rspace-online/modules/rnotes/mod.ts

/**
 * rNotes module — vault sync and browse for Obsidian and Logseq.
 *
 * Replaces the old full-editor module (~1784 lines).
 * Rich editing is now in rDocs. This module handles:
 *   - ZIP vault uploads (Obsidian / Logseq)
 *   - Automerge metadata storage (titles, tags, hashes, wikilinks)
 *   - On-demand note content served from ZIP on disk
 *   - Graph data for wikilink visualization
 *   - Browser-extension compat shims → rdocs
 */

import { Hono } from "hono";
import * as Automerge from "@automerge/automerge";
import JSZip from "jszip";
import { createHash } from "crypto";
import { mkdir, writeFile, readFile, unlink } from "fs/promises";
import { join } from "path";
import { renderShell } from "../../server/shell";
import { getModuleInfoList } from "../../shared/module";
import type { RSpaceModule, SpaceLifecycleContext } from "../../shared/module";
import { resolveDataSpace } from "../../shared/scope-resolver";
import { verifyToken, extractToken } from "../../server/auth";
import { renderLanding } from "./landing";
import { vaultSchema, vaultDocId } from "./schemas";
import type { VaultDoc, VaultNoteMeta } from "./schemas";
import type { SyncServer } from "../../server/local-first/sync-server";

// ── SyncServer ref (set during onInit) ──

let _syncServer: SyncServer | null = null;

// ── Constants ──

const VAULT_UPLOAD_DIR = "/data/files/uploads/vaults";

// ── Helpers ──

/** Find all vault docIds for a space by scanning known prefix. */
function findVaultDocIds(space: string): string[] {
	if (!_syncServer) return [];
	const prefix = `${space}:rnotes:vaults:`;
	return _syncServer.getDocIds().filter((id) => id.startsWith(prefix));
}

interface ParsedNote {
	path: string;
	title: string;
	tags: string[];
	frontmatter: Record<string, unknown>;
	contentHash: string;
	sizeBytes: number;
	wikilinks: string[];
}

/**
 * Parse a ZIP buffer and extract metadata for each .md file.
 * Returns an array of parsed note metadata objects.
 */
async function parseVaultZip(
	buffer: ArrayBuffer,
	source: "obsidian" | "logseq",
): Promise<ParsedNote[]> {
	const zip = await JSZip.loadAsync(buffer);
	const results: ParsedNote[] = [];

	for (const [relativePath, zipEntry] of Object.entries(zip.files)) {
		if (zipEntry.dir) continue;
		if (!relativePath.endsWith(".md") && !relativePath.endsWith(".markdown")) continue;
		// Skip hidden / system files
		if (relativePath.includes("/.") || relativePath.startsWith(".")) continue;

		const content = await zipEntry.async("string");
		const sizeBytes = content.length;
		const contentHash = createHash("sha256").update(content).digest("hex");

		// Parse frontmatter (YAML between --- delimiters)
		const frontmatter: Record<string, unknown> = {};
		let bodyStart = 0;
		if (content.startsWith("---")) {
			const end = content.indexOf("\n---", 3);
			if (end !== -1) {
				const fmBlock = content.slice(3, end).trim();
				bodyStart = end + 4;
				// Simple YAML key: value parser (no arrays/nested)
				for (const line of fmBlock.split("\n")) {
					const colon = line.indexOf(":");
					if (colon === -1) continue;
					const key = line.slice(0, colon).trim();
					const val = line.slice(colon + 1).trim();
					if (key) frontmatter[key] = val;
				}
			}
		}

		// Extract tags from frontmatter
		let tags: string[] = [];
		if (typeof frontmatter.tags === "string") {
			tags = frontmatter.tags
				.replace(/[\[\]]/g, "")
				.split(",")
				.map((t) => t.trim().replace(/^#/, ""))
				.filter(Boolean);
		}
		// Also pick up inline #tags from body
		const body = content.slice(bodyStart);
		const inlineTags = [...body.matchAll(/#([\w/-]+)/g)].map((m) => m[1]);
		tags = [...new Set([...tags, ...inlineTags])];

		// Derive title: first H1 heading, or frontmatter title, or filename
		let title = (frontmatter.title as string) ?? "";
		if (!title) {
			const h1 = body.match(/^#\s+(.+)/m);
			title = h1 ? h1[1].trim() : relativePath.split("/").pop()!.replace(/\.md$/i, "");
		}

		// Extract wikilinks [[Target]] and [[Target|Alias]]
		const wikilinks = [...body.matchAll(/\[\[([^\]|]+)(?:\|[^\]]+)?\]\]/g)].map((m) =>
			m[1].trim(),
		);

