Merge branch 'db-export-filters' into 'main'

Allow filtering items exported by `deno task db:export`

Closes #186

See merge request soapbox-pub/ditto!463

deno.json

@@ -48,6 +48,7 @@
     "@std/streams": "jsr:@std/streams@^0.223.0",
     "comlink": "npm:comlink@^4.4.1",
     "comlink-async-generator": "npm:comlink-async-generator@^0.0.1",
+    "commander": "npm:commander@12.1.0",
     "deno-safe-fetch/load": "https://gitlab.com/soapbox-pub/deno-safe-fetch/-/raw/v1.0.0/load.ts",
     "deno.json": "./deno.json",
     "entities": "npm:entities@^4.5.0",

deno.lock

@@ -60,6 +60,7 @@
       "npm:comlink-async-generator": "npm:comlink-async-generator@0.0.1",
       "npm:comlink-async-generator@^0.0.1": "npm:comlink-async-generator@0.0.1",
       "npm:comlink@^4.4.1": "npm:comlink@4.4.1",
+      "npm:commander@12.1.0": "npm:commander@12.1.0",
       "npm:entities@^4.5.0": "npm:entities@4.5.0",
       "npm:fast-stable-stringify@^1.0.0": "npm:fast-stable-stringify@1.0.0",
       "npm:formdata-helper@^0.3.0": "npm:formdata-helper@0.3.0",
@@ -543,6 +544,10 @@
         "integrity": "sha512-yPVavfyCcRhmorC7rWlkHn15b4wDVgVmBA7kV4QVBsF7kv/9TKJAbAXVTxvTnwP8HHKjRCJDClKbciiYS7p0DQ==",
         "dependencies": {}
       },
+      "commander@12.1.0": {
+        "integrity": "sha512-Vw8qHK3bZM9y/P10u3Vib8o/DdkvA2OtPtZvD871QKjy74Wj1WSKFILMPRPSdUSx5RFK1arlJzEtA4PkFgnbuA==",
+        "dependencies": {}
+      },
       "cross-spawn@7.0.3": {
         "integrity": "sha512-iRDPJKUPVEND7dHPO8rkbOnPpyDygcDFtWjpeWNCgy8WP2rXcxXL8TskReQl6OrB2G7+UJrags1q15Fudc7G6w==",
         "dependencies": {
@@ -1904,6 +1909,7 @@
       "npm:@scure/base@^1.1.6",
       "npm:comlink-async-generator@^0.0.1",
       "npm:comlink@^4.4.1",
+      "npm:commander@12.1.0",
       "npm:entities@^4.5.0",
       "npm:fast-stable-stringify@^1.0.0",
       "npm:formdata-helper@^0.3.0",

scripts/db-export.test.ts

@@ -0,0 +1,94 @@
+import { assertEquals, assertThrows } from '@std/assert';
+import { buildFilter } from './db-export.ts';
+
+Deno.test('buildFilter should return an empty filter when no arguments are provided', () => {
+  const filter = buildFilter({});
+  assertEquals(Object.keys(filter).length, 0);
+});
+
+Deno.test('buildFilter should correctly handle valid authors', () => {
+  const filter = buildFilter({
+    authors: ['a'.repeat(64)],
+  });
+
+  assertEquals(filter.authors, ['a'.repeat(64)]);
+});
+
+Deno.test('buildFilter throws on invalid author pubkey', () => {
+  assertThrows(
+    () => {
+      buildFilter({
+        authors: ['invalid_pubkey'],
+      });
+    },
+    Error,
+    'ERROR: Invalid pubkey invalid_pubkey supplied.',
+  );
+});
+
+Deno.test('buildFilter should correctly handle valid ids', () => {
+  const filter = buildFilter({
+    ids: ['b'.repeat(64)],
+  });
+
+  assertEquals(filter.ids, ['b'.repeat(64)]);
+});
+
+Deno.test('buildFilter should throw on invalid event IDs', () => {
+  assertThrows(
+    () => {
+      buildFilter({
+        ids: ['invalid_id'],
+      });
+    },
+    Error,
+    'ERROR: Invalid event ID invalid_id supplied.',
+  );
+});
+
+Deno.test('buildFilter should correctly handle tag shortcuts', () => {
+  const filter = buildFilter({
+    d: 'value1',
+    e: 'a'.repeat(64),
+    p: 'b'.repeat(64),
+  });
+
+  assertEquals(filter['#d'], ['value1']);
+  assertEquals(filter['#e'], ['a'.repeat(64)]);
+  assertEquals(filter['#p'], ['b'.repeat(64)]);
+});
+
+Deno.test('buildFilter should correctly handle since and until args', () => {
+  const filter = buildFilter({
+    since: 1000,
+    until: 2000,
+  });
+
+  assertEquals(filter.since, 1000);
+  assertEquals(filter.until, 2000);
+});
+
+Deno.test('buildFilter should correctly handle search field', () => {
+  const filter = buildFilter({
+    search: 'search_term',
+  });
+
+  assertEquals(filter.search, 'search_term');
+});
+
+Deno.test('buildFilter should correctly handle tag k-v pairs', () => {
+  const filter = buildFilter({
+    tags: ['tag1=value1', 'tag2=value2'],
+  });
+
+  assertEquals(filter['#tag1'], ['value1']);
+  assertEquals(filter['#tag2'], ['value2']);
+});
+
+Deno.test('buildFilter should correctly handle limit specifier', () => {
+  const filter = buildFilter({
+    limit: 10,
+  });
+
+  assertEquals(filter.limit, 10);
+});

scripts/db-export.ts

@@ -1,24 +1,158 @@
 import { Storages } from '@/storages.ts';
+import { NostrFilter } from '@nostrify/nostrify';
+import { Command, InvalidOptionArgumentError } from 'commander';
 
