from pathlib import Path
import subprocess
import re
import json
from functools import reduce
from os import system, environ as ENVIRON
import tempfile
import fire

ENVIRON["PATH"] = ENVIRON["PATH"] + ";%s" % Path.cwd().absolute()

NUM_MATCH = re.compile(r"\D*(\d+)\D*")
TOCS_TAB_MATCH = re.compile(r"(\t*)(.+)\t(\d+)")
TOCS_PDFTK_MATCH = re.compile(
    r"BookmarkBegin\nBookmarkTitle:\s*(.+)\nBookmarkLevel:\s*(\d+)\nBookmarkPageNumber:\s*(\d+)(?:$|\n)"
)


def _asn(f: Path):
    m = NUM_MATCH.match(f.stem)
    return int(m.group(1)) if m else f.stem


def _sort_and_rename(path):
    items = []
    for f in path.glob("*"):
        m = NUM_MATCH.match(f.stem)
        num = int(m.group(1)) if m else None
        items.append((f, num))
    new_names = sorted(
        ((f, f.with_stem("{:0>5d}".format(num))) for f, num in items if num),
        key=lambda j: j[1],
    )
    new_exists = [n.exists() for (o, n) in new_names]
    if all(new_exists):
        pass
    elif any(new_exists):
        # 这里没想好该怎么做
        pass
    else:
        for old_name, new_name in new_names:
            old_name.rename(new_name)


# ========目录========
def _write_tocs_pdftk(tocs):
    combine_line = lambda l, t, p: "\n".join(
        (
            "BookmarkBegin",
            "BookmarkTitle: %s" % t,
            "BookmarkLevel: %s" % l,
            "BookmarkPageNumber: %s" % p,
            "",
        )
    )
    return reduce(
        lambda i, j: i + j,
        (combine_line(*line) for line in tocs),
        "",
    )


def _read_tocs_pdftk(text):
    return [
        (int(match.group(2)), match.group(1), int(match.group(3)))
        for match in TOCS_PDFTK_MATCH.finditer(text)
    ]


def _read_tocs_tab(text):
    result = []
    for line in text.split("\n"):
        m = TOCS_TAB_MATCH.match(line)
        if m:
            result.append([len(m.group(1)) + 1, m.group(2), int(m.group(3))])
    return result


# ========对外接口========
def combine_pics(indir, outpdf, auto_search_num=False):
    indir, outpdf = Path(indir), Path(outpdf)
    outpdf.parent.mkdir(parents=True, exist_ok=True)
    # 所有合法文件对象
    if auto_search_num:
        _sort_and_rename(indir)
    outpdf = '"%s"' % outpdf.absolute()
    indir = '"%s"' % indir.joinpath("*").absolute()
    cmd_str = "mutool convert -o %s -F PDF -O compress %s" % (outpdf, indir)
    return subprocess.Popen(cmd_str, creationflags=subprocess.CREATE_NO_WINDOW)


def extract_pics(inpdf, outdir):
    inpdf, outdir = Path(inpdf), Path(outdir)
    outdir.mkdir(parents=True, exist_ok=True)
    inpdf = '"%s"' % inpdf.absolute()
    cmd_str = "mutool extract %s" % inpdf
    return subprocess.Popen(
        cmd_str, creationflags=subprocess.CREATE_NO_WINDOW, cwd=outdir.absolute()
    )


def page2PNG(inpdf, outdir, dpi=300):
    inpdf, outdir = Path(inpdf), Path(outdir)
    outdir.mkdir(parents=True, exist_ok=True)
    outpics = '"%s"' % outdir.joinpath("%05d.png").absolute()
    inpdf = '"%s"' % inpdf.absolute()
    cmd_str = "mutool convert -o %s -F PNG -O resolution=%s %s" % (outpics, dpi, inpdf)
    return subprocess.Popen(cmd_str, creationflags=subprocess.CREATE_NO_WINDOW)


def combine_pdfs(indir, outpdf, auto_search_num=False):
    indir, outpdf = Path(indir), Path(outpdf)
    outpdf.parent.mkdir(parents=True, exist_ok=True)
    if auto_search_num:
        _sort_and_rename(indir)
    outpdf = '"%s"' % outpdf.absolute()
    indir = '"%s"' % indir.joinpath("*.pdf").absolute()
    cmd_str = "pdftk %s cat output %s" % (indir, outpdf)
    return subprocess.Popen(cmd_str, creationflags=subprocess.CREATE_NO_WINDOW)


def set_tocs(inpdf, tocs, outpdf):
    inpdf, tocs, outpdf = Path(inpdf), Path(tocs), Path(outpdf)
    outpdf.parent.mkdir(parents=True, exist_ok=True)
    if tocs.suffix == ".json":
        tocs_array = json.loads(tocs.read_text("utf-8"))
    elif tocs.suffix == ".txt":
        tocs_array = _read_tocs_tab(tocs.read_text("utf-8"))
    # 这里的pdftk导出文件名不能含有中文
    with tempfile.NamedTemporaryFile(
        mode="w", encoding="utf-8", newline="\n", delete=False
    ) as tf:
        temp_path = Path(tf.name)
        tf.write(_write_tocs_pdftk(tocs_array))
    # ====
    inpdf = '"%s"' % inpdf.absolute()
    outpdf = '"%s"' % outpdf.absolute()
    cmd_str = "pdftk %s update_info_utf8 %s output %s" % (
        inpdf,
        '"%s"' % temp_path,
        outpdf,
    )
    system(cmd_str)
    temp_path.unlink()


def output_tocs(inpdf, outtocs):
    inpdf, outtocs = Path(inpdf), Path(outtocs)
    outtocs.parent.mkdir(parents=True, exist_ok=True)
    inpdf = '"%s"' % inpdf.absolute()
    # 这里的pdftk导出文件名不能含有中文
    with tempfile.NamedTemporaryFile(
        mode="w", encoding="utf-8", newline="\n", delete=False
    ) as tf:
        temp_path = Path(tf.name)
        cmd_str = "pdftk %s dump_data_utf8 output %s" % (inpdf, '"%s"' % tf.name)
        system(cmd_str)
    # ====
    tocs = _read_tocs_pdftk(temp_path.read_text("utf-8"))
    outtocs.write_text(json.dumps(tocs, ensure_ascii=False), "utf-8")


def combine_pics_in_subfolders(path, auto_search_num):
    path = Path(path)
    dirs = (d for d in path.glob("*") if d.is_dir())
    return [
        combine_pics(d, path.absolute().joinpath("%s.pdf" % d.name), auto_search_num)
        for d in dirs
    ]


if __name__ == "__main__":
    # 考虑到命令行程序需求，所有函数的参数都可以接受字符串
    fire.Fire()