Merge branch 'main' into fix/firms-drop-snpp

# Conflicts:
#	docs/CURRENT-STATE.md
This commit is contained in:
German Khokhlov committed 2026-11-09 21:41:01 +04:00
commit d90da158ea
593 files changed
+104676 -2274

No files matched your search

+181
View File
@@ -0,0 +1,181 @@
import { normalizeAcartiaSightings } from '../../src/layers/wildlifeSightings/records.js';
import { coalesceProxyRequest, readResponseJsonCapped } from './common/http.js';
import { makeRateLimiter, clientKey } from './common/rate-limit.js';
import { admitSameSiteKeyless } from './common/same-site.js';
import { providerUserAgent } from './common/userAgent.js';
const API_URL = 'https://acartia.io/api/v1/sightings/current';
// All views share the provider's seven-day snapshot.
const TTL_MS = 600_000;
const STALE_MAX_MS = 24 * 60 * 60_000;
const MAX_RESPONSE_BYTES = 2 * 1024 * 1024;
const TIMEOUT_MS = 15_000;
const COOLDOWN_MS = 60_000;
const COOLDOWN_MIN_MS = 5000;
const COOLDOWN_MAX_MS = 60 * 60_000;
const FAILURE_COOLDOWN_MS = 15_000;
/** Bound upstream Retry-After seconds or dates to five seconds–one hour. */
export function acartiaRetryCooldownMs(raw, nowMs) {
const value = typeof raw === 'string' ? raw.trim() : '';
if (value) {
const seconds = /^\d+(?:\.\d+)?$/.test(value) ? Number(value) : NaN;
const deadline =
!Number.isFinite(seconds) && /[a-z]/i.test(value)
? Date.parse(value)
: NaN;
const delay = Number.isFinite(seconds) ? seconds * 1000 : deadline - nowMs;
if (Number.isFinite(delay) && delay >= 0)
return Math.min(COOLDOWN_MAX_MS, Math.max(COOLDOWN_MIN_MS, delay));
}
return COOLDOWN_MS;
}
/** Salish Sea coverage. Precise coordinates and comments never enter cache. */
export function acartiaProxy({
fetchImpl = (...args) => globalThis.fetch(...args),
now = () => Date.now(),
cache = new Map(),
warn = (...args) => console.warn(...args),
} = {}) {
const inFlight = new Map();
const allow = makeRateLimiter({
windowMs: 60_000,
max: 60,
globalMax: 600,
now,
});
let cooldownUntil = 0;
let cooldownStatus = 429;
let warnedClamp = false;
const retrySeconds = () =>
Math.max(1, Math.ceil((cooldownUntil - now()) / 1000));
const unavailable = (status = 502) =>
Object.assign(new Error('acartia_unavailable'), {
status,
retryAfterS: retrySeconds(),
});
function coolDown(status, delay) {
const until = now() + delay;
if (until > cooldownUntil) {
cooldownUntil = until;
cooldownStatus = status;
} else if (status === 429 && now() < cooldownUntil) cooldownStatus = 429;
}
async function load() {
if (now() < cooldownUntil) throw unavailable(cooldownStatus);
const signal = AbortSignal.timeout(TIMEOUT_MS);
try {
const response = await fetchImpl(API_URL, {
signal,
redirect: 'error',
headers: {
Accept: 'application/json',
'User-Agent': providerUserAgent('wildlife-sightings'),
},
});
if (!response.ok) {
await response.body?.cancel().catch(() => {});
if (response.status === 429) {
const delay = acartiaRetryCooldownMs(
response.headers.get('retry-after'),
now(),
);
if (delay === COOLDOWN_MAX_MS && !warnedClamp) {
warnedClamp = true;
warn('[acartia-proxy] upstream cooldown capped at one hour');
}
coolDown(429, delay);
throw unavailable(429);
}
throw unavailable();
}
// This endpoint documents a JSON array, not a results/data envelope.
const payload = await readResponseJsonCapped(
response,
MAX_RESPONSE_BYTES,
signal,
);
const rows = normalizeAcartiaSightings(payload, now());
// Only a non-array is malformed. A quiet feed, or one whose rows all fall
// outside coverage or the species checks, is a valid empty snapshot.
if (!rows) throw new Error('invalid_snapshot');
return { fetchedAt: now(), rows };
} catch (error) {
if (error.status !== 429) coolDown(502, FAILURE_COOLDOWN_MS);
warn('[acartia-proxy] upstream refresh failed');
throw unavailable(error.status === 429 ? 429 : 502);
}
}
function age(entry) {
if (!entry) return Infinity;
const elapsed = now() - entry.savedAt;
return elapsed < 0 ? Infinity : elapsed;
}
async function acquire() {
const previous = cache.get('sightings');
if (age(previous) < TTL_MS) return { value: previous.value, stale: false };
if (age(previous) > STALE_MAX_MS) cache.delete('sightings');
try {
const { promise } = coalesceProxyRequest(
inFlight,
'sightings',
async () => {
const value = await load();
cache.set('sightings', { value, savedAt: now() });
return value;
},
);
return { value: await promise, stale: false };
} catch (error) {
if (age(previous) <= STALE_MAX_MS)
return { value: previous.value, stale: true };
throw error;
}
}
async function handler(req, res) {
const json = (status, value, { stale = false, retryAfterS } = {}) => {
if (res.destroyed) return;
res.writeHead(status, {
'Content-Type': 'application/json',
'Cache-Control': 'no-store',
...(status === 405 ? { Allow: 'GET' } : {}),
...(retryAfterS ? { 'Retry-After': String(retryAfterS) } : {}),
...(stale ? { 'X-Data-Stale': 'true' } : {}),
});
res.end(JSON.stringify(value));
};
if (admitSameSiteKeyless(req, res)) return;
if (req.method !== 'GET') return json(405, { error: 'method_not_allowed' });
const pathname = (req.url || '/').split('?')[0];
if (pathname !== '/sightings') return json(404, { error: 'unknown_route' });
if (!allow(clientKey(req)))
return json(429, { error: 'rate_limited' }, { retryAfterS: 60 });
try {
const { value, stale } = await acquire();
return json(200, stale ? { ...value, stale: true } : value, { stale });
} catch (error) {
return json(
error.status === 429 ? 429 : 502,
{ error: 'acartia_unavailable' },
{ retryAfterS: error.retryAfterS },
);
}
}
return {
name: 'acartia',
configureServer({ middlewares }) {
middlewares.use('/api/acartia', handler);
},
configurePreviewServer({ middlewares }) {
middlewares.use('/api/acartia', handler);
},
};
}
+419
View File
@@ -0,0 +1,419 @@
import test from 'node:test';
import assert from 'node:assert/strict';
import { acartiaProxy, acartiaRetryCooldownMs } from './acartia.js';
const NOW = Date.parse('2026-11-04T18:05:00Z');
const sighting = (id = 'synthetic-1') => ({
entry_id: id,
latitude: 48.12345,
longitude: -123.12345,
created: '2026-11-04 17:00:00',
type: 'Orca',
no_sighted: '5',
photo_url: 'https://provider.example/photos/synthetic.jpg',
data_source_name: 'Synthetic Whale Network',
data_source_entity: 'Community reports',
trusted: true,
data_source_comments: '[email protected] +1-555-000-0000',
});
function install(options = {}, hook = 'configureServer') {
let handler;
const plugin = acartiaProxy({ now: () => NOW, warn() {}, ...options });
assert.equal(plugin.name, 'acartia');
assert.equal(typeof plugin.configureServer, 'function');
assert.equal(typeof plugin.configurePreviewServer, 'function');
plugin[hook]({
middlewares: {
use(mount, callback) {
assert.equal(mount, '/api/acartia');
handler = callback;
},
},
});
return async (
url = '/sightings',
{ method = 'GET', peer = 'local', headers = {} } = {},
) => {
const res = {
headers: {},
writeHead(status, headers) {
this.status = status;
this.headers = headers;
},
setHeader(key, value) {
this.headers[key] = value;
},
end(value) {
this.status ??= this.statusCode;
this.body = JSON.parse(value);
},
};
await handler(
{ url, method, headers, socket: { remoteAddress: peer } },
res,
);
return res;
};
}
test('cache stores only supported marine mammals inside the coverage box', async () => {
const cache = new Map();
const request = install({
cache,
fetchImpl: async () =>
Response.json([
sighting('orca'),
{ ...sighting('seal'), type: 'Harbor seal' },
{ ...sighting('unknown'), type: 'Unknown' },
{ ...sighting('fish'), type: 'Whale shark' },
{ ...sighting('outside'), latitude: 37.8 },
{ ...sighting('rounded-edge'), latitude: 45.49999 },
]),
});
const result = await request();
assert.equal(result.status, 200);
for (const rows of [result.body.rows, cache.get('sightings').value.rows]) {
assert.deepEqual(
rows.map((row) => [row.stableId, row.cetacean]),
[
['acartia:orca', true],
['acartia:seal', false],
],
);
}
});
for (const hook of ['configureServer', 'configurePreviewServer']) {
test(`${hook}: documented array is normalized and coarsened before caching`, async () => {
const cache = new Map();
let calls = 0;
const request = install(
{
cache,
fetchImpl: async (url, options) => {
calls += 1;
assert.equal(url, 'https://acartia.io/api/v1/sightings/current');
assert.equal(options.redirect, 'error');
assert.ok(options.signal instanceof AbortSignal);
assert.equal(options.headers.Accept, 'application/json');
assert.equal(
options.headers['User-Agent'],
'gods-eye-view-wildlife-sightings-proxy/1.0 (+https://github.com/bilawalsidhu/gods-eye-view)',
);
return Response.json([sighting()]);
},
},
hook,
);
assert.equal(calls, 0);
assert.equal(cache.size, 0);
const result = await request(
'/sightings?url=https://attacker.example&latitude=1',
);
assert.equal(result.status, 200);
assert.equal(result.headers['Cache-Control'], 'no-store');
assert.equal(result.body.fetchedAt, NOW);
assert.equal(result.body.rows[0].stableId, 'acartia:synthetic-1');
assert.equal(result.body.rows[0].lat, 48.12);
assert.equal(result.body.rows[0].lon, -123.12);
assert.equal(result.body.rows[0].provider.name, 'Synthetic Whale Network');
const retained = cache.get('sightings');
assert.equal(retained.value.rows[0].lat, 48.12);
assert.equal(retained.value.rows[0].lon, -123.12);
for (const value of [result.body, [...cache.values()]])
assert.doesNotMatch(
JSON.stringify(value),
/private-observer|555|data_source_comments|48\.12345|123\.12345/,
);
assert.equal(calls, 1);
});
}
test('gate, method and fixed route run before cache or upstream', async () => {
const cache = new Map();
let calls = 0;
const request = install({
cache,
fetchImpl: async () => {
calls += 1;
return Response.json([sighting()]);
},
});
for (let i = 0; i < 65; i += 1) {
assert.equal(
(
await request('/sightings', {
headers: {
host: 'localhost:5173',
origin: 'https://foreign.example',
'sec-fetch-site': 'cross-site',
},
})
).status,
403,
);
}
const method = await request('/sightings', { method: 'POST' });
assert.equal(method.status, 405);
assert.equal(method.headers.Allow, 'GET');
for (const route of ['/', '/unknown', '/sightings/extra']) {
const result = await request(route);
assert.equal(result.status, 404);
assert.equal(result.body.error, 'unknown_route');
}
assert.equal(calls, 0);
assert.equal(cache.size, 0);
const sameOrigin = await request('/sightings', {
headers: {
host: 'gev.example',
origin: 'https://gev.example',
'sec-fetch-site': 'same-origin',
'x-forwarded-proto': 'https',
'x-forwarded-for': '203.0.113.9',
},
});
assert.equal(sameOrigin.status, 200);
});
test('request 61 is locally limited even with a fresh cache', async () => {
let calls = 0;
const request = install({
fetchImpl: async () => {
calls += 1;
return Response.json([sighting()]);
},
});
for (let i = 0; i < 60; i += 1) assert.equal((await request()).status, 200);
const limited = await request();
assert.equal(limited.status, 429);
assert.equal(limited.body.error, 'rate_limited');
assert.equal(limited.headers['Retry-After'], '60');
assert.equal(calls, 1);
});
test('one global snapshot single-flights and stays fresh for ten minutes', async () => {
let at = NOW;
let calls = 0;
let release;
const request = install({
now: () => at,
fetchImpl: async () => {
calls += 1;
if (calls === 1)
await new Promise((resolve) => {
release = resolve;
});
return Response.json([sighting()]);
},
});
const first = request();
const concurrent = request();
await new Promise(setImmediate);
assert.equal(calls, 1);
release();
assert.deepEqual((await first).body, (await concurrent).body);
at += 599_999;
await request();
assert.equal(calls, 1);
at += 1;
await request();
assert.equal(calls, 2);
});
test('a valid empty or fully filtered snapshot is a fresh success that clears old rows', async () => {
for (const quiet of [
[],
[null],
[{ ...sighting(), data_source_name: '' }],
[{ ...sighting(), latitude: 37.8 }],
[{ ...sighting(), type: 'Unknown' }],
]) {
let at = NOW;
let calls = 0;
let empty = false;
const cache = new Map();
const request = install({
cache,
now: () => at,
fetchImpl: async () => {
calls += 1;
return Response.json(empty ? quiet : [sighting()]);
},
});
assert.equal((await request()).body.rows.length, 1);
empty = true;
at += 600_001;
const result = await request();
assert.equal(result.status, 200);
assert.deepEqual(result.body, { fetchedAt: at, rows: [] });
assert.equal(result.headers['X-Data-Stale'], undefined);
assert.equal(result.headers['Retry-After'], undefined);
assert.deepEqual(cache.get('sightings').value.rows, []);
// The quiet feed keeps the normal ten-minute cadence, not a failure retry.
at += 599_999;
assert.equal((await request()).status, 200);
assert.equal(calls, 2);
at += 1;
await request();
assert.equal(calls, 3);
const cold = install({ fetchImpl: async () => Response.json(quiet) });
const coldResult = await cold();
assert.equal(coldResult.status, 200);
assert.deepEqual(coldResult.body, { fetchedAt: NOW, rows: [] });
}
});
test('an invented wrapped snapshot never replaces the cache', async () => {
const bad = { results: [sighting()] };
let at = NOW;
let fail = false;
const cache = new Map();
const request = install({
cache,
now: () => at,
fetchImpl: async () => Response.json(fail ? bad : [sighting()]),
});
const fresh = await request();
const retained = cache.get('sightings');
fail = true;
at += 600_001;
const stale = await request();
assert.equal(stale.status, 200);
assert.equal(stale.headers['X-Data-Stale'], 'true');
assert.equal(stale.body.stale, true);
assert.equal(stale.body.fetchedAt, fresh.body.fetchedAt);
assert.equal(cache.get('sightings'), retained);
const coldCache = new Map();
const cold = install({
cache: coldCache,
fetchImpl: async () => Response.json(bad),
});
const result = await cold();
assert.equal(result.status, 502);
assert.equal(result.body.error, 'acartia_unavailable');
assert.equal(result.headers['Retry-After'], '15');
assert.equal(coldCache.size, 0);
});
test('failure serves stale for at most 24 hours and never reports a cold success', async () => {
let at = NOW;
let fail = false;
const request = install({
now: () => at,
fetchImpl: async () => {
if (fail) throw new Error('private upstream detail');
return Response.json([sighting()]);
},
});
const first = await request();
fail = true;
at += 23 * 60 * 60_000 + 59 * 60_000;
const stale = await request();
assert.equal(stale.status, 200);
assert.equal(stale.body.stale, true);
assert.equal(stale.body.fetchedAt, first.body.fetchedAt);
at += 2 * 60_000;
const expired = await request();
assert.equal(expired.status, 502);
assert.deepEqual(expired.body, { error: 'acartia_unavailable' });
});
test('upstream 429 honors Retry-After, caps it, and spends no requests while cooling down', async () => {
assert.equal(acartiaRetryCooldownMs('2', NOW), 5000);
assert.equal(acartiaRetryCooldownMs('999999', NOW), 3_600_000);
assert.equal(
acartiaRetryCooldownMs(new Date(NOW + 120_000).toUTCString(), NOW),
120_000,
);
assert.equal(acartiaRetryCooldownMs(null, NOW), 60_000);
let at = NOW;
let calls = 0;
const warnings = [];
const request = install({
now: () => at,
warn: (message) => warnings.push(message),
fetchImpl: async () => {
calls += 1;
return new Response('limit', {
status: 429,
headers: { 'Retry-After': '999999' },
});
},
});
const first = await request();
assert.equal(first.status, 429);
assert.equal(first.body.error, 'acartia_unavailable');
assert.equal(first.headers['Retry-After'], '3600');
at += 5000;
const cooling = await request();
assert.equal(cooling.headers['Retry-After'], '3595');
assert.equal(calls, 1);
assert.equal(
warnings.filter((message) => message.includes('capped')).length,
1,
);
at += 3_600_000;
await request();
assert.equal(calls, 2);
assert.equal(
warnings.filter((message) => message.includes('capped')).length,
1,
);
});
test('upstream 429 also serves a prior snapshot stale', async () => {
let at = NOW;
let fail = false;
let calls = 0;
const request = install({
now: () => at,
fetchImpl: async () => {
calls += 1;
return fail
? new Response('limit', { status: 429 })
: Response.json([sighting()]);
},
});
const first = await request();
fail = true;
at += 600_001;
const stale = await request();
assert.equal(stale.status, 200);
assert.equal(stale.body.stale, true);
assert.equal(stale.body.fetchedAt, first.body.fetchedAt);
at += 1000;
assert.equal((await request()).body.stale, true);
assert.equal(calls, 2);
});
test('bad JSON, 5xx and capped bodies start a 15 second failure cooldown', async () => {
for (const bad of [
() => new Response('{malformed'),
() => new Response('private upstream message', { status: 503 }),
() =>
new Response('{}', {
headers: { 'Content-Length': String(2 * 1024 * 1024 + 1) },
}),
]) {
let at = NOW;
let calls = 0;
const request = install({
now: () => at,
fetchImpl: async () => {
calls += 1;
return calls === 1 ? bad() : Response.json([sighting()]);
},
});
const first = await request();
assert.equal(first.status, 502);
assert.deepEqual(first.body, { error: 'acartia_unavailable' });
assert.equal(first.headers['Retry-After'], '15');
at += 14_999;
assert.equal((await request()).headers['Retry-After'], '1');
assert.equal(calls, 1);
at += 1;
assert.equal((await request()).status, 200);
assert.equal(calls, 2);
}
});
+9
View File
@@ -83,6 +83,15 @@ export function adsbLolProxy() {
// This mount also matches its /track route, served by its own handler.
if (new URL(req.url || '/', 'http://localhost').pathname !== '/')
return next();
// After the /track hand-off, so /track keeps its own method handling.
if (req.method !== 'GET') {
res.writeHead(405, {
'Content-Type': 'application/json',
Allow: 'GET',
});
res.end(JSON.stringify({ error: 'Method Not Allowed' }));
return;
}
const now = Date.now();
if (_cache && now - _cacheAt < CACHE_MS) {
serve(res, 200, _cache, 'HIT');
+69 -22
View File
@@ -1,6 +1,7 @@
import fs from 'node:fs';
import os from 'node:os';
import { normalizeAdsbLolPointResponse } from '../../../src/data/adsbLolFallback.js';
import { encodedBody, gzipBody } from '../common/compress.js';
import {
coalesceProxyRequest,
readResponseJsonCapped,
@@ -18,6 +19,11 @@ let _openskyTokenExpiry = 0;
let _openskyTokenPromise = null;
/** @type {string|null} Cached upstream response body (JSON text). */
let _openskyCacheBody = null;
/**
* @type {Buffer|null} Gzip of `_openskyCacheBody`, written together with it.
* Null when the body is not worth compressing.
*/
let _openskyCacheGzip = null;
/** @type {number} HTTP status of the cached response. */
let _openskyCacheStatus = 0;
/** @type {number} Epoch-ms when the response was cached. */
@@ -399,8 +405,12 @@ async function fetchAdsbLolPointFallback(req) {
ADSBLOL_POINT_MAX_RESPONSE_BYTES,
);
const normalized = normalizeAdsbLolPointResponse(payload);
const body = JSON.stringify(normalized);
const record = {
body: JSON.stringify(normalized),
body,
// Compressed once per entry; ADSBLOL_POINT_CACHE_MAX bounds the
// extra memory.
gzip: await gzipBody(body),
cachedAt: Date.now(),
count: normalized.states.length,
};
@@ -429,6 +439,7 @@ async function fetchAdsbLolPointFallback(req) {
async function serveAdsbLolPointFallback(req, res, requestedMode, reason) {
const fallback = await fetchAdsbLolPointFallback(req);
if (!fallback) return false;
const encoded = encodedBody(req, fallback.body, fallback.gzip);
res.writeHead(200, {
...buildOpenSkyHeaders({
cacheStatus: fallback.cacheStatus,
@@ -439,8 +450,9 @@ async function serveAdsbLolPointFallback(req, res, requestedMode, reason) {
'X-Flight-Source': 'adsb.lol',
'X-Flight-Coverage': `${ADSBLOL_POINT_RADIUS_NM}nm regional fallback`,
'X-Flight-Count': String(fallback.count),
...encoded.headers,
});
res.end(fallback.body);
res.end(encoded.body);
return true;
}
@@ -484,6 +496,15 @@ export function openSkyProxy() {
// This mount also matches its /track route, served by its own handler.
if (new URL(req.url || '/', 'http://localhost').pathname !== '/')
return next();
// After the /track hand-off, so /track keeps its own method handling.
if (req.method !== 'GET') {
res.writeHead(405, {
'Content-Type': 'application/json',
Allow: 'GET',
});
res.end(JSON.stringify({ error: 'Method Not Allowed' }));
return;
}
res.setHeader('X-Feed-Source', 'OpenSky Network');
const requestedMode = normalizeOpenSkyAuthMode(
process.env.OPENSKY_AUTH_MODE,
@@ -514,9 +535,13 @@ export function openSkyProxy() {
reason: 'cached',
};
const isStale = now - _openskyCacheTime >= _openskyTtlMs;
res.writeHead(
_openskyCacheStatus || 200,
buildOpenSkyHeaders({
const encoded = encodedBody(
req,
_openskyCacheBody,
_openskyCacheGzip,
);
res.writeHead(_openskyCacheStatus || 200, {
...buildOpenSkyHeaders({
cacheStatus: isStale ? 'STALE' : 'HIT',
requestedMode: cachedMeta.requestedMode || requestedMode,
usedMode: cachedMeta.usedMode || 'unknown',
@@ -532,8 +557,9 @@ export function openSkyProxy() {
? (_openskyCooldownUntil - now) / 1000
: undefined,
}),
);
res.end(_openskyCacheBody);
...encoded.headers,
});
res.end(encoded.body);
return;
}
// Cooling down with nothing cached (cold start into a rate limit):
@@ -628,7 +654,11 @@ export function openSkyProxy() {
// Only cache successful responses — error responses (401/403/429/5xx)
// should not be served from cache on subsequent requests
if (upstream.ok) {
// Compress before writing either slot, so a request arriving
// mid-refresh still reads a body with its own frame.
const frame = await gzipBody(body);
_openskyCacheBody = body;
_openskyCacheGzip = frame;
_openskyCacheStatus = upstream.status;
_openskyCacheTime = now;
_openskyCacheSourceEpochMs = sourceEpochMs;
@@ -681,9 +711,13 @@ export function openSkyProxy() {
// Serve the last-good body instead of the 429 when we have one —
// the layer keeps rendering (STALE-cued) instead of dying.
if (_openskyCacheBody && _openskyCacheStatus === 200) {
res.writeHead(
200,
buildOpenSkyHeaders({
const encoded = encodedBody(
req,
_openskyCacheBody,
_openskyCacheGzip,
);
res.writeHead(200, {
...buildOpenSkyHeaders({
cacheStatus: 'STALE',
requestedMode,
usedMode,
@@ -691,8 +725,9 @@ export function openSkyProxy() {
staleSeconds: (now - _openskyCacheTime) / 1000,
retryAfterSeconds: cooldownMs / 1000,
}),
);
res.end(_openskyCacheBody);
...encoded.headers,
});
res.end(encoded.body);
return;
}
}
@@ -734,16 +769,23 @@ export function openSkyProxy() {
}
}
res.writeHead(
upstream.status,
buildOpenSkyHeaders({
// The frame belongs to this body only if this body was cached. An
// uncached error must not borrow the previous snapshot's frame.
const encoded = encodedBody(
req,
body,
_openskyCacheBody === body ? _openskyCacheGzip : null,
);
res.writeHead(upstream.status, {
...buildOpenSkyHeaders({
cacheStatus: 'MISS',
requestedMode,
usedMode,
reason,
}),
);
res.end(body);
...encoded.headers,
});
res.end(encoded.body);
} catch (e) {
console.error('[OpenSky Proxy]', e.message);
if (_openskyCacheBody) {
@@ -754,17 +796,22 @@ export function openSkyProxy() {
usedMode: 'unknown',
reason: 'cached_stale',
};
res.writeHead(
_openskyCacheStatus || 200,
buildOpenSkyHeaders({
const encoded = encodedBody(
req,
_openskyCacheBody,
_openskyCacheGzip,
);
res.writeHead(_openskyCacheStatus || 200, {
...buildOpenSkyHeaders({
cacheStatus: 'STALE',
requestedMode:
cachedMeta.requestedMode || OPENSKY_AUTH_MODE_DEFAULT,
usedMode: cachedMeta.usedMode || 'unknown',
reason: cachedMeta.reason || 'cached_stale',
}),
);
res.end(_openskyCacheBody);
...encoded.headers,
});
res.end(encoded.body);
return;
}
const requestedMode = normalizeOpenSkyAuthMode(
+187 -5
View File
@@ -1,4 +1,5 @@
import { createCctvCatalog } from './cctv/catalog.js';
import { quebec511VideoUrl, quebecBrowserVideoEnabled } from './cctv/quebec.js';
import { normalizeFeedType, isVideoFeedType } from './cctv/normalize.js';
import {
buildSyntheticCctvSvg,
@@ -18,6 +19,8 @@ import { sanitizeCctvRangeHeader } from './cctv/range.js';
import { admitSameSiteKeyless } from './common/same-site.js';
import { PROXY_SIGNALS } from '../../src/localRequestGate.mjs';
import { createHlsPuller } from './cctv/stream.js';
import { isMlitRiverAdjustmentImage } from './cctv/worldCameras.js';
import { ottawaAuthorizedFrameUrl } from './cctv/ottawa.js';
import { googleServerApiKey } from './places/google-key.js';
export { CCTV_FRAME_FETCH_TIMEOUT_MS, fetchCctvImageFromUpstream };
/**
@@ -43,6 +46,120 @@ export function cctvProxy({ sourceRoot = process.cwd() } = {}) {
const HEALTH_MAX_ENTRIES = CCTV_MAX_SOURCES_CEILING;
/** Live HLS strategies (see ./cctv/stream.js). Shared across dev and preview. */
const puller = createHlsPuller();
/**
* Bounded transient image reuse keeps cadence-limited providers (World
* stills, Ottawa's once-a-minute API) inside their refresh floor. The byte
* cap is the real bound; the entry cap only limits map overhead.
