# -*- coding: utf-8 -*-
"""Extraction automatique des « blocs configurables » d'une page HTML.

extract(body) -> (body_avec_jetons, sections)
- Chaque <section> devient une section de blocs (libellé = étiquette .sec-label ou premier titre).
- Les images, boutons (.btn) et éléments texte « feuilles » (titres, paragraphes, listes, étiquettes…)
  deviennent des champs éditables ; le HTML de structure reste dans la page avec des jetons {{B:sN.fM}}.
- Zones protégées (jamais extraites) : commentaires, <script>, <style>, <svg>, <form>, <table>, <pre>, <iframe>,
  et tout élément contenant déjà un jeton {{…}}.
"""
import re, html

INLINE_ALLOWED = r"strong|em|b|i|br|small|sup|sub|u|code|span|a|mark|abbr|time"
BLOCK_FORBIDDEN = (r"div|p|ul|ol|li|h[1-6]|section|article|header|footer|nav|form|table|thead|tbody|tr|td|th|details|summary|blockquote|svg|img|button|input|select|textarea|label|figure|figcaption|picture|iframe|script|style|dl|dt|dd|hr|video|audio|canvas|main|aside")
TEXT_TAGS = r"h1|h2|h3|h4|h5|h6|p|li|summary|blockquote|dt|dd|figcaption|small|div|span|b|strong|em|a|label"
_INNER = r"(?:(?!<(?:%s)\b)(?!</(?:%s)\b)[\s\S])*?" % (BLOCK_FORBIDDEN, BLOCK_FORBIDDEN)

PROTECT_RE = re.compile(r"<!--[\s\S]*?-->|<script\b[\s\S]*?</script>|<style\b[\s\S]*?</style>|<svg\b[\s\S]*?</svg>|<form\b[\s\S]*?</form>|<table\b[\s\S]*?</table>|<pre\b[\s\S]*?</pre>|<iframe\b[\s\S]*?</iframe>|<textarea\b[\s\S]*?</textarea>", re.I)
SECTION_RE = re.compile(r"<section\b[^>]*>[\s\S]*?</section>", re.I)
ELEM_RE = re.compile(
    r"(?P<img><img\b[^>]*>)"
    r"|(?P<btn><a\b(?P<battrs>[^>]*\bclass=\"[^\"]*\bbtn\b[^\"]*\"[^>]*)>(?P<binner>" + _INNER + r")</a>)"
    r"|(?P<txt><(?P<tag>" + TEXT_TAGS + r")\b(?P<tattrs>[^>]*)>(?P<tinner>" + _INNER + r")</(?P=tag)>)",
    re.I)

LABELS = {"h1": "Titre principal (H1)", "h2": "Titre (H2)", "h3": "Sous-titre (H3)", "h4": "Titre (H4)", "h5": "Titre (H5)", "h6": "Titre (H6)",
          "p": "Paragraphe", "li": "Élément de liste", "summary": "Question / titre dépliable", "blockquote": "Citation", "dt": "Terme", "dd": "Définition",
          "figcaption": "Légende", "small": "Précision", "div": "Texte", "span": "Texte", "b": "Texte en gras", "strong": "Texte en gras", "em": "Texte en italique", "a": "Lien", "label": "Libellé"}
CLASS_LABELS = [("sec-label", "Étiquette de section"), ("kicker", "Sur-titre"), ("when", "Étiquette (quand)"), ("big", "Chiffre"), ("r-big", "Phrase clé"), ("r-sub", "Sous-phrase"),
                ("f-tag", "Nom du forfait"), ("price", "Prix"), ("p-note", "Note sous le prix"), ("result", "Objectif"), ("more", "Texte du lien"), ("kpi", "Résultat clé"),
                ("cap", "Légende"), ("intro", "Introduction"), ("lead", "Accroche"), ("desc", "Description"), ("pick", "Badge"), ("badge-off", "Badge"), ("ib-title", "Titre du bandeau"),
                ("ib-price", "Prix du bandeau"), ("fine", "Mentions"), ("btn-note", "Note sous le bouton"), ("testi-note", "Note"), ("stars", "Étoiles"), ("avatar", "Initiales"),
                ("who", "Qui"), ("hl", "Surligné"), ("tag", "Étiquette"), ("meta", "Méta"), ("date", "Date"), ("author", "Auteur"), ("form-sub", "Sous-titre du formulaire")]

