"""
Prepare external drug markdown files into DeepMed RAG-ready dataset format.

Usage:
  python prepare_rag_drug_data.py --source "e:/AI/Thuốc" --target "e:/AI/DeepMed/backend/data"

This script:
  1) Copies and normalizes files from:
       - <source>/Thuốc nội bộ -> <target>/thông tin thuốc nội bộ
       - <source>/cảnh giác dược -> <target>/cảnh giác dược
  2) Cleans markdown noise (widget/image-only lines, redundant separators)
  3) Ensures each drug markdown begins with:
       # <DRUG_NAME>
       Hoạt chất: <value or Chưa cập nhật>
  4) Optionally removes long hash suffixes in filenames.
"""

from __future__ import annotations

import argparse
import hashlib
import os
import re
import shutil
from dataclasses import dataclass
from pathlib import Path
from typing import Dict, Iterable, List, Optional, Tuple

HASH_SUFFIX_RE = re.compile(r"\s+[0-9a-f]{20,}(?=\.md$)", re.IGNORECASE)
MULTI_SPACE_RE = re.compile(r"[ \t]{2,}")
EMPTY_LINE_RE = re.compile(r"\n{3,}")
HEADING_RE = re.compile(r"^#\s+(.+)$", re.MULTILINE)
ACTIVE_RE = re.compile(r"^\s*Hoạt\s*chất\s*:\s*(.+)$", re.IGNORECASE | re.MULTILINE)
WIDGET_LINK_RE = re.compile(r"^\s*\[https?://widgetbox\.app/.*\]\(https?://widgetbox\.app/.*\)\s*$", re.IGNORECASE)
IMAGE_LINE_RE = re.compile(r"^\s*!\[[^\]]*\]\([^\)]*\)\s*$")
HORIZONTAL_RULE_RE = re.compile(r"^\s*---\s*$")


@dataclass
class SyncStats:
    copied: int = 0
    skipped: int = 0
    errors: int = 0


def _clean_filename(name: str, remove_hash_suffix: bool = True) -> str:
    cleaned = name.strip()
    if remove_hash_suffix and cleaned.lower().endswith(".md"):
        cleaned = HASH_SUFFIX_RE.sub("", cleaned)
    cleaned = MULTI_SPACE_RE.sub(" ", cleaned)
    return cleaned


def _safe_output_path(dest_dir: Path, desired_name: str, content: str) -> Path:
    out = dest_dir / desired_name
    if not out.exists():
        return out

    digest = hashlib.md5(content.encode("utf-8", errors="ignore")).hexdigest()[:8]
    stem = out.stem
    suffix = out.suffix
    return dest_dir / f"{stem}__{digest}{suffix}"


def _normalize_newlines(text: str) -> str:
    return text.replace("\r\n", "\n").replace("\r", "\n")


def _strip_noise_lines(text: str) -> str:
    lines = []
    for raw in text.split("\n"):
        line = raw.replace("\xa0", " ").rstrip()

        if WIDGET_LINK_RE.match(line):
            continue
        if IMAGE_LINE_RE.match(line):
            continue
        if HORIZONTAL_RULE_RE.match(line):
            continue

        lines.append(line)

    normalized = "\n".join(lines)
    normalized = EMPTY_LINE_RE.sub("\n\n", normalized)
    return normalized.strip()


def _extract_heading(content: str) -> Optional[str]:
    m = HEADING_RE.search(content)
    if not m:
        return None
    return m.group(1).strip()


def _extract_active_ingredient(content: str) -> Optional[str]:
    m = ACTIVE_RE.search(content)
    if not m:
        return None
    return m.group(1).strip()


def _ensure_rag_header(content: str, fallback_name: str) -> str:
    heading = _extract_heading(content) or fallback_name
    active = _extract_active_ingredient(content) or "Chưa cập nhật"

    body = content
    body = HEADING_RE.sub("", body, count=1).lstrip("\n")
    body = ACTIVE_RE.sub("", body, count=1).lstrip("\n")

    header = f"# {heading}\n\nHoạt chất: {active}\n"
    out = f"{header}\n{body.strip()}\n"
    out = EMPTY_LINE_RE.sub("\n\n", out)
    return out


def _read_text(path: Path) -> str:
    return path.read_text(encoding="utf-8", errors="ignore")


def _write_text(path: Path, content: str) -> None:
    path.parent.mkdir(parents=True, exist_ok=True)
    path.write_text(content, encoding="utf-8", newline="\n")


