src/fuse_esdac/x_02_gadm_eu_gen.py

import csv
import json
import os
import pickle
from collections import defaultdict
from typing import Dict, Any

import matplotlib.pyplot as plt
from shapely.ops import unary_union
from shapely.wkb import dumps as wkb_dumps
from shapely.wkb import loads as wkb_loads

UNSD_EU_CSV = "src/fuse_esdac/codebooks/unsd_eu.csv"
REGION_ALIAS_JSON = "src/fuse_esdac/codebooks/unsd_eu_alias.json"
WORLD_TREE_PKL = "src/fuse_esdac/large_inputs/gadm_tree.pkl"
OUT_TREE_PKL = "src/fuse_esdac/outputs/03_photo_gadm/gadm_tree_europe.pkl"


def load_unsd_europe(csv_path: str):
    """
    Load the UNSD-based Europe CSV we built earlier.

    Expected columns (0-based index):
      0: "001"
      1: "World"
      2: "150"
      3: "Europe"
      4: primary region code (039, 151, 154, 155, 153)
      5: primary region name ("Western Europe", etc.)
      6: secondary region name (e.g. "DACH Region")
      7: secondary region ID (e.g. "DACH")
      8: country name
      9: numeric country code
      10: alpha-2 code (e.g. "GB")
      11: alpha-3 code (e.g. "GBR")
      12+: may be empty
    """
    macro_id_map = {
        "Northern Europe": "N_EU",
        "Western Europe": "W_EU",
        "Eastern Europe": "E_EU",
        "Southern Europe": "S_EU",
        "Central Europe": "C_EU",
    }

    countries = {}  # alpha3 -> dict
    macro_to_sub = defaultdict(set)  # (macro_name, macro_id) -> {(sub_name, sub_id), ...}
    sub_to_countries = defaultdict(set)  # (macro_name, macro_id, sub_name, sub_id) -> {alpha3, ...}

    with open(csv_path, "r", encoding="utf-8") as f:
        reader = csv.reader(f, delimiter=";")
        for row in reader:
            if not row:
                continue
            # skip obviously invalid rows
            if len(row) < 12:
                continue
            # very simple heuristic: first field should be "001" for our data; if not, skip
            if row[0] != "001":
                continue

            macro_name = row[5].strip()
            sub_name = row[6].strip()
            sub_id = row[7].strip()
            country_name = row[8].strip()
            alpha3 = row[11].strip()

            if not alpha3:
                continue

            if macro_name not in macro_id_map:
                raise ValueError(f"Unknown macro region name in CSV: {macro_name}")

            macro_id = macro_id_map[macro_name]

            countries[alpha3] = {
                "name": country_name,
                "macro_name": macro_name,
                "macro_id": macro_id,
                "sub_name": sub_name,
                "sub_id": sub_id,
            }

            macro_to_sub[(macro_name, macro_id)].add((sub_name, sub_id))
            sub_to_countries[(macro_name, macro_id, sub_name, sub_id)].add(alpha3)

    return countries, macro_to_sub, sub_to_countries


def build_europe_subtree(tree_world, countries, macro_to_sub, sub_to_countries):
    """
    Build a new tree that only contains Europe, with new levels:

      level 0: Europe (EU)
      level 1: macro regions (N_EU, W_EU, ...)
      level 2: subregions (BENE, DACH, BRIT, ...)
      level 3+: original GADM hierarchy (countries and below) with level offset +3

    We keep original geometry for GADM nodes and aggregate new levels from those.
    """
    # 1) Copy only European countries and their entire subtrees
    europe_country_gids = set(countries.keys())  # these are ISO alpha-3, matching GADM GID_0
    tree_eu: Dict[str, Dict[str, Any]] = {}

    def copy_subtree(root_gid: str):
        stack = [root_gid]
        while stack:
            gid = stack.pop()
            if gid in tree_eu:
                continue
            if gid not in tree_world:
                raise KeyError(f"GID {gid} from UNSD CSV not found in GADM tree")
            node = tree_world[gid]
            tree_eu[gid] = {
                "name": node["name"],
                "level": node["level"],  # temporary, will be shifted later
                "parent": node["parent"],
                "children": set(node["children"]),
                "geometry": node["geometry"],  # WKB bytes
            }
            stack.extend(node["children"])

    for alpha3 in sorted(europe_country_gids):
        copy_subtree(alpha3)

    # 2) Create Europe root
    tree_eu["EU"] = {
        "name": "Europe",
        "level": 0,
        "parent": None,
        "children": set(),
        "geometry": None,
    }

