scrapely 1.2.3 → 1.2.4

package/dist/scrapely-proxy-configuration.js

@@ -0,0 +1,559 @@
+"use strict";
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.ProxyConfiguration = void 0;
+exports.isPlatformProxySentinel = isPlatformProxySentinel;
+exports.validateProxySessionId = validateProxySessionId;
+exports.resolveCountryOptions = resolveCountryOptions;
+exports.buildPlatformProxyUrl = buildPlatformProxyUrl;
+exports.hasPlatformSentinel = hasPlatformSentinel;
+const node_http_1 = __importDefault(require("node:http"));
+const node_tls_1 = __importDefault(require("node:tls"));
+const core_1 = require("@crawlee/core");
+const configuration_js_1 = require("./configuration.js");
+const proxy_configuration_js_1 = require("./proxy-configuration.js");
+/** Max length of a proxy session ID (Apify parity). */
+const MAX_SESSION_ID_LENGTH = 50;
+/** Allowed session ID characters (Apify parity): 0-9 a-z A-Z . _ ~ */
+const SESSION_ID_REGEX = /^[a-zA-Z0-9._~]+$/;
+/** Two-letter ISO country code (Apify parity). */
+const COUNTRY_CODE_REGEX = /^[A-Z]{2}$/;
+/**
+ * ISO 3166-2 subdivision codes are 1-3 uppercase alphanumeric characters,
+ * e.g. 'CA' (California), 'NSW' (New South Wales), '9' (Wien, AT-9)
+ * (Apify parity).
+ */
+const SUBDIVISION_CODE_REGEX = /^[A-Z0-9]{1,3}$/;
+/** Status-check timeout / attempts (Apify parity). */
+const CHECK_ACCESS_REQUEST_TIMEOUT_MILLIS = 4_000;
+const CHECK_ACCESS_MAX_ATTEMPTS = 2;
+/**
+ * Sentinel values that reserve a slot for the platform proxy inside
+ * `proxyUrls` / `tieredProxyUrls` / `tieredProxyConfig[].proxyUrls`,
+ * allowing BYOP and platform proxies to be mixed in one configuration
+ * (Apify-compatible: both `'auto'` and `'http://auto'` are accepted).
+ */
+const PLATFORM_PROXY_SENTINELS = new Set(['auto', 'http://auto']);
+/**
+ * Internal placeholder that occupies a rotation slot on the Crawlee side and is
+ * translated to a real platform proxy URL lazily in `newUrl()`. It must be a
+ * valid URL so Crawlee's validation accepts it, and it must never collide with
+ * a real upstream (the `.invalid` TLD is reserved by RFC 2606).
+ */
+const PLATFORM_PROXY_PLACEHOLDER_HOST = 'scrapely-platform.invalid';
+const PLATFORM_PROXY_PLACEHOLDER = `http://${PLATFORM_PROXY_PLACEHOLDER_HOST}`;
+/**
+ * Per-tier placeholder variant so a sentinel inside `tieredProxyConfig` /
+ * `tieredProxyUrls` translates with that tier's own groups/countryCode.
+ */
+function platformPlaceholderForTier(tierIndex) {
+    return `http://t${tierIndex}.${PLATFORM_PROXY_PLACEHOLDER_HOST}`;
+}
+/** True when a resolved URL is an internal platform-proxy placeholder. */
+function isPlatformPlaceholder(url) {
+    return (url === PLATFORM_PROXY_PLACEHOLDER || (url != null && url.endsWith(`.${PLATFORM_PROXY_PLACEHOLDER_HOST}`)));
+}
+/** Returns true for platform-proxy sentinel values ('auto', 'http://auto'). */
+function isPlatformProxySentinel(url) {
+    return url != null && PLATFORM_PROXY_SENTINELS.has(url);
+}
+/**
+ * Validates a proxy session ID the same way the Apify SDK does: at most 50
+ * characters, and only `0-9`, `a-z`, `A-Z`, `.`, `_` and `~`. Throws otherwise.
