mcp-scraper 0.40.1 → 0.40.3
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +1 -1
- package/dist/bin/api-server.cjs +61501 -0
- package/dist/bin/api-server.cjs.map +1 -0
- package/dist/bin/api-server.d.cts +1 -0
- package/dist/bin/api-server.d.ts +1 -0
- package/dist/bin/api-server.js +38 -0
- package/dist/bin/api-server.js.map +1 -0
- package/dist/bin/mcp-scraper-cli.cjs +2671 -0
- package/dist/bin/mcp-scraper-cli.cjs.map +1 -0
- package/dist/bin/mcp-scraper-cli.d.cts +1 -0
- package/dist/bin/mcp-scraper-cli.d.ts +1 -0
- package/dist/bin/mcp-scraper-cli.js +742 -0
- package/dist/bin/mcp-scraper-cli.js.map +1 -0
- package/dist/bin/mcp-scraper-install.cjs +129 -0
- package/dist/bin/mcp-scraper-install.cjs.map +1 -0
- package/dist/bin/mcp-scraper-install.d.cts +1 -0
- package/dist/bin/mcp-scraper-install.d.ts +1 -0
- package/dist/bin/mcp-scraper-install.js +27 -0
- package/dist/bin/mcp-scraper-install.js.map +1 -0
- package/dist/bin/mcp-stdio-server.cjs +12600 -0
- package/dist/bin/mcp-stdio-server.cjs.map +1 -0
- package/dist/bin/mcp-stdio-server.d.cts +1 -0
- package/dist/bin/mcp-stdio-server.d.ts +1 -0
- package/dist/bin/mcp-stdio-server.js +136 -0
- package/dist/bin/mcp-stdio-server.js.map +1 -0
- package/dist/bin/paa-harvest.cjs +3830 -0
- package/dist/bin/paa-harvest.cjs.map +1 -0
- package/dist/bin/paa-harvest.d.cts +1 -0
- package/dist/bin/paa-harvest.d.ts +1 -0
- package/dist/bin/paa-harvest.js +44 -0
- package/dist/bin/paa-harvest.js.map +1 -0
- package/dist/chunk-27DUCRAZ.js +684 -0
- package/dist/chunk-27DUCRAZ.js.map +1 -0
- package/dist/chunk-345BQXZH.js +712 -0
- package/dist/chunk-345BQXZH.js.map +1 -0
- package/dist/chunk-5UN33CGU.js +89 -0
- package/dist/chunk-5UN33CGU.js.map +1 -0
- package/dist/chunk-6J57U6HA.js +499 -0
- package/dist/chunk-6J57U6HA.js.map +1 -0
- package/dist/chunk-6SZ52BQ5.js +11503 -0
- package/dist/chunk-6SZ52BQ5.js.map +1 -0
- package/dist/chunk-BBI7RGOT.js +99 -0
- package/dist/chunk-BBI7RGOT.js.map +1 -0
- package/dist/chunk-CB5C3BPB.js +135 -0
- package/dist/chunk-CB5C3BPB.js.map +1 -0
- package/dist/chunk-CS4HE6QY.js +194 -0
- package/dist/chunk-CS4HE6QY.js.map +1 -0
- package/dist/chunk-EHES33KB.js +276 -0
- package/dist/chunk-EHES33KB.js.map +1 -0
- package/dist/chunk-FQI5PFE7.js +1866 -0
- package/dist/chunk-FQI5PFE7.js.map +1 -0
- package/dist/chunk-G3P3ZDB4.js +69 -0
- package/dist/chunk-G3P3ZDB4.js.map +1 -0
- package/dist/chunk-IQTT7CAB.js +3141 -0
- package/dist/chunk-IQTT7CAB.js.map +1 -0
- package/dist/chunk-JVIW4GK2.js +3442 -0
- package/dist/chunk-JVIW4GK2.js.map +1 -0
- package/dist/chunk-K443GQY5.js +24 -0
- package/dist/chunk-K443GQY5.js.map +1 -0
- package/dist/chunk-QZXKQB7Y.js +414 -0
- package/dist/chunk-QZXKQB7Y.js.map +1 -0
- package/dist/chunk-SBLGBZZB.js +647 -0
