01 — Data download (Tier 1, Iberian baseline at HEALPix nside=128)

import json
import shutil
import zipfile
from pathlib import Path

import requests

# Paths — match exactly what the healpix_port scripts hard-code.
ROOT = Path("..").resolve()
DATA_DIR = ROOT / "data"
GBIF_DL_DIR = DATA_DIR / "gbif_dl"
REF_DIR = ROOT / "reference"
GBIF_DL_DIR.mkdir(parents=True, exist_ok=True)
REF_DIR.mkdir(parents=True, exist_ok=True)

GBIF_DL_KEY = "0006204-260423192947929"
GBIF_DL_URL = f"https://api.gbif.org/v1/occurrence/download/request/{GBIF_DL_KEY}.zip"
GBIF_DL_DOI = "10.15468/dl.3frmsq"

GBIF_ZIP_PATH = GBIF_DL_DIR / f"{GBIF_DL_KEY}.zip"
GBIF_TSV_PATH = GBIF_DL_DIR / f"{GBIF_DL_KEY}.csv"

# Conservative size threshold: the zip is ~1.9 MB, so >100 KB is "looks real".
MIN_ZIP_BYTES = 100_000
MIN_TSV_BYTES = 1_000_000


def fetch_gbif_zip() -> Path:
    if GBIF_ZIP_PATH.exists() and GBIF_ZIP_PATH.stat().st_size > MIN_ZIP_BYTES:
        print(f"  cached: {GBIF_ZIP_PATH} ({GBIF_ZIP_PATH.stat().st_size:,} bytes)")
        return GBIF_ZIP_PATH
    print(f"  fetching {GBIF_DL_URL}")
    r = requests.get(GBIF_DL_URL, stream=True, timeout=600, allow_redirects=True)
    r.raise_for_status()
    with open(GBIF_ZIP_PATH, "wb") as f:
        for chunk in r.iter_content(chunk_size=1 << 20):
            f.write(chunk)
    print(f"  saved {GBIF_ZIP_PATH} ({GBIF_ZIP_PATH.stat().st_size:,} bytes)")
    return GBIF_ZIP_PATH


def extract_gbif_tsv() -> Path:
    if GBIF_TSV_PATH.exists() and GBIF_TSV_PATH.stat().st_size > MIN_TSV_BYTES:
        print(f"  cached: {GBIF_TSV_PATH} ({GBIF_TSV_PATH.stat().st_size:,} bytes)")
        return GBIF_TSV_PATH
    with zipfile.ZipFile(GBIF_ZIP_PATH) as zf:
        members = zf.namelist()
        # SIMPLE_CSV downloads contain one occurrence file (.csv extension,
        # but tab-separated). Find it regardless of internal name.
        candidates = [m for m in members if m.endswith(".csv") or m.endswith(".tsv")]
        if not candidates:
            raise RuntimeError(
                f"No CSV/TSV inside {GBIF_ZIP_PATH}; members={members}"
            )
        # Pick the largest (avoids picking a manifest)
        candidates.sort(key=lambda m: zf.getinfo(m).file_size, reverse=True)
        member = candidates[0]
        print(f"  extracting {member} -> {GBIF_TSV_PATH}")
        with zf.open(member) as src, open(GBIF_TSV_PATH, "wb") as dst:
            shutil.copyfileobj(src, dst)
    print(f"  saved {GBIF_TSV_PATH} ({GBIF_TSV_PATH.stat().st_size:,} bytes)")
    return GBIF_TSV_PATH


print("--- 1. GBIF Iberia Bombus download ---")
fetch_gbif_zip()
extract_gbif_tsv()

FIGSHARE_API = "https://api.figshare.com/v2/articles/9956471"
BUMBLEBEE_ZIP = REF_DIR / "Bumblebee_repo.zip"
BUMBLEBEE_DIR = REF_DIR / "Bumblebee_repo_wbombusdat"

# Files the healpix_port scripts expect to exist after extraction.
EXPECTED_FILES = [
    BUMBLEBEE_DIR / "0_data" / "Kerr_et_al2015_spplist.csv",
    BUMBLEBEE_DIR / "0_data" / "bombus_err_obs.csv",
]

# CRU TS climate dir — 04_climate_tei_pei_healpix.py globs `*tmp.dat.nc`,
# `*tmn.dat.nc`, `*tmx.dat.nc`, `*pre.dat.nc` from this directory.
CRU_DIR = BUMBLEBEE_DIR / "0_ClimateData"

