#!/usr/bin/env python3
"""Vérifications du manuscrit : stubs restants, figures manquantes, scripts des figures,
citations/références non définies (log), \\synth sans encadré, équations hors chapitre, CSV de vérification.
Usage : tools/check_manuscript.py [--log build/main.log] [--no-log] [--strict]"""
import re, sys, csv, pathlib, argparse
ROOT = pathlib.Path(__file__).resolve().parents[1]
ap = argparse.ArgumentParser(); ap.add_argument("--log"); ap.add_argument("--no-log", action="store_true"); ap.add_argument("--strict", action="store_true")
a = ap.parse_args()
problems, notes = [], []
chapters = sorted((ROOT/"manuscript/chapters").glob("ch*.tex"))
apps = sorted((ROOT/"manuscript/appendices").glob("*.tex"))
texts = {p: p.read_text(encoding="utf-8") for p in chapters + apps}
# 1. stubs
stubs = [p.name for p, t in texts.items() if "\\stubchapter" in t]
notes.append(f"{len(chapters)-sum(1 for s in stubs if s.startswith('ch'))}/{len(chapters)} chapitres rédigés ; stubs restants : {', '.join(stubs) or 'aucun'}")
if a.strict and stubs: problems.append(f"stubs restants : {stubs}")
# 2. figures incluses existent + inscrites dans verification/figures.csv
figreg = {}
fcsv = ROOT/"verification/figures.csv"
if fcsv.exists():
    for row in csv.DictReader(fcsv.open(encoding="utf-8")): figreg[row["fichier"]] = row
incl = re.compile(r"\\includegraphics(?:\[[^\]]*\])?\{([^}]+)\}")
for p, t in texts.items():
    for f in incl.findall(t):
        cands = [ROOT/"assets/figures"/f, ROOT/"assets/figures"/(f+".pdf"), ROOT/"assets/figures"/(f+".png")]
        if not any(c.exists() for c in cands): problems.append(f"{p.name}: figure absente {f}")
        key = f if f.endswith((".pdf",".png")) else f+".pdf"
        if key not in figreg and not f.startswith("logo"): problems.append(f"{p.name}: figure {f} non inscrite dans verification/figures.csv")
for key, row in figreg.items():
    s = ROOT/row["script"]
    if row["script"] and not s.exists(): problems.append(f"figures.csv: script introuvable {row['script']} ({key})")
    if "schema" not in (row.get("type_donnees") or "").lower() and "tikz" not in key.lower() and not (ROOT/"assets/figures"/key).exists(): problems.append(f"figures.csv: fichier inscrit mais absent {key}")
# 3. compteurs
n_fig = sum(len(incl.findall(t)) + t.count("\\begin{tikzpicture}") for t in texts.values())
n_tab = sum(t.count("\\begin{table}") + t.count("\\begin{longtable}") for t in texts.values())
n_eq = sum(t.count("\\begin{equation}") + t.count("\\begin{align}") + sum(ln.count("\\\\") for ln in re.findall(r"\\begin\{align\}(.*?)\\end\{align\}", t, re.S)) for t in texts.values())
n_ex = sum(t.count("\\begin{exercice}") for t in texts.values())
n_words = sum(len(re.sub(r"\\[a-zA-Z]+|[{}%$&_^~]", " ", t).split()) for p, t in texts.items() if p.name.startswith("ch"))
notes.append(f"figures≈{n_fig} tableaux≈{n_tab} équations≈{n_eq} exercices={n_ex} mots(chapitres)≈{n_words}")
# 4. log LaTeX
if a.log and not a.no_log:
    lp = ROOT/a.log
    if not lp.exists(): problems.append(f"log introuvable : {a.log}")
    else:
        L = lp.read_text(errors="ignore")
        for m in sorted(set(re.findall(r"LaTeX Warning: (Reference|Citation) `([^']+)' .*undefined", L))): problems.append(f"{m[0]} non définie : {m[1]}")
        if "There were undefined references" in L: problems.append("références non définies (voir log)")
        if re.search(r"^!", L, re.M): problems.append("erreur LaTeX fatale dans le log")
        m = re.findall(r"Output written on .*? \((\d+) pages", L); notes.append(f"pages PDF : {m[-1] if m else '?'}")
        notes.append(f"overfull hbox : {L.count('Overfull \\\\hbox')}")
# 5. CSV de vérification bien formés
for name, cols in {"verification/claims.csv": ["id"], "verification/calculations.csv": ["id"], "verification/figures.csv": ["fichier"], "sources/registry.csv": ["id"]}.items():
    f = ROOT/name
    if not f.exists(): problems.append(f"{name} manquant"); continue
    r = csv.DictReader(f.open(encoding="utf-8"))
    if not r.fieldnames or any(c not in r.fieldnames for c in cols): problems.append(f"{name}: en-tête invalide")
    else: notes.append(f"{name}: {sum(1 for _ in r)} lignes")
print("\n".join("• " + n for n in notes))
if problems:
    print(f"\n{len(problems)} problème(s) :"); print("\n".join("  ✗ " + p for p in problems[:80]))
    sys.exit(1 if (a.strict or a.log) else 0)
print("\n✓ aucun problème bloquant")