-const store = await Storages.db();
-
-console.warn('Exporting events...');
-
-let count = 0;
-
-for await (const msg of store.req([{}])) {
-  if (msg[0] === 'EOSE') {
-    break;
-  }
-  if (msg[0] === 'EVENT') {
-    console.log(JSON.stringify(msg[2]));
-    count++;
-  }
-  if (msg[0] === 'CLOSED') {
-    console.error('Database closed unexpectedly');
-    break;
-  }
+interface ExportFilter {
+  authors?: string[];
+  ids?: string[];
+  kinds?: number[];
+  limit?: number;
+  search?: string;
+  /**
+   * Array of `key=value` pairs.
+   */
+  tags?: string[];
+  since?: number;
+  until?: number;
+  /**
+   * shortcut for `--tag d=<value>`
+   */
+  d?: string;
+  /**
+   * shortcut for `--tag e=<value>`
+   */
+  e?: string;
+  /**
+   * shortcut for `--tag p=<value>`
+   */
+  p?: string;
 }
 
-console.warn(`Exported ${count} events`);
-Deno.exit();
+function safeParseInt(s: string) {
+  const n = parseInt(s);
+  if (isNaN(n)) throw new InvalidOptionArgumentError('Not a number.');
+  return n;
+}
+
+function findInvalid(arr: string[], predicate = (v: string) => !/[a-f0-9]{64}/.test(v)) {
+  return arr.find(predicate);
+}
+
+function die(code: number, ...args: any[]) {
+  console.error(...args);
+  Deno.exit(code);
+}
+
+function tagFilterShortcut(name: 'd' | 'e' | 'p', value: string) {
+  const val = [value];
+  if (findInvalid(val)) throw new Error(`ERROR: Invalid value supplied for ${name}-tag.`);
+  return val;
+}
+
+export function buildFilter(args: ExportFilter) {
+  const filter: NostrFilter = {};
+  const { authors, ids, kinds, d, e, limit, p, search, since, until, tags } = args;
+  if (since) {
+    filter.since = since;
+  }
+  if (until) {
+    filter.until = until;
+  }
+  if (authors && authors.length) {
+    const invalid = findInvalid(authors);
+    if (invalid) throw new Error(`ERROR: Invalid pubkey ${invalid} supplied.`);
+    filter.authors = authors;
+  }
+  if (ids) {
+    const invalid = findInvalid(ids);
+    if (invalid) throw new Error(`ERROR: Invalid event ID ${invalid} supplied.`);
+    filter.ids = ids;
+  }
+  if (kinds && kinds.length) {
+    filter.kinds = kinds;
+  }
+  if (d) {
+    filter['#d'] = [d];
+  }
+  if (e) {
+    filter['#e'] = tagFilterShortcut('e', e);
+  }
+  if (p) {
+    filter['#p'] = tagFilterShortcut('e', p);
+  }
+  if (search) {
+    filter.search = search;
+  }
+  if (limit) {
+    filter.limit = limit;
+  }
+  if (tags) {
+    for (const val of tags) {
+      const [name, ...values] = val.split('=');
+      filter[`#${name}`] = [values.join('=')];
+    }
+  }
+
+  return filter;
+}
+
+async function exportEvents(args: ExportFilter) {
+  const store = await Storages.db();
+
+  let filter: NostrFilter = {};
+  try {
+    filter = buildFilter(args);
+  } catch (e) {
+    die(1, e.message || e.toString());
+  }
+
+  let count = 0;
+  for await (const msg of store.req([filter])) {
+    if (msg[0] === 'EOSE') {
+      break;
+    }
+    if (msg[0] === 'EVENT') {
+      console.log(JSON.stringify(msg[2]));
+      count++;
+    }
+    if (msg[0] === 'CLOSED') {
+      console.error('Database closed unexpectedly');
+      break;
+    }
+  }
+
+  console.warn(`Exported ${count} events`);
+}
+
+if (import.meta.main) {
+  const exporter = new Command()
+    .name('db:export')
+    .description('Export the specified set of events from the Ditto database, in JSONL format.')
+    .version('0.1.0')
+    .showHelpAfterError();
+
+  exporter
+    .option('-a, --authors <authors...>', 'Pubkeys of authors whose events you want to export.', [])
+    .option('-i, --ids <ids...>', 'IDs of events you want to export.', [])
+    .option(
+      '-k --kinds <kinds...>',
+      'Event kinds you want to export.',
+      (v: string, arr: number[]) => arr.concat([safeParseInt(v)]),
+      [],
+    )
+    .option(
+      '-t --tags <tag pairs...>',
+      'A list of key=value pairs of tags to search for events using. For tag values with spaces etc, simply quote the entire item, like `deno task db:export -t "name=A string with spaces in it"`.',
+      [],
+    )
+    .option('--search <search string>', 'A string to full-text search the db for.')
+    .option('-s --since <number>', 'The oldest time an exported event should be from.', safeParseInt)
+    .option('-u --until <number>', 'The newest time an exported event should be from.', safeParseInt)
+    .option('--limit <number>', 'Maximum number of events to export.', safeParseInt)
+    .option('-d <string>', 'Shortcut for `--tag d=<value>`.')
+    .option('-e <string>', 'Shortcut for `--tag e=<value>`.')
+    .option('-p <string>', 'Shortcut for `--tag p=<value>`.')
+    .action(exportEvents);
+
+  await exporter.parseAsync(Deno.args, { from: 'user' });
+}