def _label_for(tag, attrs, inner):
    m = re.search(r'class="([^"]*)"', attrs or "")
    classes = m.group(1).split() if m else []
    for c, lab in CLASS_LABELS:
        if c in classes: return lab
    return LABELS.get(tag.lower(), "Texte")

def _clean_text(s):
    return re.sub(r"\s+", " ", html.unescape(re.sub(r"<[^>]+>", "", s))).strip()

def _size_hint(attrs):
    w = re.search(r'\bwidth="(\d+)"', attrs); h = re.search(r'\bheight="(\d+)"', attrs)
    if w and h: return f"{w.group(1)} × {h.group(1)} px · WebP · < 250 Ko"
    return "largeur d'affichage × 2 · WebP · < 250 Ko"

def _extract_chunk(chunk, sec_id, fields, counters):
    """Remplace dans `chunk` les éléments extraits par des jetons ; ajoute les champs à `fields`."""
    # 1) zones protégées : on les met de côté
    keep = []
    def prot(m): keep.append(m.group(0)); return f"\x00P{len(keep)-1}\x00"
    work = PROTECT_RE.sub(prot, chunk)
    def nid(): counters["n"] += 1; return f"f{counters['n']}"
    def label_num(base):
        counters.setdefault(base, 0); counters[base] += 1
        return base if counters[base] == 1 else f"{base} {counters[base]}"
    def repl(m):
        if m.group("img"):
            tag = m.group("img")
            if "{{" in tag or "\x00" in tag: return tag
            src = re.search(r'\bsrc="([^"]*)"', tag); alt = re.search(r'\balt="([^"]*)"', tag)
            if not src: return tag
            k = nid(); fields.append({"key": k, "label": label_num("Image"), "type": "image", "value": src.group(1), "size": _size_hint(tag)})
            tag = tag.replace(src.group(0), f'src="{{{{B:{sec_id}.{k}}}}}"', 1)
            if alt is not None:
                ka = nid(); fields.append({"key": ka, "label": f"{fields[-1]['label']} — texte alternatif", "type": "text", "value": alt.group(1)})
                tag = tag.replace(alt.group(0), f'alt="{{{{B:{sec_id}.{ka}}}}}"', 1)
            return tag
        if m.group("btn"):
            attrs, inner = m.group("battrs"), m.group("binner")
            if "{{B:" in inner or "\x00" in inner or not _clean_text(inner): return m.group(0)
            href = re.search(r'\bhref="([^"]*)"', attrs)
            base = label_num("Bouton")
            k = nid(); fields.append({"key": k, "label": f"{base} — texte", "type": "text", "value": inner.strip()})
            new_attrs = attrs
            if href:
                ku = nid(); fields.append({"key": ku, "label": f"{base} — lien", "type": "url", "value": href.group(1)})
                new_attrs = attrs.replace(href.group(0), f'href="{{{{B:{sec_id}.{ku}}}}}"', 1)
            return f"<a{new_attrs}>{{{{B:{sec_id}.{k}}}}}</a>"
        tag, attrs, inner = m.group("tag"), m.group("tattrs"), m.group("tinner")
        tl = tag.lower()
        # div/span contenant des liens ou des boutons : on descend d'un niveau
        if tl in ("div", "span", "label") and re.search(r"<a\b", inner, re.I):
            return f"<{tag}{attrs}>{ELEM_RE.sub(repl, inner)}</{tag}>"
        if "{{" in inner: return m.group(0)
        # zones protégées (svg, etc.) en début/fin : on les garde dans la structure, on extrait le texte du milieu
        pre = post = ""
        mm = re.match(r"^((?:\s*\x00P\d+\x00\s*)*)([\s\S]*?)((?:\s*\x00P\d+\x00\s*)*)$", inner)
        if mm and mm.group(2) and "\x00" not in mm.group(2): pre, inner, post = mm.group(1), mm.group(2), mm.group(3)
        if "\x00" in inner:
            # conteneur mêlant zones protégées (svg…) et éléments : on descend d'un niveau
            # (ex. <div class="hero-trust"><span><svg/>Dès 119 €…</span>…) au lieu d'abandonner
            if tl in ("div", "span", "label", "li", "p"):
                return f"<{tag}{attrs}>{ELEM_RE.sub(repl, inner)}</{tag}>"
            return m.group(0)
        text = _clean_text(inner)
        if not text: return m.group(0)
        if tl == "a":
            href = re.search(r'\bhref="([^"]*)"', attrs)
            base = label_num("Lien")
            k = nid(); fields.append({"key": k, "label": f"{base} — texte", "type": "text", "value": inner.strip()})
            new_attrs = attrs
            if href:
                ku = nid(); fields.append({"key": ku, "label": f"{base} — adresse", "type": "url", "value": href.group(1)})
                new_attrs = attrs.replace(href.group(0), f'href="{{{{B:{sec_id}.{ku}}}}}"', 1)
            return f"<a{new_attrs}>{pre}{{{{B:{sec_id}.{k}}}}}{post}</a>"
        has_tags = bool(re.search(r"<[a-z]", inner, re.I))
        ftype = "html" if (has_tags or tl in ("p", "li", "blockquote", "dd") or len(text) > 110) else "text"
        base = _label_for(tag, attrs, inner)
        fld = {"key": nid(), "label": label_num(base), "type": ftype, "value": inner.strip()}
        if tl in ("h1", "h2", "h3", "h4", "h5", "h6"): fld["htag"] = tl
        k = fld["key"]; fields.append(fld)
        return f"<{tag}{attrs}>{pre}{{{{B:{sec_id}.{k}}}}}{post}</{tag}>"
    work = ELEM_RE.sub(repl, work)
    # 2) on remet les zones protégées
    return re.sub(r"\x00P(\d+)\x00", lambda m: keep[int(m.group(1))], work)

