#!/usr/bin/env python3
"""Build a published update manifest from a plugin/theme repo's curated inputs.

Inputs (all WP-native / YAML, all export-ignored from the shipped zip):
  .mu-manifest.yaml   curated base — name/homepage/author/tags + optional section
                      OVERRIDES (inline HTML, or `@path.(md|html|txt)` includes).
  .mu-release.yaml    pipeline config (passed via --config): sections.autogen,
                      sections.strip_title, channel, app_build.
  readme.txt          the WordPress readme — its `== Section ==` blocks are the
                      DEFAULT modal sections (Description / Installation / FAQ /
                      Changelog / …). The `=== Title ===` header + the tag block
                      are naturally skipped (they aren't `== … ==`).
  CHANGELOG.md        changelog fallback, used only if readme.txt has no Changelog.

SECTIONS cascade, per modal key:  .mu-manifest.yaml override  >  readme.txt block
  ( + changelog: ... > readme.txt == Changelog == > CHANGELOG.md ).

readme.txt is a markdown subset; its `= Subhead =` lines render as bold paragraphs
(not <h4>) because WP's modal puts `clear:both` on headings, which — when a section
LEADS with one — drops the body below the floated sidebar (a big top gap). For the
same reason a leading <h1–6> is demoted to <p><strong>…</strong></p>. `strip_title`
(null|bool|dict) removes a leading H1 from parsed markdown before that.

Computed fields (--computed) always win for release-critical/readme-derived keys;
the curated base wins for everything else it provides.

Usage:
  build-manifest.py --repo-root . --computed computed.json --out out.json [--config .mu-release.yaml]
"""
import argparse
import html
import json
import os
import re
import sys

try:
    import markdown as _md
    import yaml
except ImportError as e:
    sys.exit(f"build-manifest: needs python 'markdown' + 'pyyaml' ({e})")

PROTECTED = {
    "version", "slug", "download_url", "last_updated",
    "icons", "screenshot_url", "requires", "tested", "requires_php",
}
INCLUDE_RE = re.compile(r"^@(?P<path>.+\.(?P<ext>md|html|txt))$")
BASE_FILE = ".mu-manifest.yaml"

# readme.txt section name -> WP modal section key.
_KEYS = {
    "description": "description", "installation": "installation",
    "faq": "faq", "frequently asked questions": "faq",
    "changelog": "changelog", "change log": "changelog",
    "screenshots": "screenshots", "other notes": "other_notes",
    "upgrade notice": "upgrade_notice",
}


def section_key(name: str) -> str:
    n = " ".join(name.strip().lower().split())
    return _KEYS.get(n) or re.sub(r"[^a-z0-9]+", "_", n).strip("_")


def md_to_html(text: str) -> str:
    return _md.markdown(text, extensions=["extra", "sane_lists", "nl2br"])


def wporg_subheads(text: str) -> str:
    """wp.org `= Subhead =` -> bold paragraph (NOT a heading: WP's modal clears
    floats on h*, so heading-led content gaps below the sidebar)."""
    return re.sub(r"(?m)^=[ \t]+(.+?)[ \t]*=[ \t]*$", r"**\1**", text)


def strip_leading_title(md: str) -> str:
    lines = md.splitlines()
    i = 0
    while i < len(lines) and lines[i].strip() == "":
        i += 1
    if i < len(lines):
        if re.match(r"^#\s+\S", lines[i]):
            i += 1
        elif i + 1 < len(lines) and lines[i].strip() and re.match(r"^=+\s*$", lines[i + 1]):
            i += 2
    return "\n".join(lines[i:]).lstrip("\n")


def demote_leading_heading(html_str: str) -> str:
    """Demote a LEADING <h1–6> to a non-clearing <p><strong>…</strong></p> so the
    section doesn't trip the modal's `clear:both` on headings (survives wp_kses)."""
    return re.sub(
        r"^\s*<h[1-6][^>]*>(.*?)</h[1-6]>",
        r"<p><strong>\1</strong></p>",
        html_str, count=1, flags=re.S,
    )


def should_strip(key: str, strip_cfg) -> bool:
    if strip_cfg is True:
        return True
    if isinstance(strip_cfg, dict):
        return bool(strip_cfg.get(key, False))
    return False


def md_section(text: str, key: str, strip_cfg) -> str:
    if should_strip(key, strip_cfg):
        text = strip_leading_title(text)
    return md_to_html(text)


def convert(text: str, ext: str, key: str, strip_cfg) -> str:
    if ext == "md":
        return md_section(text, key, strip_cfg)
    if ext == "html":
        return text
    paras = [p.strip() for p in re.split(r"\n\s*\n", text) if p.strip()]
    return "".join("<p>" + html.escape(p).replace("\n", "<br>") + "</p>" for p in paras)


def resolve_in_repo(repo_root: str, rel: str):
    root = os.path.realpath(repo_root)
    full = os.path.realpath(os.path.join(root, rel))
    if full != root and not full.startswith(root + os.sep):
        return None
    return full if os.path.isfile(full) else None