*/
const WORLD_FRAME_CACHE_MAX_ENTRIES = 1024;
/** Retry a failed fetch sooner than a full refresh window, never under a minute. */
const THROTTLED_FRAME_FAILURE_BACKOFF_MS = 60 * 1000;
const WORLD_FRAME_CACHE_MAX_ENTRY_BYTES = 2 * 1024 * 1024;
const WORLD_FRAME_CACHE_MAX_BYTES = 32 * 1024 * 1024;
const worldFrameCache = new Map();
const worldFrameInflight = new Map();
const worldFrameNextAllowedAt = new Map();
let worldFrameCacheBytes = 0;
function getWorldFrameCache(cameraId) {
const entry = worldFrameCache.get(cameraId);
if (!entry) return null;
worldFrameCache.delete(cameraId);
worldFrameCache.set(cameraId, entry);
return entry;
}
function storeWorldFrame(cameraId, image) {
if (
!Buffer.isBuffer(image?.body) ||
image.body.length > WORLD_FRAME_CACHE_MAX_ENTRY_BYTES
)
return;
const previous = worldFrameCache.get(cameraId);
if (previous) worldFrameCacheBytes -= previous.body.length;
const entry = {
body: image.body,
contentType: image.contentType,
fetchedAt: Date.now(),
};
worldFrameCache.delete(cameraId);
worldFrameCache.set(cameraId, entry);
worldFrameCacheBytes += entry.body.length;
while (
worldFrameCache.size > WORLD_FRAME_CACHE_MAX_ENTRIES ||
worldFrameCacheBytes > WORLD_FRAME_CACHE_MAX_BYTES
) {
const oldestId = worldFrameCache.keys().next().value;
const oldest = worldFrameCache.get(oldestId);
worldFrameCache.delete(oldestId);
worldFrameCacheBytes -= oldest.body.length;
}
}
function setNextAllowedAt(cameraId, at) {
worldFrameNextAllowedAt.delete(cameraId);
worldFrameNextAllowedAt.set(cameraId, at);
while (worldFrameNextAllowedAt.size > 4096) {
worldFrameNextAllowedAt.delete(
worldFrameNextAllowedAt.keys().next().value,
);
}
}
/**
* Fetch a frame from a provider with a per-camera refresh floor. Inside the
* floor the last frame is reused; with no frame to reuse the caller gets
* `{ throttled: true }` and serves a placeholder, so a cadence-limited camera
* never falls through to the key-bearing Street View fallback. A failed
* fetch backs off for a minute (capped at the floor) instead of a full
* window.
*/
async function fetchThrottledFrame(source, cameraId, frameUrl) {
const cached = getWorldFrameCache(cameraId);
const rawRefreshMs = Number(source?.frameRefreshMs);
const refreshMs =
Number.isFinite(rawRefreshMs) && rawRefreshMs > 0
? rawRefreshMs
: 5 * 60 * 1000;
const now = Date.now();
if (cached && now - cached.fetchedAt < refreshMs) {
return { ok: true, ...cached, cached: true };
}
const inflight = worldFrameInflight.get(cameraId);
if (inflight) return inflight;
const stale = cached
? {
ok: true,
body: cached.body,
contentType: cached.contentType,
stale: true,
}
: null;
const nextAllowedAt = worldFrameNextAllowedAt.get(cameraId) || 0;
if (now < nextAllowedAt) return stale || { ok: false, throttled: true };
if (!frameUrl) return stale;
setNextAllowedAt(cameraId, now + refreshMs);
let request;
request = fetchCctvImageFromUpstream(frameUrl)
.then((image) => {
if (image?.ok) {
storeWorldFrame(cameraId, image);
return image;
}
setNextAllowedAt(
cameraId,
Date.now() + Math.min(refreshMs, THROTTLED_FRAME_FAILURE_BACKOFF_MS),
);
return stale || image;
})
.finally(() => {
if (worldFrameInflight.get(cameraId) === request)
worldFrameInflight.delete(cameraId);
});
worldFrameInflight.set(cameraId, request);
return request;
}
/** Update the health entry for a camera, evicting the oldest entry if at capacity. */
const setHealth = (cameraId, patch) => {
@@ -153,6 +270,7 @@ export function cctvProxy({ sourceRoot = process.cwd() } = {}) {
);
if (url.pathname === '/sources') {
const quebecVideo = quebecBrowserVideoEnabled();
const body = {
trimmedPacks: getCctvSources.trimmedPacks?.(sources) ?? [],
sources: sources.map((source) => ({
@@ -164,6 +282,7 @@ export function cctvProxy({ sourceRoot = process.cwd() } = {}) {
provider: source.provider,
lat: source.lat,
lon: source.lon,
frameRefreshMs: source.frameRefreshMs,
headingDeg: source.headingDeg,
headingConfidence: source.headingConfidence || '',
pitchDeg: source.pitchDeg,
@@ -174,6 +293,13 @@ export function cctvProxy({ sourceRoot = process.cwd() } = {}) {
feedType: normalizeFeedType(source.feedType),
sourceKind:
source.sourceKind || (source.url ? 'configured' : 'fallback'),
sourceUrl: source.sourceUrl || '',
browserMediaUrl:
quebecVideo && source.sourceKind === 'quebec-511-browser'
? quebec511VideoUrl(source.id)
: '',
accuracyM: source.accuracyM,
positionSource: source.positionSource || '',
poseSource: source.poseSource,
model: source.model || '',
manufacturer: source.manufacturer || '',
@@ -526,15 +652,71 @@ export function cctvProxy({ sourceRoot = process.cwd() } = {}) {
: '');
const upstreamImage =
source?.sourceKind === 'txdot-its'
? await fetchTxdotSnapshot(upstreamCandidate)
: await fetchCctvImageFromUpstream(upstreamCandidate);
source?.pack === 'world-cameras'
? await fetchThrottledFrame(source, cameraId, upstreamCandidate)
: source?.sourceKind === 'ottawa-traffic'
? await fetchThrottledFrame(
source,
cameraId,
ottawaAuthorizedFrameUrl(upstreamCandidate),
)
: source?.sourceKind === 'txdot-its'
? await fetchTxdotSnapshot(upstreamCandidate)
: await fetchCctvImageFromUpstream(upstreamCandidate, {
allowMjpeg: source?.sourceKind === 'taiwan-highway-motc',
});
if (
source?.sourceKind === 'mlit-river' &&
isMlitRiverAdjustmentImage(upstreamImage)
) {
const svg = buildSyntheticCctvSvg({
cameraId,
label,
city,
status: 'CAMERA UNDER ADJUSTMENT',
});
setHealth(cameraId, {
status: 'degraded',
sourceKind: 'snapshot',
label: source.provider || 'MLIT river cameras',
message: 'Provider camera under adjustment',
});
res.writeHead(200, {
'Content-Type': 'image/svg+xml',
'Cache-Control': 'no-store',
'X-CCTV-Source': 'provider-maintenance',
});
res.end(svg);
return;
}
if (upstreamImage?.throttled) {
// Inside the provider's refresh floor with no frame to reuse: a
// placeholder, not Street View, and no health change.
res.writeHead(200, {
'Content-Type': 'image/svg+xml',
'Cache-Control': 'no-store',
'X-CCTV-Source': 'synthetic',
});
res.end(
buildSyntheticCctvSvg({
cameraId,
label,
city,
status: 'WAITING FOR NEXT FRAME',
}),
);
return;
}
if (upstreamImage?.ok) {
setHealth(cameraId, {
status: 'ok',
status: upstreamImage.stale ? 'degraded' : 'ok',
sourceKind: 'snapshot',
label: source?.provider || 'Configured source',
message: 'Upstream snapshot active',
message: upstreamImage.stale
? 'Serving last snapshot; refresh failed'
: upstreamImage.cached
? 'Cached snapshot active'
: 'Upstream snapshot active',
});
res.writeHead(200, {
'Content-Type': upstreamImage.contentType,
+49 -12
View File
@@ -1,5 +1,6 @@
import {
CCTV_MAX_SOURCES_CEILING,
CCTV_SHARED_SOURCE_SLOTS,
DEFAULT_CCTV_MAX_SOURCES,
} from './constants.js';
@@ -23,16 +24,27 @@ export function resolveCatalogCap(raw) {
* exceed the catalog cap, the cap is filled ROUND-ROBIN — one camera from
* each pack per turn, in each pack's own order — so a lowered
* CCTV_MAX_SOURCES thins every region a little instead of deleting whichever
* pack happened to be appended last. Duplicate ids resolve last-pack-wins so
* pack happened to be appended last. A pack with a `reservedCount` keeps up to
* that many rows only from the capacity ABOVE the shared floor, so a
* reservation adds room and never shrinks the other packs' round-robin
* share; at or below the floor reserved packs are ordinary lanes. Duplicate
* ids resolve last-pack-wins so
* file/env overrides keep replacing live records; the record is then counted
* against the pack that won it.
*
* @param {Array<{name:string, sources:Array<object>}>} packs - Normalized
* packs in merge order (earlier packs lose duplicate ids to later ones).
* @param {number} maxCount - Catalog cap.
* @param {object} [options]
* @param {number} [options.sharedFloor=CCTV_SHARED_SOURCE_SLOTS] - Slots kept
* for the round-robin before reservations apply.
* @returns {{sources:Array<object>, packs:Array<{name:string, offered:number, kept:number}>}}
*/
export function allocateSourceCap(packs, maxCount) {
export function allocateSourceCap(
packs,
maxCount,
{ sharedFloor = CCTV_SHARED_SOURCE_SLOTS } = {},
) {
const owner = new Map();
for (const pack of packs) {
for (const source of pack.sources) {
@@ -40,30 +52,53 @@ export function allocateSourceCap(packs, maxCount) {
owner.set(source.id, { pack: pack.name, source });
}
}
const lanes = packs.map((pack) => ({
name: pack.name,
const lanes = packs.map((pack) => {
// Only the record that WON its id (the last occurrence anywhere, so also
// the last within this pack) takes a lane slot; earlier duplicates never
// count against the cap.
queue: pack.sources.filter(
const queue = pack.sources.filter(
(source) => source?.id && owner.get(source.id)?.source === source,
),
next: 0,
kept: 0,
}));
);
return {
name: pack.name,
queue,
reserved: Math.max(
0,
Math.min(queue.length, Math.floor(Number(pack.reservedCount) || 0)),
),
next: 0,
kept: 0,
};
});
const total = lanes.reduce((sum, lane) => sum + lane.queue.length, 0);
const limit = Math.min(total, Math.max(0, Math.floor(maxCount)));
const sources = [];
const reservedSources = [];
if (limit >= total) {
for (const lane of lanes) {
sources.push(...lane.queue);
lane.kept = lane.queue.length;
}
} else {
while (sources.length < limit) {
// Reserved packs draw only on the capacity above the shared floor, then
// every lane (reserved ones included) shares the rest round-robin.
const reserveBudget = Math.max(0, limit - Math.max(0, sharedFloor));
while (reservedSources.length < reserveBudget) {
let tookReserved = false;
for (const lane of lanes) {
if (sources.length >= limit) break;
if (reservedSources.length >= reserveBudget) break;
if (lane.kept >= lane.reserved) continue;
reservedSources.push(lane.queue[lane.next]);
lane.next += 1;
lane.kept += 1;
tookReserved = true;
}
if (!tookReserved) break;
}
while (sources.length + reservedSources.length < limit) {
for (const lane of lanes) {
if (sources.length + reservedSources.length >= limit) break;
if (lane.next >= lane.queue.length) continue;
sources.push(lane.queue[lane.next]);
lane.next += 1;
@@ -72,7 +107,9 @@ export function allocateSourceCap(packs, maxCount) {
}
}
return {
sources,
// Keep the shared round-robin sequence first so the default active camera
// remains in its established region; guaranteed regional rows follow it.
sources: [...sources, ...reservedSources],
packs: lanes.map((lane) => ({
name: lane.name,
offered: lane.queue.length,
+34 -2
View File
@@ -1,11 +1,18 @@
import fs from 'node:fs';
import path from 'node:path';
import { DEFAULT_CCTV_SOURCE_FILE, CCTV_SOURCE_CACHE_MS } from './constants.js';
import {
DEFAULT_CCTV_SOURCE_FILE,
CCTV_SOURCE_CACHE_MS,
WORLD_CCTV_SOURCE_COUNT,
} from './constants.js';
import { allocateSourceCap, resolveCatalogCap } from './cap.js';
import { loadGroundHeights, joinGroundHeights } from './groundHeights.js';
import { cameraRegion, normalizeSourceItem } from './normalize.js';
import { loadLithuaniaSources } from './lithuania.js';
import { loadOttawaSourcesFromOpenData } from './ottawa.js';
import { loadQuebecSources, QUEBEC_CAMERA_SOURCE_COUNT } from './quebec.js';
import { loadTaiwanSources } from './taiwan.js';
import { loadWorldCameraSources } from './worldCameras.js';
import {
loadAustinSourcesFromOpenData,
loadCaltransSourcesFromOpenData,
@@ -20,6 +27,7 @@ import {
loadNswSourcesFromOpenData,
loadCalgarySourcesFromOpenData,
loadCataloniaSourcesFromOpenData,
loadIcelandSourcesFromOpenData,
loadHkSourcesFromOpenData,
loadDelDOTSourcesFromOpenData,
loadKingCountySourcesFromOpenData,
@@ -75,6 +83,11 @@ const LIVE_PACKS = [
enabled: () => envEnabled('CCTV_DRIVEBC_ENABLED'),
load: loadDriveBcSourcesFromOpenData,
},
{
name: 'ottawa',
enabled: () => envEnabled('CCTV_OTTAWA_ENABLED'),
load: loadOttawaSourcesFromOpenData,
},
{
name: 'txdot',
enabled: () => envEnabled('CCTV_TXDOT_ENABLED'),
@@ -110,6 +123,11 @@ const LIVE_PACKS = [
enabled: () => envEnabled('CCTV_CATALONIA_ENABLED'),
load: loadCataloniaSourcesFromOpenData,
},
{
name: 'iceland',
enabled: () => envEnabled('CCTV_ICELAND_ENABLED'),
load: loadIcelandSourcesFromOpenData,
},
{
name: 'hk',
enabled: () => envEnabled('CCTV_HK_ENABLED'),
@@ -172,6 +190,18 @@ const LIVE_PACKS = [
enabled: () => envOptIn('CCTV_HOUSTON_ENABLED'),
load: loadHoustonSourcesFromOpenData,
},
{
name: 'quebec',
enabled: () => envEnabled('CCTV_QUEBEC_ENABLED'),
load: loadQuebecSources,
reservedCount: QUEBEC_CAMERA_SOURCE_COUNT,
},
{
name: 'world-cameras',
enabled: () => envEnabled('CCTV_WORLD_CAMERAS_ENABLED'),
load: loadWorldCameraSources,
reservedCount: WORLD_CCTV_SOURCE_COUNT,
},
{
name: 'a22',
enabled: () => envEnabled('CCTV_A22_ENABLED'),
@@ -316,7 +346,7 @@ export function createCctvCatalog({ sourceRoot = process.cwd() } = {}) {
: [];
// Live packs first so file/env overrides win on duplicate IDs; each pack
// keeps its own priority order and the catalog cap is shared fairly.
const normalizePack = (name, items) => ({
const normalizePack = (name, items, reservedCount = 0) => ({
name,
sources: items
.filter((item) => item && typeof item === 'object')
@@ -326,6 +356,7 @@ export function createCctvCatalog({ sourceRoot = process.cwd() } = {}) {
? items.available
: items.length,
region: items.region ?? null,
reservedCount,
});
const packs = [
...LIVE_PACKS.map((pack, index) =>
@@ -334,6 +365,7 @@ export function createCctvCatalog({ sourceRoot = process.cwd() } = {}) {
liveResults[index]?.status === 'fulfilled'
? liveResults[index].value
: [],
pack.reservedCount,
),
),
normalizePack('file', fromFile),
+74 -7
View File
@@ -5,15 +5,18 @@ export const DEFAULT_AUSTIN_ROWS_URL =
/** Default cap on Austin cameras after distance-based prioritization. */
export const DEFAULT_AUSTIN_MAX_SOURCES = 250;
/**
* Catalog-wide safety ceiling on served cameras. Each pack already caps
* itself (nearest-to-anchor first); this bound only matters when the packs
* together exceed it, and it is then filled round-robin across packs (see
* cap.js) so no region is silently dropped. The default per-pack caps
* together can exceed it, so a default install may be thinned evenly here.
* Slots every pack shares round-robin before any reservation applies. A
* reserved pack (Québec, World) only takes capacity above this floor, so a
* cap at or below it thins every region evenly (see cap.js).
*/
export const DEFAULT_CCTV_MAX_SOURCES = 4000;
export const CCTV_SHARED_SOURCE_SLOTS = 4000;
/**
* Catalog-wide default: the 4,000 shared slots plus room reserved for the
* 1,120 Québec and 2,230 World image locations.
*/
export const DEFAULT_CCTV_MAX_SOURCES = 7350;
/** Hard upper bound for CCTV_MAX_SOURCES; also sizes the health map. */
export const CCTV_MAX_SOURCES_CEILING = 5000;
export const CCTV_MAX_SOURCES_CEILING = 8000;
/** Reference point for Austin camera prioritization (Congress & 6th). */
export const AUSTIN_DOWNTOWN = { lat: 30.2672, lon: -97.7431 };
/** Caltrans CCTV: one JSON feed per district, identical schema statewide. */
@@ -197,6 +200,39 @@ export const DRIVEBC_ANCHORS = [
{ lat: 49.2827, lon: -123.1207 }, // Vancouver
{ lat: 48.4284, lon: -123.3656 }, // Victoria
];
/** City of Ottawa traffic cameras: documented JSON list (keyless). */
export const OTTAWA_CAMERAS_URL =
'https://traffic.ottawa.ca/map/service/camera';
/**
* Keyless canonical frame URL for one camera number. The city's Traffic Camera
* Images API needs a certificate (OTTAWA_CAMERA_CERTIFICATE), which the frame
* proxy adds at request time so it never appears in a catalog response.
*/
export const OTTAWA_IMAGE_URL = (cameraNumber) =>
`https://traffic.ottawa.ca/opendata/camera?c=${encodeURIComponent(cameraNumber)}`;
/** Pinned frame directory, for redirect containment in the media proxy. */
export const OTTAWA_IMAGE_ORIGIN = 'https://traffic.ottawa.ca/opendata/';
/** The API lets one certificate and user id fetch a camera once a minute. */
export const OTTAWA_FRAME_REFRESH_MS = 60 * 1000;
/** Alphanumeric user id this server sends with its certificate. */
export const OTTAWA_API_USER_ID = 'gev';
export const DEFAULT_OTTAWA_MAX_SOURCES = 250;
/** Prioritization anchor: downtown Ottawa (Parliament Hill). */
export const OTTAWA_ANCHORS = [{ lat: 45.4215, lon: -75.6972 }];
/**
* Taiwan Highway Bureau MJPEG servers. Each camera is one root path segment,
* so the pin is the host root: a redirect must stay on the same server.
*/
export const TAIWAN_HIGHWAY_IMAGE_ORIGINS = Object.freeze(
[1, 2, 3, 4, 5, 6, 7, 8].map((n) => `https://cctv-ss0${n}.thb.gov.tw/`),
);
/** Direct still-image directories from the World cams add-on's eligible packs. */
export const TRIPCHECK_IMAGE_ORIGIN =
'https://www.tripcheck.com/roadcams/cams/';
export const MLIT_RIVER_IMAGE_ORIGIN = 'https://cam.river.go.jp/cam/now/';
export const DEFAULT_WORLD_CCTV_SOURCE_FILE =
'config/cctv_sources.world-image-cameras.json';
export const WORLD_CCTV_SOURCE_COUNT = 2230;
/** Tallinn intersection cameras: curated catalog + public stills on ristmikud.tallinn.ee. */
export const DEFAULT_TALLINN_SOURCE_FILE = 'config/cctv_sources.tallinn.json';
export const DEFAULT_TALLINN_MAX_SOURCES = 255;
@@ -309,6 +345,37 @@ export const CATALONIA_ANCHORS = [
{ lat: 41.5636, lon: 2.0111 }, // Terrassa
];
/**
* Icelandic Road and Coastal Administration (Vegagerðin, IRCA) road cameras:
* one keyless JSON list, one row per camera view (about 500 views at roughly
* 170 stations, verified 2026-11-06). Frames are stills on the IRCA host, each
* rewritten by the publisher every few minutes. Camera images are published
* under CC BY 4.0 (IRCA terms of use, "Open data" section).