- package/dist/chunk-SBLGBZZB.js.map +1 -0
- package/dist/chunk-XELSA2MS.js +108 -0
- package/dist/chunk-XELSA2MS.js.map +1 -0
- package/dist/chunk-XU4ZLIB2.js +7 -0
- package/dist/chunk-XU4ZLIB2.js.map +1 -0
- package/dist/db-LVVU6NOK.js +253 -0
- package/dist/db-LVVU6NOK.js.map +1 -0
- package/dist/editorial-reading-room/assets/app.js +335 -0
- package/dist/editorial-reading-room/assets/index.html +131 -0
- package/dist/editorial-reading-room/assets/styles.css +1052 -0
- package/dist/extract-bundle-SKXEDG4Z.js +568 -0
- package/dist/extract-bundle-SKXEDG4Z.js.map +1 -0
- package/dist/index.cjs +4182 -0
- package/dist/index.cjs.map +1 -0
- package/dist/index.d.cts +413 -0
- package/dist/index.d.ts +413 -0
- package/dist/index.js +338 -0
- package/dist/index.js.map +1 -0
- package/dist/location-data-repository-NBTCOA7B.js +35 -0
- package/dist/location-data-repository-NBTCOA7B.js.map +1 -0
- package/dist/server-W76RUYD3.js +35716 -0
- package/dist/server-W76RUYD3.js.map +1 -0
- package/dist/site-extract-repository-6GX72U4N.js +62 -0
- package/dist/site-extract-repository-6GX72U4N.js.map +1 -0
- package/dist/worker-5RWLFSAJ.js +141 -0
- package/dist/worker-5RWLFSAJ.js.map +1 -0
- package/package.json +4 -2
|
@@ -0,0 +1,99 @@
|
|
|
1
|
+
import {
|
|
2
|
+
publicErrorMessage
|
|
3
|
+
} from "./chunk-5UN33CGU.js";
|
|
4
|
+
|
|
5
|
+
// src/api/outbound-sanitize.ts
|
|
6
|
+
var KEY_RENAMES = {
|
|
7
|
+
kernel: "browserRuntime"
|
|
8
|
+
};
|
|
9
|
+
var SANITIZED_VALUE_KEYS = /error|message/i;
|
|
10
|
+
var INTERNAL_DIAGNOSTIC_KEYS = /* @__PURE__ */ new Set([
|
|
11
|
+
"kernel_session_id",
|
|
12
|
+
"browser_session_id",
|
|
13
|
+
"kernelSessionId",
|
|
14
|
+
"browserSessionId",
|
|
15
|
+
"sessionId",
|
|
16
|
+
"kernel_delete_started",
|
|
17
|
+
"session_cleanup_started",
|
|
18
|
+
"kernelDeleteStarted",
|
|
19
|
+
"sessionCleanupStarted",
|
|
20
|
+
"kernel_delete_succeeded",
|
|
21
|
+
"session_cleanup_succeeded",
|
|
22
|
+
"kernelDeleteSucceeded",
|
|
23
|
+
"sessionCleanupSucceeded",
|
|
24
|
+
"kernel_delete_error",
|
|
25
|
+
"session_cleanup_error",
|
|
26
|
+
"kernelDeleteError",
|
|
27
|
+
"sessionCleanupError",
|
|
28
|
+
"kernelProxyId",
|
|
29
|
+
"proxyId",
|
|
30
|
+
"requestedProxyIdSuffix",
|
|
31
|
+
"retrievedProxyIdSuffix",
|
|
32
|
+
"provider",
|
|
33
|
+
"providerName",
|
|
34
|
+
"service",
|
|
35
|
+
"serviceName",
|
|
36
|
+
"vendor",
|
|
37
|
+
"vendorName",
|
|
38
|
+
"organization",
|
|
39
|
+
"organizationId",
|
|
40
|
+
"accountId"
|
|
41
|
+
]);
|
|
42
|
+
var VENDOR_URL_KEYS = /* @__PURE__ */ new Set([
|
|
43
|
+
"hosted_url",
|
|
44
|
+
"hostedUrl",
|
|
45
|
+
"live_view_url",
|
|
46
|
+
"liveViewUrl",
|
|
47
|
+
"cdp_ws_url",
|
|
48
|
+
"cdpWsUrl",
|
|
49
|
+
"browser_live_view_url"