+ */
+function validateProxySessionId(sessionId) {
+    if (sessionId.length > MAX_SESSION_ID_LENGTH) {
+        throw new Error(`Session ID must not be longer than ${MAX_SESSION_ID_LENGTH} characters.`);
+    }
+    if (!SESSION_ID_REGEX.test(sessionId)) {
+        throw new Error(`Session ID "${sessionId}" contains unsupported characters. ` +
+            'Only "a-z", "A-Z", "0-9", ".", "_" and "~" are allowed.');
+    }
+    return sessionId;
+}
+/**
+ * Resolves and validates the country/subdivision options (Apify parity).
+ * Country must be a two-letter ISO code; subdivision an ISO 3166-2 code
+ * (1-3 uppercase alphanumeric characters) and requires a country. Returns
+ * the normalized values used by the username grammar.
+ */
+function resolveCountryOptions(options) {
+    const countryCode = options.countryCode ?? options.apifyProxyCountry;
+    const subdivisionCode = options.subdivisionCode ?? options.apifyProxySubdivision;
+    if (countryCode != null && !COUNTRY_CODE_REGEX.test(countryCode)) {
+        throw new Error(`ProxyConfiguration: "countryCode" must be a two-letter ISO country code (got "${countryCode}").`);
+    }
+    if (subdivisionCode != null) {
+        if (!SUBDIVISION_CODE_REGEX.test(subdivisionCode)) {
+            throw new Error(`ProxyConfiguration: "subdivisionCode" must be a 1-3 character ISO 3166-2 subdivision code ` +
+                `(got "${subdivisionCode}").`);
+        }
+        if (!countryCode) {
+            throw new Error('ProxyConfiguration: "subdivisionCode" requires "countryCode" to be set.');
+        }
+    }
+    return { countryCode, subdivisionCode };
+}
+/**
+ * Strips URL-breaking characters from a username fragment. The gateway grammar
+ * is comma-based, so fragments themselves must never contain `,@:/?#` or
+ * whitespace (Apify behavior: the username is NOT percent-encoded).
+ */
+function sanitizeUsernameFragment(value) {
+    return value.replace(/[:@/?#\s]/g, '');
+}
+/**
+ * Canonical percent-encoding for proxy URL userinfo (username/password).
+ * Matches Python's urllib.parse.quote(safe='') exactly: escapes everything
+ * except A-Z a-z 0-9 - _ . ~ — including !'()* which encodeURIComponent
+ * leaves unencoded. Both encodings are RFC 3986-valid, but SDKs must emit
+ * byte-identical URLs.
+ */
+function encodeUserInfoPart(value) {
+    return encodeURIComponent(value).replace(/[!'()*]/g, (c) => `%${c.charCodeAt(0).toString(16).toUpperCase()}`);
+}
+/**
+ * Gateway upstream error codes (590-599, proxy-chain semantics). The gateway
+ * uses these instead of a bare 502 to describe upstream failures — the same
+ * table Apify Proxy uses. See docs/evomi-integration.md for the contract.
+ */
+const PROXY_STATUS_CODE_DESCRIPTIONS = {
+    590: 'Non Successful: upstream responded with a non-200 status code',
+    591: 'reserved for further use',
+    592: 'Status Code Out Of Range: upstream responded with status code different than 100-999',
+    593: 'Not Found: DNS lookup failed (EAI_NODATA or EAI_NONAME)',
+    594: 'Connection Refused: upstream refused connection',
+    595: 'Connection Reset: connection reset due to loss of connection or timeout',
+    596: 'Broken Pipe: trying to write on a closed socket',
+    597: 'Auth Failed: incorrect upstream credentials',
+    598: 'reserved for further use',
+    599: 'Upstream Error: generic upstream error',
+};
+/**
+ * Performs a GET through an HTTP forward proxy (the platform gateway).
+ * - http targets: absolute-form request line + `Proxy-Authorization` header
+ * - https targets: CONNECT tunnel + TLS + a raw HTTP/1.1 request
+ * Network errors reject; HTTP status codes always resolve.
