#!/usr/bin/env node
/**
 * @file program script: does a cache hold everything in a diigo RSS
 * export? Generates RSS from the cache's items (src/rss.ts) and compares
 * it, item by item keyed on link, with the export: title, tags,
 * description, and highlights + sticky notes. A program script, so it
 * reaches the filesystem and the clock itself.
 *
 *     node scripts/compare-export.ts <cache.sqlite> <export.xml> <out-dir>
 *
 * Prints counts only; writes <out-dir>/from-cache.rss and
 * <out-dir>/report.json (mismatched links, by category). diigo's export
 * is not well-formed XML, so it is read leniently, item by item.
 */
import { readFileSync, writeFileSync } from "node:fs";
import { join } from "node:path";
import { DatabaseSync } from "node:sqlite";
import { itemDescription, toRss } from "../src/rss.ts";

type Item = Record<string, unknown>;

const [dbPath, exportPath, outDir] = process.argv.slice(2);
if (!dbPath || !exportPath || !outDir) {
	console.error("usage: node scripts/compare-export.ts <cache.sqlite> <export.xml> <out-dir>");
	process.exit(2);
}

const FORBIDDEN = /[\u0000-\u0008\u000B\u000C\u000E-\u001F]/g;
const clean = (s: string) => s.replace(FORBIDDEN, "");

// the cache: every item it holds, newest update first
const db = new DatabaseSync(dbPath, { readOnly: true });
const ours = (db.prepare("select json from items order by updated desc").all() as { json: string }[]).map(
	(r) => JSON.parse(r.json) as Item,
);
writeFileSync(join(outDir, "from-cache.rss"), toRss(ours, { title: "from the cache", pubDate: new Date().toUTCString() }));

// the export, leniently
const field = (it: string, name: string) =>
	it.match(new RegExp(`<${name}><!\\[CDATA\\[([\\s\\S]*?)\\]\\]></${name}>`))?.[1] ?? "";
const theirs = [...readFileSync(exportPath, "utf8").matchAll(/<item>([\s\S]*?)<\/item>/g)].map((m) => ({
	title: clean(field(m[1], "title")),
	link: clean(field(m[1], "link")),
	description: clean(field(m[1], "description")),
}));

const byUrl = new Map<string, Item[]>();
for (const it of ours) {
	const url = String(it.url ?? "");
	byUrl.set(url, [...(byUrl.get(url) ?? []), it]);
}

const PARTS = /^Tags:([\s\S]*?);Description:([\s\S]*?);Annotations and Sticky notes:([\s\S]*)$/;
const report: Record<string, string[]> = {};
const note = (category: string, link: string) => (report[category] ??= []).push(link);
let complete = 0;

for (const t of theirs) {
	const candidates = byUrl.get(t.link);
	if (!candidates) {
		note("missing from cache", t.link);
		continue;
	}
	// a link bookmarked twice: any candidate that matches will do
	const verdicts = candidates.map((o) => {
		const problems: string[] = [];
		const title = clean(String(o.title ?? ""));
		if (title !== t.title && !(t.title.length === 255 && title.startsWith(t.title))) problems.push("title differs");
		const mine = clean(itemDescription(o)).match(PARTS);
		const yours = t.description.match(PARTS);
		if (!mine || !yours) problems.push("description unparsed");
		else {
			if (mine[1] !== yours[1]) problems.push("tags differ");
			if (mine[2] !== yours[2]) problems.push("description differs");
			if (mine[3] !== yours[3]) problems.push("annotations differ");
		}
		return problems;
	});
	const best = verdicts.reduce((a, b) => (b.length < a.length ? b : a));
	if (best.length === 0) complete += 1;
	for (const p of best) note(p, t.link);
}

const exportLinks = new Set(theirs.map((t) => t.link));
for (const url of byUrl.keys()) if (!exportLinks.has(url)) note("in cache, not in export", url);

writeFileSync(join(outDir, "report.json"), JSON.stringify(report, null, 1));
console.log({
	export: theirs.length,
	cache: ours.length,
	complete,
	...Object.fromEntries(Object.entries(report).map(([k, v]) => [k, v.length])),
});