*/
export const ICELAND_CAMERAS_URL =
'https://gagnaveita.vegagerdin.is/api/vefmyndavelar2014_1';
/** The only directory Iceland camera frames may come from. */
export const ICELAND_IMAGE_ORIGIN =
'https://www.vegagerdin.is/vgdata/vefmyndavelar/';
/** Byte ceiling for the Iceland catalog read (~150 KB observed 2026-11-06). */
export const ICELAND_MAX_CATALOG_BYTES = 2 * 1024 * 1024;
/** Ground-elevation prior in metres: the feed carries none, and the country
* runs from sea level to highland passes. The client's ground snap corrects
* this on 3D-tile stacks. */
export const ICELAND_GROUND_ELEVATION_M = 60;
/** The feed lists about 500 views; the default keeps half so the shared
* catalog cap still leaves room for denser regions. */
export const DEFAULT_ICELAND_MAX_SOURCES = 250;
/** Prioritization anchors: Reykjavík and the regional centres, so a cap keeps
* every region's nearest cameras instead of only the capital's. */
export const ICELAND_ANCHORS = [
{ lat: 64.1466, lon: -21.9426 }, // Reykjavík
{ lat: 65.6885, lon: -18.1262 }, // Akureyri
{ lat: 65.2639, lon: -14.3948 }, // Egilsstaðir
{ lat: 66.0749, lon: -23.1351 }, // Ísafjörður
{ lat: 63.9331, lon: -20.9971 }, // Selfoss
];
/** Hong Kong Transport Department traffic snapshot cameras: keyless English
* locations XML on DATA.GOV.HK; frames are stills on the TD CCTV host. */
export const DEFAULT_HK_CAMERAS_URL =
+243 -6
View File
@@ -10,10 +10,15 @@ import {
CCTV_MEDIA_MAX_BODY_BYTES,
HK_IMAGE_ORIGIN,
KING_COUNTY_IMAGE_ORIGIN,
MLIT_RIVER_IMAGE_ORIGIN,
TRIPCHECK_IMAGE_ORIGIN,
NSW_IMAGE_ORIGIN,
NSW_IMAGE_USER_AGENT,
OTTAWA_IMAGE_ORIGIN,
TALLINN_IMAGE_ORIGIN,
TARKTEE_IMAGE_ORIGIN,
ICELAND_IMAGE_ORIGIN,
TAIWAN_HIGHWAY_IMAGE_ORIGINS,
TFL_IMAGE_ORIGIN,
WARENDORF_IMAGE_ORIGINS,
WSDOT_IMAGE_ORIGIN,
@@ -523,13 +528,18 @@ for (const prefix of [
TFL_IMAGE_ORIGIN,
TALLINN_IMAGE_ORIGIN,
TARKTEE_IMAGE_ORIGIN,
ICELAND_IMAGE_ORIGIN,
...WARENDORF_IMAGE_ORIGINS,
NSW_IMAGE_ORIGIN,
OTTAWA_IMAGE_ORIGIN,
CALGARY_IMAGE_ORIGIN,
HK_IMAGE_ORIGIN,
KING_COUNTY_IMAGE_ORIGIN,
WSDOT_IMAGE_ORIGIN,
A22_IMAGE_ORIGIN,
TRIPCHECK_IMAGE_ORIGIN,
MLIT_RIVER_IMAGE_ORIGIN,
...TAIWAN_HIGHWAY_IMAGE_ORIGINS,
]) {
const host = new URL(prefix).hostname;
CCTV_PINNED_PREFIXES_BY_HOST.set(host, [
@@ -621,6 +631,216 @@ export function cctvUpstreamUserAgent(url) {
}
}
/** Largest single JPEG the MJPEG first-frame reader will buffer. */
export const CCTV_MJPEG_FRAME_MAX_BYTES = 2 * 1024 * 1024;
/**
* Read one bounded JPEG part from an open MJPEG response, then stop the stream.
*
* The first part may follow a CRLF preamble, and its delimiter may be written
* as `--<boundary>` or, when the declared boundary already starts with `--`,
* as the boundary itself. A part with Content-Length is read to that length;
* a part without one ends at the next delimiter, so a JPEG end marker inside
* the image (an EXIF thumbnail) never truncates it. Bytes are copied once
* into a preallocated buffer and every search resumes where it stopped, so a
* trickling upstream costs linear time.
*/
async function readFirstMjpegFrame(upstream, maxBytes) {
const mediaType = cctvLiveMediaType(
'mjpeg',
upstream.headers.get('content-type'),
);
const boundary = parseMediaType(mediaType)?.params.get('boundary') || '';
const frameMax = Math.min(maxBytes, CCTV_MJPEG_FRAME_MAX_BYTES);
if (
!boundary ||
!upstream.body ||
!Number.isSafeInteger(frameMax) ||
frameMax < 4
) {
try {
await upstream.body?.cancel();
} catch {
/* no-op */
}
return null;
}
const delimiters = [Buffer.from('--' + boundary, 'latin1')];
if (boundary.startsWith('--') && boundary.length > 2)
delimiters.push(Buffer.from(boundary, 'latin1'));
const lineEnd = Buffer.from('\r\n', 'ascii');
const headerTerminator = Buffer.from('\r\n\r\n', 'ascii');
const jpegStart = Buffer.from([0xff, 0xd8]);
const jpegEnd = Buffer.from([0xff, 0xd9]);
const maxPreambleBytes = 64;
const maxHeaderBytes = 8 * 1024;
const maxBufferedBytes =
maxPreambleBytes +
delimiters[0].length +
lineEnd.length +
maxHeaderBytes +
headerTerminator.length +
frameMax +
lineEnd.length +
delimiters[0].length;
const buffer = Buffer.allocUnsafe(maxBufferedBytes);
let length = 0;
let reader;
let headerStart = -1;
let delimiter = null;
let headerScanFrom = 0;
let bodyStart = -1;
let declaredLength = null;
let boundaryScanFrom = -1;
/** Match the opening delimiter after an optional CRLF preamble. */
function findOpening() {
let at = 0;
while (at < length && (buffer[at] === 0x0d || buffer[at] === 0x0a)) at++;
if (at > maxPreambleBytes) return { failed: true };
for (const candidate of delimiters) {
const line = Buffer.concat([candidate, lineEnd]);
const need = at + line.length;
if (length < need) {
const seen = buffer.subarray(at, length);
if (line.subarray(0, seen.length).equals(seen))
return { pending: true };
continue;
}
if (buffer.subarray(at, need).equals(line))
return { start: need, delimiter: candidate };
}
return { failed: true };
}
/** Parse the part headers; false rejects the stream. */
function parseHeaders(headerEnd) {
const headerText = buffer
.subarray(headerStart, headerEnd)
.toString('latin1');
const headers = new Map();
for (const line of headerText.split('\r\n')) {
const colon = line.indexOf(':');
if (colon <= 0) return false;
const name = line.slice(0, colon).trim().toLowerCase();
if (!name || headers.has(name)) return false;
headers.set(name, line.slice(colon + 1).trim());
}
if (cctvRasterImageType(headers.get('content-type')) !== 'image/jpeg')
return false;
const declared = headers.get('content-length');
if (declared !== undefined) {
if (!/^\d+$/.test(declared)) return false;
declaredLength = Number(declared);
if (
!Number.isSafeInteger(declaredLength) ||
declaredLength < jpegStart.length + jpegEnd.length ||
declaredLength > frameMax
)
return false;
}
return true;
}
/** A complete JPEG body, or null. */
function asJpeg(frame) {
return frame.length >= jpegStart.length + jpegEnd.length &&
frame.subarray(0, jpegStart.length).equals(jpegStart) &&
frame.subarray(-jpegEnd.length).equals(jpegEnd)
? Buffer.from(frame)
: null;
}
try {
reader = upstream.body.getReader();
for (;;) {
const { done, value } = await reader.read();
if (done) return null;
const chunk = Buffer.from(value);
const room = maxBufferedBytes - length;
const included = Math.min(room, chunk.length);
chunk.copy(buffer, length, 0, included);
length += included;
const overflow = included < chunk.length;
if (headerStart < 0) {
const opening = findOpening();
if (opening.failed) return null;
if (opening.pending) {
if (overflow) return null;
continue;
}
headerStart = opening.start;
delimiter = opening.delimiter;
headerScanFrom = headerStart;
}
if (bodyStart < 0) {
const headerEnd = buffer
.subarray(0, length)
.indexOf(headerTerminator, headerScanFrom);
if (headerEnd < 0) {
if (length - headerStart > maxHeaderBytes || overflow) return null;
headerScanFrom = Math.max(
headerStart,
length - headerTerminator.length + 1,
);
continue;
}
if (headerEnd - headerStart > maxHeaderBytes) return null;
if (!parseHeaders(headerEnd)) return null;
bodyStart = headerEnd + headerTerminator.length;
boundaryScanFrom = bodyStart;
}
const available = length - bodyStart;
if (available >= jpegStart.length) {
if (
!buffer
.subarray(bodyStart, bodyStart + jpegStart.length)
.equals(jpegStart)
)
return null;
}
const nextDelimiter = Buffer.concat([lineEnd, delimiter]);
const nextPart = buffer
.subarray(0, length)
.indexOf(nextDelimiter, boundaryScanFrom);
if (declaredLength !== null) {
if (nextPart >= bodyStart && nextPart - bodyStart < declaredLength)
return null;
if (available >= declaredLength)
return asJpeg(buffer.subarray(bodyStart, bodyStart + declaredLength));
} else if (nextPart >= bodyStart) {
const frame = buffer.subarray(bodyStart, nextPart);
return frame.length > frameMax ? null : asJpeg(frame);
} else if (available > frameMax) {
return null;
}
if (nextPart < 0)
boundaryScanFrom = Math.max(
bodyStart,
length - nextDelimiter.length + 1,
);
if (overflow) return null;
}
} finally {
if (reader) {
try {
await reader.cancel();
} catch {
/* already closed or aborted */
}
try {
reader.releaseLock();
} catch {
/* already released */
}
}
}
}
/**
* Fetch one upstream CCTV image within the frame-refresh budget.
*
@@ -633,6 +853,10 @@ export function cctvUpstreamUserAgent(url) {
* @param {typeof fetch} [options.fetchImpl=fetch] - Fetch implementation.
* @param {number} [options.timeoutMs=CCTV_FRAME_FETCH_TIMEOUT_MS] - Abort timeout.
* @param {number} [options.maxBytes=CCTV_FRAME_MAX_BODY_BYTES] - Snapshot byte cap.
* @param {boolean} [options.allowMjpeg=false] - Accept a multipart MJPEG
* answer and keep its first JPEG. Only sources registered as MJPEG stills
* (the Taiwan Highway Bureau feeds) set it; any other host answering
* multipart falls through to the next fallback.
* @returns {Promise<{ok:true,body:Buffer,contentType:string}|null>}
*/
export async function fetchCctvImageFromUpstream(
@@ -641,6 +865,7 @@ export async function fetchCctvImageFromUpstream(
fetchImpl = fetch,
timeoutMs = CCTV_FRAME_FETCH_TIMEOUT_MS,
maxBytes = CCTV_FRAME_MAX_BODY_BYTES,
allowMjpeg = false,
} = {},
) {
if (!url || !/^https?:\/\//i.test(url)) return null;
@@ -660,10 +885,7 @@ export async function fetchCctvImageFromUpstream(
fetchImpl,
);
if (!upstream) return null;
const contentType = cctvRasterImageType(
upstream.headers.get('content-type'),
);
if (!upstream.ok || !contentType) {
if (!upstream.ok) {
try {
await upstream.body?.cancel();
} catch {
@@ -671,9 +893,24 @@ export async function fetchCctvImageFromUpstream(
}
return null;
}
const body = await readCappedResponseBytes(upstream, maxBytes);
const upstreamType = upstream.headers.get('content-type');
const contentType = cctvRasterImageType(upstreamType);
if (contentType) {
const body = await readCappedResponseBytes(upstream, maxBytes);
if (!body) return null;
return { ok: true, body, contentType };
}
if (!allowMjpeg || !cctvLiveMediaType('mjpeg', upstreamType)) {
try {
await upstream.body?.cancel();
} catch {
/* no-op */
}
return null;
}
const body = await readFirstMjpegFrame(upstream, maxBytes);
if (!body) return null;
return { ok: true, body, contentType };
return { ok: true, body, contentType: 'image/jpeg' };
} catch {
return null;
} finally {
+15
View File
@@ -460,6 +460,17 @@ export function isLikelyCataloniaCoordinate(lat, lon) {
);
}
/** Iceland bounding box (mainland and nearby islands, with a margin). */
export function isLikelyIcelandCoordinate(lat, lon) {
return (
isPlausibleLatLon(lat, lon) &&
lat >= 63.0 &&
lat <= 66.7 &&
lon >= -25.0 &&
lon <= -13.0
);
}
/**
* Decode numeric HTML character references, decimal ("&#237;") and hex
* ("&#xED;") alike, the entity form the Catalonia feed uses for accented
@@ -639,6 +650,10 @@ export function normalizeSourceItem(item) {
feedType: normalizeFeedType(item.feedType || item.type || ''),
url: typeof item.url === 'string' ? item.url : '',
snapshotUrl: typeof item.snapshotUrl === 'string' ? item.snapshotUrl : '',
frameRefreshMs: toFiniteNumber(item.frameRefreshMs),
sourceUrl: typeof item.sourceUrl === 'string' ? item.sourceUrl : '',
accuracyM: toFiniteNumber(item.accuracyM),
positionSource: String(item.positionSource || ''),
license: String(item.license || item.licenseNote || ''),
// Per-camera attribution for feeds a partner supplies inside a pack
// (DriveBC: TransLink, city cameras). Shown beside the provider.
+185
View File
@@ -0,0 +1,185 @@
import {
CCTV_SOURCE_FETCH_TIMEOUT_MS,
DEFAULT_OTTAWA_MAX_SOURCES,
OTTAWA_ANCHORS,
OTTAWA_API_USER_ID,
OTTAWA_CAMERAS_URL,
OTTAWA_FRAME_REFRESH_MS,
OTTAWA_IMAGE_ORIGIN,
OTTAWA_IMAGE_URL,
} from './constants.js';
import {
fallbackHeadingFromId,
pinnedCameraUrl,
prioritizeSources,
toFiniteNumber,
} from './normalize.js';
let ottawaCertificateMissingLogged = false;
/** Test hook: let the next missing-certificate skip log again. */
export function _resetOttawaCertificateWarningForTest() {
ottawaCertificateMissingLogged = false;
}
/** The operator's Traffic Camera Images API certificate, trimmed. */
function ottawaCertificate(env = process.env) {
return String(env.OTTAWA_CAMERA_CERTIFICATE ?? '').trim();
}
/**
* Add the operator's certificate to a registered keyless Ottawa frame URL.
* Only a URL inside the pinned `/opendata/` directory with a numeric camera
* number is signed, so the certificate never goes to any other path or host.
*
* @param {string} url - Canonical frame URL from the catalog.
* @param {Record<string,string|undefined>} [env]
* @returns {string} The signed URL, or '' when there is no certificate or the
* URL is not a registered Ottawa frame.
*/
export function ottawaAuthorizedFrameUrl(url, env = process.env) {
const certificate = ottawaCertificate(env);
const pinned = pinnedCameraUrl(url, OTTAWA_IMAGE_ORIGIN);
if (!certificate || !pinned) return '';
const parsed = new URL(pinned);
const cameraNumber = parsed.searchParams.get('c') || '';
if (parsed.pathname !== '/opendata/camera' || !/^\d+$/.test(cameraNumber))
return '';
const signed = new URL(OTTAWA_IMAGE_URL(cameraNumber));
signed.searchParams.set('certificate', certificate);
signed.searchParams.set('id', OTTAWA_API_USER_ID);
return signed.href;
}
/**
* City of Ottawa envelope: Kanata to Orléans, Barrhaven to the river, with
* slack for the MTO cameras on the 416/417 that the city feed also carries.
*/
const OTTAWA_BOUNDS = Object.freeze({
minLat: 45.1,
maxLat: 45.65,
minLon: -76.2,
maxLon: -75.1,
});
/** Whether a point is inside the city feed's service area. */
function isOttawaCoordinate(lat, lon) {
return (
Number.isFinite(lat) &&
Number.isFinite(lon) &&
lat >= OTTAWA_BOUNDS.minLat &&
lat <= OTTAWA_BOUNDS.maxLat &&
lon >= OTTAWA_BOUNDS.minLon &&
lon <= OTTAWA_BOUNDS.maxLon
);
}
/**
* Load the City of Ottawa's traffic cameras.
*
* Camera images come from the city's Traffic Camera Images API, which needs a
* certificate from the city's registration form (OTTAWA_CAMERA_CERTIFICATE)
* and allows one fetch per camera per minute for a certificate and user id.
* Without a certificate the pack is skipped, with no request and one log line.
* The camera list is the documented `map/service/camera` JSON: a camera
* number, English and French names, coordinates and the owner (CITY or MTO).
* The feed carries no bearing, no elevation and no status field, so headings
* use the same deterministic low-confidence fallback as the other
* heading-less packs and the client's ground snap refines the elevation
* prior.
*
* @param {object} [_options]
* @param {Record<string,string|undefined>} [env]
* @returns {Promise<Array<object>>} Cameras nearest downtown first, or [] on
* a missing certificate or any upstream or payload failure.
*/
export async function loadOttawaSourcesFromOpenData(
_options = {},
env = process.env,
) {
if (!ottawaCertificate(env)) {
if (!ottawaCertificateMissingLogged) {
ottawaCertificateMissingLogged = true;
console.warn(
'[CCTV] Ottawa cameras skipped: the city camera image API needs a certificate (set OTTAWA_CAMERA_CERTIFICATE)',
);
}
return [];
}
try {
const resp = await fetch(OTTAWA_CAMERAS_URL, {
headers: { Accept: 'application/json' },
signal: AbortSignal.timeout(CCTV_SOURCE_FETCH_TIMEOUT_MS),
});
if (!resp.ok) {
console.warn('[CCTV] Ottawa camera download failed:', resp.status);
return [];
}
const payload = await resp.json();
const rows = Array.isArray(payload?.cameras) ? payload.cameras : null;
if (!rows) return [];
const cameras = [];
const seen = new Set();
for (const row of rows) {
const number = toFiniteNumber(row?.camera_number);
if (!Number.isSafeInteger(number) || number <= 0 || seen.has(number))
continue;
const lat = toFiniteNumber(row?.latitude);
const lon = toFiniteNumber(row?.longitude);
if (!isOttawaCoordinate(lat, lon)) continue;
seen.add(number);
const cameraId = `ottawa-${number}`;
const imageUrl = OTTAWA_IMAGE_URL(number);
// The docs name the field camera_owner; the live feed sends cameraOwner.
const owner = String(row?.camera_owner || row?.cameraOwner || '')
.trim()
.toUpperCase();
cameras.push({
id: cameraId,
name: String(row?.name || '').trim() || `Ottawa camera ${number}`,
city: 'Ottawa',
cityId: 'ottawa',
provider: 'City of Ottawa',
lat,
lon,
headingDeg: fallbackHeadingFromId(cameraId),
headingConfidence: 'low',
pitchDeg: -18,
fovDeg: 44,
rangeM: 145,
mountHeightM: 8,
groundElevationM: 0,
feedType: 'image',
url: imageUrl,
snapshotUrl: imageUrl,
frameRefreshMs: OTTAWA_FRAME_REFRESH_MS,
sourceKind: 'ottawa-traffic',
license: 'City of Ottawa Traffic Camera Images API (registered access)',
credit:
owner === 'MTO'
? 'Ministry of Transportation of Ontario (MTO) camera'
: '',
});
}
const maxRaw = Number(
process.env.CCTV_OTTAWA_MAX_SOURCES || DEFAULT_OTTAWA_MAX_SOURCES,
);
const maxCount = Number.isFinite(maxRaw)
? Math.max(8, Math.min(500, Math.floor(maxRaw)))
: DEFAULT_OTTAWA_MAX_SOURCES;
const prioritized = prioritizeSources(cameras, maxCount, OTTAWA_ANCHORS);
console.log(
`[CCTV] Loaded Ottawa camera sources: ${cameras.length} published (using nearest ${prioritized.length})`,
);
return prioritized;
} catch (error) {
console.warn(
'[CCTV] Ottawa camera download error:',
error?.message || error,
);
return [];
}
}
+138
View File
@@ -0,0 +1,138 @@
import fs from 'node:fs';
import path from 'node:path';
import { cameraDisplayCode, fallbackHeadingFromId } from './normalize.js';
const QUEBEC_CATALOG_PATH = 'public/quebec-cameras.json';
export const QUEBEC_CAMERA_SOURCE_COUNT = 1120;
/**
* Whether the operator opted in to the browser-direct Québec 511 video. Off by
* default: the camera locations are CC BY 4.0, but whether Québec 511's
* service terms allow another site to embed its clips is unconfirmed, so an
* unset flag lists the cameras with a link to the 511 viewer page only.
*/
export function quebecBrowserVideoEnabled(env = process.env) {
return String(env.CCTV_QUEBEC_VIDEO_ENABLED || '').trim() === '1';
}
/** Build the direct browser-only MP4 URL for one namespaced MTQ camera id. */
export function quebec511VideoUrl(cameraId) {
const match = /^qc-511-(\d+)$/.exec(String(cameraId || ''));
return match
? `https://www.quebec511.info/Carte/Fenetres/camera.ashx?id=${match[1]}&format=mp4`
: '';
}
function safeViewerUrl(value) {
try {
const url = new URL(String(value || ''));
return url.protocol === 'https:' && !url.username && !url.password
? url.href
: '';
} catch {
return '';
}
}
function sourceId(camera) {
const id = String(camera?.id || '').trim();
if (camera?.kind === 'traffic' && /^\d+$/.test(id)) return `qc-511-${id}`;
if (camera?.kind === 'montreal' && /^mtl-\d+(?:-\d+)*$/.test(id))
return `qc-mtl-${id.slice(4)}`;
if (camera?.kind === 'webcam' && /^web-\d+$/.test(id))
return `qc-web-${id.slice(4)}`;
return '';
}
/**
* Read the checked-in, no-key Québec location catalog and adapt it to the
* regular CCTV source schema. MTQ video stays browser-only; other rows link to
* their operator page because this dataset has no direct image URL for them.
*/
export function loadQuebecSources({ sourceRoot = process.cwd() } = {}) {
const filename = path.resolve(sourceRoot, QUEBEC_CATALOG_PATH);
let cameras;
try {
const payload = JSON.parse(fs.readFileSync(filename, 'utf8'));
cameras = Array.isArray(payload?.cameras) ? payload.cameras : [];
} catch (error) {
console.warn('[CCTV] Québec camera catalog unavailable:', error?.message);
return [];
}
const sources = [];
const seen = new Set();
for (const camera of cameras) {
const id = sourceId(camera);
const lat = Number(camera.lat);
const lon = Number(camera.lon);
if (
!id ||
seen.has(id) ||
!Number.isFinite(lat) ||
!Number.isFinite(lon) ||
lat < 44.9 ||
lat > 62.6 ||
lon < -79.9 ||
lon > -57.1
)
continue;
seen.add(id);
const name = String(camera.nameEn || camera.name || id).trim();
const isTraffic = camera.kind === 'traffic';
const isMontreal = camera.kind === 'montreal';
const sourceKind = isTraffic
? 'quebec-511-browser'
: isMontreal
? 'quebec-montreal-osm'
: 'quebec-webcam-link';
const provider = isTraffic
? 'Québec 511 / MTQ'
: isMontreal
? 'Ville de Montréal / OpenStreetMap'
: String(camera.operator || 'Québec webcam');
sources.push({
id,
name: name.slice(0, 180),
city: String(camera.region || 'Québec'),
cityId: 'quebec',
provider,
lat,
lon,
headingDeg: fallbackHeadingFromId(id),
headingConfidence: 'low',
pitchDeg: -18,
fovDeg: 74,
rangeM: 400,
mountHeightM: 8,
groundElevationM: 0,
feedType: 'image',
// No frame URL is registered: the server never fetches 511 media. With
// CCTV_QUEBEC_VIDEO_ENABLED=1 the browser plays the 511 MP4 in the DOM
// overlay after a play click; otherwise the camera links to 511.
url: '',
snapshotUrl: '',
sourceKind,
sourceUrl: safeViewerUrl(camera.viewerUrl),
accuracyM: Number.isFinite(Number(camera.accuracyM))
? Number(camera.accuracyM)
: null,
positionSource: String(camera.positionSource || ''),
license: isTraffic
? 'Locations: Données Québec / MTQ, CC BY 4.0. Video: hosted by Québec 511 under its terms of use'
: isMontreal
? 'OpenStreetMap contributors, ODbL 1.0'
: 'Operator terms apply; location provenance shown per camera',
credit: isTraffic
? 'Locations: MTQ · CC BY 4.0 · Video: Québec 511'
: isMontreal
? 'OpenStreetMap contributors · ODbL'
: provider,
code: cameraDisplayCode(name.toUpperCase()),
});
}
return sources;
}
+175
View File
@@ -134,6 +134,12 @@ import {
HOUSTON_MAX_CATALOG_BYTES,
HOUSTON_GROUND_ELEVATION_M,
HOUSTON_ANCHORS,
ICELAND_CAMERAS_URL,
ICELAND_IMAGE_ORIGIN,
ICELAND_MAX_CATALOG_BYTES,
ICELAND_GROUND_ELEVATION_M,
DEFAULT_ICELAND_MAX_SOURCES,
ICELAND_ANCHORS,
} from './constants.js';
import {
toFiniteNumber,
@@ -152,6 +158,7 @@ import {
isLikelyCalgaryCoordinate,
isLikelyFloridaCoordinate,
isLikelyCataloniaCoordinate,
isLikelyIcelandCoordinate,
decodeNumericEntities,
isLikelyNycCoordinate,
isLikelyHoustonCoordinate,
@@ -2029,6 +2036,174 @@ export async function loadCataloniaSourcesFromOpenData() {
}
}
/**
* Compass headings for Icelandic direction words, as they appear after "til"
* (genitive: "til vesturs"), "í" (accusative: "í vestur") or "séð" ("seen",
* as in "séð austur").