+ */
+async function requestThroughProxy(targetUrl, proxyUrl, timeoutMs) {
+    const target = new URL(targetUrl);
+    const proxy = new URL(proxyUrl);
+    const credentials = `${decodeURIComponent(proxy.username)}:${decodeURIComponent(proxy.password)}`;
+    const proxyAuthorization = `Basic ${Buffer.from(credentials).toString('base64')}`;
+    if (target.protocol === 'http:') {
+        return await new Promise((resolve, reject) => {
+            const req = node_http_1.default.request({
+                host: proxy.hostname,
+                port: Number(proxy.port) || 80,
+                method: 'GET',
+                path: targetUrl,
+                headers: {
+                    Host: target.host,
+                    'Proxy-Authorization': proxyAuthorization,
+                    Accept: 'application/json',
+                },
+                setHost: false,
+            }, (res) => {
+                let body = '';
+                res.setEncoding('utf8');
+                res.on('data', (chunk) => {
+                    body += chunk;
+                });
+                res.on('end', () => resolve({ statusCode: res.statusCode ?? 0, body }));
+            });
+            req.setTimeout(timeoutMs, () => req.destroy(new Error(`request timed out after ${timeoutMs}ms`)));
+            req.on('error', reject);
+            req.end();
+        });
+    }
+    // HTTPS target: CONNECT tunnel through the gateway, then TLS + raw HTTP/1.1.
+    return await new Promise((resolve, reject) => {
+        const connectReq = node_http_1.default.request({
+            host: proxy.hostname,
+            port: Number(proxy.port) || 80,
+            method: 'CONNECT',
+            path: `${target.hostname}:${target.port || 443}`,
+            headers: {
+                Host: `${target.hostname}:${target.port || 443}`,
+                'Proxy-Authorization': proxyAuthorization,
+            },
+            setHost: false,
+        });
+        connectReq.setTimeout(timeoutMs, () => connectReq.destroy(new Error(`CONNECT timed out after ${timeoutMs}ms`)));
+        connectReq.on('error', reject);
+        connectReq.on('connect', (res, socket) => {
+            if (res.statusCode !== 200) {
+                socket.destroy();
+                resolve({ statusCode: res.statusCode ?? 0, body: '' });
+                return;
+            }
+            const tlsSocket = node_tls_1.default.connect({ socket, servername: target.hostname }, () => {
+                tlsSocket.write(`GET ${target.pathname}${target.search} HTTP/1.1\r\n` +
+                    `Host: ${target.host}\r\n` +
+                    'Accept: application/json\r\n' +
+                    'Connection: close\r\n\r\n');
+            });
+            let raw = '';
+            tlsSocket.setEncoding('utf8');
+            tlsSocket.on('data', (chunk) => {
+                raw += chunk;
+            });
+            tlsSocket.on('end', () => {
+                const separatorIndex = raw.indexOf('\r\n\r\n');
+                const head = separatorIndex >= 0 ? raw.slice(0, separatorIndex) : raw;
+                const responseBody = separatorIndex >= 0 ? raw.slice(separatorIndex + 4) : '';
+                const statusCode = Number.parseInt(head.split(' ')[1] ?? '', 10);
+                resolve({ statusCode: Number.isFinite(statusCode) ? statusCode : 0, body: responseBody });
+            });
+            tlsSocket.on('error', reject);
+        });
+        connectReq.end();
+    });
+}
+/**
+ * Build a platform proxy URL using the Apify username grammar:
+ *   http://GROUPS-G1+G2,country-US,session-a1b2c3:PASSWORD@HOST:PORT
+ *
+ * `groups` (joined with `+`), `country` and `session` fragments are
+ * comma-joined. When the server injected a subaccount username
+ * (SCRAPELY_PROXY_USERNAME / APIFY_PROXY_USERNAME) it is prepended so the
+ * gateway can attribute traffic to the right Scrapely tenant; without it the
+ * username falls back to Apify's `auto` when no fragments apply. The password
+ * and gateway host/port come from the environment credentials injected by the
+ * server.
+ */
+function buildPlatformProxyUrl(options, env, sessionId) {
+    const groups = options.groups?.length ? options.groups : options.apifyProxyGroups;
+    const { countryCode, subdivisionCode } = resolveCountryOptions(options);
+    const password = options.password ?? env.password;
+    // Apify username grammar: groups, then session, then country — with the
+    // subdivision merged into the country fragment (ISO 3166-2: `country-US_CA`).