|
|
50
|
+
]);
|
|
51
|
+
var VENDOR_URL_RE = /\b(?:wss?|https?):\/\/[^\s"'<>]*\bonkernel\.com[^\s"'<>]*/gi;
|
|
52
|
+
function redactVendorUrls(value) {
|
|
53
|
+
return value.replace(VENDOR_URL_RE, "[browser-service]");
|
|
54
|
+
}
|
|
55
|
+
function sanitizeOutboundDiagnostics(value, parentKey = "") {
|
|
56
|
+
if (typeof value === "string") {
|
|
57
|
+
let out = redactVendorUrls(value);
|
|
58
|
+
if (SANITIZED_VALUE_KEYS.test(parentKey)) {
|
|
59
|
+
out = publicErrorMessage("service_unavailable");
|
|
60
|
+
}
|
|
61
|
+
return out;
|
|
62
|
+
}
|
|
63
|
+
if (Array.isArray(value)) return value.map((v) => sanitizeOutboundDiagnostics(v, parentKey));
|
|
64
|
+
if (value !== null && typeof value === "object") {
|
|
65
|
+
const out = {};
|
|
66
|
+
for (const [key, val] of Object.entries(value)) {
|
|
67
|
+
if (INTERNAL_DIAGNOSTIC_KEYS.has(key)) continue;
|
|
68
|
+
const renamed = KEY_RENAMES[key] ?? key;
|
|
69
|
+
if (VENDOR_URL_KEYS.has(key)) {
|
|
70
|
+
out[renamed] = null;
|
|
71
|
+
continue;
|
|
72
|
+
}
|
|
73
|
+
out[renamed] = sanitizeOutboundDiagnostics(val, key);
|
|
74
|
+
}
|
|
75
|
+
return out;
|
|
76
|
+
}
|
|
77
|
+
return value;
|
|
78
|
+
}
|
|
79
|
+
function sanitizeAttempts(attempts) {
|
|
80
|
+
return attempts.map((a) => sanitizeOutboundDiagnostics(a));
|
|
81
|
+
}
|
|
82
|
+
function sanitizeHarvestResult(result) {
|
|
83
|
+
const diagnostics = result?.diagnostics;
|
|
84
|
+
if (!diagnostics?.debug) return result;
|
|
85
|
+
return {
|
|
86
|
+
...result,
|
|
87
|
+
diagnostics: {
|
|
88
|
+
...diagnostics,
|
|
89
|
+
debug: sanitizeOutboundDiagnostics(diagnostics.debug)
|
|
90
|
+
}
|
|
91
|
+
};
|
|
92
|
+
}
|
|
93
|
+
|
|
94
|
+
export {
|
|
95
|
+
sanitizeOutboundDiagnostics,
|
|
96
|
+
sanitizeAttempts,
|
|
97
|
+
sanitizeHarvestResult
|
|
98
|
+
};
|
|
99
|
+
//# sourceMappingURL=chunk-BBI7RGOT.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"sources":["../src/api/outbound-sanitize.ts"],"sourcesContent":["import { publicErrorMessage } from '../errors.js'\n\nconst KEY_RENAMES: Record<string, string> = {\n kernel: 'browserRuntime',\n}\n\nconst SANITIZED_VALUE_KEYS = /error|message/i\nconst INTERNAL_DIAGNOSTIC_KEYS = new Set([\n 'kernel_session_id',\n 'browser_session_id',\n 'kernelSessionId',\n 'browserSessionId',\n 'sessionId',\n 'kernel_delete_started',\n 'session_cleanup_started',\n 'kernelDeleteStarted',\n 'sessionCleanupStarted',\n 'kernel_delete_succeeded',\n 'session_cleanup_succeeded',\n 'kernelDeleteSucceeded',\n 'sessionCleanupSucceeded',\n 'kernel_delete_error',\n 'session_cleanup_error',\n 'kernelDeleteError',\n 'sessionCleanupError',\n 'kernelProxyId',\n 'proxyId',\n 'requestedProxyIdSuffix',\n 'retrievedProxyIdSuffix',\n 'provider',\n 'providerName',\n 'service',\n 'serviceName',\n 'vendor',\n 'vendorName',\n 'organization',\n 'organizationId',\n 'accountId',\n])\n\nconst VENDOR_URL_KEYS = new Set([\n 'hosted_url',\n 'hostedUrl',\n 'live_view_url',\n 'liveViewUrl',\n 'cdp_ws_url',\n 'cdpWsUrl',\n 'browser_live_view_url',\n])\nconst VENDOR_URL_RE = /\\b(?:wss?