"""Reproduce the BnF 2025 translation-language analysis.

Python 3.11+, pandas and openpyxl. No participant or private data.
Run: python analysis.py --source bnf-books-2025.xlsx --out results
Download the source from SOURCE_URL; no source workbook is redistributed here.
The output unit is a bibliographic record, not a sale, new work or reader.
"""
import argparse
import hashlib
import json
from pathlib import Path
import pandas as pd

SOURCE_URL = 'https://static.data.gouv.fr/resources/observatoire-du-depot-legal/20261019-084134/observatoire-du-depot-legal-2025-indicateurs-livres.xlsx'

def analyze(source, out):
    df = pd.read_excel(source, sheet_name='Données Observatoire 2025')
    df.columns = df.columns.str.strip()
    assert len(df) == 67498 and df['NNB'].nunique() == len(df)
    assert set(df['Année de parution BNF']) == {25}
    translations = df[df['Code de traduction'].fillna('').str.strip().eq('1 - Traduction (avec ou sans texte original)')].copy()
    translations['original_label'] = translations['Langue originale'].fillna('').str.strip().replace('', '[non renseigné]')
    total = len(translations)
    assert total == 14394
    rows = []
    for label, group in translations.groupby('original_label'):
        rows.append({'original_language_label': label, 'translation_records': len(group),
                     'share_all_translation_records_pct': 100 * len(group) / total,
                     'fiction_records': int(group['Genre de la publication'].eq('Fiction').sum()),
                     'documentary_records': int(group['Genre de la publication'].eq('Documentaire').sum()),
                     'french_publication_records': int(group['Langue de publication'].eq('français').sum())})
    result = pd.DataFrame(rows).sort_values(['translation_records','original_language_label'],ascending=[False,True])
    assert result.translation_records.sum() == total
    genres = []
    for label in ['Fiction','Documentaire']:
        g = translations[translations['Genre de la publication'].eq(label)]
        count = int(g.original_label.isin(['anglais','japonais']).sum())
        genres.append({'genre':label,'translation_records':len(g),'english_japanese_records':count,'english_japanese_share_pct':100*count/len(g)})
    top2 = int(translations.original_label.isin(['anglais','japonais']).sum())
    french = translations[translations['Langue de publication'].eq('français')]
    french_top2 = int(french.original_label.isin(['anglais','japonais']).sum())
    summary = {'source_url':SOURCE_URL,'source_sha256':hashlib.sha256(Path(source).read_bytes()).hexdigest(),
               'all_bibliographic_records':len(df),'translation_records':total,
               'known_original_label_records':int(translations['Langue originale'].notna().sum()),
               'missing_original_label_records':int(translations['Langue originale'].isna().sum()),
               'distinct_nonempty_original_labels':int(translations['Langue originale'].nunique()),
               'english_japanese_records':top2,'english_japanese_share_pct':100*top2/total,
               'french_publication_translation_records':len(french),'french_publication_top2_records':french_top2,
               'french_publication_top2_share_pct':100*french_top2/len(french),
               'genres':genres,
               'genre_gap_percentage_points':genres[0]['english_japanese_share_pct']-genres[1]['english_japanese_share_pct'],
               'scope':'Records published in Bibliographie nationale française – Livres in 2025, translation code 1; all publication languages. Not deposits received in 2025, new titles, all French-language translations, sales or unique works.'}
    out.mkdir(parents=True,exist_ok=True)
    result.to_csv(out/'results.csv',index=False,encoding='utf-8')
    pd.DataFrame(genres).to_csv(out/'genre-comparison.csv',index=False,encoding='utf-8')
    (out/'results-summary.json').write_text(json.dumps(summary,ensure_ascii=False,indent=2)+'\n',encoding='utf-8')
    return summary

if __name__ == '__main__':
    p=argparse.ArgumentParser();p.add_argument('--source',required=True,type=Path);p.add_argument('--out',default='results',type=Path)
    args=p.parse_args(); print(json.dumps(analyze(args.source,args.out),ensure_ascii=False,indent=2))