+    const usernameParts = [];
+    if (groups?.length) {
+        usernameParts.push(`groups-${groups.map(sanitizeUsernameFragment).join('+')}`);
+    }
+    if (sessionId) {
+        usernameParts.push(`session-${sanitizeUsernameFragment(validateProxySessionId(sessionId))}`);
+    }
+    if (subdivisionCode) {
+        usernameParts.push(`country-${sanitizeUsernameFragment(countryCode)}_${sanitizeUsernameFragment(subdivisionCode)}`);
+    }
+    else if (countryCode) {
+        usernameParts.push(`country-${sanitizeUsernameFragment(countryCode)}`);
+    }
+    const fragments = usernameParts.join(',');
+    const username = env.username
+        ? sanitizeUsernameFragment(env.username) + (fragments ? `,${fragments}` : '')
+        : fragments || 'auto';
+    return `http://${username}:${encodeUserInfoPart(password)}@${env.hostname}:${env.port}`;
+}
+function urlListHasSentinel(urls) {
+    return !!urls?.some((url) => isPlatformProxySentinel(url));
+}
+function replaceSentinelUrls(urls, placeholder = PLATFORM_PROXY_PLACEHOLDER) {
+    return urls.map((url) => (isPlatformProxySentinel(url) ? placeholder : url));
+}
+/** True when any proxy URL list reserves a platform slot via the 'auto' sentinel. */
+function hasPlatformSentinel(options) {
+    return (urlListHasSentinel(options.proxyUrls) ||
+        urlListHasSentinel(options.tieredProxyUrls?.flat()) ||
+        !!options.tieredProxyConfig?.some((tier) => urlListHasSentinel(tier.proxyUrls)));
+}
+function urlListHasRealUrl(urls) {
+    return !!urls?.some((url) => url != null && !isPlatformProxySentinel(url));
+}
+/**
+ * True when the configuration contains real (non-sentinel) custom proxies:
+ * a BYOP entry in any list or a custom newUrlFunction. Sentinel-only lists do
+ * not count — they are platform slots and may legally combine with
+ * groups/countryCode.
+ */
+function hasRealCustomProxyOptions(options) {
+    return (urlListHasRealUrl(options.proxyUrls) ||
+        typeof options.newUrlFunction === 'function' ||
+        !!options.tieredProxyUrls?.some((tier) => urlListHasRealUrl(tier)) ||
+        !!options.tieredProxyConfig?.some((tier) => urlListHasRealUrl(tier.proxyUrls)));
+}
+/**
+ * Maps declarative tiers to URL lists handed to Crawlee. Tiers with explicit
+ * `proxyUrls` pass through (sentinels become per-tier placeholders); tiers
+ * without any URLs become per-tier placeholders too, so their platform
+ * options (groups/countryCode/password) and the session ID are applied
+ * lazily in `newUrl()` — mirroring Apify's `_generateTieredProxyUrls` while
+ * preserving session stickiness.
+ */
+function mapTieredProxyConfig(tieredProxyConfig, globalOptions, env, placeholderOptions) {
+    return tieredProxyConfig.map((tier, index) => {
+        if (tier.newUrlFunction) {
+            throw new Error('[Scrapely] tieredProxyConfig tiers with newUrlFunction are not supported. Use tieredProxyUrls instead.');
+        }
+        if (tier.proxyUrls) {
+            if (tier.proxyUrls.some((url) => isPlatformProxySentinel(url))) {
+                // Register the tier's own platform options so the placeholder
+                // translates with this tier's groups/countryCode, not globals.
+                placeholderOptions.set(platformPlaceholderForTier(index), { ...globalOptions, ...tier });
+            }
+            return replaceSentinelUrls(tier.proxyUrls, platformPlaceholderForTier(index));
+        }
+        if (!env) {
+            // Defensive backstop: the constructor already rejects this case with
+            // a more detailed message; keep it in case of direct invocation.
