This repository has no description
13 kB
339 lines
1/**
2 * island:discover
3 *
4 * Find connected components in the global network.cosmik.connection graph.
5 * Uses SQLite to cache network data so we don't rebuild each time.
6 *
7 * Usage:
8 * bun island:discover # find all connected components
9 * bun island:discover --refresh # force re-fetch all DIDs
10 * bun island:discover <island-id> # show full details for an island
11 * bun island:discover --min-size 5 # only show components with >= 5 nodes
12 * bun island:discover --json # machine-readable output
13 */
14
15import { resolve, dirname } from 'node:path';
16import { fileURLToPath } from 'node:url';
17import { createHash } from 'node:crypto';
18import { Database } from 'bun:sqlite';
19import { loadDotEnv } from './cli-utils.js';
20import { islandId, domainFromUrl, computeCentroid, type Component } from './island-shared.js';
21
22const __dirname = dirname(fileURLToPath(import.meta.url));
23await loadDotEnv(resolve(__dirname, '..', '.env'));
24
25const RELAY = process.env.ATPROTO_RELAY ?? 'https://bsky.network';
26const CARRY_DIR = process.env.CARRY_DIR ?? '/home/nandi/.local/share/carry-vault';
27const DB_PATH = process.env.ISLAND_DETECT_DB ?? resolve(CARRY_DIR, '.island-detect.db');
28const COLLECTION = 'network.cosmik.connection';
29const CACHE_TTL_MS = 60 * 60 * 1000; // 1 hour
30
31// --- SQLite ---
32
33function openDb(): Database {
34 const db = new Database(DB_PATH, { create: true });
35 db.exec('PRAGMA journal_mode = WAL');
36 db.exec('PRAGMA synchronous = NORMAL');
37 db.exec(`
38 CREATE TABLE IF NOT EXISTS edges (
39 source TEXT NOT NULL,
40 target TEXT NOT NULL,
41 relation TEXT,
42 did TEXT NOT NULL,
43 fetched_at TEXT NOT NULL,
44 PRIMARY KEY (source, target, did)
45 );
46 CREATE TABLE IF NOT EXISTS dids (
47 did TEXT PRIMARY KEY,
48 handle TEXT,
49 pds TEXT,
50 edge_count INTEGER DEFAULT 0,
51 fetched_at TEXT
52 );
53 CREATE INDEX IF NOT EXISTS idx_edges_source ON edges(source);
54 CREATE INDEX IF NOT EXISTS idx_edges_target ON edges(target);
55 `);
56 return db;
57}
58
59// --- Network discovery ---
60
61async function discoverDIDs(): Promise<string[]> {
62 const dids: string[] = [];
63 let cursor = '';
64 while (true) {
65 const url = new URL(`${RELAY}/xrpc/com.atproto.sync.listReposByCollection`);
66 url.searchParams.set('collection', COLLECTION);
67 url.searchParams.set('limit', '1000');
68 if (cursor) url.searchParams.set('cursor', cursor);
69 const res = await fetch(url.toString());
70 if (!res.ok) break;
71 const data = await res.json() as { repos: Array<{ did: string }>; cursor?: string };
72 for (const r of data.repos) dids.push(r.did);
73 cursor = data.cursor ?? '';
74 if (!cursor) break;
75 await new Promise((r) => setTimeout(r, 200));
76 }
77 return dids;
78}
79
80async function resolvePDS(did: string): Promise<string | null> {
81 try {
82 const res = await fetch(`https://plc.directory/${did}`);
83 if (!res.ok) return null;
84 const doc = await res.json() as { service?: Array<{ id: string; serviceEndpoint: string }> };
85 return doc.service?.find((s) => s.id === '#atproto_pds')?.serviceEndpoint ?? null;
86 } catch { return null; }
87}
88
89async function resolveHandle(did: string): Promise<string | null> {
90 try {
91 const res = await fetch(`https://plc.directory/${did}`);
92 if (!res.ok) return null;
93 const doc = await res.json() as { alsoKnownAs?: string[] };
94 return doc.alsoKnownAs?.find((h) => h.startsWith('at://'))?.replace('at://', '') ?? null;
95 } catch { return null; }
96}
97
98async function fetchConnectionRecords(pds: string, did: string): Promise<Array<{ source: string; target: string; relation?: string; atUri: string }>> {
99 const records: Array<{ source: string; target: string; relation?: string; atUri: string }> = [];
100 let cursor = '';
101 while (true) {
102 const url = new URL(`${pds}/xrpc/com.atproto.repo.listRecords`);
103 url.searchParams.set('repo', did);
104 url.searchParams.set('collection', COLLECTION);
105 url.searchParams.set('limit', '100');
106 if (cursor) url.searchParams.set('cursor', cursor);
107 const res = await fetch(url.toString());
108 if (!res.ok) break;
109 const data = await res.json() as {
110 records: Array<{ uri: string; value: { source?: string; target?: string; connectionType?: string } }>;
111 cursor?: string;
112 };
113 for (const r of data.records) {
114 if (r.value.source && r.value.target) {
115 records.push({ source: r.value.source, target: r.value.target, relation: r.value.connectionType, atUri: r.uri });
116 }
117 }
118 cursor = data.cursor ?? '';
119 if (!cursor) break;
120 await new Promise((r) => setTimeout(r, 100));
121 }
122 return records;
123}
124
125// --- Cache operations ---
126
127function isCached(db: Database, did: string): boolean {
128 const row = db.query('SELECT fetched_at FROM dids WHERE did = ?').get(did) as { fetched_at: string } | null;
129 if (!row) return false;
130 return Date.now() - new Date(row.fetched_at).getTime() < CACHE_TTL_MS;
131}
132
133function upsertDid(db: Database, did: string, handle: string | null, pds: string | null, edgeCount: number): void {
134 const now = new Date().toISOString();