*/
const ICELAND_DIRECTION_HEADINGS = {
norður: 0,
norðaustur: 45,
austur: 90,
suðaustur: 135,
suður: 180,
suðvestur: 225,
vestur: 270,
norðvestur: 315,
};
const ICELAND_DIRECTION_WORD =
'(norðaustur|norðvestur|suðaustur|suðvestur|norður|suður|austur|vestur)s?';
/** A facing, plus a second one joined by "og" ("til austurs og vesturs"). */
const ICELAND_DIRECTION_RE = new RegExp(
`(?<![\\p{L}])(?:til|í|séð)\\s+${ICELAND_DIRECTION_WORD}` +
`(?:\\s+og\\s+(?:til\\s+|í\\s+)?${ICELAND_DIRECTION_WORD})?(?![\\p{L}])`,
'gu',
);
/**
* The compass heading an Icelandic view description states, or NaN. Only an
* explicit "til <átt>", "í <átt>" or "séð <átt>" counts: a bare place name or
* a road direction ("Miklabraut austur") is not a facing. A description naming
* two different directions ("til norðurs og til suðurs", "til austurs og
* vesturs") is ambiguous and yields NaN, and "niður á veg" (down onto the
* road) states none.
*
* @param {string} text - The feed's `Skyring` field.
* @returns {number} Heading in degrees [0..360), or NaN.
*/
export function icelandHeadingFromDescription(text) {
const headings = new Set();
for (const match of String(text || '')
.toLowerCase()
.matchAll(ICELAND_DIRECTION_RE)) {
for (const word of [match[1], match[2]]) {
if (word) headings.add(ICELAND_DIRECTION_HEADINGS[word]);
}
}
return headings.size === 1 ? headings.values().next().value : NaN;
}
/**
* One Vegagerðin camera row -> one catalog source, or null. A row is one camera
* view; a station (`Maelist_nr`) carries several. The id comes from the frame
* file name, which is unique per view in the live feed.
*
* @param {object} row - Element of the `vefmyndavelar2014_1` array.
* @returns {?object}
*/
export function icelandCameraToSource(row) {
const url = pinnedCameraUrl(row?.Slod, ICELAND_IMAGE_ORIGIN);
if (!url || !/\.jpe?g$/i.test(new URL(url).pathname)) return null;
// Numbers only: Number(null) and Number('') are 0.
const lat = typeof row.Breidd === 'number' ? row.Breidd : NaN;
const lon = typeof row.Lengd === 'number' ? row.Lengd : NaN;
if (!isLikelyIcelandCoordinate(lat, lon)) return null;
const stem = new URL(url).pathname
.slice(new URL(ICELAND_IMAGE_ORIGIN).pathname.length)
.replace(/\.jpe?g$/i, '')
.toLowerCase();
if (!/^[a-z0-9_-]+$/.test(stem)) return null;
const cameraId = `is-vegagerdin-${stem}`;
const description = String(row.Skyring || '')
.replace(/\s+/g, ' ')
// Letter lookarounds, not \b: \b is ASCII-only and would join "Þverá
// Árnessýslu" while missing "séð séð".
.replace(/(?<!\p{L})(\p{L}+) \1(?!\p{L})/giu, '$1')
.trim();
const station = String(row.Myndavel || '').trim();
const heading = icelandHeadingFromDescription(description);
const hasHeading = Number.isFinite(heading);
return {
id: cameraId,
name: description || station || `Vegagerðin ${row.Maelist_nr ?? stem}`,
city: 'Iceland',
cityId: 'iceland',
provider: 'Vegagerðin (IRCA)',
lat,
lon,
headingDeg: hasHeading ? heading : fallbackHeadingFromId(cameraId),
headingConfidence: hasHeading ? 'high' : 'low',
pitchDeg: hasHeading ? -24 : -18,
fovDeg: hasHeading ? 56 : 44,
rangeM: hasHeading ? 210 : 145,
mountHeightM: hasHeading ? 10 : 8,
groundElevationM: ICELAND_GROUND_ELEVATION_M,
feedType: 'image',
url,
snapshotUrl: url,
sourceKind: 'iceland-vegagerdin',
license:
'Icelandic Road and Coastal Administration (IRCA) — road camera images, CC BY 4.0',
code: cameraDisplayCode(station.toUpperCase() || description),
};
}
/**
* Fetch Icelandic road cameras (Vegagerðin / IRCA), keyless: one JSON list.
* Frames are stills on the IRCA host and the loader registers only links
* inside that host's camera directory (ICELAND_IMAGE_ORIGIN). The list decides
* ids, coordinates, labels and which frame path is proxied, so it is fetched
* over https, redirects are refused and the body read is capped at
* ICELAND_MAX_CATALOG_BYTES.
*
* @returns {Promise<Array<object>>} Normalized camera source objects.
*/
export async function loadIcelandSourcesFromOpenData() {
try {
const resp = await fetch(ICELAND_CAMERAS_URL, {
headers: {
Accept: 'application/json',
'User-Agent': 'gods-eye-view-cctv-proxy/1.0',
},
redirect: 'manual',
signal: AbortSignal.timeout(CCTV_SOURCE_FETCH_TIMEOUT_MS),
});
// A response this loader will not read still owns its transport until the
// body is released, so every rejection path cancels before returning.
const discard = async () => {
try {
await resp.body?.cancel();
} catch {
/* no-op */
}
return [];
};
if (resp.status >= 300 && resp.status < 400) {
console.warn(
'[CCTV] Iceland catalog redirected; redirects are not followed',
);
return discard();
}
if (!resp.ok) {
console.warn('[CCTV] Iceland camera download failed:', resp.status);
return discard();
}
const body = await readResponseJsonCapped(resp, ICELAND_MAX_CATALOG_BYTES);
const rows = Array.isArray(body) ? body : [];
const cameras = rows.map(icelandCameraToSource).filter(Boolean);
const unique = Array.from(
new Map(cameras.map((camera) => [camera.id, camera])).values(),
);
const maxRaw = Number(
process.env.CCTV_ICELAND_MAX_SOURCES || DEFAULT_ICELAND_MAX_SOURCES,
);
const maxCount = Number.isFinite(maxRaw)
? Math.max(8, Math.min(600, Math.floor(maxRaw)))
: DEFAULT_ICELAND_MAX_SOURCES;
const prioritized = prioritizeSources(unique, maxCount, ICELAND_ANCHORS);
console.log(
`[CCTV] Loaded Iceland camera sources: ${unique.length} (using nearest ${prioritized.length})`,
);
return prioritized;
} catch (error) {
console.warn(
'[CCTV] Iceland camera download error:',
error?.message || error,
);
return [];
}
}
/**
* Read one XML child tag's text from an `<image>` block.
*
+3 -1
View File
@@ -5,7 +5,9 @@ export const HLS_LIMITS = Object.freeze({
sessions: 2,
leasesPerSession: 8,
playlistBytes: 256 * 1024,
segmentBytes: 4 * 1024 * 1024,
// High-resolution cameras can exceed 4 MiB in a single ten-second TS segment.
// Retained data still shares the independent 24 MiB session budget below.
segmentBytes: 8 * 1024 * 1024,
sessionBytes: 24 * 1024 * 1024,
segments: 12,
pollMs: 2000,
+158 -11
View File
@@ -1,4 +1,7 @@
import { CCTV_SOURCE_FETCH_TIMEOUT_MS } from './constants.js';
import {
CCTV_SOURCE_FETCH_TIMEOUT_MS,
TAIWAN_HIGHWAY_IMAGE_ORIGINS,
} from './constants.js';
import {
cameraDisplayCode,
fallbackHeadingFromId,
@@ -10,6 +13,8 @@ import { readResponseTextCapped } from '../common/http.js';
* standard, published keyless and refreshed daily (~1,800 cameras). */
export const TAIWAN_FREEWAY_CCTV_URL =
'https://tisvcloud.freeway.gov.tw/history/motc20/CCTV.xml';
export const TAIWAN_HIGHWAY_CCTV_URL =
'https://cctv-maintain.thb.gov.tw/opendataCCTVs.xml';
/** The whole network is ~1.2 MB. */
export const TAIWAN_MAX_CATALOG_BYTES = 6 * 1024 * 1024;
const DEFAULT_TAIWAN_MAX_SOURCES = 600;
@@ -35,6 +40,9 @@ const STREAM_TO_FRAME = new Map([
]);
const DIRECTIONS = { N: 'N', S: 'S', E: 'E', W: 'W' };
const TAIWAN_HIGHWAY_CAMERA_HOSTS = new Set(
TAIWAN_HIGHWAY_IMAGE_ORIGINS.map((origin) => new URL(origin).hostname),
);
/** Prioritization anchors, north to south, so a cap keeps the whole island
* rather than only greater Taipei. */
@@ -211,6 +219,74 @@ export function taiwanCameraToSource(row) {
};
}
/**
* One Highway Bureau row -> a source whose first MJPEG frame is served as a
* snapshot by the existing camera-frame proxy.
*
* @param {object} row - Output of parseTaiwanCctvXml.
* @returns {?object}
*/
export function taiwanHighwayCameraToSource(row) {
if (!row || typeof row !== 'object') return null;
const { lat, lon } = row;
if (!isLikelyTaiwanCoordinate(lat, lon)) return null;
let url;
try {
url = new URL(String(row.streamUrl ?? ''));
} catch {
return null;
}
if (
url.protocol !== 'https:' ||
url.username ||
url.password ||
url.port ||
url.search ||
url.hash ||
!TAIWAN_HIGHWAY_CAMERA_HOSTS.has(url.hostname) ||
!/^\/[A-Za-z0-9._+~-]{1,100}$/.test(url.pathname) ||
url.pathname.includes('..')
)
return null;
const rawId = String(row.cctvId ?? '').trim();
if (!/^[A-Za-z0-9-]{1,80}$/.test(rawId)) return null;
const cameraId = 'tw-thb-' + rawId.toLowerCase();
const road = String(row.roadName ?? '').trim();
const direction = DIRECTIONS[String(row.direction ?? '').trim()] || '';
const mile = String(row.mile ?? '').trim();
const section = [row.sectionStart, row.sectionEnd]
.map((part) => String(part ?? '').trim())
.filter(Boolean)
.join('→');
const name =
(road || mile || section
? [road, direction, mile, section].filter(Boolean).join(' ')
: '') || rawId;
return {
id: cameraId,
name: name.slice(0, 160),
city: 'Taiwan / 臺灣',
cityId: 'taiwan',
provider: 'Highway Bureau, MOTC',
lat,
lon,
headingDeg: fallbackHeadingFromId(cameraId),
headingConfidence: 'low',
pitchDeg: -18,
fovDeg: 44,
rangeM: 145,
mountHeightM: 8,
feedType: 'mjpeg',
url: url.href,
snapshotUrl: url.href,
sourceKind: 'taiwan-highway-motc',
license: 'Highway Bureau, MOTC open data',
code: cameraDisplayCode((road + ' ' + mile).trim().toUpperCase() || rawId),
};
}
let lastGood = null;
let lastGoodAt = 0;
let retryAfter = 0;
@@ -237,6 +313,49 @@ function taiwanSourceCap() {
: DEFAULT_TAIWAN_MAX_SOURCES;
}
/** Fetch the public provincial-road CCTV table as a no-key fallback. */
async function loadTaiwanHighwaySources() {
const signal = AbortSignal.timeout(CCTV_SOURCE_FETCH_TIMEOUT_MS);
const response = await fetch(TAIWAN_HIGHWAY_CCTV_URL, {
headers: { Accept: 'application/xml,text/xml' },
redirect: 'manual',
signal,
});
if (!response.ok) {
await response.body?.cancel().catch(() => {});
throw new Error(
response.status >= 300 && response.status < 400
? 'Highway Bureau catalog redirected; redirects are not followed'
: 'Highway Bureau catalog HTTP ' + response.status,
);
}
const xml = await readResponseTextCapped(
response,
TAIWAN_MAX_CATALOG_BYTES,
signal,
);
const cameras = new Map();
for (const row of parseTaiwanCctvXml(xml)) {
const camera = taiwanHighwayCameraToSource(row);
if (camera && !cameras.has(camera.id)) cameras.set(camera.id, camera);
}
if (!cameras.size)
throw new Error('no usable Highway Bureau cameras in the table');
const sources = prioritizeSources(
[...cameras.values()],
taiwanSourceCap(),
ANCHORS,
);
console.log(
'[CCTV] Loaded Taiwan Highway Bureau fallback: ' +
cameras.size +
' (using ' +
sources.length +
')',
);
return sources;
}
/**
* Fetch Taiwan Freeway Bureau cameras from the keyless MOTC 2.0 XML table.
* One bounded request per shared catalog refresh. A failed download serves the
@@ -262,7 +381,7 @@ export async function loadTaiwanSources() {
throw new Error(
response.status >= 300 && response.status < 400
? 'catalog redirected; redirects are not followed'
: `HTTP ${response.status}`,
: 'HTTP ' + response.status,
);
}
const xml = await readResponseTextCapped(
@@ -282,7 +401,11 @@ export async function loadTaiwanSources() {
ANCHORS,
);
console.log(
`[CCTV] Loaded Taiwan camera sources: ${cameras.size} (using ${sources.length})`,
'[CCTV] Loaded Taiwan camera sources: ' +
cameras.size +
' (using ' +
sources.length +
')',
);
lastGood = sources;
lastGoodAt = Date.now();
@@ -291,13 +414,37 @@ export async function loadTaiwanSources() {
} catch (error) {
retryAfter = Date.now() + TAIWAN_RETRY_AFTER_FAILURE_MS;
const fallback = fallbackSources();
console.warn(
`[CCTV] Taiwan camera download failed (${error?.message || error}); ` +
(fallback.length
? `serving the last good ${fallback.length} cameras`
: 'no Taiwan cameras') +
', next attempt in 1 h',
);
return fallback;
if (fallback.length) {
console.warn(
'[CCTV] Taiwan Freeway camera download failed (' +
(error?.message || error) +
'); serving the last good ' +
fallback.length +
' cameras, next attempt in 1 h',
);
return fallback;
}
try {
const sources = await loadTaiwanHighwaySources();
lastGood = sources;
lastGoodAt = Date.now();
console.warn(
'[CCTV] Taiwan Freeway camera download failed (' +
(error?.message || error) +
'); using ' +
sources.length +
' Highway Bureau cameras as fallback, next attempt in 1 h',
);
return sources;
} catch (fallbackError) {
console.warn(
'[CCTV] Taiwan Freeway camera download failed (' +
(error?.message || error) +
'); Highway Bureau fallback failed (' +
(fallbackError?.message || fallbackError) +
'); no Taiwan cameras, next attempt in 1 h',
);
return [];
}
}
}
+132
View File
@@ -0,0 +1,132 @@
import { createHash } from 'node:crypto';
import fs from 'node:fs';
import path from 'node:path';
import {
DEFAULT_WORLD_CCTV_SOURCE_FILE,
MLIT_RIVER_IMAGE_ORIGIN,
TRIPCHECK_IMAGE_ORIGIN,
WORLD_CCTV_SOURCE_COUNT,
} from './constants.js';
import { pinnedCameraUrl } from './normalize.js';
const SOURCE_POLICIES = Object.freeze({
'tripcheck-odot': {
idPrefix: 'world-or-',
origin: TRIPCHECK_IMAGE_ORIGIN,
bounds: [41, 47, -125, -116],
filename: /^\/roadcams\/cams\/[^/]+\.jpe?g$/i,
frameRefreshMs: 5 * 60 * 1000,
},
'mlit-river': {
idPrefix: 'world-jp-river-',
origin: MLIT_RIVER_IMAGE_ORIGIN,
bounds: [24, 46, 122, 154],
filename: /^\/cam\/now\/\d+\.jpg$/i,
// The provider asks clients to avoid frequent repeated image collection.
frameRefreshMs: 10 * 60 * 1000,
},
});
// MLIT currently serves this fixed JPEG on cameras whose images are paused
// for maintenance (verified against two Yamagata cameras on 2026-11-02).
const MLIT_ADJUSTMENT_IMAGE_SHA256 =
'dbd2c3790eb75edf9ca44162f7979f2a3799090e23f2c5142f63211a035cb191';
export function isMlitRiverAdjustmentImage(image) {
return (
image?.ok === true &&
Buffer.isBuffer(image.body) &&
createHash('sha256').update(image.body).digest('hex') ===
MLIT_ADJUSTMENT_IMAGE_SHA256
);
}
function safeFrameUrl(value, policy) {
const normalized = pinnedCameraUrl(value, policy.origin);
if (!normalized) return '';
const url = new URL(normalized);
return !url.search && !url.hash && policy.filename.test(url.pathname)
? url.href
: '';
}
function inBounds(lat, lon, [minLat, maxLat, minLon, maxLon]) {
return (
Number.isFinite(lat) &&
Number.isFinite(lon) &&
lat >= minLat &&
lat <= maxLat &&
lon >= minLon &&
lon <= maxLon
);
}
/**
* Load only the new image-backed networks from the add-on export. Existing
* Canadian networks, Québec entries, JMA placeholder images, and NZTA's
* scraped Journey Planner rows are deliberately excluded. Vegagerðin (Iceland)
* rows are excluded too: the Iceland pack loads the full live list of those
* cameras, so a World copy would draw each one twice.
*/
export function loadWorldCameraSources({ sourceRoot = process.cwd() } = {}) {
const filename = path.resolve(sourceRoot, DEFAULT_WORLD_CCTV_SOURCE_FILE);
let rows;
try {
const payload = JSON.parse(fs.readFileSync(filename, 'utf8'));
rows = Array.isArray(payload) ? payload : [];
} catch (error) {
console.warn('[CCTV] World camera catalog unavailable:', error?.message);
return [];
}
const seen = new Set();
const sources = [];
for (const row of rows) {
const sourceKind = String(row?.sourceKind || '');
const policy = SOURCE_POLICIES[sourceKind];
if (!policy) continue;
const id = String(row.id || '').trim();
const lat = Number(row.lat);
const lon = Number(row.lon);
const url = safeFrameUrl(row.url || row.snapshotUrl, policy);
const snapshotUrl = safeFrameUrl(row.snapshotUrl || row.url, policy);
if (
!id.startsWith(policy.idPrefix) ||
seen.has(id) ||
!inBounds(lat, lon, policy.bounds) ||
!url ||
!snapshotUrl ||
url !== snapshotUrl
)
continue;
seen.add(id);
sources.push({
id,
name: String(row.name || id)
.trim()
.slice(0, 180),
city: String(row.city || row.country || 'World'),
cityId: String(row.cityId || ''),
provider: String(row.provider || 'World traffic camera'),
lat,
lon,
headingDeg: Number.isFinite(Number(row.headingDeg))
? Number(row.headingDeg)
: undefined,
headingConfidence: String(row.headingConfidence || 'low'),
feedType: 'image',
sourceKind,
url,
snapshotUrl,
frameRefreshMs: policy.frameRefreshMs,
license: String(row.license || ''),
credit: String(row.credit || row.provider || ''),
});
}
console.log(
`[CCTV] Loaded World camera image sources: ${sources.length}/${WORLD_CCTV_SOURCE_COUNT}`,
);
return sources;
}
+81
View File
@@ -0,0 +1,81 @@
import { gzip } from 'node:zlib';
import { promisify } from 'node:util';
const gzipAsync = promisify(gzip);
/**
* Opt-in gzip for text bodies a proxy already holds in memory.
*
* Not a middleware: `cctv/media.js` streams bodies and answers range requests,
* and several providers forward binary upstream types that gzip only grows.
* `encodedBody` returns the body with its headers so the two cannot disagree.
*/
/** Skip compression for bodies smaller than 1 KiB. */
const MIN_COMPRESSED_BYTES = 1024;
/**
* RFC 9110 section 12.4.2 `qvalue`. Matching the whole token keeps `q=1oops`
* and `q=2` unreadable instead of rounding them into acceptance.
*/
const QVALUE = /^(?:0(?:\.\d{0,3})?|1(?:\.0{0,3})?)$/;
/**
* Whether `Accept-Encoding` offers gzip. Anything unreadable (no header, a bad
* q value, `x-gzip`) reads as no: every client can decode identity.
*
* @param {import('node:http').IncomingMessage} request
* @returns {boolean}
*/
export function acceptsGzip(request) {
const header = request?.headers?.['accept-encoding'];
if (typeof header !== 'string') return false;
return header.split(',').some((part) => {
const [token, ...parameters] = part.split(';');
if (token.trim().toLowerCase() !== 'gzip') return false;
const quality = parameters
.map((parameter) => parameter.trim().toLowerCase())
.find((parameter) => parameter.startsWith('q='));
if (quality === undefined) return true;
// `q=0` names gzip and refuses it.
const value = quality.slice(2);
return QVALUE.test(value) && Number(value) > 0;
});
}
/**
* Gzip a body once for a caller that caches it. Async, so the ~860 KB OpenSky
* snapshot does not block other requests while it compresses.
*
* Returns null, meaning "serve identity", when the body is under 1 KiB or gzip
* does not shrink it.
*
* @param {string|Buffer} body - the exact bytes the identity response sends.
* @returns {Promise<Buffer|null>}
*/
export async function gzipBody(body) {
const raw = Buffer.isBuffer(body) ? body : Buffer.from(body);
if (raw.byteLength < MIN_COMPRESSED_BYTES) return null;
const compressed = await gzipAsync(raw);
return compressed.byteLength < raw.byteLength ? compressed : null;
}
/**
* Pick the body and the headers that describe it. `Vary` is on both branches
* so a shared cache never serves one encoding to a client that asked for the
* other.
*
* @param {import('node:http').IncomingMessage} request
* @param {string|Buffer} identity - the uncompressed body.
* @param {Buffer|null} compressed - `gzipBody(identity)`, or null.
* @returns {{body: string|Buffer, headers: Record<string,string>}}
*/
export function encodedBody(request, identity, compressed) {
if (compressed && acceptsGzip(request)) {
return {
body: compressed,
headers: { 'Content-Encoding': 'gzip', Vary: 'Accept-Encoding' },
};
}
return { body: identity, headers: { Vary: 'Accept-Encoding' } };
}
+4
View File
@@ -0,0 +1,4 @@
/** Identify each provider while keeping the project contact URL consistent. */
export function providerUserAgent(name) {
return `gods-eye-view-${name}-proxy/1.0 (+https://github.com/bilawalsidhu/gods-eye-view)`;
}
+370
View File
@@ -0,0 +1,370 @@
import path from 'node:path';
import { promises as fs } from 'node:fs';
import { randomUUID } from 'node:crypto';
import { normalizeInatObservations } from '../../../src/layers/wildlifeSightings/records.js';
import { readResponseJsonCapped } from '../common/http.js';
import { clampInt } from '../common/query.js';
import { providerUserAgent } from '../common/userAgent.js';
import { projectInatGrid, validInatGridTile } from './grid.js';
import { createInatQueue } from './queue.js';
const DAILY_BUDGET = 8000;
const DAY_MS = 24 * 60 * 60_000;
// One page is capped independently of upstream Content-Length.
const MAX_RESPONSE_BYTES = 8 * 1024 * 1024;
const TIMEOUT_MS = 20_000;
const API_URL = 'https://api.inaturalist.org/v1';
// Request only the fields used by the public observation projection.
const OBSERVATIONS_URL = 'https://api.inaturalist.org/v2/observations';
const OBSERVATION_FIELDS =
'(id:!t,uri:!t,location:!t,geojson:all,geoprivacy:!t,taxon_geoprivacy:!t,' +
'obscured:!t,quality_grade:!t,observed_on:!t,' +
'time_observed_at:!t,faves_count:!t,' +
'num_identification_agreements:!t,' +
'photos:(url:!t,attribution:!t,license_code:!t),' +
'taxon:(name:!t,preferred_common_name:!t,iconic_taxon_name:!t,' +
'threatened:!t))';
const COOLDOWN_MS = 60_000;
const FAILURE_COOLDOWN_MS = 15_000;
/** Seconds and HTTP dates are accepted; a missing value earns one minute. */
export function inatRetryCooldownMs(raw, nowMs) {
const value = typeof raw === 'string' ? raw.trim() : '';
if (value) {
const seconds = /^\d+(?:\.\d+)?$/.test(value) ? Number(value) : NaN;
const deadline =
!Number.isFinite(seconds) && /[a-z]/i.test(value)
? Date.parse(value)
: NaN;
const delay = Number.isFinite(seconds) ? seconds * 1000 : deadline - nowMs;
if (Number.isFinite(delay) && delay >= 0) return delay;
}
return COOLDOWN_MS;
}
function upstreamError(code, status, retryAfterS, upstreamStatus) {
return Object.assign(new Error(code), {
code,
status,
retryAfterS,
upstreamStatus,
});
}
function observationFilters(params) {
const query = new URLSearchParams({
verifiable: 'true',
captive: 'false',
// d1 is an observation DATE, unlike creation-time created_d1.
d1: params.observedD1,
});
if (params.taxa !== 'whales') query.set('photos', 'true');
if (params.quality === 'research') query.set('quality_grade', 'research');
if (params.taxa === 'whales') query.set('taxon_id', '152871');
else if (['Aves', 'Mammalia', 'Insecta', 'Plantae'].includes(params.taxa))
query.set('iconic_taxa', params.taxa);
return query;
}
/**
* Shared iNaturalist governor. Only request counters are persisted; observation
* payloads are projected immediately and are never written to disk.