    # 3) Create macro-region nodes (level 1)
    macro_nodes = {}  # macro_id -> gid (=macro_id)
    for (macro_name, macro_id) in sorted({(m, mid) for (m, mid) in macro_to_sub.keys()}):
        gid = macro_id  # use macro_id as GID
        macro_nodes[macro_id] = gid
        tree_eu[gid] = {
            "name": macro_name,
            "level": 1,
            "parent": "EU",
            "children": set(),
            "geometry": None,
        }
        tree_eu["EU"]["children"].add(gid)

    # 4) Create subregion nodes (level 2)
    sub_nodes = {}  # (macro_id, sub_id) -> gid (=sub_id)
    for (macro_name, macro_id), subregions in macro_to_sub.items():
        macro_gid = macro_nodes[macro_id]
        for (sub_name, sub_id) in sorted(subregions):
            gid = sub_id  # use sub_id as GID
            sub_nodes[(macro_id, sub_id)] = gid
            tree_eu[gid] = {
                "name": sub_name,
                "level": 2,
                "parent": macro_gid,
                "children": set(),
                "geometry": None,
            }
            tree_eu[macro_gid]["children"].add(gid)

    # 5) Attach country nodes under subregions; they will be level 3 after shift
    # First, identify which nodes are countries in the copied GADM tree: level == 0
    country_nodes_in_eu = {gid for gid, node in tree_eu.items() if node["level"] == 0 and gid != "EU"}
    # By construction, these should be exactly europe_country_gids
    missing = europe_country_gids - country_nodes_in_eu
    if missing:
        raise RuntimeError(f"Countries present in CSV but not in GADM subtree: {missing}")

    for alpha3 in sorted(europe_country_gids):
        node = tree_eu[alpha3]
        info = countries[alpha3]
        macro_id = info["macro_id"]
        sub_id = info["sub_id"]
        sub_gid = sub_nodes[(macro_id, sub_id)]

        # Re-parent this country from None -> subregion
        node["parent"] = sub_gid
        tree_eu[sub_gid]["children"].add(alpha3)

    # 6) Shift levels of all original GADM nodes by +3
    reserved_gids = {"EU"} | set(macro_nodes.values()) | set(sub_nodes.values())
    for gid, node in tree_eu.items():
        if gid in reserved_gids:
            continue
        # This includes all GADM-derived nodes: countries and below
        node["level"] = node["level"] + 3

    # 7) Compute geometry for subregions, macro-regions, and Europe
    # All existing GADM nodes already have WKB geometry

    # Subregions: union of country geometries
    for (macro_id, sub_id), sub_gid in sub_nodes.items():
        country_alpha3s = sub_to_countries[
            (countries[next(iter(sub_to_countries[(macro_id, sub_id)].__iter__()))]["macro_name"],  # ugly, ignore
             macro_id,
             tree_eu[sub_gid]["name"],
             sub_id)] if False else None  # placeholder to avoid confusion
        # simpler: recompute country set for this (macro_id, sub_id)
        # to avoid the above monstrosity:
        country_alpha3s = {
            a3
            for (m_name, m_id, s_name, s_id), a3s in sub_to_countries.items()
            if m_id == macro_id and s_id == sub_id
            for a3 in a3s
        }

        geoms = []
        for alpha3 in country_alpha3s:
            geom_wkb = tree_eu[alpha3]["geometry"]
            geoms.append(wkb_loads(geom_wkb))
        if not geoms:
            raise RuntimeError(f"No country geometries for subregion {sub_gid}")
        merged = unary_union(geoms)
        tree_eu[sub_gid]["geometry"] = wkb_dumps(merged, hex=False)

    # Macro-regions: union of subregion geometries
    for macro_id, macro_gid in macro_nodes.items():
        geoms = []
        for sub_gid in tree_eu[macro_gid]["children"]:
            geom_wkb = tree_eu[sub_gid]["geometry"]
            if geom_wkb is None:
                raise RuntimeError(f"Missing geometry for subregion {sub_gid}")
            geoms.append(wkb_loads(geom_wkb))
        merged = unary_union(geoms)
        tree_eu[macro_gid]["geometry"] = wkb_dumps(merged, hex=False)

    # Europe: union of macro-region geometries
    geoms = []
    for macro_gid in tree_eu["EU"]["children"]:
        geom_wkb = tree_eu[macro_gid]["geometry"]
        if geom_wkb is None:
            raise RuntimeError(f"Missing geometry for macro region {macro_gid}")
        geoms.append(wkb_loads(geom_wkb))
    merged = unary_union(geoms)
    tree_eu["EU"]["geometry"] = wkb_dumps(merged, hex=False)