+            throw new Error('[Scrapely] tieredProxyConfig tiers without proxyUrls require platform proxy credentials ' +
+                '(SCRAPELY_PROXY_* or APIFY_PROXY_* env), which were not injected into this environment.');
+        }
+        // URL-less tiers also become per-tier placeholders so newUrl(sessionId)
+        // resolves them lazily with the tier's groups/countryCode AND the
+        // session ID — building a concrete URL here would lose session
+        // stickiness (the gateway would rotate IPs for repeated sessions).
+        placeholderOptions.set(platformPlaceholderForTier(index), { ...globalOptions, ...tier });
+        return [platformPlaceholderForTier(index)];
+    });
+}
+function toCrawleeOptions(options, env, placeholderOptions) {
+    const crawleeOptions = {};
+    if (options.proxyUrls) {
+        crawleeOptions.proxyUrls = replaceSentinelUrls(options.proxyUrls);
+    }
+    if (options.newUrlFunction) {
+        crawleeOptions.newUrlFunction = options.newUrlFunction;
+    }
+    if (options.tieredProxyUrls) {
+        crawleeOptions.tieredProxyUrls = options.tieredProxyUrls.map((tier, index) => {
+            const hasSentinel = tier.some((url) => isPlatformProxySentinel(url));
+            if (hasSentinel) {
+                placeholderOptions.set(platformPlaceholderForTier(index), options);
+            }
+            return replaceSentinelUrls(tier, platformPlaceholderForTier(index));
+        });
+    }
+    else if (options.tieredProxyConfig) {
+        crawleeOptions.tieredProxyUrls = mapTieredProxyConfig(options.tieredProxyConfig, options, env, placeholderOptions);
+    }
+    return crawleeOptions;
+}
+function hasApifyOnlyOptions(options) {
+    const groups = options.groups?.length ? options.groups : options.apifyProxyGroups;
+    const { countryCode, subdivisionCode } = resolveCountryOptions(options);
+    return !!(groups?.length || countryCode || subdivisionCode || options.password);
+}
+function hasCustomProxyOptions(options) {
+    return !!((options.proxyUrls && options.proxyUrls.length > 0) ||
+        typeof options.newUrlFunction === 'function' ||
+        (options.tieredProxyUrls && options.tieredProxyUrls.length > 0) ||
+        (options.tieredProxyConfig && options.tieredProxyConfig.length > 0));
+}
+/**
+ * True when platform URLs will be generated either way (pure platform mode,
+ * a sentinel slot, or a declarative tier that expands to platform URLs).
+ */
+function needsPlatformProxyUrls(options) {
+    return (!hasCustomProxyOptions(options) ||
+        hasPlatformSentinel(options) ||
+        !!options.tieredProxyConfig?.some((tier) => !tier.proxyUrls));
+}
+/**
+ * Apify-compatible ProxyConfiguration.
+ *
+ * Modes:
+ *  - Platform proxy: when the server injected APIFY_PROXY_* or SCRAPELY_PROXY_*
+ *    credentials (Evomi-backed) and no custom proxies were requested, the
+ *    configuration routes through the platform gateway using the Apify
+ *    username grammar (groups/country/session fragments).
+ *  - BYOP: custom proxyUrls / newUrlFunction / tieredProxyUrls are passed
+ *    through to Crawlee unchanged.
+ *  - Mixed: the 'auto' (or 'http://auto') sentinel inside proxyUrls /
+ *    tieredProxyUrls / tieredProxyConfig[].proxyUrls reserves a slot for the
+ *    platform proxy alongside BYOP entries. The sentinel is translated to a
+ *    concrete platform URL lazily on every newUrl() call, so Crawlee's
+ *    rotation and session-stickiness semantics apply to the mix unchanged.
+ */
+class ProxyConfiguration extends core_1.ProxyConfiguration {
+    config;
+    platformEnv;
+    platformOptions;
+    /** Platform options to use when translating each per-tier placeholder. */
+    tierPlaceholderOptions;
+    constructor(options = {}, config = configuration_js_1.Configuration.getGlobalConfig()) {
+        const { proxyUrls, newUrlFunction } = options;
+        const platformEnv = (0, proxy_configuration_js_1.getPlatformProxyEnv)();
+        // Validate country/subdivision options eagerly (Apify parity: bad
+        // values throw at construction, not on the first newUrl() call).