*/
export function createInatClient({
fetchImpl = (...args) => globalThis.fetch(...args),
now = () => Date.now(),
cacheDir = path.join(process.cwd(), '.gev-cache', 'inaturalist'),
warn = (...args) => console.warn(...args),
queueOptions,
} = {}) {
const enqueue = createInatQueue(queueOptions);
let budget = null;
let budgetLoaded = false;
let budgetTail = Promise.resolve();
let cooldownUntil = 0;
// An unavailable UTFGrid tile must never place observation queries into a
// 502 cooldown. Only an upstream 429 is a provider-wide restriction.
const failures = new Map();
function dailyLimit() {
return clampInt(
process.env.INATURALIST_DAILY_BUDGET,
1,
Number.MAX_SAFE_INTEGER,
DAILY_BUDGET,
);
}
function day() {
return new Date(now()).toISOString().slice(0, 10);
}
function currentBudget() {
const date = day();
if (budget?.date !== date) budget = { date, count: 0 };
return budget;
}
async function readBudgetOnce() {
if (budgetLoaded) return;
budgetLoaded = true;
if (cacheDir === null) return;
try {
const value = JSON.parse(
await fs.readFile(path.join(cacheDir, 'budget.json'), 'utf8'),
);
if (
typeof value?.date === 'string' &&
/^\d{4}-\d{2}-\d{2}$/.test(value.date) &&
Number.isSafeInteger(value.count) &&
value.count >= 0
)
budget = { date: value.date, count: value.count };
} catch (error) {
if (error.code !== 'ENOENT')
warn('[inaturalist-proxy] budget read failed');
}
}
async function persistBudget(value) {
if (cacheDir === null) return;
const destination = path.join(cacheDir, 'budget.json');
const temporary = `${destination}.${randomUUID()}.tmp`;
try {
await fs.mkdir(cacheDir, { recursive: true });
await fs.writeFile(temporary, JSON.stringify(value), 'utf8');
await fs.rename(temporary, destination);
} catch {
await fs.rm(temporary, { force: true }).catch(() => {});
warn('[inaturalist-proxy] budget write failed');
}
}
function retrySeconds(until = cooldownUntil) {
return Math.max(0, Math.ceil((until - now()) / 1000));
}
function checkCooldown(resource) {
if (now() < cooldownUntil)
throw upstreamError('inaturalist_rate_limited', 429, retrySeconds(), 429);
const failure = failures.get(resource);
if (failure && now() < failure.until)
throw upstreamError(
'inaturalist_unavailable',
502,
retrySeconds(failure.until),
failure.status,
);
failures.delete(resource);
}
// Reserve and atomically persist one attempt before issuing it. A recovered
// tail prevents one disk/reservation failure from poisoning future calls.
function reserveAttempt(signal, resource) {
const result = budgetTail.then(async () => {
signal?.throwIfAborted();
checkCooldown(resource);
await readBudgetOnce();
signal?.throwIfAborted();
checkCooldown(resource);
const value = currentBudget();
if (value.count >= dailyLimit()) {
const retry = Math.max(
1,
Math.ceil((Date.parse(`${day()}T00:00:00Z`) + DAY_MS - now()) / 1000),
);
throw upstreamError('inaturalist_budget_exhausted', 503, retry);
}
value.count += 1;
await persistBudget({ ...value });
});
budgetTail = result.catch(() => {});
return result;
}
async function requestProjected(url, project, resource, { signal } = {}) {
signal?.throwIfAborted();
checkCooldown(resource);
return enqueue(() => dispatch(url, project, resource, signal), {
signal,
priority: resource === 'observations' ? 0 : 1,
});
}
async function dispatch(url, project, resource, signal) {
signal?.throwIfAborted();
await reserveAttempt(signal, resource);
signal?.throwIfAborted();
checkCooldown(resource);
const timeout = AbortSignal.timeout(TIMEOUT_MS);
const requestSignal = signal ? AbortSignal.any([signal, timeout]) : timeout;
let upstreamStatus;
let retryDelay = FAILURE_COOLDOWN_MS;
try {
const response = await fetchImpl(url, {
signal: requestSignal,
redirect: 'error',
headers: {
Accept: 'application/json',
'User-Agent': providerUserAgent('wildlife-sightings'),
},
});
upstreamStatus = response.status;
if (!response.ok) {
await response.body?.cancel().catch(() => {});
if (response.status === 429) {
const delay = inatRetryCooldownMs(
response.headers.get('retry-after'),
now(),
);
cooldownUntil = Math.max(cooldownUntil, now() + delay);
throw upstreamError(
'inaturalist_rate_limited',
429,
retrySeconds(),
429,
);
}
if (response.headers.has('retry-after'))
retryDelay = inatRetryCooldownMs(
response.headers.get('retry-after'),
now(),
);
throw upstreamError('inaturalist_unavailable', 502);
}
const payload = await readResponseJsonCapped(
response,
MAX_RESPONSE_BYTES,
requestSignal,
);
return project(payload);
} catch (error) {
// Log only safe diagnostic categories, never a URL, body, status text or
// exception message (each can contain request/private information).
const reason = signal?.aborted
? 'abort'
: timeout.aborted || error.name === 'TimeoutError'
? 'timeout'
: upstreamStatus && upstreamStatus >= 400
? 'http'
: error.code === 'RESPONSE_TOO_LARGE'
? 'response-too-large'
: upstreamStatus
? 'invalid-payload'
: 'network';
warn(
`[inaturalist-proxy] upstream refresh failed resource=${resource} status=${upstreamStatus ?? 'none'} reason=${reason}`,
);
if (signal?.aborted) throw error;
// Network, abort, schema and size failures are not upstream cooldown
// signals. A later healthy request may recover immediately.
if (upstreamStatus >= 500)
failures.set(resource, {
until: now() + retryDelay,
status: upstreamStatus,
});
const failure = failures.get(resource);
throw upstreamError(
error.status === 429
? 'inaturalist_rate_limited'
: 'inaturalist_unavailable',
error.status === 429 ? 429 : 502,
error.status === 429
? retrySeconds()
: failure
? retrySeconds(failure.until)
: undefined,
upstreamStatus === 429 || upstreamStatus >= 500
? upstreamStatus
: undefined,
);
}
}
/** Request one bounded observation sample and project its public fields. */
async function getObservations(params, options = {}) {
const query = observationFilters(params);
// View samples stay stable in the route cache until their TTL expires.
query.set('order_by', params.scope === 'view' ? 'random' : 'observed_on');
if (params.scope !== 'view') query.set('order', 'desc');
query.set('per_page', '200');
query.set('fields', OBSERVATION_FIELDS);
if (params.scope === 'view') {
query.set('swlat', String(params.box.south));
query.set('swlng', String(params.box.west));
query.set('nelat', String(params.box.north));
query.set('nelng', String(params.box.east));
}
return requestProjected(
`${OBSERVATIONS_URL}?${query}`,
(payload) => {
if (
!Array.isArray(payload?.results) ||
!Number.isSafeInteger(payload.total_results) ||
payload.total_results < 0
)
throw upstreamError('inaturalist_unavailable', 502);
const rows = normalizeInatObservations(payload.results, {
nowMs: now(),
windowMs: (params.taxa === 'whales' ? 7 : 2) * DAY_MS,
});
// Empty views need total_results: 0. A wholly invalid page must not
// replace good data. Valid rows excluded by the finer rolling-time
// cutoff are a healthy empty result of the upstream DATE over-fetch.
if (
!rows ||
(!rows.length &&
(payload.results.length || payload.total_results) &&
!normalizeInatObservations(payload.results)?.length)
)
throw upstreamError('inaturalist_unavailable', 502);
return {
rows,
totalResults: payload.total_results,
...(payload.total_results > payload.results.length
? { partial: true, rawRowCount: payload.total_results }
: {}),
};
},
'observations',
options,
);
}
/** Decode one validated density tile without observation identifiers. */
async function getGrid(params, options = {}) {
if (!validInatGridTile(params)) throw new TypeError('Invalid grid tile');
const query = observationFilters(params);
return requestProjected(
`${API_URL}/grid/${params.z}/${params.x}/${params.y}.grid.json?${query}`,
(payload) => projectInatGrid(payload, params),
'grid',
options,
);
}
return {
getObservations,
getGrid,
/** Current provider-wide or resource-specific retry information. */
cooldownState(resource) {
if (now() < cooldownUntil)
return {
upstreamStatus: 429,
retryAfterS: retrySeconds(),
};
const failure = failures.get(resource);
if (failure && now() < failure.until)
return {
upstreamStatus: failure.status,
retryAfterS: retrySeconds(failure.until),
};
return undefined;
},
/** The current UTC day's reserved upstream attempts and configured limit. */
budgetState() {
const value = currentBudget();
return { ...value, budget: dailyLimit() };
},
};
}
@@ -0,0 +1,163 @@
import test from 'node:test';
import assert from 'node:assert/strict';
import { createInatClient } from './client.js';
import { syntheticGrid } from './grid.fixture.mjs';
const NOW = Date.parse('2026-11-04T18:05:00Z');
const observation = (id, time) => ({
id,
geojson: { coordinates: [-122.3, 47.6] },
quality_grade: 'research',
time_observed_at: time,
observed_on: time.slice(0, 10),
// Upload age does not determine whether an observation is recent.
created_at: '2026-11-04T18:00:00Z',
taxon: { name: 'Synthetic taxon', iconic_taxon_name: 'Aves' },
});
function clientFor(results) {
return createInatClient({
now: () => NOW,
cacheDir: null,
queueOptions: { intervalMs: 0 },
warn() {},
fetchImpl: async () =>
Response.json({ results, total_results: results.length }),
});
}
test('iNaturalist drops ancient and boundary-expired observations uploaded today', async () => {
const client = clientFor([
observation(1, '2010-11-04T18:00:00Z'),
observation(2, '2026-11-02T18:04:59Z'),
observation(3, '2026-11-02T18:05:00Z'),
]);
const result = await client.getObservations({
scope: 'global',
taxa: 'all',
observedD1: '2026-11-02',
});
assert.deepEqual(
result.rows.map((row) => row.stableId),
['inat:3'],
);
});
test('a valid date-filtered page can be empty after the exact recency cutoff', async () => {
const client = clientFor([observation(1, '2026-11-02T06:00:00Z')]);
const result = await client.getObservations({
scope: 'global',
taxa: 'all',
observedD1: '2026-11-02',
});
assert.deepEqual(result.rows, []);
assert.equal(result.totalResults, 1);
});
const params = {
scope: 'view',
box: { south: 47, west: -123, north: 48, east: -122 },
taxa: 'all',
observedD1: '2026-11-02',
};
test('observations and tiles share pacing, and observation priority wins a cold burst', async () => {
let at = 0;
const calls = [];
const client = createInatClient({
cacheDir: null,
now: () => NOW + at,
queueOptions: {
now: () => at,
wait: async (ms) => {
at += ms;
},
},
fetchImpl: async (url) => {
const isGrid = new URL(url).pathname.includes('/grid/');
calls.push([isGrid ? 'grid' : 'observations', at]);
return Response.json(
isGrid ? syntheticGrid() : { results: [], total_results: 0 },
);
},
});
const tiles = Array.from({ length: 3 }, (_, x) =>
client.getGrid({ ...params, z: 2, x, y: 1 }),
);
await Promise.all([...tiles, client.getObservations(params)]);
assert.deepEqual(calls, [
['observations', 0],
['grid', 0],
['grid', 0],
['grid', 0],
]);
assert.equal(client.budgetState().count, 4);
});
test('upstream status and safe failure reasons are logged without upstream messages or URLs', async () => {
for (const [response, expected] of [
[
() =>
new Response('request-secret', {
status: 503,
statusText: 'request-secret',
}),
/status=503 reason=http/,
],
[
() => Promise.reject(new DOMException('request-secret', 'TimeoutError')),
/status=none reason=timeout/,
],
[
() => Response.json({ unexpected: 'request-secret' }),
/status=200 reason=invalid-payload/,
],
]) {
const warnings = [];
const client = createInatClient({
cacheDir: null,
now: () => NOW,
queueOptions: { intervalMs: 0 },
warn: (message) => warnings.push(message),
fetchImpl: response,
});
await assert.rejects(client.getObservations(params), { status: 502 });
assert.equal(warnings.length, 1);
assert.match(warnings[0], expected);
assert.doesNotMatch(warnings[0], /request-secret|https?:|swlat|d1=/);
}
});
test('a dispatched abort is diagnosed, spends only its attempt, and leaves the next request healthy', async () => {
const warnings = [];
const controller = new AbortController();
let started;
const didStart = new Promise((resolve) => {
started = resolve;
});
let calls = 0;
const client = createInatClient({
cacheDir: null,
now: () => NOW,
queueOptions: { intervalMs: 0 },
warn: (message) => warnings.push(message),
fetchImpl: async (url, { signal }) => {
calls += 1;
if (calls > 1) return Response.json({ results: [], total_results: 0 });
started();
return new Promise((resolve, reject) =>
signal.addEventListener('abort', () => reject(signal.reason), {
once: true,
}),
);
},
});
const pending = client.getObservations(params, { signal: controller.signal });
const failed = assert.rejects(pending, { name: 'AbortError' });
await didStart;
controller.abort();
await failed;
assert.match(warnings[0], /resource=observations status=none reason=abort/);
await client.getObservations(params);
assert.equal(calls, 2);
});
@@ -0,0 +1,51 @@
/** Build a synthetic UTFGrid with logical cell counts and repeated hit pixels. */
export function syntheticGrid({
z = 2,
x = 1,
y = 1,
cells = [
{ row: 1, column: 51, count: 6, iconic: 3 },
{ row: 3, column: 49, count: 12, iconic: 47170 },
],
} = {}) {
const pixels = Array.from({ length: 64 }, () => Array(64).fill(' '));
const keys = [''];
const data = {};
cells.forEach((cell, index) => {
const key = String(900_000_000 + index);
keys.push(key);
const value = index + 1;
let code = value + 32;
if (code >= 34) code += 1;
if (code >= 92) code += 1;
const character = String.fromCharCode(code);
for (const [row, column] of [
[cell.row, cell.column],
[cell.row - 1, cell.column],
[cell.row + 1, cell.column],
[cell.row, cell.column - 1],
[cell.row, cell.column + 1],
]) {
if (row >= 0 && row < 64 && column >= 0 && column < 64)
pixels[row][column] = character;
}
data[key] = {
captive: null,
cellCount: cell.count,
geoprivacy: null,
id: Number(key),
latitude:
(Math.atan(
Math.sinh(Math.PI * (1 - 2 * ((y + cell.row / 64) / 2 ** z))),
) *
180) /
Math.PI,
longitude: ((x + cell.column / 64) / 2 ** z) * 360 - 180,
private_location: null,
quality_grade: 'needs_id',
'taxon.iconic_taxon_id': cell.iconic,
uuid: `00000000-0000-4000-8000-${String(index).padStart(12, '0')}`,
};
});
return { grid: pixels.map((row) => row.join('')), keys, data };
}
+127
View File
@@ -0,0 +1,127 @@
// iNaturalist's operational /v1/grid/{z}/{x}/{y}.grid.json endpoint is not
// covered by its v1 Swagger contract. Validate the entire UTFGrid shape before
// accepting any cell; callers preserve the last good sample on schema drift.
import {
wildlifeGridCellId,
WILDLIFE_GRID_MAX_ZOOM,
} from '../../../src/layers/wildlifeSightings/gridViewport.js';
const INAT_GRID_SIZE = 64;
export const INAT_GRID_DECODE_MEASURE = 'wildlife:server:tile-decode';
const grades = new Set(['research', 'needs_id', 'casual']);
const object = (value) =>
!!value && typeof value === 'object' && !Array.isArray(value);
/** Accept only integer XYZ coordinates within the supported zoom range. */
export function validInatGridTile({ z, x, y } = {}) {
return (
Number.isInteger(z) &&
z >= 0 &&
z <= WILDLIFE_GRID_MAX_ZOOM &&
Number.isInteger(x) &&
Number.isInteger(y) &&
x >= 0 &&
y >= 0 &&
x < 2 ** z &&
y < 2 ** z
);
}
/** Pure allow-list projection. A UTFGrid key is an observation identifier and
* its repeated pixels form a dilated hit area, not additional occupied cells.
* Neither those keys nor representative/private fields cross this boundary. */
export function projectInatGrid(payload, tile) {
const start = `${INAT_GRID_DECODE_MEASURE}:start`;
const end = `${INAT_GRID_DECODE_MEASURE}:end`;
performance.mark(start);
try {
return decodeInatGrid(payload, tile);
} finally {
performance.mark(end);
// Keep the latest sample inspectable without retaining every tile forever.
// PerformanceObservers still receive each individual decode measurement.
performance.clearMeasures(INAT_GRID_DECODE_MEASURE);
performance.measure(INAT_GRID_DECODE_MEASURE, start, end);
performance.clearMarks(start);
performance.clearMarks(end);
}
}
function decodeInatGrid(payload, tile) {
const invalid = () => {
throw new TypeError('Invalid iNaturalist UTFGrid');
};
if (
!validInatGridTile(tile) ||
!object(payload) ||
!Array.isArray(payload.grid) ||
payload.grid.length !== INAT_GRID_SIZE ||
!payload.grid.every(
(row) => typeof row === 'string' && row.length === INAT_GRID_SIZE,
) ||
!Array.isArray(payload.keys) ||
!payload.keys.length ||
payload.keys.length > INAT_GRID_SIZE ** 2 + 1 ||
payload.keys[0] !== '' ||
!payload.keys.every(
(key) => typeof key === 'string' && key.length <= 128,
) ||
new Set(payload.keys).size !== payload.keys.length ||
!object(payload.data)
)
invalid();
const dataKeys = Object.keys(payload.data);
if (dataKeys.length !== payload.keys.length - 1) invalid();
for (const key of payload.keys.slice(1)) {
if (!key || !Object.hasOwn(payload.data, key)) invalid();
const value = payload.data[key];
if (
!object(value) ||
!Number.isSafeInteger(value.cellCount) ||
value.cellCount <= 0 ||
!Number.isFinite(value.latitude) ||
Math.abs(value.latitude) > 90 ||
!Number.isFinite(value.longitude) ||
Math.abs(value.longitude) > 180 ||
!grades.has(value.quality_grade) ||
!(
value['taxon.iconic_taxon_id'] === null ||
(Number.isSafeInteger(value['taxon.iconic_taxon_id']) &&
value['taxon.iconic_taxon_id'] > 0)
)
)
invalid();
}
const occupied = new Set();
for (const line of payload.grid) {
for (let column = 0; column < INAT_GRID_SIZE; column += 1) {
let code = line.charCodeAt(column);
if (code < 32 || code === 34 || code === 92) invalid();
if (code >= 93) code -= 1;
if (code >= 35) code -= 1;
const index = code - 32;
if (index < 0 || index >= payload.keys.length) invalid();
if (index) occupied.add(payload.keys[index]);
}
}
if (occupied.size !== dataKeys.length) invalid();
const cells = new Map();
for (const key of occupied) {
const value = payload.data[key];
const id = wildlifeGridCellId(value.latitude, value.longitude, tile.z);
const cell = {
id,
cellCount: value.cellCount,
lat: value.latitude,
lon: value.longitude,
};
// Two different representatives claiming one logical cell are a contract
// change; summing or choosing one silently would fabricate the count.
