#!/usr/bin/env python3
# lfs-sysvbook -- graft the SysV init flavour onto a systemd-only LFS book.
#
# LFS 13.0 dropped the SysV edition.  The difference between the two flavours
# was never the 80+ packages -- it is five SysV-only sections, two
# systemd-only sections, chapter 9, the fstab, and a handful of configure
# flags.  This tool takes the last SysV book (the DONOR, e.g. 12.4) and a new
# systemd book (the TARGET, e.g. 13.0-systemd) and produces a hybrid book
# plus its wget-list and md5sums, stored under a new version label so the
# `lfs` tool builds from it like any other book:
#
#   lfs fetch 12.4
#   lfs fetch 13.0-systemd
#   lfs-sysvbook make                      # -> 13.0-sysv in the lfs store
#   lfs-sysvbook check --book 13.0-sysv
#   lfs --book 13.0-sysv run
#
# What `make` does to the target book:
#   ch8: Systemd section  -> donor "Udev from Systemd" (pinned donor version)
#        D-Bus section    -> removed (BLFS provides it on SysV)
#        Sysklogd,SysVinit-> inserted after E2fsprogs, from the donor
#   ch9: replaced wholesale with the donor's (bootscripts, sysv network, ...)
#   10.2 fstab            -> donor's (SysV mounts /proc, /sys, ... via fstab)
#   7.6  passwd/group     -> systemd-* account lines filtered out
#   flags: Man-DB, Procps-ng, Util-linux get their SysV configure flags
#   wget-list/md5sums: target's, minus removed sections' files, plus the
#        donor files the transplanted sections reference (URLs and MD5s are
#        read from each book's own chapter 3, so no network is needed)
#
# Transplanted sections keep the DONOR's package versions (e.g. udev from
# systemd-257.8) -- those are the instructions that are known to work.
# Everything else builds at the target's versions.

import argparse
import hashlib
import os
import re
import sys

try:
    from bs4 import BeautifulSoup
except ImportError:
    sys.stderr.write(
        "lfs-sysvbook needs beautifulsoup4 (same as the lfs tool):\n"
        "      pip install beautifulsoup4\n"
        "  or on the built system:\n"
        "      packagemanager pip install beautifulsoup4\n")
    sys.exit(2)

VERSION = "1.0.0"


def _build_id():
    """Fingerprint of this file's own contents -- cannot go stale."""
    try:
        with open(os.path.abspath(__file__), "rb") as fh:
            return hashlib.md5(fh.read()).hexdigest()[:7]
    except OSError:
        return "unknown"


# --------------------------------------------------------------------------- #
# the store -- same layout as the lfs tool
# --------------------------------------------------------------------------- #
def store_dir():
    return os.environ.get("LFS_STORE") or "/usr/share/lfs"


def books_dir():
    return os.path.join(store_dir(), "books")


def book_path(ver):
    return os.path.join(books_dir(), f"LFS-BOOK-{ver}-NOCHUNKS.html")


def wget_list_path(ver):
    return os.path.join(books_dir(), f"wget-list-{ver}.txt")


def md5sums_path(ver):
    return os.path.join(books_dir(), f"md5sums-{ver}.txt")


def load_soup(path):
    with open(path, "rb") as f:
        html = f.read()
    for parser in ("lxml", "html.parser"):
        try:
            return BeautifulSoup(html, parser)
        except Exception:
            continue
    return BeautifulSoup(html, "html.parser")


# --------------------------------------------------------------------------- #
# what makes a SysV book a SysV book -- all flavour knowledge lives here
# --------------------------------------------------------------------------- #
# Sections removed from the target (systemd-only), by anchor id.
REMOVE = ["ch-system-dbus"]

# Target section replaced by a donor section: (target id, donor id).
REPLACE = [
    ("ch-system-systemd", "ch-system-udev"),      # udev built from systemd src
    ("chapter-config",    "chapter-config"),      # whole chapter 9
    ("ch-bootable-fstab", "ch-bootable-fstab"),   # SysV mounts via fstab
]

# Donor sections inserted after a target section: (after-target-id, [donor ids]).
INSERT_AFTER = [
    ("ch-system-e2fsprogs", ["ch-system-sysklogd", "ch-system-sysvinit"]),
]

