)
            return doc, f"recovered after strict error: {type(exc).__name__}"
        except Exception as exc2:
            return None, f"{type(exc).__name__} | recover: {type(exc2).__name__}: {exc2}"


def explore(path):
    rec = {"file": os.path.basename(path)}
    doc, note = load(path)
    if doc is None:
        rec["error"] = note
        return rec
    rec["note"] = note
    msp = doc.modelspace()
    layers = collections.Counter()
    counts = collections.Counter()
    for e in msp:
        counts[e.dxftype()] += 1
        layers[e.dxf.layer] += 1
    rec["layers"] = dict(layers.most_common())
    rec["entity_counts"] = dict(counts.most_common())
    rec["inserts"] = dict(collections.Counter(e.dxf.name for e in msp if e.dxftype() == "INSERT").most_common(30))
    dims, texts = [], []
    for e in msp:
        t = e.dxftype()
        if t == "DIMENSION":
            try:
                val = round(float(e.get_measurement()), 3)
            except Exception:
                val = None
            dims.append({"layer": e.dxf.layer, "value": val, "text": e.dxf.get("text", "")})
        elif t in ("TEXT", "MTEXT"):
            s = e.dxf.text if t == "TEXT" else e.plain_text()
            if s and s.strip():
                pos = e.dxf.insert
                texts.append({"layer": e.dxf.layer, "x": round(pos.x, 2), "y": round(pos.y, 2),
                              "text": s.strip().replace("\n", " | ")[:200]})
    rec["dimensions"] = dims[:200]
    rec["texts_total"] = len(texts)
    rec["texts"] = texts[:MAX_TEXT]
    return rec


def main():
    files = sorted(glob.glob(os.path.join(RAG_04, "C0060304-00[4-9]*.dxf")))
    out = [explore(f) for f in files]
    with open(OUT_JSON, "w", encoding="utf-8") as fh:
        json.dump(out, fh, ensure_ascii=False, indent=1)
    for r in out:
        print("==", r["file"])
        if "error" in r:
            print("  ERROR", r["error"])
            continue
        print("  note:", r["note"])
        print("  entities:", r["entity_counts"])
        print("  layers:", dict(list(r["layers"].items())[:25]))
        print("  inserts:", r["inserts"])
        print("  dims:", len(r["dimensions"]), "texts:", r["texts_total"])


if __name__ == "__main__":
    main()