if (cells.has(id)) invalid();
cells.set(id, cell);
}
const rows = [...cells.values()];
const totalResults = rows.reduce((total, cell) => total + cell.cellCount, 0);
if (!Number.isSafeInteger(totalResults)) invalid();
return { cells: rows, totalResults };
}
+374
View File
@@ -0,0 +1,374 @@
import test from 'node:test';
import assert from 'node:assert/strict';
import { projectInatGrid } from './grid.js';
import { syntheticGrid } from './grid.fixture.mjs';
import { installInatProxy, NOW } from './proxy.fixture.mjs';
import { createInatClient } from './client.js';
const tile = { z: 2, x: 1, y: 1 };
const gridUrl = '/grid?z=2&x=1&y=1&taxa=Aves&quality=research';
const install = (options, hook) => installInatProxy(gridUrl, options, hook);
test('UTFGrid projection validates the full shape and counts each logical occupied cell once', () => {
const raw = syntheticGrid();
const projected = projectInatGrid(raw, tile);
assert.equal(raw.grid.length, 64);
assert.ok(raw.grid.every((row) => row.length === 64));
assert.equal(
raw.grid.join('').split('!').length - 1,
5,
'one key is repeated by its hit area',
);
assert.equal(projected.cells.length, 2);
assert.equal(projected.totalResults, 18);
assert.deepEqual(projected.cells[0], {
id: 'grid:2/1/1:1:51',
cellCount: 6,
lat: raw.data[raw.keys[1]].latitude,
lon: raw.data[raw.keys[1]].longitude,
});
for (const cell of projected.cells)
assert.deepEqual(Object.keys(cell).sort(), [
'cellCount',
'id',
'lat',
'lon',
]);
for (const privateField of [
'900000000',
'uuid',
'geoprivacy',
'private_location',
'captive',
'quality_grade',
])
assert.ok(!JSON.stringify(projected).includes(privateField));
});
test('legitimate blank grid and null/unknown iconic ids remain safe', () => {
assert.deepEqual(projectInatGrid(syntheticGrid({ cells: [] }), tile), {
cells: [],
totalResults: 0,
});
const raw = syntheticGrid({
cells: [{ row: 1, column: 1, count: 1, iconic: null }],
});
const projected = projectInatGrid(raw, tile);
assert.equal(projected.cells[0].cellCount, 1);
raw.data[raw.keys[1]]['taxon.iconic_taxon_id'] = 999999;
assert.deepEqual(projectInatGrid(raw, tile), projected);
});
test('schema drift anywhere fails closed, including unused, malformed and orphan cells', () => {
const changes = [
(raw) => {
raw.grid.pop();
},
(raw) => {
raw.grid[0] = '';
},
(raw) => {
raw.grid[0] = '~'.repeat(64);
},
(raw) => {
raw.keys[0] = 'not-empty';
},
(raw) => {
raw.keys.push(raw.keys[1]);
},
(raw) => {
delete raw.data[raw.keys[1]];
},
(raw) => {
raw.data.extra = {};
},
(raw) => {
raw.data[raw.keys[1]].cellCount = -1;
},
(raw) => {
raw.data[raw.keys[1]].latitude = '47';
},
(raw) => {
raw.data[raw.keys[1]].longitude = 181;
},
(raw) => {
raw.data[raw.keys[1]].quality_grade = 'new-grade';
},
(raw) => {
delete raw.data[raw.keys[1]]['taxon.iconic_taxon_id'];
},
(raw) => {
raw.data[raw.keys[1]]['taxon.iconic_taxon_id'] = 'Aves';
},
(raw) => {
raw.grid = Array(64).fill(' '.repeat(64));
},
];
for (const change of changes) {
const raw = syntheticGrid();
change(raw);
assert.throws(
() => projectInatGrid(raw, tile),
/Invalid iNaturalist UTFGrid/,
);
}
});
for (const hook of ['configureServer', 'configurePreviewServer']) {
test(`${hook}: fixed grid URL shares filters, caps and allow-list cache`, async () => {
const calls = [];
const cache = new Map();
const request = install(
{
cache,
fetchImpl: async (url, options) => {
calls.push(new URL(url));
assert.equal(options.redirect, 'error');
assert.ok(options.signal instanceof AbortSignal);
assert.match(
options.headers['User-Agent'],
/^gods-eye-view-wildlife-sightings-proxy/,
);
return Response.json(syntheticGrid());
},
},
hook,
);
assert.equal(calls.length, 0);
const response = await request(
`${gridUrl}&url=https://attacker.test&per_page=999`,
);
assert.equal(response.status, 200);
assert.equal(response.body.cells.length, 2);
assert.equal(response.body.fetchedAt, NOW);
assert.equal(response.headers['Cache-Control'], 'no-store');
assert.equal(
calls[0].origin + calls[0].pathname,
'https://api.inaturalist.org/v1/grid/2/1/1.grid.json',
);
assert.deepEqual(Object.fromEntries(calls[0].searchParams), {
verifiable: 'true',
captive: 'false',
d1: '2026-11-02',
photos: 'true',
quality_grade: 'research',
iconic_taxa: 'Aves',
});
await request(gridUrl);
assert.equal(calls.length, 1, 'fresh stable tile/filter key is cached');
assert.ok(!JSON.stringify([...cache]).includes('private_location'));
assert.ok(!JSON.stringify([...cache]).includes('900000000'));
const whale = await request('/grid?z=2&x=1&y=1&taxa=whales&confirmed=true');
assert.equal(whale.status, 200);
assert.equal(calls[1].searchParams.get('taxon_id'), '152871');
assert.equal(calls[1].searchParams.get('d1'), '2026-10-28');
assert.equal(calls[1].searchParams.has('photos'), false);
});
}
test('grid admission rejects cross-site/method/path/tile before acquisition', async () => {
const request = install({
fetchImpl() {
assert.fail('invalid grid request acquired');
},
});
assert.equal(
(
await request(gridUrl, {
headers: {
host: 'localhost:5173',
origin: 'https://other.test',
'sec-fetch-site': 'cross-site',
},
})
).status,
403,
);
assert.equal((await request('/grid', { method: 'POST' })).status, 405);
assert.equal((await request('/grid/extra')).status, 404);
for (const query of [
'',
'z=2&x=4&y=0',
'z=-1&x=0&y=0',
'z=23&x=0&y=0',
'z=2&x=1.5&y=0',
'z=2&x=&y=0',
'z=2&x=0&y=4',
]) {
const result = await request(`/grid?${query}`);
assert.equal(result.status, 400, query);
assert.equal(result.body.error, 'invalid_tile');
}
});
test('grid refresh single-flight, stale fallback and hour-key expiry keep the original age', async () => {
let time = NOW;
let calls = 0;
let resolve;
let mode = 'pending';
const request = install({
now: () => time,
fetchImpl: async () => {
calls += 1;
if (mode === 'pending')
return new Promise((done) => {
resolve = done;
});
if (mode === 'bad') return Response.json({ grid: [] });
return Response.json(syntheticGrid());
},
});
const first = request();
const second = request();
for (let i = 0; i < 100 && !resolve; i += 1) await new Promise(setImmediate);
assert.equal(calls, 1);
resolve(Response.json(syntheticGrid()));
assert.equal((await first).status, 200);
assert.equal((await second).status, 200);
time += 900_001;
mode = 'bad';
const stale = await request();
assert.equal(stale.status, 200);
assert.equal(stale.body.stale, true);
assert.equal(stale.body.fetchedAt, NOW);
assert.equal(stale.headers['X-Data-Stale'], 'true');
time = NOW + 3_600_001;
const expired = await request();
assert.equal(expired.status, 502);
assert.equal(expired.body.error, 'inaturalist_unavailable');
});
test('grid and observations share the upstream cooldown and daily reservations', async () => {
let calls = 0;
const client = createInatClient({
now: () => NOW,
cacheDir: null,
queueOptions: { intervalMs: 0 },
warn() {},
fetchImpl: async () => {
calls += 1;
return new Response('', {
status: 429,
headers: { 'Retry-After': '120' },
});
},
});
await assert.rejects(
client.getGrid({ ...tile, taxa: 'all', observedD1: '2026-11-02' }),
(error) => error.status === 429 && error.retryAfterS === 120,
);
await assert.rejects(
client.getObservations({
scope: 'global',
taxa: 'all',
observedD1: '2026-11-02',
}),
(error) => error.status === 429,
);
assert.equal(calls, 1);
assert.equal(client.budgetState().count, 1);
});
test('grid body cap and upstream failures return fixed errors; only real 5xx cools down', async () => {
for (const [response, retryAfter] of [
[() => new Response('x'.repeat(8 * 1024 * 1024 + 1)), undefined],
[() => new Response('', { status: 503 }), '15'],
]) {
const request = install({ fetchImpl: async () => response() });
const result = await request();
assert.equal(result.status, 502);
assert.deepEqual(result.body, { error: 'inaturalist_unavailable' });
assert.equal(result.headers['Retry-After'], retryAfter);
}
});
test('a failed tile preserves last good cells and cannot send unrelated observations into 502', async () => {
let at = NOW;
let failGrid = false;
const calls = [];
const warnings = [];
const request = install({
now: () => at,
warn: (message) => warnings.push(message),
fetchImpl: async (url) => {
const isGrid = new URL(url).pathname.includes('/grid/');
calls.push(isGrid ? 'grid' : 'observations');
if (isGrid && failGrid)
return new Response('private upstream text', {
status: 503,
headers: { 'Retry-After': '45' },
});
return Response.json(
isGrid ? syntheticGrid() : { results: [], total_results: 0 },
);
},
});
const first = await request();
at += 900_001;
failGrid = true;
const failed = await request();
assert.equal(failed.status, 200);
assert.equal(failed.headers['X-Data-Stale'], 'true');
assert.equal(failed.body.stale, true);
assert.equal(failed.body.fetchedAt, first.body.fetchedAt);
assert.deepEqual(failed.body.cells, first.body.cells);
assert.equal(
(await request('/observations?scope=global&taxa=whales')).status,
200,
);
assert.deepEqual(calls, ['grid', 'grid', 'observations']);
assert.match(warnings[0], /resource=grid status=503 reason=http/);
assert.doesNotMatch(warnings[0], /private upstream text/);
// The tile cooldown also preserves its data without another upstream try.
const cooling = await request();
assert.equal(cooling.body.stale, true);
assert.equal(cooling.headers['Retry-After'], '45');
assert.equal(cooling.headers['X-Upstream-Status'], '503');
assert.equal(calls.length, 3);
at += 45_000;
failGrid = false;
const refreshing = await request();
assert.equal(refreshing.body.stale, true);
await new Promise(setImmediate);
const recovered = await request();
assert.equal(recovered.status, 200);
assert.equal(recovered.body.stale, undefined);
assert.equal(
calls.length,
4,
'a failed queue task does not poison later tile work',
);
});
test('twelve queued tiles leave observation capacity and give the next dispatch to observations', async () => {
const calls = [];
let release;
const request = install({
fetchImpl: (url) => {
const path = new URL(url).pathname;
calls.push(path);
if (calls.length === 1)
return new Promise((resolve) => {
release = resolve;
});
return Response.json(
path.includes('/grid/')
? syntheticGrid()
: { results: [], total_results: 0 },
);
},
});
const first = request('/grid?z=4&x=0&y=1');
for (let i = 0; i < 100 && !release; i += 1) await new Promise(setImmediate);
assert.ok(release);
const tiles = Array.from({ length: 11 }, (_, i) =>
request(`/grid?z=4&x=${i + 1}&y=1`),
);
const observation = request('/observations?scope=global');
const queuedTile = request('/grid?z=4&x=12&y=1');
release(Response.json(syntheticGrid()));
const results = await Promise.all([first, observation, ...tiles, queuedTile]);
assert.ok(results.every((result) => result.status === 200));
assert.equal(calls.length, 14);
assert.equal(calls[1], '/v2/observations');
});
@@ -0,0 +1,343 @@
import { createInatClient } from './client.js';
import { coalesceProxyRequest } from '../common/http.js';
import { admitSameSiteKeyless } from '../common/same-site.js';
import { requiredFiniteQueryNumber } from '../common/query.js';
import { snapWildlifeBox } from '../../../src/layers/wildlifeSightings/bounds.js';
import { validInatGridTile } from './grid.js';
// The upstream advertises five-minute freshness; global samples cost less often.
const TTL_MS = 300_000;
const GLOBAL_TTL_MS = 600_000;
const INAT_GRID_CACHE_TTL_MS = 15 * 60_000;
const STALE_MAX_MS = 3_600_000;
const INAT_CACHE_MAX = 400;
const HOUR_MS = 3_600_000;
const TAXA = Object.freeze([
'all',
'Aves',
'Mammalia',
'Insecta',
'Plantae',
'whales',
]);
function requestParams(pathname, query, nowMs) {
const grid = pathname === '/grid';
const scope = grid ? 'grid' : (query.get('scope') ?? 'view');
if (!grid && scope !== 'view' && scope !== 'global')
return { error: 'invalid_scope' };
const quality =
query.get('quality') === 'research' ||
(grid && query.get('confirmed') === 'true')
? 'research'
: 'all';
const taxa = TAXA.includes(query.get('taxa')) ? query.get('taxa') : 'all';
let box = null;
let tile;
if (grid) {
tile = Object.fromEntries(
['z', 'x', 'y'].map((name) => [
name,
requiredFiniteQueryNumber(query, name),
]),
);
if (!validInatGridTile(tile)) return { error: 'invalid_tile' };
} else if (scope === 'view') {
const south = requiredFiniteQueryNumber(query, 'south');
const west = requiredFiniteQueryNumber(query, 'west');
const north = requiredFiniteQueryNumber(query, 'north');
const east = requiredFiniteQueryNumber(query, 'east');
box = snapWildlifeBox({ south, west, north, east });
if (!box) return { error: 'invalid_bbox' };
}
const windowMs = (taxa === 'whales' ? 7 * 24 : 48) * HOUR_MS;
const windowKey = new Date(Math.floor((nowMs - windowMs) / HOUR_MS) * HOUR_MS)
.toISOString()
.replace('.000Z', 'Z');
const bounds = box ? `${box.south},${box.west},${box.north},${box.east}` : '';
const baseKey = grid
? `grid:v1:${tile.z}/${tile.x}/${tile.y}:${quality}:${taxa}`
: `v1:${scope}:${bounds}:${quality}:${taxa}`;
return {
scope,
...(grid
? { ...tile, waitForFresh: query.get('refresh') === 'wait' }
: { box }),
quality,
taxa,
observedD1: windowKey.slice(0, 10),
baseKey,
key: `${baseKey}:${windowKey}`,
};
}
/** A shared-provider route for dev and preview; acquisition starts on demand. */
export function inaturalistProxy({
fetchImpl = (...args) => globalThis.fetch(...args),
now = () => Date.now(),
cacheDir,
cache = new Map(),
warn = (...args) => console.warn(...args),
queueOptions,
} = {}) {
const inFlight = new Map();
const consumers = new Map();
let client = null;
function age(entry) {
if (!entry) return Infinity;
const elapsed = now() - entry.savedAt;
return elapsed < 0 ? Infinity : elapsed;
}
function joinFlight(promise, group, signal) {
group.count += 1;
return new Promise((resolve, reject) => {
let released = false;
function release() {
if (released) return;
released = true;
signal?.removeEventListener('abort', abort);
group.count -= 1;
if (!group.count && !group.background && !group.settled) {
group.controller.abort();
// A fresh consumer must not join an abandoned queued generation.
if (inFlight.get(group.key) === promise) inFlight.delete(group.key);
if (consumers.get(group.key) === group) consumers.delete(group.key);
}
}
function abort() {
release();
reject(signal.reason);
}
signal?.addEventListener('abort', abort, { once: true });
if (signal?.aborted) abort();
promise.then(
(value) => {
release();
resolve(value);
},
(error) => {
release();
reject(error);
},
);
});
}
function refresh(params, background = false) {
// Global samples share pending refreshes across hourly window changes.
const flightKey = params.scope === 'global' ? params.baseKey : params.key;
let group = consumers.get(flightKey);
if (!group) {
group = {
key: flightKey,
controller: new AbortController(),
count: 0,
settled: false,
background,
};
consumers.set(flightKey, group);
}
group.background ||= background;
const { promise, shared } = coalesceProxyRequest(
inFlight,
flightKey,
async () => {
client ??= createInatClient({
fetchImpl,
now,
cacheDir,
warn,
queueOptions,
});
const options = { signal: group.controller.signal };
const result = await (params.scope === 'grid'
? client.getGrid(params, options)
: client.getObservations(params, options));
const value = {
fetchedAt: now(),
...(params.scope === 'grid'
? {}
: { scope: params.scope, box: params.box }),
...result,
};
cache.delete(params.key);
cache.set(params.key, {
value,
savedAt: now(),
baseKey: params.baseKey,
});
while (cache.size > INAT_CACHE_MAX)
cache.delete(cache.keys().next().value);
return value;
},
);
if (!shared) {
const settle = () => {
group.settled = true;
if (consumers.get(flightKey) === group) consumers.delete(flightKey);
};
void promise.then(settle, settle);
}
return { promise, group };
}
async function acquire(params, signal) {
signal?.throwIfAborted();
for (const [key, entry] of cache) {
if (age(entry) > STALE_MAX_MS) cache.delete(key);
}
const current = cache.get(params.key);
const ttl =
params.scope === 'grid'
? INAT_GRID_CACHE_TTL_MS
: params.scope === 'global'
? GLOBAL_TTL_MS
: TTL_MS;
if (current && age(current) < ttl) {
cache.delete(params.key);
cache.set(params.key, current);
return { value: current.value, stale: false };
}
// Window-hour rollover can reuse prior cells without claiming a new age.
let previous = current;
let previousKey = params.key;
if (!previous) {
for (const [key, entry] of cache) {
if (
entry.baseKey === params.baseKey &&
(!previous || entry.savedAt > previous.savedAt)
) {
previous = entry;
previousKey = key;
}
}
}
if (params.scope === 'global' && age(previous) < GLOBAL_TTL_MS)
return { value: previous.value, stale: false };
if (
params.scope === 'grid' &&
!params.waitForFresh &&
age(previous) <= STALE_MAX_MS
) {
cache.delete(previousKey);
cache.set(previousKey, previous);
// Retain one background consumer after this HTTP response closes. A stale
// tile is useful immediately; waiting for the token queue defeats SWR.
const cooling = client?.cooldownState('grid');
if (!cooling) {
const { promise } = refresh(params, true);
void promise.catch(() => {});
}
return {
value: previous.value,
stale: true,
...cooling,
};
}
try {
const { promise, group } = refresh(params);
return { value: await joinFlight(promise, group, signal), stale: false };
} catch (error) {
if (signal?.aborted) throw error;
if (age(previous) <= STALE_MAX_MS)
return {
value: previous.value,
stale: true,
retryAfterS: error.retryAfterS,
upstreamStatus: error.upstreamStatus,
};
throw error;
}
}
async function handler(req, res) {
const json = (
status,
value,
{ stale = false, retryAfterS, upstreamStatus } = {},
) => {
if (res.destroyed) return;
res.writeHead(status, {
'Content-Type': 'application/json',
'Cache-Control': 'no-store',
...(status === 405 ? { Allow: 'GET' } : {}),
...(Number.isFinite(retryAfterS)
? { 'Retry-After': String(retryAfterS) }
: {}),
...(upstreamStatus
? { 'X-Upstream-Status': String(upstreamStatus) }
: {}),
...(stale ? { 'X-Data-Stale': 'true' } : {}),
});
res.end(JSON.stringify(value));
};
if (admitSameSiteKeyless(req, res)) return;
if (req.method !== 'GET') return json(405, { error: 'method_not_allowed' });
const url = String(req.url || '/');
const queryAt = url.indexOf('?');
const pathname = queryAt < 0 ? url : url.slice(0, queryAt);
if (pathname !== '/observations' && pathname !== '/grid')
return json(404, { error: 'unknown_route' });
const params = requestParams(
pathname,
new URLSearchParams(queryAt < 0 ? '' : url.slice(queryAt + 1)),
now(),
);
if (params.error) return json(400, { error: params.error });
const controller = new AbortController();
const abort = () => controller.abort();
// IncomingMessage.close also fires after a complete request body. The
// response close identifies the abandoned browser fetch, while aborted
// covers a request cut short before it reached the response stage.
req.once?.('aborted', abort);
res.once?.('close', abort);
try {
const { value, stale, retryAfterS, upstreamStatus } = await acquire(
params,
controller.signal,
);
return json(200, stale ? { ...value, stale: true } : value, {
stale,
retryAfterS,
upstreamStatus,
});
} catch (error) {
if (controller.signal.aborted) return;
const status =
error.code === 'inaturalist_budget_exhausted'
? 503
: error.status === 429
? 429
: 502;
const code =
error.code === 'inaturalist_budget_exhausted'
? error.code
: status === 429
? 'inaturalist_rate_limited'
: 'inaturalist_unavailable';
return json(
status,
{ error: code },
{
retryAfterS: error.retryAfterS,
upstreamStatus: error.upstreamStatus,
},
);
} finally {
req.removeListener?.('aborted', abort);
res.removeListener?.('close', abort);
}
}
return {
name: 'inaturalist',
configureServer({ middlewares }) {
middlewares.use('/api/inaturalist', handler);
},
configurePreviewServer({ middlewares }) {
middlewares.use('/api/inaturalist', handler);
},
};
}
@@ -0,0 +1,893 @@
import test from 'node:test';
import assert from 'node:assert/strict';
import { mkdtemp, readFile, readdir, rm } from 'node:fs/promises';
import os from 'node:os';
import path from 'node:path';
import { installInatProxy, NOW } from './proxy.fixture.mjs';
import { createInatClient, inatRetryCooldownMs } from './client.js';
const observation = (id = 1) => ({
id,
uri: `https://www.inaturalist.org/observations/${id}`,
geojson: { coordinates: [-122.351, 47.613] },
time_observed_at: '2026-11-04T17:00:00Z',
observed_on: '2026-11-04',
created_at: '2026-11-04T17:01:00Z',
quality_grade: 'research',
license_code: 'cc-by',
positional_accuracy: 12,
faves_count: 2,
num_identification_agreements: 3,
taxon: {
name: 'Calypte anna',
preferred_common_name: "Anna's hummingbird",
iconic_taxon_name: 'Aves',
},
photos: [
{
url: 'https://inaturalist-open-data.s3.amazonaws.com/photos/123/square.jpeg',
license_code: 'cc-by',
attribution: 'Synthetic photographer, CC BY',
},
],
user: { login: 'must-never-leave-server' },
});
const envelope = (rows = [observation()], total = rows.length) => ({
total_results: total,
page: 1,
per_page: 200,
results: rows,
});
const boxUrl = (index = 0, extra = '') =>
`/observations?south=47.51&west=${-122.49 + index * 0.2}&north=47.69&east=${-122.31 + index * 0.2}${extra}`;
function install(options = {}, hook = 'configureServer') {
return installInatProxy(
boxUrl(),
{
// Serial dispatch makes queued cancellation and coalescing observable.
queueOptions: { intervalMs: 0, windowMs: 0, maxInFlight: 1 },
...options,
},
hook,
);
}
async function until(check) {
for (let i = 0; i < 100; i += 1) {
if (check()) return;
await new Promise(setImmediate);
}
assert.fail('expected fake upstream calls were not dispatched');
}
for (const hook of ['configureServer', 'configurePreviewServer']) {
test(`${hook}: actual v2 envelope, fixed query, projection and privacy`, async () => {
const calls = [];
const cache = new Map();
const request = install(
{
cache,
fetchImpl: async (url, options) => {
calls.push(new URL(url));
assert.equal(options.redirect, 'error');
assert.ok(options.signal instanceof AbortSignal);
assert.equal(options.headers.Accept, 'application/json');
assert.equal(
options.headers['User-Agent'],
'gods-eye-view-wildlife-sightings-proxy/1.0 (+https://github.com/bilawalsidhu/gods-eye-view)',
);
return Response.json(envelope());
},
},
hook,
);
assert.equal(calls.length, 0, 'construction and mounting do not acquire');
const result = await request(
boxUrl(0, '&url=https://attacker.example&per_page=999'),
);
assert.equal(result.status, 200);
assert.equal(result.headers['Cache-Control'], 'no-store');
assert.equal(result.body.rows[0].stableId, 'inat:1');
assert.equal(result.body.fetchedAt, NOW);
assert.equal(result.body.scope, 'view');
assert.deepEqual(result.body.box, {
south: 47.5,
west: -122.5,
north: 47.7,
east: -122.3,
});
const query = calls[0].searchParams;
assert.equal(
calls[0].origin + calls[0].pathname,
'https://api.inaturalist.org/v2/observations',
);
// Only the projected fields are requested, never user or identification data.