135 db.query(`
136 INSERT INTO dids (did, handle, pds, edge_count, fetched_at)
137 VALUES (?, ?, ?, ?, ?)
138 ON CONFLICT(did) DO UPDATE SET handle=?, pds=?, edge_count=?, fetched_at=?
139 `).run(did, handle, pds, edgeCount, now, handle, pds, edgeCount, now);
140}
141
142function upsertEdges(db: Database, did: string, records: Array<{ source: string; target: string; relation?: string; atUri: string }>): void {
143 const now = new Date().toISOString();
144 // Add at_uri column if missing
145 try { db.exec('ALTER TABLE edges ADD COLUMN at_uri TEXT'); } catch {}
146 const insert = db.query(`
147 INSERT INTO edges (source, target, relation, did, at_uri, fetched_at)
148 VALUES (?, ?, ?, ?, ?, ?)
149 ON CONFLICT(source, target, did) DO UPDATE SET relation=?, at_uri=?, fetched_at=?
150 `);
151 // Delete stale edges for this DID before inserting
152 db.query('DELETE FROM edges WHERE did = ?').run(did);
153 for (const r of records) {
154 insert.run(r.source, r.target, r.relation ?? null, did, r.atUri, now, r.relation ?? null, r.atUri, now);
155 }
156}
157
158// --- Connected components ---
159
160function findComponents(db: Database): Component[] {
161 // Build adjacency list from all edges
162 const adj = new Map<string, Set<string>>();
163 const edgeDids = new Map<string, Set<string>>(); // node -> which DIDs have edges touching it
164
165 const rows = db.query('SELECT source, target, did FROM edges').all() as Array<{ source: string; target: string; did: string }>;
166 for (const row of rows) {
167 const s = row.source;
168 const t = row.target;
169 if (!adj.has(s)) adj.set(s, new Set());
170 if (!adj.has(t)) adj.set(t, new Set());
171 adj.get(s)!.add(t);
172 adj.get(t)!.add(s);
173 if (!edgeDids.has(s)) edgeDids.set(s, new Set());
174 if (!edgeDids.has(t)) edgeDids.set(t, new Set());
175 edgeDids.get(s)!.add(row.did);
176 edgeDids.get(t)!.add(row.did);
177 }
178
179 // BFS to find components
180 const visited = new Set<string>();
181 const components: Component[] = [];
182
183 for (const node of adj.keys()) {
184 if (visited.has(node)) continue;
185 const compNodes = new Set<string>();
186 const compDids = new Set<string>();
187 const queue = [node];
188 while (queue.length > 0) {
189 const current = queue.pop()!;
190 if (visited.has(current)) continue;
191 visited.add(current);
192 compNodes.add(current);
193 for (const did of edgeDids.get(current) ?? []) compDids.add(did);
194 for (const neighbor of adj.get(current) ?? []) {
195 if (!visited.has(neighbor)) queue.push(neighbor);
196 }
197 }
198 const lexmin = [...compNodes].sort()[0];
199 const centroid = computeCentroid(adj, compNodes);
200 components.push({ nodes: compNodes, dids: compDids, lexmin, centroid, islandId: islandId(centroid) });
201 }
202
203 return components.sort((a, b) => b.nodes.size - a.nodes.size);
204}
205
206// --- Main ---
207
208async function main(): Promise<void> {
209 const args = process.argv.slice(2);
210 const wantJson = args.includes('--json');
211 const refresh = args.includes('--refresh');
212 const targetId = args.find((a) => !a.startsWith('--'));
213 const minSizeArg = args.find((a) => a.startsWith('--min-size='));
214 const minSize = minSizeArg ? Number(minSizeArg.split('=')[1]) : 3;
215
216 const db = openDb();
217
218 console.error('Discovering network DIDs...');
219 const networkDids = await discoverDIDs();
220 console.error(`Found ${networkDids.length} DIDs`);
221
222 // Fetch edges for each DID (use cache unless --refresh)
223 let fetched = 0;
224 let cached = 0;
225 for (let i = 0; i < networkDids.length; i++) {
226 const did = networkDids[i];
227 if (!refresh && isCached(db, did)) {
228 cached++;
229 continue;
230 }
231 console.error(`[${i + 1}/${networkDids.length}] Fetching ${did}...`);
232 const pds = await resolvePDS(did);
233 if (!pds) { console.error(' Could not resolve PDS'); continue; }
234 const handle = await resolveHandle(did);
235 const records = await fetchConnectionRecords(pds, did);
236 upsertDid(db, did, handle, pds, records.length);