|https?):\\/\\/[^\\s\"'<>]*\\bonkernel\\.com[^\\s\"'<>]*/gi\n\nfunction redactVendorUrls(value: string): string {\n return value.replace(VENDOR_URL_RE, '[browser-service]')\n}\n\nexport function sanitizeOutboundDiagnostics<T>(value: T, parentKey = ''): T {\n if (typeof value === 'string') {\n let out = redactVendorUrls(value)\n if (SANITIZED_VALUE_KEYS.test(parentKey)) {\n out = publicErrorMessage('service_unavailable')\n }\n return out as T\n }\n if (Array.isArray(value)) return value.map(v => sanitizeOutboundDiagnostics(v, parentKey)) as T\n if (value !== null && typeof value === 'object') {\n const out: Record<string, unknown> = {}\n for (const [key, val] of Object.entries(value as Record<string, unknown>)) {\n if (INTERNAL_DIAGNOSTIC_KEYS.has(key)) continue\n const renamed = KEY_RENAMES[key] ?? key\n if (VENDOR_URL_KEYS.has(key)) {\n out[renamed] = null\n continue\n }\n out[renamed] = sanitizeOutboundDiagnostics(val, key)\n }\n return out as T\n }\n return value\n}\n\nexport function sanitizeAttempts<T>(attempts: T[]): T[] {\n return attempts.map(a => sanitizeOutboundDiagnostics(a))\n}\n\nexport function sanitizeHarvestResult<T extends Record<string, unknown>>(result: T): T {\n const diagnostics = result?.diagnostics as Record<string, unknown> | undefined\n if (!diagnostics?.debug) return result\n return {\n ...result,\n diagnostics: {\n ...diagnostics,\n debug: sanitizeOutboundDiagnostics(diagnostics.debug),\n },\n }\n}\n"],"mappings":";;;;;AAEA,IAAM,cAAsC;AAAA,EAC1C,QAAQ;AACV;AAEA,IAAM,uBAAuB;AAC7B,IAAM,2BAA2B,oBAAI,IAAI;AAAA,EACvC;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF,CAAC;AAED,IAAM,kBAAkB,oBAAI,IAAI;AAAA,EAC9B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF,CAAC;AACD,IAAM,gBAAgB;AAEtB,SAAS,iBAAiB,OAAuB;AAC/C,SAAO,MAAM,QAAQ,eAAe,mBAAmB;AACzD;AAEO,SAAS,4BAA+B,OAAU,YAAY,IAAO;AAC1E,MAAI,OAAO,UAAU,UAAU;AAC7B,QAAI,MAAM,iBAAiB,KAAK;AAChC,QAAI,qBAAqB,KAAK,SAAS,GAAG;AACxC,YAAM,mBAAmB,qBAAqB;AAAA,IAChD;AACA,WAAO;AAAA,EACT;AACA,MAAI,MAAM,QAAQ,KAAK,EAAG,QAAO,MAAM,IAAI,OAAK,4BAA4B,GAAG,SAAS,CAAC;AACzF,MAAI,UAAU,QAAQ,OAAO,UAAU,UAAU;AAC/C,UAAM,MAA+B,CAAC;AACtC,eAAW,CAAC,KAAK,GAAG,KAAK,OAAO,QAAQ,KAAgC,GAAG;AACzE,UAAI,yBAAyB,IAAI,GAAG,EAAG;AACvC,YAAM,UAAU,YAAY,GAAG,KAAK;AACpC,UAAI,gBAAgB,IAAI,GAAG,GAAG;AAC5B,YAAI,OAAO,IAAI;AACf;AAAA,MACF;AACA,UAAI,OAAO,IAAI,4BAA4B,KAAK,GAAG;AAAA,IACrD;AACA,WAAO;AAAA,EACT;AACA,SAAO;AACT;AAEO,SAAS,iBAAoB,UAAoB;AACtD,SAAO,SAAS,IAAI,OAAK,4BAA4B,CAAC,CAAC;AACzD;AAEO,SAAS,sBAAyD,QAAc;AACrF,QAAM,cAAc,QAAQ;AAC5B,MAAI,CAAC,aAAa,MAAO,QAAO;AAChC,SAAO;AAAA,IACL,GAAG;AAAA,IACH,aAAa;AAAA,MACX,GAAG;AAAA,MACH,OAAO,4BAA4B,YAAY,KAAK;AAAA,IACtD;AAAA,EACF;AACF;","names":[]}