assert.match(query.get('fields'), /^\(id:!t,uri:!t,.*photos:\(url:!t/);
assert.doesNotMatch(query.get('fields'), /user|identifications/);
assert.equal(query.get('swlat'), '47.5');
assert.equal(query.get('swlng'), '-122.5');
assert.equal(query.get('nelat'), '47.7');
assert.equal(query.get('nelng'), '-122.3');
assert.equal(query.get('d1'), '2026-11-02');
assert.equal(query.has('created_d1'), false);
assert.equal(query.get('per_page'), '200');
assert.equal(query.get('verifiable'), 'true');
assert.equal(query.get('photos'), 'true');
assert.equal(query.get('captive'), 'false');
assert.equal(query.get('order_by'), 'random');
assert.equal(query.has('order'), false);
assert.doesNotMatch(
JSON.stringify(result.body),
/must-never-leave-server|"user"/,
);
assert.doesNotMatch(
JSON.stringify([...cache.values()]),
/must-never-leave-server|"user"/,
);
});
}
test('gate, method, route and query validation run before quota, cache or upstream', async () => {
const cache = new Map();
let calls = 0;
const request = install({
cache,
fetchImpl: async () => {
calls += 1;
return Response.json(envelope());
},
});
for (let i = 0; i < 32; i += 1) {
const response = await request('/observations?scope=global', {
headers: {
host: 'localhost:5173',
origin: 'https://foreign.example',
'sec-fetch-site': 'cross-site',
},
});
assert.equal(response.status, 403);
}
const method = await request('/observations', { method: 'POST' });
assert.equal(method.status, 405);
assert.equal(method.headers.Allow, 'GET');
for (const route of ['/', '/unknown', '/observations/extra', '//[invalid'])
assert.equal((await request(route)).body.error, 'unknown_route');
assert.equal(
(await request('/observations?scope=planet')).body.error,
'invalid_scope',
);
for (const query of [
'',
'south=NaN&west=0&north=1&east=1',
'south=&west=0&north=1&east=1',
'south=0&west=0&north=0&east=1',
'south=0&west=1&north=1&east=0',
'south=-91&west=0&north=1&east=1',
'south=0&west=-181&north=1&east=1',
'south=0&west=0&north=41&east=1',
'south=0&west=0&north=1&east=41',
]) {
const result = await request(`/observations?${query}`);
assert.equal(result.status, 400, query);
assert.equal(result.body.error, 'invalid_bbox', query);
}
assert.equal(cache.size, 0);
assert.equal(calls, 0);
const allowed = await request('/observations?scope=global', {
headers: {
host: 'gev.example',
origin: 'https://gev.example',
'sec-fetch-site': 'same-origin',
'x-forwarded-proto': 'https',
'x-forwarded-for': '203.0.113.9',
},
});
assert.equal(allowed.status, 200);
});
test('cached requests never exhaust a local request admission limiter', async () => {
let calls = 0;
const request = install({
fetchImpl: async () => {
calls += 1;
return Response.json(envelope());
},
});
for (let i = 0; i < 30; i += 1) assert.equal((await request()).status, 200);
const cached = await request();
assert.equal(cached.status, 200);
assert.equal(cached.headers['Retry-After'], undefined);
assert.equal(calls, 1);
});
test('proxy rejects oversized snapped spans before acquisition and preserves canonical client boxes', async () => {
const urls = [];
const request = install({
fetchImpl: async (url) => {
urls.push(new URL(url));
return Response.json(envelope());
},
});
for (const query of [
'south=10.2&west=0&north=50.1&east=1',
'south=0&west=10.2&north=1&east=50.1',
]) {
const result = await request(`/observations?${query}`);
assert.equal(result.status, 400);
assert.equal(result.body.error, 'invalid_bbox');
}
assert.equal(urls.length, 0);
const canonical =
'/observations?south=47.4&west=-122.3&north=47.6&east=-122.1';
const first = await request(canonical);
assert.equal(first.status, 200);
assert.deepEqual(first.body.box, {
south: 47.4,
west: -122.3,
north: 47.6,
east: -122.1,
});
assert.equal(urls[0].searchParams.get('swlat'), '47.4');
assert.equal(urls[0].searchParams.get('swlng'), '-122.3');
const second = await request(canonical);
assert.deepEqual(second.body, first.body);
assert.equal(urls.length, 1, 'canonical boxes share their cache key');
});
test('view snapping, whales, research and global queries obey the requested policy', async () => {
const urls = [];
const request = install({
fetchImpl: async (url) => {
urls.push(new URL(url));
return Response.json(envelope());
},
});
await request(
'/observations?south=40.2&west=-123.7&north=46.1&east=-121.4&quality=research&taxa=Aves',
);
assert.equal(urls[0].searchParams.get('swlat'), '40');
assert.equal(urls[0].searchParams.get('nelat'), '47');
assert.equal(urls[0].searchParams.get('per_page'), '200');
assert.equal(urls[0].searchParams.get('quality_grade'), 'research');
assert.equal(urls[0].searchParams.get('iconic_taxa'), 'Aves');
await request(boxUrl(0, '&taxa=whales'));
const whales = urls[1].searchParams;
assert.equal(whales.get('taxon_id'), '152871');
assert.equal(whales.get('d1'), '2026-10-28');
assert.equal(whales.has('created_d1'), false);
assert.equal(whales.has('photos'), false);
assert.equal(whales.has('iconic_taxa'), false);
await request('/observations?scope=global&quality=typo&taxa=typo&south=bad');
const global = urls[2].searchParams;
for (const field of [
'swlat',
'swlng',
'nelat',
'nelng',
'quality_grade',
'iconic_taxa',
])
assert.equal(global.has(field), false, field);
assert.equal(global.get('per_page'), '200');
assert.equal(global.get('d1'), '2026-11-02');
assert.equal(global.has('created_d1'), false);
assert.equal(global.get('order_by'), 'observed_on');
assert.equal(global.get('order'), 'desc');
await request('/observations?scope=global&taxa=whales');
const globalWhales = urls[3].searchParams;
assert.equal(globalWhales.get('d1'), '2026-10-28');
assert.equal(globalWhales.has('created_d1'), false);
assert.equal(globalWhales.get('order_by'), 'observed_on');
assert.equal(globalWhales.get('order'), 'desc');
});
test('fresh cache hits and global ten-minute TTL spend no upstream quota', async () => {
let at = NOW;
let calls = 0;
const request = install({
now: () => at,
fetchImpl: async () => {
calls += 1;
return Response.json(envelope());
},
});
const first = await request();
at += 299_999;
assert.deepEqual((await request()).body, first.body);
at += 1;
await request();
assert.equal(calls, 2);
await request('/observations?scope=global');
at += 599_999;
await request('/observations?scope=global');
assert.equal(calls, 3);
at += 1;
await request('/observations?scope=global');
assert.equal(calls, 4);
});
test('the same snapped view and window hour returns one cached random sample until expiry', async () => {
let at = NOW;
let draws = 0;
const cache = new Map();
const request = install({
cache,
now: () => at,
fetchImpl: async (url) => {
const query = new URL(url).searchParams;
assert.equal(query.get('order_by'), 'random');
assert.equal(query.get('per_page'), '200');
return Response.json(envelope([observation(++draws)]));
},
});
const first = await request();
for (const elapsed of [1, 1000, 60_000, 299_999]) {
at = NOW + elapsed;
const next = await request();
assert.deepEqual(next.body, first.body);
}
assert.equal(draws, 1);
assert.match(cache.keys().next().value, /2026-11-02T18:00:00Z$/);
at = NOW + 300_000;
const refresh = await request();
assert.equal(refresh.body.rows[0].stableId, 'inat:2');
assert.equal(draws, 2);
});
test('disconnected queued views are dropped while a remaining coalesced consumer keeps its request', async () => {
const calls = [];
const releases = [];
const cache = new Map();
const request = install({
cache,
fetchImpl: (url, { signal }) => {
calls.push({ url: new URL(url), signal });
return new Promise((resolve) => releases.push(resolve));
},
});
const first = request(boxUrl(0));
await until(() => releases.length === 1);
const superseded = new AbortController();
const oldView = request(boxUrl(1), { signal: superseded.signal });
const oneConsumer = new AbortController();
const one = request(boxUrl(2), { signal: oneConsumer.signal });
const other = request(boxUrl(2));
superseded.abort();
oneConsumer.abort();
await Promise.all([oldView, one]);
// Rejoining the abandoned key creates a healthy generation; the old
// coalescing promise must not later delete or cancel this replacement.
const replacement = request(boxUrl(1));
releases[0](Response.json(envelope()));
await until(() => releases.length === 2);
assert.equal(calls[1].url.searchParams.get('swlng'), '-122.1');
assert.equal(calls[1].signal.aborted, false);
releases[1](Response.json(envelope([observation(2)])));
await until(() => releases.length === 3);
assert.equal(calls[2].url.searchParams.get('swlng'), '-122.3');
releases[2](Response.json(envelope([observation(3)])));
const results = await Promise.all([first, other, replacement]);
assert.ok(results.every((result) => result.status === 200));
assert.equal(
calls.length,
3,
'the abandoned queued generation never dispatched',
);
assert.equal(cache.size, 3);
assert.equal(results[1].body.rows[0].stableId, 'inat:2');
assert.equal(results[2].body.rows[0].stableId, 'inat:3');
});
test('observation windows reject old sightings uploaded today and keep recent observations', async () => {
for (const [taxa, recentTime, expectedD1] of [
['all', '2026-11-04T17:00:00Z', '2026-11-02'],
['whales', '2026-10-30T17:00:00Z', '2026-10-28'],
]) {
let upstreamQuery;
const request = install({
fetchImpl: async (url) => {
upstreamQuery = new URL(url).searchParams;
return Response.json(
envelope([
{
...observation(80),
time_observed_at: '2010-06-01T12:00:00Z',
observed_on: '2010-06-01',
},
{
...observation(81),
time_observed_at: recentTime,
observed_on: recentTime.slice(0, 10),
},
]),
);
},
});
const result = await request(boxUrl(0, `&taxa=${taxa}`));
assert.equal(result.status, 200, taxa);
assert.equal(upstreamQuery.get('d1'), expectedD1);
assert.equal(upstreamQuery.has('created_d1'), false);
assert.equal(upstreamQuery.get('order_by'), 'random');
assert.deepEqual(
result.body.rows.map((row) => row.stableId),
['inat:81'],
);
}
});
test('same-key requests single-flight and a seventeenth distinct in-flight key queues', async () => {
const releases = [];
const request = install({
fetchImpl: () => new Promise((resolve) => releases.push(resolve)),
});
const a = request();
const b = request();
await until(() => releases.length === 1);
releases.shift()(Response.json(envelope()));
assert.deepEqual((await a).body, (await b).body);
const pending = Array.from({ length: 16 }, (_, i) => request(boxUrl(i + 1)));
await until(() => releases.length === 1);
pending.push(request(boxUrl(18)));
for (let i = 0; i < 17; i += 1) {
await until(() => releases.length > i);
releases[i](Response.json(envelope()));
}
assert.ok((await Promise.all(pending)).every(({ status }) => status === 200));
});
test('a fresh global sample survives the hourly observation-window rollover without another fetch', async () => {
let at = Date.parse('2026-11-04T18:59:00Z');
let calls = 0;
const request = install({
now: () => at,
fetchImpl: async () => {
calls += 1;
return Response.json(envelope());
},
});
const first = await request('/observations?scope=global');
at += 60_000;
const nextHour = await request('/observations?scope=global');
assert.deepEqual(nextHour.body, first.body);
assert.equal(nextHour.body.stale, undefined);
assert.equal(calls, 1);
at += 9 * 60_000;
await request('/observations?scope=global');
assert.equal(calls, 2);
});
test('pending global refresh coalesces across the hourly observation-window rollover with a busy queue', async () => {
let at = Date.parse('2026-11-04T18:59:59Z');
const releases = [];
const cache = new Map();
const request = install({
cache,
now: () => at,
fetchImpl: (url) =>
new Promise((resolve) =>
releases.push({
resolve,
url: new URL(url),
}),
),
});
const first = request(
'/observations?scope=global&quality=research&taxa=Aves',
);
const views = Array.from({ length: 15 }, (_, i) => request(boxUrl(i)));
await until(() => releases.length === 1);
assert.equal(releases[0].url.searchParams.get('d1'), '2026-11-02');
at = Date.parse('2026-11-04T19:00:01Z');
const nextHour = request(
'/observations?scope=global&quality=research&taxa=Aves',
);
await new Promise(setImmediate);
assert.equal(
releases.length,
1,
'the same global query joins the pending refresh',
);
const distinct = request(
'/observations?scope=global&quality=research&taxa=Mammalia',
);
for (let i = 0; i < 17; i += 1) {
await until(() => releases.length > i);
releases[i].resolve(Response.json(envelope()));
}
const [original, joined] = await Promise.all([first, nextHour]);
assert.equal(original.status, 200);
assert.equal((await distinct).status, 200);
assert.deepEqual(joined.body, original.body);
assert.ok((await Promise.all(views)).every(({ status }) => status === 200));
const globalEntries = [...cache.entries()].filter(
([, entry]) => entry.value.scope === 'global',
);
assert.equal(globalEntries.length, 2);
assert.match(globalEntries[0][0], /2026-11-02T18:00:00Z$/);
assert.equal(
(await request('/observations?scope=global&quality=research&taxa=Aves'))
.status,
200,
);
assert.equal(
releases.length,
17,
'the accepted global answer keeps its ten-minute freshness',
);
});
test('view refreshes retain distinct hourly keys while a previous-hour request is pending', async () => {
let at = Date.parse('2026-11-04T18:59:59Z');
const releases = [];
const request = install({
now: () => at,
fetchImpl: () => new Promise((resolve) => releases.push(resolve)),
});
const first = request();
await until(() => releases.length === 1);
at = Date.parse('2026-11-04T19:00:01Z');
const nextHour = request();
await new Promise(setImmediate);
assert.equal(
releases.length,
1,
'the next view queues behind the current call',
);
releases[0](Response.json(envelope()));
await until(() => releases.length === 2);
releases[1](Response.json(envelope()));
assert.ok(
(await Promise.all([first, nextHour])).every(
({ status }) => status === 200,
),
);
});
test('upstream minute ceiling waits for request 61 across client addresses', async () => {
let at = NOW;
let calls = 0;
const request = install({
now: () => at,
queueOptions: {
now: () => at,
intervalMs: 0,
wait: async (ms) => {
at += ms;
},
},
fetchImpl: async () => {
calls += 1;
return Response.json(envelope());
},
});
for (let i = 0; i < 60; i += 1)
assert.equal(
(await request(boxUrl(i), { peer: `10.0.0.${i}` })).status,
200,
);
const queued = await request(boxUrl(60), { peer: '10.0.1.1' });
assert.equal(queued.status, 200);
assert.equal(queued.headers['Retry-After'], undefined);
assert.equal(at, NOW + 60_000);
assert.equal(calls, 61);
});
test('views refresh after many prior requests without fabricating an upstream limit', async () => {
let at = NOW;
let calls = 0;
const request = install({
now: () => at,
fetchImpl: async () => {
calls += 1;
return Response.json(envelope());
},
});
const first = await request();
at += 300_001;
for (let i = 1; i <= 60; i += 1)
await request(boxUrl(i), { peer: `10.0.0.${i}` });
const stale = await request();
assert.equal(stale.status, 200);
assert.equal(stale.body.stale, undefined);
assert.notEqual(stale.body.fetchedAt, first.body.fetchedAt);
assert.equal(calls, 62);
});
test('bounded samples preserve partial metadata, and well-formed empty views are valid', async () => {
let at = NOW;
let fail = false;
const request = install({
now: () => at,
fetchImpl: async () =>
fail
? new Response('unavailable', { status: 503 })
: Response.json(envelope([observation()], 500)),
});
const first = await request();
assert.equal(first.body.partial, true);
assert.equal(first.body.rawRowCount, 500);
fail = true;
at += 59 * 60_000;
const stale = await request();
assert.equal(
stale.status,
200,
'hourly window changes still permit stale fallback',
);
assert.equal(stale.headers['X-Data-Stale'], 'true');
assert.equal(stale.body.stale, true);
assert.equal(stale.body.fetchedAt, first.body.fetchedAt);
assert.equal(stale.body.partial, true);
at += 2 * 60_000;
const expired = await request();
assert.equal(expired.status, 502);
assert.equal(expired.body.error, 'inaturalist_unavailable');
const empty = install({
fetchImpl: async () => Response.json(envelope([], 0)),
});
const result = await empty();
assert.equal(result.status, 200);
assert.deepEqual(result.body.rows, []);
});
test('bad envelopes, JSON and capped bodies fail closed without inventing a cooldown', async () => {
for (const bad of [
() => Response.json({ total_results: 1, results: {} }),
() => Response.json({ total_results: 1, observations: [observation()] }),
() => Response.json({ total_results: '1', results: [observation()] }),
() => Response.json(envelope([], 1)),
() => Response.json(envelope([{ ...observation(), id: null }], 1)),
() => new Response('{malformed'),
() =>
new Response('{}', {
headers: { 'Content-Length': String(8 * 1024 * 1024 + 1) },
}),
]) {
let at = NOW;
let calls = 0;
const request = install({
now: () => at,
fetchImpl: async () => {
calls += 1;
return calls === 1 ? bad() : Response.json(envelope());
},
});
const first = await request();
assert.equal(first.status, 502);
assert.equal(first.body.error, 'inaturalist_unavailable');
assert.equal(first.headers['Retry-After'], undefined);
assert.equal((await request(boxUrl(1))).status, 200);
assert.equal(calls, 2);
}
});
test('a wholly invalid iNat page cannot replace a previously valid snapshot', async () => {
let at = NOW;
let fail = false;
const cache = new Map();
const request = install({
cache,
now: () => at,
fetchImpl: async () =>
Response.json(
fail
? envelope([{ ...observation(), quality_grade: 'casual' }])
: envelope(),
),
});
const first = await request();
const original = cache.values().next().value;
fail = true;
at += 300_001;
const stale = await request();
assert.equal(stale.status, 200);
assert.equal(stale.body.stale, true);
assert.equal(stale.body.fetchedAt, first.body.fetchedAt);
assert.equal(cache.values().next().value, original);
});
test('429 cooldown honors seconds, HTTP dates, and long upstream deadlines', async () => {
assert.equal(inatRetryCooldownMs(null, NOW), 60_000);
assert.equal(inatRetryCooldownMs('0', NOW), 0);
assert.equal(inatRetryCooldownMs('2', NOW), 2000);
assert.equal(inatRetryCooldownMs('999999', NOW), 999_999_000);
assert.equal(inatRetryCooldownMs('7200', NOW), 7_200_000);
assert.equal(
inatRetryCooldownMs(new Date(NOW + 7_200_000).toUTCString(), NOW),
7_200_000,
);
let at = NOW;
const releases = [];
const warnings = [];
const request = install({
now: () => at,
warn: (message) => warnings.push(message),
fetchImpl: () => new Promise((resolve) => releases.push(resolve)),
});
const first = request();
const later = request(boxUrl(1));
await until(() => releases.length === 1);
releases[0](
new Response('limit', {
status: 429,
headers: { 'Retry-After': '999999' },
}),
);
assert.equal((await first).headers['Retry-After'], '999999');
assert.equal((await later).headers['Retry-After'], '999999');
at += 1000;
assert.equal((await request(boxUrl(1))).headers['Retry-After'], '999998');
at += 5000;
const cooling = await request(boxUrl(2));
assert.equal(cooling.status, 429);
assert.equal(cooling.headers['Retry-After'], '999993');
assert.equal(
releases.length,
1,
'queued calls do not dispatch during cooldown',
);
assert.equal(
warnings.filter((message) => message.includes('capped')).length,
0,
);
});
for (const [upstreamStatus, expectedStatus, code, retryAfterS] of [
[429, 429, 'inaturalist_rate_limited', 120],
[503, 502, 'inaturalist_unavailable', 120],
]) {
test(`concurrent reservations honor an immediate upstream ${upstreamStatus} before dispatch`, async () => {
let calls = 0;
const warnings = [];
const client = createInatClient({
now: () => 0,
cacheDir: null,
queueOptions: { intervalMs: 0 },
warn: (message) => warnings.push(message),
fetchImpl: async () => {
calls += 1;
return new Response('synthetic failure', {
status: upstreamStatus,
headers: { 'Retry-After': '120' },
});
},
});
const params = {
scope: 'global',
quality: 'all',
taxa: 'all',
observedD1: '1969-12-30',
};
const results = await Promise.allSettled(
Array.from({ length: 3 }, () => client.getObservations(params)),
);
assert.equal(
calls,
1,
'pending reservations must not dispatch during cooldown',
);
for (const result of results) {
assert.equal(result.status, 'rejected');
assert.equal(result.reason.status, expectedStatus);
assert.equal(result.reason.code, code);
assert.equal(result.reason.retryAfterS, retryAfterS);
}
assert.equal(
warnings.length,
1,
'only the dispatched refresh logs a failure',
);
await assert.rejects(client.getObservations(params), {
status: expectedStatus,
code,
retryAfterS,
});
assert.equal(calls, 1);
});
}
test('budget is read lazily, persisted as counters only, enforced across restart and UTC rollover', async (t) => {
const parent = await mkdtemp(path.join(os.tmpdir(), 'gev-inat-budget-'));
t.after(() => rm(parent, { recursive: true, force: true }));
const cacheDir = path.join(parent, 'not-created');
const before = process.env.INATURALIST_DAILY_BUDGET;
process.env.INATURALIST_DAILY_BUDGET = '2';
t.after(() =>
before === undefined
? delete process.env.INATURALIST_DAILY_BUDGET
: (process.env.INATURALIST_DAILY_BUDGET = before),
);
let at = NOW;
let calls = 0;
const options = {
now: () => at,
cacheDir,
warn() {},
fetchImpl: async () => {
calls += 1;
return Response.json(envelope());
},
};
const client = createInatClient(options);
assert.deepEqual(
await readdir(parent),
[],
'construction performs no directory write',
);
const params = {
scope: 'global',
quality: 'all',
taxa: 'all',
observedD1: '2026-11-02',
};
await Promise.all([
client.getObservations(params),
client.getObservations(params),
]);
assert.deepEqual(
JSON.parse(await readFile(path.join(cacheDir, 'budget.json'), 'utf8')),
{ date: '2026-11-04', count: 2 },
);
assert.deepEqual(
await readdir(cacheDir),
['budget.json'],
'atomic temporary files are retired',
);
const restarted = createInatClient(options);
await assert.rejects(restarted.getObservations(params), {
status: 503,
code: 'inaturalist_budget_exhausted',
});
assert.equal(calls, 2);
at = Date.parse('2026-11-05T00:00:00Z');
await restarted.getObservations(params);
assert.deepEqual(restarted.budgetState(), {
date: '2026-11-05',
count: 1,
budget: 2,
});
assert.deepEqual(
JSON.parse(await readFile(path.join(cacheDir, 'budget.json'), 'utf8')),
{ date: '2026-11-05', count: 1 },
);
assert.equal(calls, 3);
});
test('memory cache is an LRU bounded to 400 projected entries', async () => {
let at = NOW;
const cache = new Map();
const request = install({
cache,
now: () => at,
fetchImpl: async () => Response.json(envelope()),
});
for (let i = 0; i < 400; i += 1) {
if (i && i % 20 === 0) at += 61_000;
await request(boxUrl(i), { peer: String(i) });
}
assert.equal(cache.size, 400);
const firstKey = cache.keys().next().value;
const secondKey = [...cache.keys()][1];
await request(boxUrl(0));
await request(boxUrl(400));
assert.equal(cache.size, 400);
assert.equal(cache.has(firstKey), true);
assert.equal(cache.has(secondKey), false);
});
test('aborted client calls spend no budget and never fetch', async () => {
const controller = new AbortController();
controller.abort();
const client = createInatClient({
now: () => NOW,
cacheDir: null,
queueOptions: { intervalMs: 0 },
fetchImpl: () => assert.fail('aborted call fetched'),
});
await assert.rejects(
client.getObservations({}, { signal: controller.signal }),
{ name: 'AbortError' },
);
assert.equal(client.budgetState().count, 0);
});
@@ -0,0 +1,70 @@
import assert from 'node:assert/strict';
import { EventEmitter } from 'node:events';
import { inaturalistProxy } from './observations.js';
export const NOW = Date.parse('2026-11-04T18:05:00Z');
/** Mount the proxy over synthetic HTTP objects, without opening a socket. */
export function installInatProxy(
defaultUrl,
options = {},
hook = 'configureServer',
) {
let handler;
const plugin = inaturalistProxy({
now: () => NOW,
cacheDir: null,
queueOptions: { intervalMs: 0, windowMs: 0 },
warn() {},
...options,
});
assert.equal(plugin.name, 'inaturalist');
assert.equal(typeof plugin.configureServer, 'function');
assert.equal(typeof plugin.configurePreviewServer, 'function');
plugin[hook]({
middlewares: {
use(mount, callback) {
assert.equal(mount, '/api/inaturalist');
handler = callback;
},
},
});
return async (
url = defaultUrl,
{ method = 'GET', peer = 'local', headers = {}, signal } = {},
) => {
const res = Object.assign(new EventEmitter(), {
headers: {},
writeHead(status, headers) {
this.status = status;
Object.assign(this.headers, headers);
},
setHeader(key, value) {
this.headers[key] = value;
},
end(value) {
this.status ??= this.statusCode;
this.body = JSON.parse(value);
},
});
const abort = () => {
res.destroyed = true;
res.emit('close');
};
signal?.addEventListener('abort', abort, { once: true });
try {
await handler(
Object.assign(new EventEmitter(), {
url,
method,
headers,
socket: { remoteAddress: peer },
}),
res,
);
} finally {
signal?.removeEventListener('abort', abort);
}
return res;
};
}
+109
View File
@@ -0,0 +1,109 @@
import { setTimeout as delay } from 'node:timers/promises';
// A four-request burst keeps a cold view responsive; thereafter every provider
// call shares the same one-per-second token supply. The rolling minute ceiling
// queues the few requests a token bucket alone could allow above 60/min.
const INAT_INTERVAL_MS = 1000;
const INAT_BURST = 4;
const INAT_WINDOW_MS = 60_000;
const INAT_MAX_PER_WINDOW = 60;
const INAT_MAX_IN_FLIGHT = 4;
/** Observation queries outrank tiles waiting for a token. Aborted consumers
* leave the queue before an attempt or daily-budget reservation is spent. */
export function createInatQueue({
now = () => performance.now(),
wait = (ms) => delay(ms),
intervalMs = INAT_INTERVAL_MS,
burst = INAT_BURST,
windowMs = INAT_WINDOW_MS,
maxInFlight = INAT_MAX_IN_FLIGHT,
} = {}) {
const pending = [];
const dispatchedAt = [];
let tokens = burst;
let refilledAt = null;
let running = false;
let inFlight = 0;
let slotFreed = null;
function refill() {
const at = now();
if (refilledAt !== null && intervalMs > 0)
tokens = Math.min(
burst,
tokens + Math.max(0, at - refilledAt) / intervalMs,
);
if (intervalMs <= 0) tokens = burst;
refilledAt = at;
}
function schedule() {
if (running) return;
running = true;
// A microtask also lets same-turn observation queries take priority.
void Promise.resolve().then(drain);
}
async function drain() {
try {
while (pending.length) {
if (inFlight >= maxInFlight) {
await new Promise((resolve) => (slotFreed = resolve));
continue;
}
refill();
const at = now();
while (dispatchedAt.length && dispatchedAt[0] <= at - windowMs)
dispatchedAt.shift();
const tokenWait = tokens < 1 ? (1 - tokens) * intervalMs : 0;
const windowWait =
windowMs > 0 && dispatchedAt.length >= INAT_MAX_PER_WINDOW
? dispatchedAt[0] + windowMs - at
: 0;
if (tokenWait > 0 || windowWait > 0) {
await wait(Math.ceil(Math.max(tokenWait, windowWait)));
continue;
}
// Stable sorting preserves FIFO order within each priority.
