import MarkdownIt from "markdown-it";
import { type Html, trusted } from "./html.ts";
import { pathUrl } from "./layout.ts";

/** Where a Markdown file lives, so its relative links and images can point at the repo's raw files. */
export type MarkdownBase = { repo: string; dir: string; at: string | null };

const SCHEME = /^([a-z][a-z0-9+.-]*):/i;
const ALLOWED_SCHEMES = ["http", "https", "mailto"];

// §25: raw HTML is off, so markdown-it's own escaping is the whole sanitizer.
const md = new MarkdownIt({ html: false, linkify: true });
md.validateLink = (url) => {
	const scheme = SCHEME.exec(url.trim());
	return !scheme || ALLOWED_SCHEMES.includes((scheme[1] as string).toLowerCase());
};
md.core.ruler.push("kepr_raw_links", (state) => {
	const base = state.env as MarkdownBase;
	const visit = (tokens: typeof state.tokens) => {
		for (const token of tokens) {
			const attr = token.type === "link_open" ? "href" : token.type === "image" ? "src" : null;
			const url = attr && token.attrGet(attr);
			if (attr && typeof url === "string") token.attrSet(attr, rawUrl(url, base));
			if (token.children) visit(token.children);
		}
	};
	visit(state.tokens);
});

function rawUrl(url: string, base: MarkdownBase): string {
	if (SCHEME.test(url) || url.startsWith("#") || url.startsWith("//")) return url;
	// A leading `/` means the repo root, as it does on a code host.
	const resolved = new URL(url, `http://repo/${base.dir ? `${pathUrl(base.dir)}/` : ""}`);
	const at = base.at ? `?at=${base.at}` : "";
	return `/${base.repo}/raw${resolved.pathname}${at}${resolved.hash}`;
}

export function renderMarkdown(source: string, base: MarkdownBase): Html {
	return trusted(md.render(source, base));
}

/** The README's first paragraph (§21.4), as markup and as plain text for link previews. */
export function firstParagraph(
	source: string,
	base: MarkdownBase,
): { html: Html; text: string } | null {
	const tokens = md.parse(source, base);
	const index = tokens.findIndex((token) => token.type === "paragraph_open");
	const inline = tokens[index + 1];
	if (index < 0 || !inline?.children) return null;
	const text = inline.children
		.filter((token) => token.type === "text" || token.type === "code_inline")
		.map((token) => token.content)
		.join("");
	return { html: trusted(md.renderer.renderInline(inline.children, md.options, base)), text };
}

/** One line of Markdown with links and images flattened away, for a row that is itself a link. */
export function renderLine(source: string, base: MarkdownBase): Html {
	const tokens = md.parseInline(source, base)[0]?.children ?? [];
	const kept = tokens.filter((token) => !/^(link_open|link_close|image)$/.test(token.type));
	return trusted(md.renderer.renderInline(kept, md.options, base));
}