|
|
@@ -0,0 +1,135 @@
|
|
|
1
|
+
// src/schemas.ts
|
|
2
|
+
import { z } from "zod";
|
|
3
|
+
var DEFAULT_PROXY_MODE = "none";
|
|
4
|
+
var DEFAULT_MAPS_PROXY_MODE = "none";
|
|
5
|
+
var HarvestOptionsSchema = z.object({
|
|
6
|
+
query: z.string().min(1),
|
|
7
|
+
location: z.string().optional(),
|
|
8
|
+
gl: z.string().length(2).default("us"),
|
|
9
|
+
hl: z.string().length(2).default("en"),
|
|
10
|
+
device: z.enum(["desktop", "mobile"]).default("desktop"),
|
|
11
|
+
proxyMode: z.enum(["location", "configured", "none"]).default(DEFAULT_PROXY_MODE),
|
|
12
|
+
proxyZip: z.string().regex(/^\d{5}$/).optional(),
|
|
13
|
+
keepDefaultProxy: z.boolean().optional(),
|
|
14
|
+
requireUsEgress: z.boolean().optional(),
|
|
15
|
+
maxAttempts: z.number().int().min(1).max(12).optional(),
|
|
16
|
+
debug: z.boolean().default(false),
|
|
17
|
+
depth: z.number().int().min(1).max(30).default(3),
|
|
18
|
+
maxQuestions: z.number().int().min(1).max(1e3).default(100),
|
|
19
|
+
headless: z.boolean().default(false),
|
|
20
|
+
profileDir: z.string().optional(),
|
|
21
|
+
proxy: z.string().url().optional(),
|
|
22
|
+
kernelApiKey: z.string().optional(),
|
|
23
|
+
kernelProxyId: z.string().optional(),
|
|
24
|
+
kernelProxyResolution: z.unknown().optional(),
|
|
25
|
+
outputDir: z.string().default("./paa-output"),
|
|
26
|
+
format: z.enum(["json", "csv", "both"]).default("both"),
|
|
27
|
+
serpOnly: z.boolean().default(false),
|
|
28
|
+
pages: z.number().int().min(1).max(2).default(1),
|
|
29
|
+
recency: z.enum(["day", "week", "month", "year"]).optional(),
|
|
30
|
+
softDeadlineMs: z.number().optional()
|
|
31
|
+
});
|
|
32
|
+
function recencyToTbs(recency) {
|
|
33
|
+
switch (recency) {
|
|
34
|
+
case "day":
|
|
35
|
+
return "qdr:d";
|
|
36
|
+
case "week":
|
|
37
|
+
return "qdr:w";
|
|
38
|
+
case "month":
|
|
39
|
+
return "qdr:m";
|
|
40
|
+
case "year":
|
|
41
|
+
return "qdr:y";
|
|
42
|
+
default:
|
|
43
|
+
return void 0;
|
|
44
|
+
}
|
|
45
|
+
}
|
|
46
|
+
var MapsPlaceOptionsSchema = z.object({
|
|
47
|
+
businessName: z.string().min(1),
|
|
48
|
+
location: z.string().min(1),
|
|
49
|
+
gl: z.string().length(2).default("us"),
|
|
50
|
+
hl: z.string().length(2).default("en"),
|
|
51
|
+
includeReviews: z.boolean().default(false),
|
|
52
|
+
maxReviews: z.number().int().min(1).max(500).default(50),
|
|
53
|
+
includeServices: z.boolean().default(false),
|
|
54
|
+
kernelApiKey: z.string().optional(),
|
|
55
|
+
kernelProxyId: z.string().optional(),
|
|
56
|
+
headless: z.boolean().default(true)
|
|
57
|
+
});
|
|
58
|
+
var MapsSearchOptionsSchema = z.object({