pending.sort((a, b) => a.priority - b.priority);
const job = pending.shift();
job.signal?.removeEventListener('abort', job.abort);
if (job.signal?.aborted) {
job.reject(job.signal.reason);
continue;
}
tokens -= 1;
if (windowMs > 0) dispatchedAt.push(at);
inFlight += 1;
// A rejected task releases its slot without stopping the queue.
void new Promise((resolve) => resolve(job.task()))
.then(job.resolve, job.reject)
.finally(() => {
inFlight -= 1;
const wake = slotFreed;
slotFreed = null;
wake?.();
});
}
} finally {
running = false;
}
}
return function enqueue(task, { signal, priority = 1 } = {}) {
if (signal?.aborted) return Promise.reject(signal.reason);
return new Promise((resolve, reject) => {
const job = { task, signal, priority, resolve, reject };
job.abort = () => {
const index = pending.indexOf(job);
if (index >= 0) pending.splice(index, 1);
signal.removeEventListener('abort', job.abort);
reject(signal.reason);
};
signal?.addEventListener('abort', job.abort, { once: true });
pending.push(job);
schedule();
});
};
}
+111
View File
@@ -0,0 +1,111 @@
import test from 'node:test';
import assert from 'node:assert/strict';
import { createInatQueue } from './queue.js';
async function until(check) {
for (let i = 0; i < 100; i += 1) {
if (check()) return;
await new Promise(setImmediate);
}
assert.fail('queue did not reach its expected state');
}
test('shared governor paces a four-call burst within the rolling minute ceiling', async () => {
let at = 0;
const waits = [];
const queue = createInatQueue({
now: () => at,
wait: async (ms) => {
waits.push(ms);
at += ms;
},
});
const dispatched = [];
await Promise.all(
Array.from({ length: 65 }, () => queue(() => dispatched.push(at))),
);
assert.equal(dispatched.length, 65);
assert.deepEqual(dispatched.slice(0, 7), [0, 0, 0, 0, 1000, 2000, 3000]);
assert.deepEqual(waits.slice(0, 3), [1000, 1000, 1000]);
for (const time of dispatched)
assert.ok(
dispatched.filter((other) => other > time - 60_000 && other <= time)
.length <= 60,
);
assert.equal(dispatched[60], 60_000);
});
test('an observation arriving during the pacing wait overtakes queued tiles', async () => {
let at = 0;
const waits = [];
const queue = createInatQueue({
burst: 1,
now: () => at,
wait: (ms) => new Promise((resolve) => waits.push({ ms, resolve })),
});
const dispatched = [];
await queue(() => dispatched.push('first tile'));
const tile = queue(() => dispatched.push('waiting tile'));
await until(() => waits.length === 1);
const observation = queue(() => dispatched.push('observation'), {
priority: 0,
});
at += waits[0].ms;
waits[0].resolve();
await observation;
assert.deepEqual(dispatched, ['first tile', 'observation']);
await until(() => waits.length === 2);
at += waits[1].ms;
waits[1].resolve();
await tile;
assert.deepEqual(dispatched, ['first tile', 'observation', 'waiting tile']);
});
test('superseded tasks leave before dispatch, and a rejection cannot poison the tail', async () => {
const queue = createInatQueue({ intervalMs: 0 });
const dispatched = [];
let release;
const first = queue(
() =>
new Promise((resolve) => {
release = resolve;
}),
);
await until(() => release);
const controller = new AbortController();
const superseded = queue(() => assert.fail('superseded tile dispatched'), {
signal: controller.signal,
});
const dropped = assert.rejects(superseded, { name: 'AbortError' });
controller.abort();
await dropped;
const failed = queue(() => {
dispatched.push('failed');
throw new Error('synthetic');
});
const failure = assert.rejects(failed, /synthetic/);
const succeeding = queue(() => dispatched.push('recovered'));
release();
await Promise.all([first, failure, succeeding]);
assert.deepEqual(dispatched, ['failed', 'recovered']);
});
test('a slow call does not hold later calls; at most four run at once', async () => {
const queue = createInatQueue({ intervalMs: 0, windowMs: 0 });
const releases = [];
const started = [];
const slow = (id) => () =>
new Promise((resolve) => {
started.push(id);
releases.push(() => resolve(id));
});
const jobs = [1, 2, 3, 4, 5].map((id) => queue(slow(id)));
await until(() => started.length === 4);
assert.deepEqual(started, [1, 2, 3, 4], 'four start without waiting');
releases[1]();
assert.equal(await jobs[1], 2, 'a later call finishes before the first');
await until(() => started.length === 5);
assert.deepEqual(started, [1, 2, 3, 4, 5], 'a freed slot starts the next');
for (const release of releases) release();
assert.deepEqual(await Promise.all(jobs), [1, 2, 3, 4, 5]);
});
+218
View File
@@ -0,0 +1,218 @@
import test from 'node:test';
import assert from 'node:assert/strict';
import { createInatClient } from './client.js';
import { installInatProxy, NOW } from './proxy.fixture.mjs';
import { syntheticGrid } from './grid.fixture.mjs';
import { INAT_GRID_DECODE_MEASURE } from './grid.js';
const gridUrl = '/grid?z=2&x=1&y=1&taxa=Aves';
const params = {
scope: 'global',
taxa: 'all',
observedD1: '2026-11-02',
};
const tick = () => new Promise(setImmediate);
const install = (options, hook) => installInatProxy(gridUrl, options, hook);
test('tile cache lasts fifteen minutes, then returns stale before revalidation finishes', async () => {
let at = NOW;
let calls = 0;
let release;
const request = install({
now: () => at,
fetchImpl: () => {
calls += 1;
return calls === 1
? Response.json(syntheticGrid())
: new Promise((resolve) => {
release = resolve;
});
},
});
const first = await request();
at += 899_999;
const freshPromise = request();
await tick();
assert.equal(calls, 1, 'a fresh tile never spends another upstream attempt');
assert.deepEqual((await freshPromise).body, first.body);
at += 1;
const stalePromise = request();
let resolved = false;
void stalePromise.then(() => {
resolved = true;
});
await tick();
try {
assert.equal(
resolved,
true,
'stale cells publish while refresh is pending',
);
const stale = await stalePromise;
assert.equal(stale.body.stale, true);
assert.equal(stale.body.fetchedAt, NOW);
assert.deepEqual(stale.body.cells, first.body.cells);
await request();
assert.equal(calls, 2, 'all stale consumers share one background refresh');
const refreshPromise = request(`${gridUrl}&refresh=wait`);
let refreshed = false;
void refreshPromise.then(() => {
refreshed = true;
});
await tick();
assert.equal(
refreshed,
false,
'a background consumer can await the same refresh',
);
release(Response.json(syntheticGrid()));
const fresh = await refreshPromise;
assert.equal(fresh.body.stale, undefined);
assert.equal(fresh.body.fetchedAt, at);
assert.equal(
calls,
2,
'waiting for fresh cells never duplicates upstream work',
);
release = null;
} finally {
release?.(Response.json(syntheticGrid()));
}
});
test('more than sixteen queued views and thirty local requests never report local rate limiting', async () => {
let release;
let calls = 0;
const request = install({
fetchImpl: async () => {
calls += 1;
if (calls === 1)
await new Promise((resolve) => {
release = resolve;
});
return Response.json(syntheticGrid());
},
});
const pending = Array.from({ length: 40 }, (_, x) =>
request(`/grid?z=6&x=${x}&y=1`),
);
await tick();
release();
const results = await Promise.all(pending);
assert.ok(results.every((result) => result.status === 200));
assert.equal(calls, 40);
});
test('only real upstream 429 or 5xx failures start cooldown', async () => {
for (const bad of [
() => Promise.reject(new Error('network')),
() => Response.json({ malformed: true }),
() => new Response('', { status: 404 }),
]) {
let calls = 0;
const client = createInatClient({
now: () => NOW,
cacheDir: null,
queueOptions: { intervalMs: 0, windowMs: 0 },
warn() {},
fetchImpl: () => {
calls += 1;
return calls === 1
? bad()
: Response.json({ results: [], total_results: 0 });
},
});
await assert.rejects(client.getObservations(params), { status: 502 });
await client.getObservations(params);
assert.equal(
calls,
2,
'a non-cooldown failure must not hold the next attempt',
);
}
});
test('actual upstream cooldown is marked separately from a local budget stop', async () => {
for (const upstream of [429, 503]) {
let calls = 0;
const request = install({
fetchImpl: () => {
calls += 1;
return new Response('', {
status: upstream,
headers: { 'Retry-After': '120' },
});
},
});
const first = await request();
assert.equal(first.headers['X-Upstream-Status'], String(upstream));
assert.equal(first.headers['Retry-After'], '120');
await request();
assert.equal(calls, 1);
}
});
test('upstream Retry-After zero survives the HTTP response and permits an immediate next request', async () => {
for (const upstream of [429, 503]) {
let calls = 0;
const request = install({
fetchImpl: () => {
calls += 1;
return calls === 1
? new Response('', {
status: upstream,
headers: { 'Retry-After': '0' },
})
: Response.json(syntheticGrid());
},
});
const first = await request();
assert.equal(first.headers['X-Upstream-Status'], String(upstream));
assert.equal(first.headers['Retry-After'], '0');
assert.equal((await request()).status, 200);
assert.equal(calls, 2);
}
});
test('a local daily budget stop has a distinct code and no upstream cooldown marker', async (t) => {
const before = process.env.INATURALIST_DAILY_BUDGET;
process.env.INATURALIST_DAILY_BUDGET = '1';
t.after(() => {
if (before === undefined) delete process.env.INATURALIST_DAILY_BUDGET;
else process.env.INATURALIST_DAILY_BUDGET = before;
});
let calls = 0;
const request = install({
fetchImpl: () => {
calls += 1;
return Response.json(syntheticGrid());
},
});
assert.equal((await request()).status, 200);
const stopped = await request('/grid?z=2&x=0&y=1');
assert.equal(stopped.status, 503);
assert.equal(stopped.body.error, 'inaturalist_budget_exhausted');
assert.equal(stopped.headers['X-Upstream-Status'], undefined);
assert.equal((await request()).status, 200, 'cached data remains available');
assert.equal(calls, 1);
});
test('cached tiles avoid UTFGrid decode and decode timing entries stay bounded', async () => {
const request = install({
fetchImpl: () => Response.json(syntheticGrid()),
});
await request();
const sample = performance.getEntriesByName(INAT_GRID_DECODE_MEASURE)[0];
assert.ok(sample.duration >= 0);
for (let i = 0; i < 10; i += 1) await request();
assert.equal(
performance.getEntriesByName(INAT_GRID_DECODE_MEASURE)[0],
sample,
);
for (let x = 0; x < 4; x += 1) await request(`/grid?z=3&x=${x}&y=1`);
assert.equal(
performance.getEntriesByName(INAT_GRID_DECODE_MEASURE).length,
1,
);
});
+4
View File
@@ -29,6 +29,8 @@ import { firePerimetersProxy } from './firePerimeters.js';
import { gnssIntegrityProxy } from './gnssIntegrity.js';
import { ghostMapsProxy } from './ghostMaps.js';
import { droneAirspaceProxy } from './droneAirspace.js';
import { inaturalistProxy } from './inaturalist/observations.js';
import { acartiaProxy } from './acartia.js';
import { cycloneProxy } from './cyclones.js';
import { windProxy } from './wind.js';
@@ -81,6 +83,8 @@ function localProviderPlugins({
gnssIntegrityProxy(),
ghostMapsProxy(),
droneAirspaceProxy(),
inaturalistProxy(),
acartiaProxy(),
keySetupEndpoint(),
];
}
+30 -1
View File
@@ -64,6 +64,30 @@ function validMessages(messages) {
return !pending.length;
}
/**
* A tool schema without the `pattern`s Chat Completions refuses: it answers
* 400 for a whole request whose tool uses Unicode property classes (`\p{L}`),
* as `resolve_area`'s area ids do. Every argument is still checked against
* the action's own contract before the tool runs, so the model loses a hint,
* not a rule. Realtime voice accepts these patterns and keeps them.
*/
export function chatSchema(schema) {
if (Array.isArray(schema)) return schema.map(chatSchema);
if (!object(schema)) return schema;
return Object.fromEntries(
Object.entries(schema)
.filter(
([key, value]) =>
!(
key === 'pattern' &&
typeof value === 'string' &&
/\\[pP]\{/.test(value)
),
)
.map(([key, value]) => [key, chatSchema(value)]),
);
}
export function createOpenAiChatHandler({
annotationGuidance,
tools = GEV_REALTIME_TOOLS,
@@ -126,7 +150,12 @@ export function createOpenAiChatHandler({
tools: offeredTools(tools, { overpass }).map(
({ name, description, parameters }) => ({
type: 'function',
function: { name, description, parameters, strict: false },
function: {
name,
description,
parameters: chatSchema(parameters),
strict: false,
},
}),
),
},
+6 -2
View File
@@ -135,7 +135,7 @@ export const ACTION_DESCRIPTIONS = {
},
control_cockpit: {
description:
'Explicit Cockpit requests only: enter from a selected/tracked aircraft (establishes Contacts itself; no set_context_mode), exit, status, or next/previous. Explicit exit-and-show voice: voicePanel preview/settings (Voice Settings), no confirmation.',
'Explicit Cockpit requests only: enter from a selected/tracked aircraft (needs only its own Flights/Military layer on, not Contacts; opens Contacts itself, no set_context_mode), exit, status, or next/previous. Explicit exit-and-show voice: voicePanel preview/settings (Voice Settings), no confirmation.',
$position: 1,
parameters: {
properties: {
@@ -523,7 +523,7 @@ export const ACTION_DESCRIPTIONS = {
properties: {
direction: {
description:
'Required for pan/tilt/rotate; orbit/rotate_globe default right.',
'Pan/rotate/tilt steps; orbit/rotate_globe default right.',
$position: 2,
},
mode: {
@@ -531,6 +531,10 @@ export const ACTION_DESCRIPTIONS = {
'once = bounded eased nudge; continuous = until stop/manual input. rotate_globe defaults continuous; other motions default once.',
$position: 2,
},
pitchDeg: {
description: 'Absolute; -45 default, -89 top-down.',
$position: 1,
},
},
},
},
+5
View File
@@ -227,6 +227,11 @@ export function installRouteMiddleware(
// only shape that asks the upstream for them. A response WITHOUT steps is
// byte for byte what this endpoint has always returned.
middlewares.use('/api/route', async (req, res) => {
if (req.method !== 'GET') {
res.writeHead(405, { 'Content-Type': 'application/json', Allow: 'GET' });
res.end(JSON.stringify({ error: 'Method Not Allowed' }));
return;
}
const fail = (msg) => {
res.writeHead(200, { 'Content-Type': 'application/json' });
res.end(JSON.stringify({ ok: false, error: msg }));
+130 -74
View File
@@ -1,6 +1,9 @@
import path from 'node:path';
import { promises as fsp } from 'node:fs';
import { celestrakTleUrl } from '../../../src/data/spaceProviderRequests.js';
import {
celestrakSatcatUrl,
celestrakTleUrl,
} from '../../../src/data/spaceProviderRequests.js';
import { providerCooldownMs } from './cooldown.js';
import { admitSameSiteKeyless } from '../common/same-site.js';
import { readResponseTextCapped } from '../common/http.js';
@@ -22,84 +25,131 @@ import { readResponseTextCapped } from '../common/http.js';
* 6 h; on upstream failure the freshest stale copy is served (a stale TLE
* beats an empty satellites layer). Pattern mirrors openSkyProxy's
* cache+serve-stale. Adapted from skylight's TleStore (MIT).
*
* The same plugin serves CelesTrak SATCAT records for the satellites layer's
* groups at `/api/celestrak-satcat/<group>` (object type, operational status
* and owner, which file each object under a class). The catalog changes
* slowly, so SATCAT is cached for a day. CelesTrak throttles the client, not
* the endpoint, so a 403 or 429 holds both routes off; any other failure
* holds off only the route that failed, so a bad SATCAT answer never stops
* TLE refreshes.
*/
export function celestrakProxy() {
const TLE_TTL_MS = 6 * 3600_000;
const SATCAT_TTL_MS = 24 * 3600_000;
const CACHE_DIR = path.join(process.cwd(), '.gev-cache');
const groups = new Set([
/** The satellites layer's core groups (src/layers/satellites/policy.js). */
const coreGroups = [
'stations',
'visual',
'gps-ops',
'glo-ops',
'galileo',
'geo',
'starlink',
// Space Missions (`getActiveTle`) reads the full active catalog.
'active',
]);
let cooldownUntil = 0;
const mem = new Map(); // group -> { at: epochMs, body: string }
const inflight = new Map(); // group -> Promise<{at, body}|null>
const diskPath = (group) => path.join(CACHE_DIR, `celestrak-${group}.json`);
async function readDisk(group) {
try {
const parsed = JSON.parse(await fsp.readFile(diskPath(group), 'utf8'));
if (typeof parsed?.body === 'string' && Number.isFinite(parsed?.at))
return parsed;
} catch {
/* no disk cache yet */
}
return null;
}
async function writeDisk(group, entry) {
try {
await fsp.mkdir(CACHE_DIR, { recursive: true });
await fsp.writeFile(diskPath(group), JSON.stringify(entry), 'utf8');
} catch (err) {
console.warn('[celestrak-proxy] cache write failed');
}
}
'resource',
'weather',
'planet',
'sar',
'science',
];
const feeds = [
{
route: '/api/celestrak',
ttlMs: TLE_TTL_MS,
cacheHeader: 'x-tle-cache',
cachedAtHeader: 'X-TLE-Cached-At',
diskName: (group) => `celestrak-${group}.json`,
upstreamUrl: celestrakTleUrl,
groups: new Set([
...coreGroups,
'starlink',
// Space Missions (`getActiveTle`) reads the full active catalog.
'active',
]),
// An upstream error page parses to zero TLEs — treat as failure, keep cache.
validate: (body) => /^1 /m.test(body),
invalid: 'no TLE lines in response',
},
{
route: '/api/celestrak-satcat',
ttlMs: SATCAT_TTL_MS,
cacheHeader: 'x-satcat-cache',
cachedAtHeader: 'X-SATCAT-Cached-At',
diskName: (group) => `celestrak-satcat-${group}.json`,
upstreamUrl: celestrakSatcatUrl,
groups: new Set(coreGroups),
// A real answer, even for an empty group, starts with the CSV header.
validate: (body) => /^OBJECT_NAME,/.test(body),
invalid: 'no SATCAT header in response',
},
];
// CelesTrak throttling (403/429) is per client, so it holds every feed
// off; any other failure holds off only the feed that failed, so a bad
// SATCAT answer never blocks the TLEs the layer cannot do without.
let hostCooldownUntil = 0;
/** The largest group (`active`) measures ~2.7 MB; this bounds a runaway body. */
const CELESTRAK_MAX_RESPONSE_BYTES = 16 * 1024 * 1024;
async function fetchUpstream(group) {
const url = celestrakTleUrl(group);
const signal = AbortSignal.timeout(20000);
const res = await fetch(url.toString(), {
signal,
// CelesTrak 403s bulk groups (e.g. `active`) unless the request carries a
// descriptive User-Agent with a contact point.
headers: {
'User-Agent':
'gods-eye-view-celestrak-proxy/1.0 (+https://github.com/bilawalsidhu/gods-eye-view)',
},
});
if (!res.ok) {
// Group refreshes run in parallel: a later failure without Retry-After
// must never shorten a longer deadline an earlier one already set.
cooldownUntil = Math.max(
cooldownUntil,
Date.now() + providerCooldownMs(res),
);
void res.body?.cancel().catch(() => {});
throw new Error(`HTTP ${res.status}`);
}
const body = await readResponseTextCapped(
res,
CELESTRAK_MAX_RESPONSE_BYTES,
signal,
);
// An upstream error page parses to zero TLEs — treat as failure, keep cache.
if (!/^1 /m.test(body)) throw new Error('no TLE lines in response');
return { at: Date.now(), body };
}
function createFeedHandler(feed) {
let feedCooldownUntil = 0;
const cooldownUntil = () => Math.max(hostCooldownUntil, feedCooldownUntil);
const mem = new Map(); // group -> { at: epochMs, body: string }
const inflight = new Map(); // group -> Promise<{at, body}|null>
const diskPath = (group) => path.join(CACHE_DIR, feed.diskName(group));
const installMiddleware = (server) => {
server.middlewares.use('/api/celestrak', async (req, res) => {
async function readDisk(group) {
try {
const parsed = JSON.parse(await fsp.readFile(diskPath(group), 'utf8'));
if (typeof parsed?.body === 'string' && Number.isFinite(parsed?.at))
return parsed;
} catch {
/* no disk cache yet */
}
return null;
}
async function writeDisk(group, entry) {
try {
await fsp.mkdir(CACHE_DIR, { recursive: true });
await fsp.writeFile(diskPath(group), JSON.stringify(entry), 'utf8');
} catch (err) {
console.warn('[celestrak-proxy] cache write failed');
}
}
async function fetchUpstream(group) {
const url = feed.upstreamUrl(group);
const signal = AbortSignal.timeout(20000);
const res = await fetch(url.toString(), {
signal,
// CelesTrak 403s bulk groups (e.g. `active`) unless the request carries a
// descriptive User-Agent with a contact point.
headers: {
'User-Agent':
'gods-eye-view-celestrak-proxy/1.0 (+https://github.com/bilawalsidhu/gods-eye-view)',
},
});
if (!res.ok) {
// Group refreshes run in parallel: a later failure without Retry-After
// must never shorten a longer deadline an earlier one already set.
const until = Date.now() + providerCooldownMs(res);
if (res.status === 403 || res.status === 429)
hostCooldownUntil = Math.max(hostCooldownUntil, until);
else feedCooldownUntil = Math.max(feedCooldownUntil, until);
void res.body?.cancel().catch(() => {});
throw new Error(`HTTP ${res.status}`);
}
const body = await readResponseTextCapped(
res,
CELESTRAK_MAX_RESPONSE_BYTES,
signal,
);
if (!feed.validate(body)) throw new Error(feed.invalid);
return { at: Date.now(), body };
}
return async (req, res) => {
const group = String(req.url || '')
.replace(/^\//, '')
.split('?')[0];
@@ -111,7 +161,7 @@ export function celestrakProxy() {
res.end();
return;
}
if (!groups.has(group)) {
if (!feed.groups.has(group)) {
res.writeHead(400, { 'Content-Type': 'text/plain' });
res.end('invalid group');
return;
@@ -123,15 +173,15 @@ export function celestrakProxy() {
if (res.headersSent) return;
res.writeHead(status, {
'Content-Type': 'text/plain',
'x-tle-cache': cacheStatus,
[feed.cacheHeader]: cacheStatus,
'Cache-Control': 'no-store',
...(entry
? { 'X-TLE-Cached-At': new Date(entry.at).toISOString() }
? { [feed.cachedAtHeader]: new Date(entry.at).toISOString() }
: {}),
...(cooldownUntil > 0 && Date.now() < cooldownUntil
...(cooldownUntil() > 0 && Date.now() < cooldownUntil()
? {
'Retry-After': String(
Math.ceil((cooldownUntil - Date.now()) / 1000),
Math.ceil((cooldownUntil() - Date.now()) / 1000),
),
}
: {}),
@@ -145,11 +195,11 @@ export function celestrakProxy() {
entry = await readDisk(group);
if (entry) mem.set(group, entry);
}
if (entry && now - entry.at < TLE_TTL_MS) {
if (entry && now - entry.at < feed.ttlMs) {
send(200, entry.body, 'HIT', entry);
return;
}
if (Date.now() < cooldownUntil) {
if (Date.now() < cooldownUntil()) {
send(
entry ? 200 : 502,
entry?.body || 'celestrak upstream cooling down',
@@ -169,8 +219,8 @@ export function celestrakProxy() {
return fresh;
})
.catch((err) => {
if (Date.now() >= cooldownUntil)
cooldownUntil = Date.now() + 60000;
if (Date.now() >= feedCooldownUntil)
feedCooldownUntil = Date.now() + 60000;
console.warn(
'[celestrak-proxy] refresh failed — serving cache if any',
);
@@ -191,7 +241,13 @@ export function celestrakProxy() {
console.error('[celestrak-proxy] request failed');
send(500, 'celestrak proxy error', 'ERROR');
}
});
};
}
const handlers = feeds.map((feed) => [feed.route, createFeedHandler(feed)]);
const installMiddleware = (server) => {
for (const [route, handler] of handlers)
server.middlewares.use(route, handler);
};
return {
name: 'celestrak-proxy',