    # 8) Final sanity: all nodes must have geometry
    for gid, node in tree_eu.items():
        if node["geometry"] is None:
            raise RuntimeError(f"No geometry for node {gid} ({node['name']})")

    # 9) Attach aliases to nodes (EU / macro / sub)
    if not os.path.exists(REGION_ALIAS_JSON):
        raise FileNotFoundError(f"Region alias file not found: {REGION_ALIAS_JSON}")

    with open(REGION_ALIAS_JSON, "r", encoding="utf-8") as f:
        alias_map = json.load(f)

    # First: enforce that all level <= 2 nodes have aliases defined
    for gid, node in tree_eu.items():
        if node["level"] <= 2:
            if gid not in alias_map:
                raise KeyError(
                    f"Missing aliases for region GID '{gid}' "
                    f"(level={node['level']}, name='{node['name']}')"
                )

    # Second: attach aliases to nodes
    for gid, aliases in alias_map.items():
        if gid not in tree_eu:
            raise KeyError(f"Alias GID '{gid}' not found in Europe tree")

        # normalize: store as sorted unique list
        tree_eu[gid]["aliases"] = sorted(set(aliases))

    # Third: ensure lower-level nodes have an aliases field (empty list)
    for gid, node in tree_eu.items():
        if "aliases" not in node:
            node["aliases"] = []

    return tree_eu


def main():
    # Load world tree
    with open(WORLD_TREE_PKL, "rb") as f:
        tree_world = pickle.load(f)

    # Load UNSD Europe table
    countries, macro_to_sub, sub_to_countries = load_unsd_europe(UNSD_EU_CSV)

    # Build Europe tree
    tree_eu = build_europe_subtree(tree_world, countries, macro_to_sub, sub_to_countries)

    # Save to disk
    os.makedirs(os.path.dirname(OUT_TREE_PKL), exist_ok=True)
    with open(OUT_TREE_PKL, "wb") as f:
        pickle.dump(tree_eu, f)

    print(f"✅ Europe tree built and saved to: {OUT_TREE_PKL}")
    print(f"Nodes: {len(tree_eu)}")
    # Small debug output
    print("Root:", "EU", tree_eu["EU"]["name"])
    print("Level-1 children:", tree_eu["EU"]["children"])


plot_times = 0


def check():
    IN_FILE = "src/fuse_esdac/outputs/03_photo_gadm/gadm_tree_europe.pkl"

    with open(IN_FILE, "rb") as f:
        tree = pickle.load(f)

    # 1) print first 4 levels as tree: EU -> macro -> subregion -> country
    def tree_print(gid, max_level=3, prefix=""):
        node = tree[gid]
        name = node["name"] or "?"
        print(f"{prefix}{gid} ({name})")
        if node["level"] >= max_level:
            return
        # sort children by gid
        for child in sorted(node["children"]):
            cnode = tree[child]
            if cnode["level"] <= max_level:
                tree_print(child, max_level, prefix + "├─")

    print("=== Europe Hierarchy (levels 0-3) ===")
    tree_print("EU", max_level=3)

    # 2) plot level<=2 boundaries
    # Level→Color mapping (explicit because user asked)
    LEVEL_COLORS = {
        0: "black",
        1: "red",
        2: "green",
        3: "blue",
        4: "magenta",
        5: "cyan",
        6: "orange"
    }

    def get_color(time: int):
        return LEVEL_COLORS[time % len(LEVEL_COLORS)]

    def plot_boundary_by_level(geom):
        global plot_times
        color = get_color(plot_times)
        plot_times += 1

        # Only outer rings
        if geom.geom_type == "Polygon":
            x, y = geom.exterior.xy
            plt.plot(x, y, color=color, lw=0.5)
        elif geom.geom_type == "MultiPolygon":
            for poly in geom.geoms:
                x, y = poly.exterior.xy
                plt.plot(x, y, color=color, lw=0.5)

    plt.figure()

    # Draw all nodes at level <= 2
    for gid, node in tree.items():
        level = node["level"]
        if level == 1:  # Europe, macro, subregions
            print(f"{gid} ({node['name']})")
            geom = wkb_loads(node["geometry"])
            plot_boundary_by_level(geom)

    plt.title("Europe boundaries (level-based coloring)")
    plt.gca().set_aspect("equal", adjustable="datalim")
    plt.show()


if __name__ == "__main__":
    main()
    check()