Gatsby integration¶

"""Pre-build: generates SEO JSON manifest for Gatsby."""
from __future__ import annotations

import json
import re
import sys
from pathlib import Path

import yaml
from seoslug import (
    SEOConfig,
    URLPolicy,
    SEOEntity,
    OGImage,
    Robots,
    build_seo_payload,
)

PROJECT_ROOT = Path(__file__).resolve().parent.parent
CONTENT_DIR = PROJECT_ROOT / "content"
SITE_URL = "https://yoursite.com"
SITE_NAME = "Your Site"

SEO_CONFIG = SEOConfig(
    canonical_host="yoursite.com",
    public_base_url=SITE_URL,
    url_policy=URLPolicy(
        enforce_https=True,
        lowercase_paths=True,
        trailing_slash="always",
    ),
    site_name=SITE_NAME,
    default_og_image=OGImage(
        url=f"{SITE_URL}/images/default-og.png",
        width=1200,
        height=630,
    ),
    publisher_name="Your Name",
    title_template="{title} | Your Site",
    default_robots=Robots(index=True, follow=True),
    locale="en_US",
    twitter_site="@yourhandle",
)

FM_RE = re.compile(
    r"^-{3}[ \r\t]*?\n(.*?\r?\n)(?:\.{3}|-{3})[ \r\t]*\n",
    re.UNICODE | re.DOTALL,
)


def route_path_from_file(filepath: Path) -> str:
    rel = filepath.relative_to(CONTENT_DIR)
    parts = list(rel.parts)
    if parts[-1] in ("index.md", "_index.md"):
        parts.pop()
        if not parts:
            return "/"
        return "/" + "/".join(parts) + "/"
    stem = Path(*parts).with_suffix("")
    return "/" + str(stem) + "/"


def extract_excerpt(body: str, max_chars: int = 160) -> str:
    body = re.sub(r"^#\s+.*\n?", "", body, count=1).strip()
    para = re.split(r"\n\s*\n", body, maxsplit=1)[0].strip()
    clean = re.sub(r"[`*_\[\]()>|#{}]", "", para)
    clean = re.sub(r"\s+", " ", clean).strip()
    if not clean:
        return ""
    if len(clean) <= max_chars:
        return clean
    break_at = clean.rfind(" ", 0, max_chars)
    return clean[:break_at] + "..." if break_at > 0 else clean[:max_chars] + "..."


def main() -> int:
    manifest: dict[str, dict] = {}
    for md_file in sorted(CONTENT_DIR.rglob("*.md")):
        text = md_file.read_text(encoding="utf-8")
        meta: dict = {}
        body = text
        if text.startswith("---"):
            m = FM_RE.match(text)
            if m:
                try:
                    parsed = yaml.safe_load(m.group(1))
                    if isinstance(parsed, dict):
                        meta = parsed
                except Exception:
                    pass
                body = text[m.end():].lstrip("\n")

        route = route_path_from_file(md_file)
        title = meta.get("title") or ""
        if not title:
            m = re.search(r"^#\s+(.+)", body, re.MULTILINE)
            if m:
                title = re.sub(r"[`*_]", "", m.group(1).strip())
        description = meta.get("description") or extract_excerpt(body)

        entity = SEOEntity(
            entity_type="page",
            title=title or SITE_NAME,
            excerpt=description or None,
        )
        payload = build_seo_payload(entity, route, SEO_CONFIG)
        if payload is None:
            continue
        manifest[route] = payload.to_dict()

    output = PROJECT_ROOT / "static" / "seo-manifest.json"
    output.write_text(json.dumps(manifest, indent=2, ensure_ascii=False))
    print(f"Written {output} ({len(manifest)} routes)")
    return 0


if __name__ == "__main__":
    sys.exit(main())