#!/usr/bin/env python
"""Reproducible retrieval of the Martins, Schaerer & Hillier (2005, A&A 436, 1049)
O-star parameter calibration.

NOTE ON PROVENANCE: these are model-calibration tables printed only in the paper
body; they were never ingested into VizieR (bibcode 2005A&A...436.1049M returns
"No catalogue or table was specified or found", and Vizier.find_catalogs turns up
nothing). astroquery.vizier therefore cannot retrieve them. Instead we pull the
immutable arXiv e-print source (astro-ph/0503346) and parse the six deluxetable
blocks by their \\label. arXiv e-prints are version-pinned, so this is reproducible.

Tables: tab_V / tab_III / tab_I  (theoretical Teff scale)
        tab_V_obs / tab_III_obs / tab_I_obs  (observational Teff scale)
Row columns: ST & Teff & logg & Mv & BC & logL & R & Mspec & q0 & q1 & Q0 & Q1
Writes martins2005.csv with (scale, lumclass, SpType, Teff, logg, Mv, BC,
logL_Lsun, R_Rsun, Mspec_Msun, logq0, logq1, logQ0_s, logQ1_s).
"""
import io
import os
import re
import csv
import tarfile
import urllib.request

ARXIV_ID = 'astro-ph/0503346'
SRC_URL = f'https://arxiv.org/e-print/{ARXIV_ID}'
CACHE = 'martins2005_src.tar.gz'

LABELS = {  # label -> (scale, lumclass)
    'tab_V': ('theoretical', 'V'), 'tab_III': ('theoretical', 'III'),
    'tab_I': ('theoretical', 'I'), 'tab_V_obs': ('observational', 'V'),
    'tab_III_obs': ('observational', 'III'), 'tab_I_obs': ('observational', 'I')}
COLS = ['Teff', 'logg', 'Mv', 'BC', 'logL_Lsun', 'R_Rsun', 'Mspec_Msun',
        'logq0', 'logq1', 'logQ0_s', 'logQ1_s']


def get_source_text():
    if not os.path.exists(CACHE):
        req = urllib.request.Request(SRC_URL, headers={'User-Agent': 'Mozilla/5.0'})
        with urllib.request.urlopen(req, timeout=60) as r:
            data = r.read()
        with open(CACHE, 'wb') as fh:
            fh.write(data)
    with tarfile.open(CACHE, 'r:gz') as tf:
        texname = [m for m in tf.getnames() if m.endswith('.tex')][0]
        return tf.extractfile(texname).read().decode('latin-1')


def parse_tables(tex):
    rows = []
    for label, (scale, lc) in LABELS.items():
        # the table block: from \label{label} to the next \end{table*}
        m = re.search(r'\\label\{' + re.escape(label) + r'\}(.*?)\\end\{table\*\}',
                      tex, re.S)
        if not m:
            raise ValueError(f'table {label} not found in source')
        for line in m.group(1).splitlines():
            if '&' not in line or '\\hline' in line:
                continue
            cells = [c.strip().rstrip('\\').strip() for c in line.split('&')]
            cells = [c for c in cells if c != '']
            if len(cells) < 12:
                continue
            st = cells[0]
            if not re.match(r'^\d', st):
                continue
            vals = []
            for c in cells[1:12]:
                c = c.replace('x', '')          # one stray '3.19x' typo in tab_I_obs
                vals.append(float(c))
            rows.append(dict(scale=scale, lumclass=lc, SpType=f'O{st}',
                             **dict(zip(COLS, vals))))
    return rows


def main():
    rows = parse_tables(get_source_text())
    with open('martins2005.csv', 'w', newline='') as fh:
        w = csv.DictWriter(fh, fieldnames=['scale', 'lumclass', 'SpType'] + COLS)
        w.writeheader()
        w.writerows(rows)
    print(f'wrote martins2005.csv ({len(rows)} rows, '
          f'{len(set((r["scale"], r["lumclass"]) for r in rows))} tables)')
    return rows


if __name__ == '__main__':
    main()