docs/convert_md.py

"""
Convert file Markdown (.md) sang Word (.docx) / HTML / PDF bang pandoc.

CACH DUNG: chi can BAM CHAY (Run trong VSCode/PyCharm, hoac double-click).
Khong can go lenh / khong can nhap args. Chinh cau hinh o khoi CONFIG ben duoi.

CAI 1 LAN (neu chua co): python -m pip install pypandoc_binary
"""

# =====================================================================
#  CONFIG  --  SUA TRUC TIEP O DAY ROI BAM CHAY
# =====================================================================

# File can convert. Co the la:
#   - 1 ten file:      "methodology_rewrite.md"
#   - nhieu file:      ["methodology_rewrite.md", "bai2.md"]
#   - tat ca file .md: "ALL"
INPUT = "methodology_rewrite.md"

# Dinh dang dau ra: "docx" | "html" | "pdf"
OUTPUT_FORMAT = "docx"

# Ten file dau ra. De None thi tu dat trung ten file goc (chi dung khi INPUT la 1 file).
OUTPUT_NAME = None

# GOP NHIEU FILE THANH 1 BAO CAO:
#   - De COMBINE = [] (rong)  -> dung che do INPUT o tren (convert tung file rieng).
#   - Liet ke file theo dung thu tu -> gop thanh 1 file COMBINE_OUTPUT duy nhat.
COMBINE = ["report_front.md", "report_part1.md", "methodology_rewrite.md", "report_part4_5.md"]
COMBINE_OUTPUT = "report.docx"

# =====================================================================
#  (Khong can sua tu day tro xuong)
# =====================================================================

import sys
from pathlib import Path

HERE = Path(__file__).resolve().parent


def ensure_pypandoc():
    try:
        import pypandoc  # noqa: F401
        return
    except ImportError:
        print("[!] Chua co pypandoc. Dang cai 'pypandoc_binary' (kem san pandoc)...")
        import subprocess
        subprocess.check_call(
            [sys.executable, "-m", "pip", "install", "--quiet", "pypandoc_binary"]
        )
        print("[+] Cai xong.")


def convert_one(md_path: Path, to_fmt: str, out_path: Path | None) -> Path:
    import pypandoc

    if not md_path.exists():
        raise FileNotFoundError(f"Khong tim thay file: {md_path}")

    if out_path is None:
        out_path = md_path.with_suffix("." + to_fmt)

    # --resource-path: de pandoc tim duoc anh (figures/...) du chay tu thu muc nao
    extra_args = ["--toc", "--toc-depth=3", "--resource-path", str(md_path.parent)]
    if to_fmt == "html":
        extra_args += ["--standalone", "--mathjax", "--embed-resources"]
    elif to_fmt == "pdf":
        extra_args += ["-V", "geometry:margin=2.5cm"]

    pypandoc.convert_file(
        str(md_path), to_fmt, outputfile=str(out_path), extra_args=extra_args
    )
    return out_path


def combine_files(names, out_name) -> Path:
    """Gop nhieu file .md (theo thu tu) thanh 1 file dau ra duy nhat."""
    import pypandoc

    parts = []
    for n in names:
        p = (HERE / n) if not Path(n).is_absolute() else Path(n)
        if not p.exists():
            raise FileNotFoundError(f"Khong tim thay file: {p}")
        parts.append(p.read_text(encoding="utf-8"))
    text = "\n\n".join(parts)

    to_fmt = Path(out_name).suffix.lstrip(".") or "docx"
    out_path = (HERE / out_name) if not Path(out_name).is_absolute() else Path(out_name)
    extra_args = ["--toc", "--toc-depth=3", "--resource-path", str(HERE)]
    if to_fmt == "html":
        extra_args += ["--standalone", "--mathjax", "--embed-resources"]
    pypandoc.convert_text(text, to_fmt, format="markdown",
                          outputfile=str(out_path), extra_args=extra_args)
    return out_path


def resolve_inputs(inp) -> list[Path]:
    if isinstance(inp, str) and inp.upper() == "ALL":
        return sorted(HERE.glob("*.md"))
    if isinstance(inp, (list, tuple)):
        return [(HERE / x) if not Path(x).is_absolute() else Path(x) for x in inp]
    p = Path(inp)
    return [p if p.is_absolute() else HERE / p]


def main():
    ensure_pypandoc()

    # Che do gop nhieu file thanh 1 bao cao (uu tien neu COMBINE khong rong)
    if COMBINE:
        try:
            out = combine_files(COMBINE, COMBINE_OUTPUT)
            print(f"[+] Gop {len(COMBINE)} file -> {out.name}")
        except Exception as e:  # noqa: BLE001
            print(f"[x] Loi khi gop: {e}")
        return

    md_files = resolve_inputs(INPUT)
    if not md_files:
        print("[!] Khong tim thay file .md nao de convert.")
        return

    many = len(md_files) > 1
    for md in md_files:
        out = None
        if OUTPUT_NAME and not many:
            out = Path(OUTPUT_NAME)
            if not out.is_absolute():
                out = HERE / out
        try:
            result = convert_one(md, OUTPUT_FORMAT, out)
            print(f"[+] {md.name} -> {result.name}")
        except Exception as e:  # noqa: BLE001
            print(f"[x] Loi voi {md.name}: {e}")
            if OUTPUT_FORMAT == "pdf":
                print("    -> PDF can mot LaTeX engine (MiKTeX/TeX Live). "
                      "Thu doi OUTPUT_FORMAT='html' roi in PDF tu trinh duyet.")


if __name__ == "__main__":
    main()
    # Giu cua so terminal mo khi double-click tren Windows
    try:
        if sys.stdin and sys.stdin.isatty():
            input("\nXong. Nhan Enter de dong...")
    except Exception:
        pass