# Configure-flag edits inside target sections that exist in both flavours:
# (anchor id, description, pattern, replacement, done-marker).
# The marker makes each rule idempotent and lets `check` verify it took.
FLAG_PATCHES = [
    ("ch-system-procps",
     "Procps-ng: drop --with-systemd",
     r"[ \t]*\\\n[ \t]*--with-systemd\b",
     "",
     None),          # done when the pattern no longer matches
    ("ch-system-man-db",
     "Man-DB: do not install systemd units/tmpfiles",
     r"(--with-grap=\S+)",
     r"\1             \\\n"
     r"            --with-systemdtmpfilesdir=            \\\n"
     r"            --with-systemdsystemunitdir=",
     "--with-systemdtmpfilesdir"),
    ("ch-system-util-linux",
     "Util-linux: build without systemd support",
     r"(?m)^([ \t]*)(ADJTIME_PATH=)",
     r"\1--without-systemd     \\\n"
     r"\1--without-systemdsystemunitdir        \\\n"
     r"\1\2",
     "--without-systemd"),
]

# 7.6: account lines that only a systemd system needs.
CREATEFILES_ID = "ch-tools-createfiles"
SYSTEMD_ACCOUNT_LINE = re.compile(r"(?m)^systemd-[^\n]*\n?")

FILE_TOKEN = re.compile(r"([A-Za-z0-9_.+-]+\.(?:tar\.[a-z0-9]+|tgz|patch))")


# --------------------------------------------------------------------------- #
# DOM helpers -- a "unit" is a section's titlepage div + its content div
# --------------------------------------------------------------------------- #
def unit(soup, anchor_id, book_label):
    """(titlepage div, content div) of the section whose heading carries
    <a id=anchor_id>.  In NOCHUNKS books these are two sibling divs."""
    a = soup.find("a", id=anchor_id)
    if a is None:
        raise SystemExit(f"! section '{anchor_id}' not found in {book_label}.\n"
                         f"  The book layout changed -- this tool needs an "
                         f"update for it.")
    tp = None
    for p in a.parents:
        if p.name == "div" and "titlepage" in (p.get("class") or []):
            tp = p
            break
    if tp is None:
        raise SystemExit(f"! '{anchor_id}' in {book_label} has no titlepage "
                         f"div -- unexpected book layout.")
    content = tp.find_next_sibling("div")
    if content is None or "titlepage" in (content.get("class") or []):
        raise SystemExit(f"! '{anchor_id}' in {book_label} has no content "
                         f"div -- unexpected book layout.")
    return tp, content


def unit_text(tp, content):
    return tp.get_text(" ", strip=True) + "\n" + content.get_text("\n")


def title_of(tp):
    return re.sub(r"\s+", " ",
                  tp.get_text(" ", strip=True)).replace("\u00a0", " ")


def _title_token(tp):
    """Name-Version token from the section heading, same rule as `lfs`:
    first whitespace token that ends in -<digits...>."""
    t = tp.get_text(" ", strip=True).replace("\u00a0", " ")
    t = re.sub(r"^\d+(\.\d+)*\.\s*", "", t.strip())
    tok_re = re.compile(r"^[A-Za-z][A-Za-z0-9+_.]*"
                        r"(?:(?:::|[-_])[A-Za-z][A-Za-z0-9+_.]*)*"
                        r"[-_][0-9][0-9.]*(?:-[0-9]+)*$")
    for tok in t.split():
        if tok_re.match(tok):
            return tok
    return None


def files_of(tp, content):
    """Tarballs/patches a section needs: its title token, the title tokens of
    any sections inside it (a transplanted chapter 9 carries LFS-Bootscripts
    only in a heading), and every file name its text mentions."""
    toks = set()
    t = _title_token(tp)
    if t:
        toks.add(("token", t))
    for h in content.find_all(re.compile(r"^h\d$")):
        t = _title_token(h)
        if t:
            toks.add(("token", t))
    for f in FILE_TOKEN.findall(unit_text(tp, content)):
        toks.add(("file", os.path.basename(f)))
    return toks


def _norm(s):
    return re.sub(r"[^a-z0-9]", "", s.lower())