|
|
59
|
+
query: z.string().min(1),
|
|
60
|
+
location: z.string().optional(),
|
|
61
|
+
gl: z.string().length(2).default("us"),
|
|
62
|
+
hl: z.string().length(2).default("en"),
|
|
63
|
+
maxResults: z.number().int().min(1).max(50).default(10),
|
|
64
|
+
includeServices: z.boolean().default(false),
|
|
65
|
+
proxyMode: z.enum(["location", "configured", "none"]).default(DEFAULT_MAPS_PROXY_MODE),
|
|
66
|
+
proxyZip: z.string().regex(/^\d{5}$/).optional(),
|
|
67
|
+
serpRedirect: z.boolean().optional(),
|
|
68
|
+
forceDirectEgress: z.boolean().optional(),
|
|
69
|
+
debug: z.boolean().default(false),
|
|
70
|
+
kernelApiKey: z.string().optional(),
|
|
71
|
+
kernelProxyId: z.string().optional(),
|
|
72
|
+
kernelProxyResolution: z.unknown().optional(),
|
|
73
|
+
headless: z.boolean().default(true)
|
|
74
|
+
});
|
|
75
|
+
var RawPAAItemSchema = z.object({
|
|
76
|
+
question: z.string().min(1),
|
|
77
|
+
answer: z.string().optional(),
|
|
78
|
+
sourceTitle: z.string().optional(),
|
|
79
|
+
sourceSite: z.string().optional(),
|
|
80
|
+
sourceCite: z.string().optional()
|
|
81
|
+
});
|
|
82
|
+
var RawMapsOverviewSchema = z.object({
|
|
83
|
+
name: z.string().nullable(),
|
|
84
|
+
rating: z.string().nullable(),
|
|
85
|
+
reviewCount: z.string().nullable(),
|
|
86
|
+
category: z.string().nullable(),
|
|
87
|
+
address: z.string().nullable(),
|
|
88
|
+
hoursSummary: z.string().nullable(),
|
|
89
|
+
phone: z.string().nullable(),
|
|
90
|
+
phoneDisplay: z.string().nullable(),
|
|
91
|
+
website: z.string().nullable(),
|
|
92
|
+
plusCode: z.string().nullable(),
|
|
93
|
+
bookingUrl: z.string().nullable()
|
|
94
|
+
});
|
|
95
|
+
var RawMapsHoursRowSchema = z.object({
|
|
96
|
+
day: z.string(),
|
|
97
|
+
hours: z.string()
|
|
98
|
+
});
|
|
99
|
+
var RawMapsReviewStatsSchema = z.object({
|
|
100
|
+
reviewHistogram: z.array(z.object({
|
|
101
|
+
stars: z.number(),
|
|
102
|
+
count: z.string()
|
|
103
|
+
})),
|
|
104
|
+
reviewTopics: z.array(z.object({
|
|
105
|
+
label: z.string(),
|
|
106
|
+
count: z.string()
|
|
107
|
+
}))
|
|
108
|
+
});
|
|
109
|
+
var RawMapsReviewCardSchema = z.object({
|
|
110
|
+
reviewId: z.string(),
|
|
111
|
+
author: z.string().nullable(),
|
|
112
|
+
stars: z.string().nullable(),
|
|
113
|
+
date: z.string().nullable(),
|
|
114
|
+
text: z.string().nullable(),
|
|
115
|
+
ownerResponse: z.string().nullable()
|
|
116
|
+
});
|
|
117
|
+
var RawMapsAboutAttributeSchema = z.object({
|
|
118
|
+
section: z.string(),
|
|
119
|
+
attribute: z.string()
|
|
120
|
+
});
|
|
121
|
+
|
|
122
|
+
export {
|
|
123
|
+
DEFAULT_PROXY_MODE,
|
|
124
|
+
DEFAULT_MAPS_PROXY_MODE,
|
|
125
|
+
HarvestOptionsSchema,
|
|
126
|
+
recencyToTbs,
|
|
127
|
+
MapsPlaceOptionsSchema,