		// Logseq: strip "pages/" prefix if present
		const normalizedPath =
			source === "logseq" && relativePath.startsWith("pages/")
				? relativePath.slice(6)
				: relativePath;

		results.push({
			path: normalizedPath,
			title,
			tags,
			frontmatter,
			contentHash,
			sizeBytes,
			wikilinks,
		});
	}

	return results;
}

// ── Routes ──

const routes = new Hono();

// GET / — render landing or app shell
routes.get("/", (c) => {
	const space = c.req.param("space") || "demo";
	if (!space || space === "rnotes.online") {
		return c.html(renderLanding());
	}
	return c.html(
		renderShell({
			title: `${space} — rNotes | rSpace`,
			moduleId: "rnotes",
			spaceSlug: space,
			modules: getModuleInfoList(),
			theme: "dark",
			body: `<folk-notes-app space="${space}"></folk-notes-app>`,
			scripts: `<script type="module" src="/modules/rnotes/folk-notes-app.js?v=9"></script>`,
		}),
	);
});

// POST /api/vault/upload — accept ZIP + metadata, parse, store
routes.post("/api/vault/upload", async (c) => {
	const token = extractToken(c.req.raw.headers);
	if (!token) return c.json({ error: "Unauthorized" }, 401);
	try {
		await verifyToken(token);
	} catch {
		return c.json({ error: "Invalid token" }, 401);
	}

	const space = c.req.param("space") || "demo";
	const dataSpace = resolveDataSpace("rnotes", space);

	let formData: FormData;
	try {
		formData = await c.req.formData();
	} catch {
		return c.json({ error: "Expected multipart/form-data" }, 400);
	}

	const file = formData.get("file") as File | null;
	const name = (formData.get("name") as string | null) ?? "My Vault";
	const source = ((formData.get("source") as string | null) ?? "obsidian") as
		| "obsidian"
		| "logseq";

	if (!file) return c.json({ error: "Missing file" }, 400);
	if (!file.name.endsWith(".zip")) return c.json({ error: "File must be a .zip" }, 400);

	const buffer = await file.arrayBuffer();
	const vaultId = crypto.randomUUID();

	// Parse ZIP metadata
	let notes: ParsedNote[];
	try {
		notes = await parseVaultZip(buffer, source);
	} catch (err) {
		return c.json({ error: `Failed to parse ZIP: ${(err as Error).message}` }, 400);
	}

	if (notes.length === 0) {
		return c.json({ error: "No markdown files found in ZIP" }, 400);
	}

	// Store ZIP to disk
	await mkdir(VAULT_UPLOAD_DIR, { recursive: true });
	const zipPath = join(VAULT_UPLOAD_DIR, `${vaultId}.zip`);
	await writeFile(zipPath, Buffer.from(buffer));

	// Build Automerge doc
	const docId = vaultDocId(dataSpace, vaultId);
	const now = Date.now();
	const totalSize = notes.reduce((acc, n) => acc + n.sizeBytes, 0);

	const doc = Automerge.change(Automerge.init<VaultDoc>(), "init vault", (d) => {
		d.meta = {
			module: "rnotes",
			collection: "vaults",
			version: 1,
			spaceSlug: dataSpace,
			createdAt: now,
		};
		d.vault = {
			id: vaultId,
			name,
			source,
			totalNotes: notes.length,
			totalSizeBytes: totalSize,
			lastSyncedAt: now,
			createdAt: now,
		};
		d.notes = {};
		d.wikilinks = {};
		for (const n of notes) {
			d.notes[n.path] = {
				path: n.path,
				title: n.title,
				tags: n.tags,
				contentHash: n.contentHash,
				sizeBytes: n.sizeBytes,
				lastModifiedAt: now,
				syncStatus: "synced",
				frontmatter: n.frontmatter as Record<string, any>,
			} satisfies VaultNoteMeta;
			if (n.wikilinks.length > 0) {
				d.wikilinks[n.path] = n.wikilinks;
			}
		}
	});

	_syncServer!.setDoc(docId, doc);

	return c.json({
		vaultId,
		name,
		source,
		totalNotes: notes.length,
		totalSizeBytes: totalSize,
	});
});

// GET /api/vault/list — list all vaults for a space
routes.get("/api/vault/list", (c) => {
	const space = c.req.param("space") || "demo";
	const dataSpace = resolveDataSpace("rnotes", space);

	const docIds = findVaultDocIds(dataSpace);
	const vaults = docIds
		.map((id) => _syncServer!.getDoc<VaultDoc>(id))
		.filter((d): d is VaultDoc => !!d)
		.map((d) => d.vault);

	return c.json({ vaults });
});