def match_urls(url_map, wanted, label):
    """URLs from {basename: (url, md5)} that a section's tokens refer to.
    A 'file' token matches its exact basename; a 'token' (Name-Version)
    matches a basename whose stem normalises to the same string."""
    def stem(name):
        s = re.sub(r"\.(tar\.[a-z0-9]+|tgz|patch)$", "", name)
        return _norm(re.sub(r"-src$", "", s))

    out, missing = {}, []
    for kind, val in sorted(wanted):
        hit = None
        if kind == "file" and val in url_map:
            hit = val
        else:
            # a title token, or prose naming the tarball with the wrong
            # compression suffix (12.4 prose says .tar.xz, github ships .gz)
            for base in url_map:
                if stem(base) == stem(val):
                    hit = base
                    break
        if hit:
            out[hit] = url_map[hit]
        elif kind == "file":
            missing.append((val, label))
    return out, missing


# --------------------------------------------------------------------------- #
# chapter 3 of each book carries every download URL and MD5 -- no network
# --------------------------------------------------------------------------- #
def url_map_from_book(path):
    """{basename: (url, md5)} from the book's own package/patch lists."""
    with open(path, encoding="utf-8", errors="replace") as f:
        html = f.read()
    urls = [(m.start(), m.group(1)) for m in
            re.finditer(r'href=\s*"(https?://[^"]+?'
                        r'\.(?:tar\.[a-z0-9]+|tgz|patch))"', html)]
    md5s = [(m.start(), m.group(1)) for m in
            re.finditer(r'MD5 sum:\s*(?:<[^>]+>\s*)*([0-9a-f]{32})', html)]
    out, mi = {}, 0
    for i, (upos, url) in enumerate(urls):
        nxt = urls[i + 1][0] if i + 1 < len(urls) else len(html)
        while mi < len(md5s) and md5s[mi][0] < upos:
            mi += 1
        md5 = md5s[mi][1] if mi < len(md5s) and md5s[mi][0] < nxt else None
        out.setdefault(os.path.basename(url), (url, md5))
    return out


# --------------------------------------------------------------------------- #
# mini command parser -- same rules as `lfs` (for check + patching)
# --------------------------------------------------------------------------- #
def _is_admonition(tag):
    cls = set(getattr(tag, "get", lambda _k: None)("class") or [])
    return bool({"admon", "note", "tip", "caution", "warning",
                 "important"} & cls)


def runnable_commands(scope):
    """Command blocks the lfs tool would execute from this tag's subtree."""
    out = []
    for pre in scope.find_all("pre"):
        classes = pre.get("class") or []
        if not any(c in classes for c in ("userinput", "root", "install")):
            continue
        if any(_is_admonition(p) for p in pre.parents):
            continue
        out.append(pre.get_text("").replace("\u00a0", " ").rstrip())
    return out


def patch_pre_text(scope, pattern, repl):
    """Apply a regex to the text nodes of runnable command blocks."""
    n = 0
    for pre in scope.find_all("pre"):
        classes = pre.get("class") or []
        if not any(c in classes for c in ("userinput", "root", "install")):
            continue
        for node in list(pre.find_all(string=True)):
            new = re.sub(pattern, repl, str(node))
            if new != str(node):
                node.replace_with(new)
                n += 1
    return n


# --------------------------------------------------------------------------- #
# make
# --------------------------------------------------------------------------- #
def resolve_input(ver, file_arg, what):
    if file_arg:
        if not os.path.isfile(file_arg):
            raise SystemExit(f"! {what} file not found: {file_arg}")
        return os.path.basename(file_arg), file_arg
    p = book_path(ver)
    if not os.path.isfile(p):
        raise SystemExit(f"! {what} book {ver} is not in the store.\n"
                         f"      lfs fetch {ver}\n"
                         f"  or pass a local file with --{what}-file.")
    return ver, p


def cmd_make(args):
    d_label, d_path = resolve_input(args.donor, args.donor_file, "donor")
    t_label, t_path = resolve_input(args.target, args.target_file, "target")
    out_ver = args.out or (t_label.replace("-systemd", "") + "-sysv"
                           if "-systemd" in t_label else t_label + "-sysv")
    print(f"donor : {d_label}  ({d_path})")
    print(f"target: {t_label}  ({t_path})")
    print(f"out   : {out_ver}")

    donor = load_soup(d_path)
    target = load_soup(t_path)
    warnings = []

