import hashlib
import re
from collections import Counter
from typing import Any, Dict, Iterable, List


STOPWORDS = {
    "的", "了", "和", "与", "及", "或", "在", "中", "对", "为", "是", "有",
    "进行", "使用", "相关", "根据", "如下", "可以", "需要", "应当", "以及",
}


def enhance_slices(
    slices: Iterable[Dict[str, Any]],
    file_info: Dict[str, Any],
    max_slices: int,
) -> List[Dict[str, Any]]:
    refs: List[Dict[str, Any]] = []
    filename = str(file_info.get("filename") or file_info.get("name") or "")
    kb_id = str(file_info.get("kb_id") or "")
    file_id = str(file_info.get("file_id") or file_info.get("id") or "")

    for idx, raw in enumerate(list(slices)[:max_slices]):
        text = _slice_text(raw)
        if not text.strip():
            continue
        slice_id = str(raw.get("id") or raw.get("slice_id") or raw.get("uuid") or f"{file_id}:{idx}")
        title_path = _infer_title_path(text, filename)
        keywords = extract_keywords(text, limit=12)
        refs.append({
            "kb_id": str(raw.get("kb_id") or kb_id),
            "file_id": str(raw.get("file_id") or file_id),
            "slice_id": slice_id,
            "ordinal": int(raw.get("ordinal") or raw.get("index") or idx),
            "filename": str(raw.get("filename") or filename),
            "title_path": title_path,
            "keywords": keywords,
            "brief": _brief(text),
            "chunk_type": _infer_chunk_type(text),
            "content_hash": hashlib.sha1(text.encode("utf-8", errors="ignore")).hexdigest(),
            "_text": text,
        })
    return refs


def extract_keywords(text: str, limit: int = 10) -> List[str]:
    text = normalize_text(text)
    candidates = re.findall(r"[\u4e00-\u9fff]{2,12}|[A-Za-z0-9][A-Za-z0-9_\-]{1,30}", text)
    normalized: List[str] = []
    for token in candidates:
        token = token.strip("：:，,。.;；、（）()[]【】")
        if len(token) < 2 or token in STOPWORDS:
            continue
        if re.fullmatch(r"\d+", token):
            continue
        normalized.append(token)
    counts = Counter(normalized)
    return [term for term, _ in counts.most_common(limit)]


def normalize_text(text: str) -> str:
    return re.sub(r"\s+", " ", text or "").strip()


def _slice_text(raw: Dict[str, Any]) -> str:
    for key in ("content", "text", "page_content", "slice_content", "markdown"):
        value = raw.get(key)
        if isinstance(value, str) and value.strip():
            return value
    return ""


def _brief(text: str, limit: int = 360) -> str:
    return normalize_text(text)[:limit]


def _infer_title_path(text: str, filename: str) -> List[str]:
    path = []
    if filename:
        path.append(filename)
    lines = [line.strip() for line in (text or "").splitlines() if line.strip()]
    for line in lines[:6]:
        line = re.sub(r"^#+\s*", "", line).strip()
        if 2 <= len(line) <= 50 and not line.endswith(("。", ".", "；", ";")):
            path.append(line)
            break
    return path[:4] if path else []


def _infer_chunk_type(text: str) -> str:
    compact = normalize_text(text)
    if "|" in text and "---" in text:
        return "table"
    if re.search(r"(步骤|流程|操作|检查|确认|打开|关闭|启动|停止)", compact):
        return "procedure"
    if re.search(r"(故障|异常|报警|失效|原因|处理|排查)", compact):
        return "fault"
    if re.search(r"!\[.*?\]\(.*?\)", text):
        return "image"
    return "text"