// GET /api/vault/:vaultId/status — sync status for a vault
routes.get("/api/vault/:vaultId/status", (c) => {
	const space = c.req.param("space") || "demo";
	const dataSpace = resolveDataSpace("rnotes", space);
	const vaultId = c.req.param("vaultId");

	const docId = vaultDocId(dataSpace, vaultId);
	const doc = _syncServer!.getDoc<VaultDoc>(docId);
	if (!doc) return c.json({ error: "Vault not found" }, 404);

	const notes = Object.values(doc.notes);
	const synced = notes.filter((n) => n.syncStatus === "synced").length;
	const conflicts = notes.filter((n) => n.syncStatus === "conflict").length;
	const localModified = notes.filter((n) => n.syncStatus === "local-modified").length;

	return c.json({
		vaultId,
		name: doc.vault.name,
		source: doc.vault.source,
		totalNotes: doc.vault.totalNotes,
		lastSyncedAt: doc.vault.lastSyncedAt,
		synced,
		conflicts,
		localModified,
	});
});

// GET /api/vault/:vaultId/notes — list notes (with ?folder= and ?search=)
routes.get("/api/vault/:vaultId/notes", (c) => {
	const space = c.req.param("space") || "demo";
	const dataSpace = resolveDataSpace("rnotes", space);
	const vaultId = c.req.param("vaultId");
	const { folder, search } = c.req.query();

	const docId = vaultDocId(dataSpace, vaultId);
	const doc = _syncServer!.getDoc<VaultDoc>(docId);
	if (!doc) return c.json({ error: "Vault not found" }, 404);

	let notes = Object.values(doc.notes);

	if (folder) {
		const prefix = folder.endsWith("/") ? folder : `${folder}/`;
		notes = notes.filter((n) => n.path.startsWith(prefix));
	}
	if (search) {
		const term = search.toLowerCase();
		notes = notes.filter(
			(n) =>
				n.title.toLowerCase().includes(term) ||
				n.path.toLowerCase().includes(term) ||
				n.tags.some((t) => t.toLowerCase().includes(term)),
		);
	}

	return c.json({
		vaultId,
		total: notes.length,
		notes: notes.sort((a, b) => b.lastModifiedAt - a.lastModifiedAt),
	});
});

// GET /api/vault/:vaultId/note/:path+ — read note content from ZIP
routes.get("/api/vault/:vaultId/note/*", async (c) => {
	const space = c.req.param("space") || "demo";
	const dataSpace = resolveDataSpace("rnotes", space);
	const vaultId = c.req.param("vaultId");

	// Extract wildcard path segment after /note/
	const url = new URL(c.req.url);
	const prefix = `/rnotes/api/vault/${vaultId}/note/`;
	const altPrefix = `/api/vault/${vaultId}/note/`;
	let notePath = url.pathname;
	if (notePath.includes(prefix)) {
		notePath = decodeURIComponent(notePath.slice(notePath.indexOf(prefix) + prefix.length));
	} else if (notePath.includes(altPrefix)) {
		notePath = decodeURIComponent(notePath.slice(notePath.indexOf(altPrefix) + altPrefix.length));
	}

	const docId = vaultDocId(dataSpace, vaultId);
	const doc = _syncServer!.getDoc<VaultDoc>(docId);
	if (!doc) return c.json({ error: "Vault not found" }, 404);
	if (!doc.notes[notePath]) return c.json({ error: "Note not found" }, 404);

	const zipPath = join(VAULT_UPLOAD_DIR, `${vaultId}.zip`);
	let zipBuffer: Buffer;
	try {
		zipBuffer = await readFile(zipPath);
	} catch {
		return c.json({ error: "Vault ZIP not found on disk" }, 404);
	}

	const zip = await JSZip.loadAsync(zipBuffer);
	const zipEntry = zip.file(notePath);
	if (!zipEntry) {
		// Logseq may store under pages/ prefix
		const altEntry = zip.file(`pages/${notePath}`);
		if (!altEntry) return c.json({ error: "File not in ZIP" }, 404);
		const content = await altEntry.async("string");
		return new Response(content, {
			headers: { "Content-Type": "text/markdown; charset=utf-8" },
		});
	}
	const content = await zipEntry.async("string");
	return new Response(content, {
		headers: { "Content-Type": "text/markdown; charset=utf-8" },
	});
});

// GET /api/vault/:vaultId/graph — wikilinks for graph visualization
routes.get("/api/vault/:vaultId/graph", (c) => {
	const space = c.req.param("space") || "demo";
	const dataSpace = resolveDataSpace("rnotes", space);
	const vaultId = c.req.param("vaultId");

	const docId = vaultDocId(dataSpace, vaultId);
	const doc = _syncServer!.getDoc<VaultDoc>(docId);
	if (!doc) return c.json({ error: "Vault not found" }, 404);