    # -- collect the file needs BEFORE surgery moves nodes around ----------- #
    donor_urls = url_map_from_book(d_path)
    target_urls = url_map_from_book(t_path)

    drop_files = set()
    for tid in REMOVE + [t for t, _d in REPLACE if t != "chapter-config"]:
        # chapter-config (ch9) references no downloads in the systemd book
        drop_files |= files_of(*unit(target, tid, t_label))
    dropped, _ = match_urls(target_urls, drop_files, t_label)

    add_files = set()
    for _tid, did in REPLACE:
        add_files |= files_of(*unit(donor, did, d_label))
    for _after, dids in INSERT_AFTER:
        for did in dids:
            add_files |= files_of(*unit(donor, did, d_label))
    added, missing = match_urls(donor_urls, add_files, d_label)
    added = {b: v for b, v in added.items()
             if b not in target_urls or b in dropped}

    # -- surgery ------------------------------------------------------------ #
    print("\nsections:")
    for tid in REMOVE:
        tp, content = unit(target, tid, t_label)
        title = title_of(tp)
        tp.decompose()
        content.decompose()
        print(f"  removed   {title}   [{tid}]")

    for tid, did in REPLACE:
        t_tp, t_content = unit(target, tid, t_label)
        d_tp, d_content = unit(donor, did, d_label)
        title_old = title_of(t_tp)
        title_new = title_of(d_tp)
        t_tp.replace_with(d_tp.extract())
        t_content.replace_with(d_content.extract())
        print(f"  replaced  {title_old}\n"
              f"        ->  {title_new}   [{tid}]")

    for after_id, dids in INSERT_AFTER:
        _tp, anchor_content = unit(target, after_id, t_label)
        at = anchor_content
        for did in dids:
            d_tp, d_content = unit(donor, did, d_label)
            title = title_of(d_tp)
            at.insert_after(d_content.extract())
            at.insert_after(d_tp.extract())
            at = d_content
            print(f"  inserted  {title}   [after {after_id}]")

    # -- 7.6: filter systemd account lines out of /etc/passwd, /etc/group -- #
    _tp, cf = unit(target, CREATEFILES_ID, t_label)
    n = patch_pre_text(cf, SYSTEMD_ACCOUNT_LINE, "")
    print(f"\n7.6 passwd/group: systemd-* account lines removed "
          f"({n} heredoc(s) touched)" if n else
          "\n7.6 passwd/group: nothing to remove")
    if not n:
        warnings.append("7.6: no systemd-* account lines found -- "
                        "book layout may have changed")

    # -- configure-flag patches --------------------------------------------- #
    print("flags:")
    for aid, desc, pat, repl, marker in FLAG_PATCHES:
        _tp, content = unit(target, aid, t_label)
        text = "\n".join(runnable_commands(content))
        if marker and marker in text:
            print(f"  ok        {desc} (already applied)")
            continue
        n = patch_pre_text(content, pat, repl)
        if n:
            print(f"  patched   {desc}")
        else:
            print(f"  !! NOT applied: {desc}")
            warnings.append(f"flag patch did nothing: {desc} -- check the "
                            f"section by hand")

    # -- stale TOC entries pointing at removed sections --------------------- #
    ids = {t.get("id") for t in target.find_all(id=True)}
    pruned = 0
    for a in target.find_all("a", href=True):
        href = a["href"]
        if href.startswith("#") and href[1:] not in ids:
            dt = a.find_parent("dt")
            if dt is not None:
                dt.decompose()
                pruned += 1
    if pruned:
        print(f"toc: {pruned} entries for removed sections pruned")

    # -- serialize + relabel ------------------------------------------------ #
    # The target's release label is read from the book itself (the
    # `echo <label> > /etc/lfs-release` command), so a --target-file with any
    # filename still gets /etc/lfs-release, os-release and the titles renamed.
    html = str(target)
    m = re.search(r"echo\s+([0-9][\w.+-]*)\s*(?:&gt;|>)\s*/etc/lfs-release",
                  html)
    if m and m.group(1) != out_ver:
        html = html.replace(m.group(1), out_ver)
        print(f"relabelled: {m.group(1)} -> {out_ver}")
    elif not m:
        warnings.append("could not find the release label in the target "
                        "book -- titles and /etc/lfs-release keep its name")