def _section_label(chunk, n):
    m = re.search(r'class="sec-label"[^>]*>([\s\S]*?)</span>', chunk)
    if m and _clean_text(m.group(1)): return _clean_text(m.group(1))
    m = re.search(r"<h[12]\b[^>]*>([\s\S]*?)</h[12]>", chunk)
    if m and _clean_text(m.group(1)): return _clean_text(m.group(1))[:60]
    m = re.search(r"<!--\s*([^-]{2,60}?)\s*-->", chunk)
    if m: return m.group(1).strip()
    return f"Section {n}"

def extract(body):
    """Retourne (body_avec_jetons, sections)."""
    if "{{B:" in body: raise ValueError("Cette page contient déjà des jetons de blocs.")
    sections = []; out = []; pos = 0; n = 0
    def handle(chunk, label):
        nonlocal n
        n += 1; sec_id = f"s{n}"; fields = []; counters = {"n": 0}
        new = _extract_chunk(chunk, sec_id, fields, counters)
        if fields: sections.append({"id": sec_id, "label": label, "fields": fields}); return new
        n -= 1; return chunk
    for m in SECTION_RE.finditer(body):
        before = body[pos:m.start()]
        if _clean_text(PROTECT_RE.sub("", before)): out.append(handle(before, "Contenu hors section"))
        else: out.append(before)
        chunk = m.group(0)
        out.append(handle(chunk, _section_label(chunk, n + 1)))
        pos = m.end()
    tail = body[pos:]
    if _clean_text(PROTECT_RE.sub("", tail)): out.append(handle(tail, "Contenu" if not sections else "Contenu hors section"))
    else: out.append(tail)
    return "".join(out), sections