def _iter_files(root: Path) -> Iterable[Path]:
    for p in root.rglob("*"):
        if p.is_file() and p.suffix.lower() in {".md", ".pdf", ".docx", ".txt", ".csv", ".xlsx", ".xls"}:
            yield p


def _normalize_markdown_for_drug(md_text: str, fallback_drug_name: str) -> str:
    text = _normalize_newlines(md_text)
    text = _strip_noise_lines(text)
    text = _ensure_rag_header(text, fallback_drug_name)
    return text


def _sync_folder(
    source: Path,
    target: Path,
    remove_hash_suffix: bool,
    drug_mode: bool,
    flatten_tree: bool = False,
) -> SyncStats:
    stats = SyncStats()

    if not source.exists():
        print(f"[WARN] Source folder not found: {source}")
        return stats

    target.mkdir(parents=True, exist_ok=True)

    seen_names: Dict[str, int] = {}

    for src in _iter_files(source):
        try:
            rel = src.relative_to(source)
            clean_name = _clean_filename(src.name, remove_hash_suffix=remove_hash_suffix)

            base_target_dir = target if flatten_tree else (target / rel.parent)
            base_target_dir.mkdir(parents=True, exist_ok=True)

            if src.suffix.lower() == ".md":
                raw = _read_text(src)
                if drug_mode:
                    fallback = Path(clean_name).stem
                    cleaned = _normalize_markdown_for_drug(raw, fallback)
                else:
                    cleaned = _strip_noise_lines(_normalize_newlines(raw)) + "\n"

                out_path = _safe_output_path(base_target_dir, clean_name, cleaned)
                if out_path.exists():
                    existing = _read_text(out_path)
                    if existing == cleaned:
                        stats.skipped += 1
                        continue
                _write_text(out_path, cleaned)
            else:
                out_path = base_target_dir / clean_name
                if out_path.exists() and out_path.stat().st_size == src.stat().st_size:
                    stats.skipped += 1
                    continue
                shutil.copy2(src, out_path)

            seen_names[out_path.name] = seen_names.get(out_path.name, 0) + 1
            stats.copied += 1
        except Exception as exc:
            stats.errors += 1
            print(f"[ERROR] Failed: {src} -> {exc}")

    dup_count = sum(1 for n in seen_names.values() if n > 1)
    if dup_count:
        print(f"[WARN] Found {dup_count} duplicate output names; hash suffix applied.")

    return stats


def run(source_root: Path, target_root: Path, remove_hash_suffix: bool = True) -> Tuple[SyncStats, SyncStats]:
    source_drug = source_root / "Thuốc nội bộ"
    source_alert = source_root / "cảnh giác dược"

    target_drug = target_root / "thông tin thuốc nội bộ"
    target_alert = target_root / "cảnh giác dược"

    print("=" * 72)
    print(f"Source root: {source_root}")
    print(f"Target root: {target_root}")
    print("=" * 72)

    drug_stats = _sync_folder(
        source=source_drug,
        target=target_drug,
        remove_hash_suffix=remove_hash_suffix,
        drug_mode=True,
    )

    alert_stats = _sync_folder(
        source=source_alert,
        target=target_alert,
        remove_hash_suffix=remove_hash_suffix,
        drug_mode=False,
        flatten_tree=True,
    )

    print("=" * 72)
    print("Sync completed")
    print(
        "Drug files: copied={0.copied}, skipped={0.skipped}, errors={0.errors} | "
        "Alert files: copied={1.copied}, skipped={1.skipped}, errors={1.errors}".format(
            drug_stats, alert_stats
        )
    )
    print("=" * 72)

    return drug_stats, alert_stats


def parse_args() -> argparse.Namespace:
    parser = argparse.ArgumentParser(description="Prepare Thuốc files for DeepMed RAG")
    parser.add_argument(
        "--source",
        default=r"e:/AI/Thuốc",
        help="Path to external Thuốc folder",
    )
    parser.add_argument(
        "--target",
        default=r"e:/AI/DeepMed/backend/data",
        help="Path to DeepMed backend data folder",
    )
    parser.add_argument(
        "--keep-hash-suffix",
        action="store_true",
        help="Keep long hash suffixes in markdown filenames",
    )
    return parser.parse_args()


def main() -> None:
    args = parse_args()
    source_root = Path(args.source).resolve()
    target_root = Path(args.target).resolve()

    run(
        source_root=source_root,
        target_root=target_root,
        remove_hash_suffix=not args.keep_hash_suffix,
    )


if __name__ == "__main__":
    main()