    # -- wget-list + md5sums ------------------------------------------------ #
    # Chapter 3 sometimes offers an alternative download (linux-6.18.tar.xz +
    # patch instead of linux-6.18.10.tar.xz); alternatives carry no MD5.  Keep
    # an MD5-less entry only when no checksummed entry for the same package
    # exists.
    def name_stem(base):
        return re.match(r"[A-Za-z_+-]+", base).group(0).lower()

    entries = [(b, u, m) for b, (u, m) in target_urls.items()
               if b not in dropped]
    entries += [(b, u, m) for b, (u, m) in added.items()]
    with_md5 = {name_stem(b) for b, _u, m in entries if m}
    wget, md5 = [], []
    for base, url, m in entries:
        if not m and name_stem(base) in with_md5:
            print(f"  (skipped alternative download: {base})")
            continue
        wget.append(url)
        if m:
            md5.append(f"{m}  {base}")

    print("\ndownloads:")
    for b in sorted(dropped):
        print(f"  -  {b}")
    for b in sorted(added):
        print(f"  +  {b}")
    for f, where in missing:
        warnings.append(f"'{f}' is referenced but has no download URL in "
                        f"{where} -- add it to the wget-list by hand if the "
                        f"build asks for it")

    # -- write -------------------------------------------------------------- #
    outdir = args.outdir or books_dir()
    try:
        os.makedirs(outdir, exist_ok=True)
    except PermissionError:
        raise SystemExit(f"! cannot create {outdir} -- run as root or pass "
                         f"--outdir somewhere writable.")
    paths = {
        "book":      os.path.join(outdir, f"LFS-BOOK-{out_ver}-NOCHUNKS.html"),
        "wget-list": os.path.join(outdir, f"wget-list-{out_ver}.txt"),
        "md5sums":   os.path.join(outdir, f"md5sums-{out_ver}.txt"),
    }
    try:
        with open(paths["book"], "w", encoding="utf-8") as f:
            f.write(html)
        with open(paths["wget-list"], "w") as f:
            f.write("\n".join(wget) + "\n")
        with open(paths["md5sums"], "w") as f:
            f.write("\n".join(md5) + "\n")
    except PermissionError:
        raise SystemExit(f"! cannot write into {outdir} -- run as root or "
                         f"pass --outdir somewhere writable.")
    for k, p in paths.items():
        os.chmod(p, 0o644)
        print(f"  {k:9s} -> {p}")

    if warnings:
        print("\nwarnings:")
        for w in warnings:
            print(f"  ! {w}")
    print(f"\nverify it:   lfs-sysvbook check --book {out_ver}"
          + ("" if not args.outdir else f" --file {paths['book']}"))
    if not args.outdir:
        print(f"build it:    lfs --book {out_ver} run")
    return 0


# --------------------------------------------------------------------------- #
# check
# --------------------------------------------------------------------------- #
def cmd_check(args):
    if args.file:
        path, label = args.file, os.path.basename(args.file)
        wl = args.wget_list
    else:
        ver = args.book
        if not ver:
            raise SystemExit("! pass --book <version> or --file <html>")
        path, label, wl = book_path(ver), ver, wget_list_path(ver)
    if not os.path.isfile(path):
        raise SystemExit(f"! no book at {path}")
    soup = load_soup(path)
    ok = True

    def result(good, msg):
        nonlocal ok
        print(("  ok   " if good else "  FAIL ") + msg)
        ok = ok and good

    print(f"checking {label}:")

    ids = {t.get("id") for t in soup.find_all(id=True)}
    for aid in ("ch-system-udev", "ch-system-sysklogd", "ch-system-sysvinit",
                "ch-config-bootscripts", "ch-config-usage"):
        result(aid in ids, f"SysV section present: {aid}")
    for aid in ("ch-system-dbus", "ch-config-systemd-custom"):
        result(aid not in ids, f"systemd section absent: {aid}")
    # ch-system-systemd must now be udev's heading, not a Systemd build
    a = soup.find("a", id="ch-system-systemd")
    if a is not None:
        h = a.find_parent(re.compile(r"^h\d$"))
        result(h is not None and "Udev" in h.get_text(),
               "no Systemd package section")