	// Build nodes + edges for graph viz
	const notes = Object.values(doc.notes);
	const nodeSet = new Set(notes.map((n) => n.path));
	const nodes = notes.map((n) => ({
		id: n.path,
		label: n.title,
		tags: n.tags,
	}));
	const edges: Array<{ source: string; target: string }> = [];

	for (const [sourcePath, targets] of Object.entries(doc.wikilinks)) {
		for (const target of targets) {
			// Try to find the matching note path (title match or path match)
			const targetPath =
				[...nodeSet].find(
					(p) =>
						p === target ||
						p.replace(/\.md$/i, "") === target ||
						p.split("/").pop()?.replace(/\.md$/i, "") === target,
				) ?? target;
			edges.push({ source: sourcePath, target: targetPath });
		}
	}

	return c.json({ vaultId, nodes, edges });
});

// DELETE /api/vault/:vaultId — remove vault doc + ZIP
routes.delete("/api/vault/:vaultId", async (c) => {
	const token = extractToken(c.req.raw.headers);
	if (!token) return c.json({ error: "Unauthorized" }, 401);
	try {
		await verifyToken(token);
	} catch {
		return c.json({ error: "Invalid token" }, 401);
	}

	const space = c.req.param("space") || "demo";
	const dataSpace = resolveDataSpace("rnotes", space);
	const vaultId = c.req.param("vaultId");

	const docId = vaultDocId(dataSpace, vaultId);
	const doc = _syncServer!.getDoc<VaultDoc>(docId);
	if (!doc) return c.json({ error: "Vault not found" }, 404);

	// Remove Automerge doc
	(_syncServer as any).deleteDoc?.(docId);

	// Remove ZIP from disk (best-effort)
	const zipPath = join(VAULT_UPLOAD_DIR, `${vaultId}.zip`);
	try {
		await unlink(zipPath);
	} catch {
		// Ignore if already gone
	}

	return c.json({ ok: true, vaultId });
});

// ── Browser extension compat shims ──
// Old extension POSTed to /api/notes and GET /api/notebooks.
// Redirect to rdocs equivalents so existing installs keep working.

routes.post("/api/notes", (c) => {
	const space = c.req.param("space") || "demo";
	return c.redirect(`/${space}/rdocs/api/notes`, 301);
});

routes.get("/api/notebooks", (c) => {
	const space = c.req.param("space") || "demo";
	return c.redirect(`/${space}/rdocs/api/notebooks`, 301);
});

// ── Module definition ──

export const notesModule: RSpaceModule = {
	id: "rnotes",
	name: "rNotes",
	icon: "🔗",
	description: "Vault sync and browse for Obsidian and Logseq",
	standaloneDomain: "rnotes.online",
	routes,

	scoping: {
		defaultScope: "space",
		userConfigurable: false,
	},

	docSchemas: [
		{
			pattern: "{space}:rnotes:vaults:{vaultId}",
			description: "Vault metadata — notes, tags, wikilinks (content in ZIP on disk)",
			init: vaultSchema.init,
		},
	],

	landingPage: renderLanding,

	onboardingActions: [
		{
			label: "Upload Vault ZIP",
			icon: "📦",
			description: "Export your Obsidian or Logseq vault as a ZIP and upload it here",
			type: "upload",
			upload: { accept: ".zip", endpoint: "/api/vault/upload" },
		},
	],

	outputPaths: [
		{
			path: "vaults",
			name: "Vaults",
			icon: "🗂️",
			description: "Synced Obsidian and Logseq vaults",
		},
	],

	async onInit({ syncServer }) {
		_syncServer = syncServer;
	},

	async onSpaceCreate(_ctx: SpaceLifecycleContext) {
		// Vaults are user-uploaded — no auto-create needed
	},
};

export default notesModule;

// ── MI Integration ──

export interface MIVaultNote {
	title: string;
	path: string;
	vaultName: string;
	tags: string[];
	lastModifiedAt: number;
}

/**
 * Return recently modified vault notes for the MI system prompt.
 * Queries all vault docs for the space and returns the most recent entries.
 */
export function getRecentVaultNotesForMI(space: string, limit = 10): MIVaultNote[] {
	if (!_syncServer) return [];

	const docIds = findVaultDocIds(space);
	const allNotes: MIVaultNote[] = [];

	for (const docId of docIds) {
		const doc = _syncServer.getDoc<VaultDoc>(docId);
		if (!doc) continue;
		const vaultName = doc.vault.name;
		for (const note of Object.values(doc.notes)) {
			allNotes.push({
				title: note.title,
				path: note.path,
				vaultName,
				tags: Array.from(note.tags),
				lastModifiedAt: note.lastModifiedAt,
			});
		}
	}

	return allNotes
		.sort((a, b) => b.lastModifiedAt - a.lastModifiedAt)
		.slice(0, limit);
}