This repository has no description
4.1 kB
126 lines
1/**
2 * Shared island utilities: canonical ID, component types, BFS, DB access.
3 */
4
5import { resolve } from 'node:path';
6import { createHash } from 'node:crypto';
7import { Database } from 'bun:sqlite';
8
9const CARRY_DIR = process.env.CARRY_DIR ?? '/home/nandi/.local/share/carry-vault';
10const DB_PATH = process.env.ISLAND_DETECT_DB ?? resolve(CARRY_DIR, '.island-detect.db');
11
12export interface Component {
13 nodes: Set<string>;
14 dids: Set<string>;
15 lexmin: string;
16 islandId: string;
17}
18
19export interface EdgeRow {
20 source: string;
21 target: string;
22 relation: string | null;
23 did: string;
24}
25
26// --- Canonical island ID: first 12 hex chars of SHA-256(lexmin) ---
27
28export function islandId(lexmin: string): string {
29 return createHash('sha256').update(lexmin).digest('hex').slice(0, 12);
30}
31
32// --- SQLite ---
33
34export function openDb(): Database {
35 return new Database(DB_PATH, { readonly: true });
36}
37
38// --- Connected components (BFS) ---
39
40export function findComponents(db: Database): Component[] {
41 const adj = new Map<string, Set<string>>();
42 const edgeDids = new Map<string, Set<string>>();
43
44 const rows = db.query('SELECT source, target, did FROM edges').all() as Array<{ source: string; target: string; did: string }>;
45 for (const row of rows) {
46 const s = row.source;
47 const t = row.target;
48 if (!adj.has(s)) adj.set(s, new Set());
49 if (!adj.has(t)) adj.set(t, new Set());
50 adj.get(s)!.add(t);
51 adj.get(t)!.add(s);
52 if (!edgeDids.has(s)) edgeDids.set(s, new Set());
53 if (!edgeDids.has(t)) edgeDids.set(t, new Set());
54 edgeDids.get(s)!.add(row.did);
55 edgeDids.get(t)!.add(row.did);
56 }
57
58 const visited = new Set<string>();
59 const components: Component[] = [];
60
61 for (const node of adj.keys()) {
62 if (visited.has(node)) continue;
63 const compNodes = new Set<string>();
64 const compDids = new Set<string>();
65 const queue = [node];
66 while (queue.length > 0) {
67 const current = queue.pop()!;
68 if (visited.has(current)) continue;
69 visited.add(current);
70 compNodes.add(current);
71 for (const did of edgeDids.get(current) ?? []) compDids.add(did);
72 for (const neighbor of adj.get(current) ?? []) {
73 if (!visited.has(neighbor)) queue.push(neighbor);
74 }
75 }
76 const lexmin = [...compNodes].sort()[0];
77 components.push({ nodes: compNodes, dids: compDids, lexmin, islandId: islandId(lexmin) });
78 }
79
80 return components.sort((a, b) => b.nodes.size - a.nodes.size);
81}
82
83/** Find a component by its island ID (lexmin hash). */
84export function findComponentById(db: Database, targetId: string): Component | undefined {
85 return findComponents(db).find((c) => c.islandId === targetId);
86}
87
88/** Fetch edges for a component from the DB. */
89export function fetchComponentEdges(db: Database, component: Component): EdgeRow[] {
90 const nodePlaceholders = [...component.nodes].map(() => '?').join(',');
91 const nodeValues = [...component.nodes];
92 const edges = db.query(
93 `SELECT source, target, relation, did FROM edges WHERE source IN (${nodePlaceholders}) OR target IN (${nodePlaceholders})`,
94 ).all(...nodeValues, ...nodeValues) as EdgeRow[];
95
96 // Deduplicate (same source+target from multiple DIDs)
97 const seen = new Set<string>();
98 const unique: EdgeRow[] = [];
99 for (const e of edges) {
100 const key = `${e.source}::${e.target}`;
101 if (!seen.has(key)) {
102 seen.add(key);
103 unique.push(e);
104 }
105 }
106 return unique;
107}
108
109/** Resolve handles for a set of DIDs. */
110export function resolveDidHandles(db: Database, dids: Set<string>): Map<string, string | null> {
111 const handles = new Map<string, string | null>();
112 if (dids.size === 0) return handles;
113 const placeholders = [...dids].map(() => '?').join(',');
114 const rows = db.query(`SELECT did, handle FROM dids WHERE did IN (${placeholders})`).all(...dids) as Array<{ did: string; handle: string | null }>;
115 for (const r of rows) handles.set(r.did, r.handle);
116 return handles;
117}
118
119/** Extract domain from a URL for display. */
120export function domainFromUrl(url: string): string {
121 try {
122 return new URL(url).hostname.replace(/^www\./, '');
123 } catch {
124 return url;
125 }
126}