+        resolveCountryOptions(options);
+        if (hasRealCustomProxyOptions(options) && hasApifyOnlyOptions(options)) {
+            throw new Error('Cannot combine custom proxies with Apify Proxy! ' +
+                'It is not allowed to set "options.proxyUrls" or "options.newUrlFunction" combined with ' +
+                '"options.groups" or "options.apifyProxyGroups" and "options.countryCode" or ' +
+                '"options.apifyProxyCountry", "options.subdivisionCode" or "options.apifyProxySubdivision".');
+        }
+        if (needsPlatformProxyUrls(options) && !platformEnv) {
+            throw new Error('[Scrapely] The platform proxy was requested (via the "auto" sentinel, tieredProxyConfig tiers without ' +
+                'proxyUrls, or by relying on the default platform configuration), but no platform proxy credentials ' +
+                '(SCRAPELY_PROXY_* or APIFY_PROXY_*) were injected into this environment. ' +
+                'The platform proxy is unavailable in local runs — remove the platform proxy entries to run locally, ' +
+                'or run the actor on Scrapely with proxy access enabled.');
+        }
+        // Pure platform proxy mode activates only when credentials were
+        // injected AND the user did not bring any proxy list/function at all.
+        const purePlatformMode = !!platformEnv && !hasCustomProxyOptions(options);
+        const tierPlaceholderOptions = new Map();
+        super({
+            ...toCrawleeOptions(options, platformEnv, tierPlaceholderOptions),
+            ...(purePlatformMode && platformEnv
+                ? {
+                    newUrlFunction: (sessionId) => buildPlatformProxyUrl(options, platformEnv, sessionId != null ? validateProxySessionId(String(sessionId)) : undefined),
+                }
+                : {}),
+            ['validateRequired']: false,
+        });
+        this.config = config;
+        this.platformEnv = needsPlatformProxyUrls(options) ? platformEnv : null;
+        this.platformOptions = options;
+        this.tierPlaceholderOptions = tierPlaceholderOptions;
+        if (proxyUrls && newUrlFunction) {
+            this._throwCannotCombineCustomMethods();
+        }
+    }
+    /**
+     * Returns a new proxy URL. Translates the internal platform-proxy
+     * placeholder to a concrete platform URL (applying groups/countryCode and
+     * session stickiness) whenever Crawlee's rotation lands on it; BYOP
+     * entries pass through unchanged. Session IDs are validated like in the
+     * Apify SDK (max 50 chars, only `0-9 a-z A-Z . _ ~`).
+     */
+    async newUrl(sessionId, options) {
+        const sid = sessionId != null ? validateProxySessionId(String(sessionId)) : undefined;
+        const url = await super.newUrl(sid, options);
+        const platformEnv = this.platformEnv;
+        if (url != null && isPlatformPlaceholder(url) && platformEnv) {
+            // Per-tier placeholders translate with the tier's own options;
+            // the generic one falls back to the top-level configuration.
+            const platformOptions = url === PLATFORM_PROXY_PLACEHOLDER
+                ? this.platformOptions
+                : this.tierPlaceholderOptions.get(url) ?? this.platformOptions;
+            return buildPlatformProxyUrl(platformOptions, platformEnv, sid);
+        }
+        return url;
+    }
+    /**
+     * Returns a new ProxyInfo. When the resolved URL points at the platform
+     * gateway, `groups`, `countryCode` and `subdivisionCode` are attached
+     * (Apify parity).
+     */
+    async newProxyInfo(sessionId, options) {
+        const sid = sessionId != null ? validateProxySessionId(String(sessionId)) : undefined;
+        const info = await super.newProxyInfo(sid, options);
+        if (info && this.platformEnv && info.hostname === this.platformEnv.hostname) {
+            const groups = this.platformOptions.groups?.length
+                ? this.platformOptions.groups
+                : this.platformOptions.apifyProxyGroups;
+            const { countryCode, subdivisionCode } = resolveCountryOptions(this.platformOptions);
+            return {
+                ...info,
+                ...(groups?.length ? { groups } : {}),
+                ...(countryCode ? { countryCode } : {}),
+                ...(subdivisionCode ? { subdivisionCode } : {}),
+            };
+        }
+        return info;
+    }
+    /**
+     * Fetches the gateway status payload (Apify format) by sending
+     * `GET {statusUrl}/?format=json` *through the platform gateway*, so the
+     * proxy handshake itself authenticates the run's credentials. Retries
+     * connection errors (Apify parity: 4s timeout, 2 attempts); returns
+     * undefined when the endpoint stays unreachable.