# Conservative threshold for the zip: > 100 MB means "actually large".
MIN_BUMBLEBEE_ZIP_BYTES = 100_000_000


def figshare_download_url() -> str:
    print(f"  querying {FIGSHARE_API}")
    r = requests.get(FIGSHARE_API, timeout=120)
    r.raise_for_status()
    meta = r.json()
    files = meta.get("files", [])
    if not files:
        raise RuntimeError(f"No files in figshare article: {meta}")
    target = next(
        (f for f in files if f["name"].lower().startswith("bumblebee")),
        files[0],
    )
    print(f"  resolved {target['name']} ({target['size']:,} bytes)")
    return target["download_url"]


def fetch_bumblebee_zip() -> Path:
    if BUMBLEBEE_ZIP.exists() and BUMBLEBEE_ZIP.stat().st_size > MIN_BUMBLEBEE_ZIP_BYTES:
        print(f"  cached: {BUMBLEBEE_ZIP} ({BUMBLEBEE_ZIP.stat().st_size:,} bytes)")
        return BUMBLEBEE_ZIP
    url = figshare_download_url()
    print(f"  fetching {url} (this is large, ~1.1 GB)")
    r = requests.get(url, stream=True, timeout=3600, allow_redirects=True)
    r.raise_for_status()
    written = 0
    with open(BUMBLEBEE_ZIP, "wb") as f:
        for chunk in r.iter_content(chunk_size=1 << 20):
            f.write(chunk)
            written += len(chunk)
            if written % (50 << 20) < (1 << 20):  # progress every ~50 MB
                print(f"    {written / (1 << 20):,.0f} MB")
    print(f"  saved {BUMBLEBEE_ZIP} ({BUMBLEBEE_ZIP.stat().st_size:,} bytes)")
    return BUMBLEBEE_ZIP


def extract_bumblebee_zip() -> Path:
    if all(p.exists() for p in EXPECTED_FILES) and CRU_DIR.exists():
        print(f"  cached: extracted tree under {BUMBLEBEE_DIR}")
        return BUMBLEBEE_DIR
    print(f"  extracting {BUMBLEBEE_ZIP} -> {REF_DIR}")
    with zipfile.ZipFile(BUMBLEBEE_ZIP) as zf:
        zf.extractall(REF_DIR)
    missing = [p for p in EXPECTED_FILES if not p.exists()]
    if missing:
        raise RuntimeError(
            f"Extraction completed but expected files are missing: {missing}. "
            f"Inspect {BUMBLEBEE_DIR} to discover the correct sub-paths."
        )
    if not CRU_DIR.exists():
        raise RuntimeError(
            f"Expected CRU TS climate directory not found at {CRU_DIR}. "
            f"04_climate_tei_pei_healpix.py needs *.tmp.dat.nc / *.tmn.dat.nc / "
            f"*.tmx.dat.nc / *.pre.dat.nc files in this directory."
        )
    print(f"  validated expected files under {BUMBLEBEE_DIR}")
    return BUMBLEBEE_DIR


print("\n--- 2. Soroye Figshare deposit (Kerr species list + CRU TS) ---")
fetch_bumblebee_zip()
extract_bumblebee_zip()

if CRU_DIR.exists():
    cru_files = sorted(CRU_DIR.glob("*.dat.nc"))
    print(f"  CRU TS NetCDFs in {CRU_DIR.name}: {len(cru_files)}")
    for p in cru_files[:8]:
        print(f"    {p.name}")
    if len(cru_files) > 8:
        print(f"    ... and {len(cru_files) - 8} more")

GBIF_META_PATH = DATA_DIR / "gbif_bombus_iberia_metadata.json"

gbif_meta = {
    "download_key": GBIF_DL_KEY,
    "doi": GBIF_DL_DOI,
    "doi_url": f"https://doi.org/{GBIF_DL_DOI}",
    "citation": (
        "GBIF.org (2026-04-25) GBIF Occurrence Download "
        f"https://doi.org/{GBIF_DL_DOI}"
    ),
    "source_url": GBIF_DL_URL,
    "format": "SIMPLE_CSV (tab-separated)",
    "local_path": str(GBIF_TSV_PATH.relative_to(ROOT)),
    "license": "CC-BY-NC-4.0 (per individual GBIF datasets)",
    "upstream_repo": {
        "name": "annefou/weatherxbiodiversity",
        "version": "v0.2.1",
        "concept_doi": "10.5281/zenodo.19756173",
    },
    "cross_substrate_reference": {
        "name": "annefou/weatherxbiodiversity-projection",
        "commit": "b7cdd47",
        "substrate": "HEALPix nside=64 NESTED (WGS84)",
        "framing": (
            "Internal cross-substrate reference for this nside=128 fit. "
            "The reference repo runs the same pipeline at nside=64 (~92 km cells)."
        ),
    },
}
with open(GBIF_META_PATH, "w") as f:
    json.dump(gbif_meta, f, indent=2)
print(f"\n--- 3. Wrote {GBIF_META_PATH}")
print(f"    citation: {gbif_meta['citation']}")

artefacts = [
    ("GBIF zip",     GBIF_ZIP_PATH),
    ("GBIF TSV",     GBIF_TSV_PATH),
    ("Bumblebee zip", BUMBLEBEE_ZIP),
    ("Kerr species list", EXPECTED_FILES[0]),
    ("err-obs table", EXPECTED_FILES[1]),
    ("CRU TS dir",   CRU_DIR),
    ("citation JSON", GBIF_META_PATH),
]
for name, p in artefacts:
    if p.exists():
        size = p.stat().st_size if p.is_file() else sum(
            f.stat().st_size for f in p.rglob("*") if f.is_file()
        )
        print(f"  ok    {name:<22} {size:>15,} bytes  {p.relative_to(ROOT)}")
    else:
        print(f"  MISS  {name:<22} {'?':>15}        {p.relative_to(ROOT)}")