"""Normalize cumulative online filters without rewriting frozen report assets."""
import json
import re

from .cluster_targets import digest


def online_page(page,root,proposal_keys):
    if '待独立核验' in page:
        verification=root/'verification.json'
        if verification.exists():
            seal=json.loads(verification.read_text())
            if seal.get('state')=='passed' and seal.get('proposals_sha256')==digest(root/'proposals.jsonl'):
                page=page.replace('（待独立核验）','（已独立核验）')
                page=page.replace('（待独立核验，尚未发布）','（已独立核验）')
                if (seal.get('publication_version',0)>=14 and (root/'manifest.json').is_file() and
                        seal.get('manifest_sha256')==digest(root/'manifest.json')):
                    page=re.sub(r'(<h1>[^<>]*)草稿\s*·\s*待独立核验，尚未发布(</h1>)',
                        r'\1 · 已独立核验\2',page)
                if seal.get('publication_version',0)>=10:
                    # Keep wide class tables scrollable inside their own panel.
                    # Clip only the document's small outer overflow; frozen HTML
                    # bytes and original asset hashes remain untouched.
                    page=page.replace('</style>','html{overflow-x:clip}body{overflow-x:clip}</style>',1)
    if 'id="only-supplement"' not in page:return page
    manifest=json.loads((root/'manifest.json').read_text());new_file=root/'new_proposals.jsonl'
    if digest(new_file)!=manifest['new_proposals_sha256']:raise ValueError('Online filter inventory changed')
    new={json.loads(line)['target_key'] for line in new_file.read_text().splitlines()}
    if not new<=proposal_keys:raise ValueError('Unknown online new-batch target')
    seen=set()
    def card(match):
        text=match[0];key_match=re.search(r'id="target-([0-9a-f]{24})"',text)
        if not key_match:raise ValueError('Cumulative card has no target identity')
        key=key_match[1];seen.add(key)
        text=re.sub(r'data-batch="[^"]*"','data-batch="'+('new' if key in new else 'previous')+'"',text)
        if key in new:
            cls=re.search(r'data-class="([0-9]+)"',text)
            if cls:text=re.sub(r'<h2>.*?</h2>',f'<h2>类别 {cls[1]} · 【本轮新增机器候选】待人工复核</h2>',text,count=1,flags=re.S)
        return text
    result=re.sub(r'<article\b[^>]*>.*?</article>',card,page,flags=re.S)
    if seen!=proposal_keys:raise ValueError('Online card inventory differs')
    result=result.replace('only-supplement','only-new').replace("a.dataset.batch!=='supplement'","a.dataset.batch!=='new'")
    result=result.replace('只看本轮复用六份','只看本轮新增候选')
    result=result.replace('第六版 · 既有人工轮廓补充计数（待独立核验）','第六版 · 人工轮廓和机器候选（已独立核验）')
    return result