+     */
+    async fetchProxyStatus() {
+        const env = this.platformEnv;
+        if (!env?.statusUrl)
+            return undefined;
+        const statusUrl = env.statusUrl;
+        const url = new URL(statusUrl);
+        url.searchParams.set('format', 'json');
+        const targetUrl = url.toString();
+        const proxyUrl = buildPlatformProxyUrl(this.platformOptions, env);
+        for (let attempt = 1; attempt <= CHECK_ACCESS_MAX_ATTEMPTS; attempt++) {
+            try {
+                const { statusCode, body } = await requestThroughProxy(targetUrl, proxyUrl, CHECK_ACCESS_REQUEST_TIMEOUT_MILLIS);
+                if (statusCode === 407) {
+                    return {
+                        connected: false,
+                        connectionError: 'Platform proxy authentication failed (HTTP 407). Check the injected ' +
+                            'SCRAPELY_PROXY_*/APIFY_PROXY_* credentials.',
+                        isManInTheMiddle: false,
+                    };
+                }
+                if (statusCode >= 590 && statusCode <= 599) {
+                    const description = PROXY_STATUS_CODE_DESCRIPTIONS[statusCode] ?? 'upstream error';
+                    return {
+                        connected: false,
+                        connectionError: `Platform proxy upstream error (HTTP ${statusCode}: ${description}).`,
+                        isManInTheMiddle: false,
+                    };
+                }
+                if (statusCode >= 200 && statusCode < 300) {
+                    try {
+                        const parsed = JSON.parse(body);
+                        return {
+                            connected: parsed.connected === true,
+                            connectionError: parsed.connectionError ?? '',
+                            isManInTheMiddle: parsed.isManInTheMiddle === true,
+                        };
+                    }
+                    catch {
+                        // Malformed JSON — fall through and retry.
+                    }
+                }
+            }
+            catch {
+                // Retry connection errors (Apify behavior).
+            }
+        }
+        return undefined;
+    }
+    /**
+     * Verifies the proxy configuration. For platform proxy mode with an
+     * injected status URL, performs the Apify-format access check through the
+     * gateway (`GET {statusUrl}/?format=json`, best-effort, never fatal);
+     * BYOP mode only verifies that a test URL can be generated.
+     */
+    async initialize(options = {}) {
+        const { checkAccess = true } = options;
+        if (!checkAccess) {
+            return true;
+        }
+        // Apify-parity access check: GET {statusUrl}/?format=json *through the
+        // platform gateway* so the proxy handshake itself authenticates the
+        // run's credentials. Best-effort — never hard fails the run.
+        if (checkAccess && this.platformEnv?.statusUrl) {
+            const status = await this.fetchProxyStatus();
+            if (!status) {
+                console.warn('[Scrapely] Platform proxy access check timed out. Watch out for errors with status code 407. ' +
+                    "If you see some, it most likely means you don't have access to either all or some of the proxies you're trying to use.");
+            }
+            else {
+                this.isManInTheMiddle = status.isManInTheMiddle;
+                if (!status.connected) {
+                    console.warn(`[Scrapely] Platform proxy access check failed: ${status.connectionError}`);
+                }
+                else {
+                    console.log('[Scrapely] Platform proxy access check succeeded.');
+                }
+            }
+        }
+        try {
+            const testUrl = await this.newUrl('test.session');
+            if (testUrl) {
+                console.log('[Scrapely] Proxy configuration initialized. Test proxy URL available.');
+            }
+            return true;
+        }
+        catch (err) {
+            console.warn(`[Scrapely] Proxy configuration verification failed: ${err.message}`);
+            return true;
+        }
+    }
+}
+exports.ProxyConfiguration = ProxyConfiguration;