    # no runnable command may use systemd machinery
    bad = []
    for pre in soup.find_all("pre"):
        classes = pre.get("class") or []
        if not any(c in classes for c in ("userinput", "root", "install")):
            continue
        if any(_is_admonition(p) for p in pre.parents):
            continue
        text = pre.get_text("")
        for pat in (r"\bsystemctl\b", r"--with-systemd(?![a-z-])",
                    r"\bsystemd-analyze\b"):
            if re.search(pat, text):
                bad.append((pat, text.strip().splitlines()[0][:60]))
    result(not bad, "no systemd commands in runnable blocks"
           + ("" if not bad else "  " + "; ".join(f"{t} in '{l}...'"
                                                  for t, l in bad[:3])))

    # flag patches took
    for aid, desc, _pat, _repl, marker in FLAG_PATCHES:
        if not marker:
            continue
        a = soup.find("a", id=aid)
        if a is None:
            result(False, f"section {aid} missing")
            continue
        tp = next(p for p in a.parents
                  if p.name == "div" and "titlepage" in (p.get("class") or []))
        content = tp.find_next_sibling("div")
        result(marker in "\n".join(runnable_commands(content)), desc)

    # 7.6 heredocs carry no systemd accounts
    a = soup.find("a", id=CREATEFILES_ID)
    if a is not None:
        tp = next(p for p in a.parents
                  if p.name == "div" and "titlepage" in (p.get("class") or []))
        content = tp.find_next_sibling("div")
        result(not SYSTEMD_ACCOUNT_LINE.search(
                   "\n".join(runnable_commands(content))),
               "7.6: no systemd accounts in /etc/passwd and /etc/group")

    # every file a runnable command references is downloadable
    if wl and os.path.isfile(wl):
        with open(wl) as f:
            have = {os.path.basename(u.strip()) for u in f if u.strip()}
        need = set()
        for pre in soup.find_all("pre"):
            classes = pre.get("class") or []
            if not any(c in classes for c in ("userinput", "root", "install")):
                continue
            if any(_is_admonition(p) for p in pre.parents):
                continue
            for tok in FILE_TOKEN.findall(pre.get_text("")):
                need.add(os.path.basename(tok))
        # names a build creates itself are not downloads
        miss = {n for n in need if n not in have
                and not n.startswith("lfs-temp-tools-")   # 7.13 creates it
                and not any(h.startswith(n.split(".tar")[0]) for h in have)}
        result(not miss, "wget-list covers every file the commands reference"
               + ("" if not miss else ": missing " + ", ".join(sorted(miss))))
    else:
        print(f"  (no wget-list at {wl} -- file coverage not checked)")

    print("PASS" if ok else "FAIL")
    return 0 if ok else 1


# --------------------------------------------------------------------------- #
def main():
    ap = argparse.ArgumentParser(
        prog="lfs-sysvbook",
        description="Graft the SysV flavour of an old LFS book onto a new "
                    "systemd-only release.")
    ap.add_argument("--version", action="version",
                    version=f"lfs-sysvbook {VERSION} (build {_build_id()})")
    sub = ap.add_subparsers(dest="cmd", required=True)

    p = sub.add_parser("make", help="produce the hybrid book + wget-list + "
                                    "md5sums")
    p.add_argument("--donor", default="12.4",
                   help="cached SysV book version (default: 12.4, the last "
                        "SysV release)")
    p.add_argument("--target", default="13.0-systemd",
                   help="cached systemd book version (default: 13.0-systemd)")
    p.add_argument("--donor-file", help="local donor html instead of the store")
    p.add_argument("--target-file", help="local target html instead of the store")
    p.add_argument("--out", help="version label for the result "
                                 "(default: target with -systemd -> -sysv)")
    p.add_argument("--outdir", help="write here instead of the lfs store")
    p.set_defaults(func=cmd_make)

    p = sub.add_parser("check", help="verify a hybrid book is SysV-clean")
    p.add_argument("--book", help="version label in the lfs store")
    p.add_argument("--file", help="local html file instead")
    p.add_argument("--wget-list", help="wget-list to check coverage against "
                                       "(with --file)")
    p.set_defaults(func=cmd_check)

    args = ap.parse_args()
    sys.exit(args.func(args))


if __name__ == "__main__":
    main()