def read_text(repo_root: str, name: str):
    p = os.path.join(repo_root, name)
    return open(p, encoding="utf-8").read() if os.path.isfile(p) else None


def parse_readme_sections(repo_root: str) -> dict:
    """Split readme.txt on `== Name ==` lines (two equals; the `=== Title ===`
    header and the tag block above the first section are skipped)."""
    text = read_text(repo_root, "readme.txt")
    if not text:
        return {}
    parts = re.split(r"(?m)^==[ \t]+(.+?)[ \t]+==[ \t]*$", text)
    out = {}
    for i in range(1, len(parts) - 1, 2):
        name, body = parts[i].strip(), parts[i + 1].strip()
        if name and body:
            out[name] = body
    return out


def screenshot_captions(repo_root: str) -> dict:
    """Numbered captions from the readme `== Screenshots ==` block (N -> caption)."""
    for name, body in parse_readme_sections(repo_root).items():
        if section_key(name) == "screenshots":
            caps = {}
            for line in body.splitlines():
                m = re.match(r"^\s*(\d+)\.\s+(.*\S)", line)
                if m:
                    caps[int(m.group(1))] = m.group(2).strip()
            return caps
    return {}


def build_screenshots(urls: dict, repo_root: str) -> dict:
    """Pair release-provided screenshot URLs {N: src} with the readme's numbered
    captions into the WP `screenshots` field ({N: {src, caption}})."""
    caps = screenshot_captions(repo_root)
    return {
        str(n): {"src": urls[str(n)], "caption": caps.get(n, "")}
        for n in sorted(int(k) for k in urls)
    }


def build_sections(base: dict, repo_root: str, autogen: bool, strip_cfg) -> dict:
    resolved = {}
    # 1. curated overrides (.mu-manifest.yaml sections).
    if isinstance(base.get("sections"), dict):
        for name, body in base["sections"].items():
            key = section_key(str(name))
            if isinstance(body, str):
                one = body.strip()
                m = INCLUDE_RE.match(one) if "\n" not in one else None
                if m:
                    full = resolve_in_repo(repo_root, m.group("path"))
                    if full:
                        resolved[key] = convert(open(full, encoding="utf-8").read(), m.group("ext"), key, strip_cfg)
                        continue
                    sys.stderr.write(f"build-manifest: section '{key}' include '{m.group('path')}' not found — literal\n")
                resolved[key] = body
            else:
                resolved[key] = str(body)
    if autogen:
        # 2. readme.txt == Section == blocks (the WP-native default).
        for name, body in parse_readme_sections(repo_root).items():
            key = section_key(name)
            if key == "screenshots":
                continue  # not a text tab — handled as the `screenshots` gallery.
            if key not in resolved:
                resolved[key] = md_section(wporg_subheads(body), key, strip_cfg)
        # 3. changelog fallback -> CHANGELOG.md.
        if "changelog" not in resolved:
            cl = read_text(repo_root, "CHANGELOG.md")
            if cl and cl.strip():
                resolved["changelog"] = md_section(cl, "changelog", strip_cfg)
    # No section may LEAD with a heading (modal clear:both gap fix).
    return {k: demote_leading_heading(v) for k, v in resolved.items()}


def main() -> None:
    ap = argparse.ArgumentParser()
    ap.add_argument("--repo-root", required=True)
    ap.add_argument("--computed", required=True)
    ap.add_argument("--out", required=True)
    ap.add_argument("--config", default="")
    args = ap.parse_args()

    repo_root = args.repo_root
    cfg = {}
    if args.config and os.path.isfile(args.config):
        cfg = yaml.safe_load(open(args.config, encoding="utf-8")) or {}
    sec_cfg = cfg.get("sections") or {}
    autogen = bool(sec_cfg.get("autogen", True))
    strip_cfg = sec_cfg.get("strip_title")

    base = {}
    base_path = os.path.join(repo_root, BASE_FILE)
    if os.path.isfile(base_path):
        base = yaml.safe_load(open(base_path, encoding="utf-8")) or {}
        if not isinstance(base, dict):
            sys.exit(f"build-manifest: {BASE_FILE} must be a mapping")

    computed = json.load(open(args.computed, encoding="utf-8"))
    shots = computed.pop("_screenshots", None)  # {N: src} from release.sh (channel URLs)
    sections = build_sections(base, repo_root, autogen, strip_cfg)

    result = dict(computed)
    for k, v in base.items():
        if k == "sections" or k in PROTECTED:
            continue
        result[k] = v
    if sections:
        result["sections"] = sections
    if shots:
        result["screenshots"] = build_screenshots(shots, repo_root)

    with open(args.out, "w", encoding="utf-8") as fh:
        json.dump(result, fh, indent=2, ensure_ascii=False)
        fh.write("\n")
    sys.stderr.write(f"build-manifest: wrote {args.out} (sections: {', '.join(sections) or 'none'})\n")


if __name__ == "__main__":
    main()