237 upsertEdges(db, did, records);
238 fetched++;
239 }
240 console.error(`Fetched: ${fetched}, Cached: ${cached}`);
241
242 // Stats
243 const totalEdges = (db.query('SELECT COUNT(*) as c FROM edges').get() as { c: number }).c;
244 const totalDids = (db.query('SELECT COUNT(*) as c FROM dids').get() as { c: number }).c;
245 const uniqueUrls = new Set<string>();
246 const urlRows = db.query('SELECT DISTINCT source FROM edges UNION SELECT DISTINCT target FROM edges').all() as Array<{ source: string }>;
247 for (const r of urlRows) uniqueUrls.add(r.source);
248
249 console.error(`Database: ${totalEdges} edges, ${totalDids} DIDs, ${uniqueUrls.size} unique URLs`);
250
251 // Find connected components
252 const components = findComponents(db);
253 const filtered = components.filter((c) => c.nodes.size >= minSize);
254
255 // Show a specific island by ID
256 if (targetId) {
257 const comp = components.find((c) => c.islandId === targetId);
258 if (!comp) {
259 console.error(`Island ${targetId} not found.`);
260 // Suggest similar IDs
261 const similar = filtered.filter((c) => c.islandId.startsWith(targetId.slice(0, 4)));
262 if (similar.length > 0) {
263 console.error(`Similar IDs:`);
264 for (const s of similar.slice(0, 5)) console.error(` ${s.islandId} ${s.nodes.size} nodes ${domainFromUrl(s.centroid ?? s.lexmin)}`);
265 } else {
266 console.error(`Run without args to list available islands.`);
267 }
268 process.exit(1);
269 }
270 const didPlaceholders = [...comp.dids].map(() => '?').join(',');
271 const didRows = db.query(`SELECT did, handle, edge_count FROM dids WHERE did IN (${didPlaceholders})`).all(...comp.dids) as Array<{ did: string; handle: string | null; edge_count: number }>;
272 if (wantJson) {
273 console.log(JSON.stringify({
274 id: comp.islandId,
275 centroid: comp.centroid,
276 lexmin: comp.lexmin,
277 nodeCount: comp.nodes.size,
278 didCount: comp.dids.size,
279 nodes: [...comp.nodes].sort(),
280 dids: didRows,
281 }, null, 2));
282 } else {
283 console.log(`Island ${comp.islandId}: ${comp.nodes.size} nodes, ${comp.dids.size} DIDs`);
284 console.log(`Centroid: ${comp.centroid}`);
285 console.log('\nNodes:');
286 for (const url of [...comp.nodes].sort()) console.log(` ${url}`);
287 console.log('\nDIDs:');
288 for (const d of didRows) console.log(` ${d.handle ?? d.did} (${d.edge_count} edges)`);
289 }
290 db.close();
291 return;
292 }
293
294 if (wantJson) {
295 console.log(JSON.stringify({
296 totalEdges,
297 totalDids,
298 uniqueUrls: uniqueUrls.size,
299 componentCount: filtered.length,
300 components: filtered.map((c) => ({
301 id: c.islandId,
302 centroid: c.centroid,
303 lexmin: c.lexmin,
304 nodeCount: c.nodes.size,
305 didCount: c.dids.size,
306 sampleNodes: [...c.nodes].sort().slice(0, 10),
307 dids: [...c.dids],
308 })),
309 }, null, 2));
310 db.close();
311 return;
312 }
313
314 console.log(`\nIsland discover: ${totalDids} DIDs, ${totalEdges} edges, ${uniqueUrls.size} unique URLs`);
315 console.log(`Connected components (>= ${minSize} nodes): ${filtered.length}\n`);
316
317 for (const comp of filtered.slice(0, 30)) {
318 const domain = domainFromUrl(comp.centroid);
319 console.log(`${comp.islandId} ${comp.nodes.size} nodes, ${comp.dids.size} DIDs ${domain}`);
320 }
321
322 // Show which component we're in (if we can determine our DID)
323 const ourDid = process.env.BLUESKY_IDENTIFIER
324 ? (db.query('SELECT did FROM dids WHERE handle = ? OR did LIKE ?').get(process.env.BLUESKY_IDENTIFIER, `%${process.env.BLUESKY_IDENTIFIER}%`) as { did: string } | null)
325 : null;
326 if (ourDid) {
327 const ourComp = components.find((c) => c.dids.has(ourDid.did));
328 if (ourComp) {
329 console.log(`\nOur island: ${ourComp.islandId} (${ourComp.nodes.size} nodes, shared with ${ourComp.dids.size - 1} other DIDs)`);
330 }
331 }
332
333 db.close();
334}
335
336main().catch((err) => {
337 console.error('Fatal:', err instanceof Error ? err.message : String(err));
338 process.exit(1);
339});