|
|
128
|
+
MapsSearchOptionsSchema,
|
|
129
|
+
RawPAAItemSchema,
|
|
130
|
+
RawMapsOverviewSchema,
|
|
131
|
+
RawMapsHoursRowSchema,
|
|
132
|
+
RawMapsReviewStatsSchema,
|
|
133
|
+
RawMapsAboutAttributeSchema
|
|
134
|
+
};
|
|
135
|
+
//# sourceMappingURL=chunk-CB5C3BPB.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"sources":["../src/schemas.ts"],"sourcesContent":["import { z } from 'zod'\n\nexport const DEFAULT_PROXY_MODE = 'none' as const\nexport const DEFAULT_MAPS_PROXY_MODE = 'none' as const\n\nexport const HarvestOptionsSchema = z.object({\n query: z.string().min(1),\n location: z.string().optional(),\n gl: z.string().length(2).default('us'),\n hl: z.string().length(2).default('en'),\n device: z.enum(['desktop', 'mobile']).default('desktop'),\n proxyMode: z.enum(['location', 'configured', 'none']).default(DEFAULT_PROXY_MODE),\n proxyZip: z.string().regex(/^\\d{5}$/).optional(),\n keepDefaultProxy: z.boolean().optional(),\n requireUsEgress: z.boolean().optional(),\n maxAttempts: z.number().int().min(1).max(12).optional(),\n debug: z.boolean().default(false),\n depth: z.number().int().min(1).max(30).default(3),\n maxQuestions: z.number().int().min(1).max(1000).default(100),\n headless: z.boolean().default(false),\n profileDir: z.string().optional(),\n proxy: z.string().url().optional(),\n kernelApiKey: z.string().optional(),\n kernelProxyId: z.string().optional(),\n kernelProxyResolution: z.unknown().optional(),\n outputDir: z.string().default('./paa-output'),\n format: z.enum(['json', 'csv', 'both']).default('both'),\n serpOnly: z.boolean().default(false),\n pages: z.number().int().min(1).max(2).default(1),\n recency: z.enum(['day', 'week', 'month', 'year']).optional(),\n softDeadlineMs: z.number().optional(),\n})\n\nexport function recencyToTbs(recency: 'day' | 'week' | 'month' | 'year' | undefined): string | undefined {\n switch (recency) {\n case 'day': return 'qdr:d'\n case 'week': return 'qdr:w'\n case 'month': return 'qdr:m'\n case 'year': return 'qdr:y'\n default: return undefined\n }\n}\n\nexport const MapsPlaceOptionsSchema = z.object({\n businessName: z.string().min(1),\n location: z.string().min(1),\n gl: z.string().length(2).default('us'),\n hl: z.string().length(2).default('en'),\n includeReviews: z.boolean().default(false),\n maxReviews: z.number().int().min(1).max(500).default(50),\n includeServices: z.boolean().default(false),\n kernelApiKey: z.string().optional(),\n kernelProxyId: z.string().optional(),\n headless: z.boolean().default(true),\n})\n\nexport const MapsSearchOptionsSchema = z.object({\n query: z.string().min(1),\n location: z.string().optional(),\n gl: z.string().length(2).default('us'),\n hl: z.string().length(2).default('en'),\n maxResults: z.number().int().min(1).max(50).default(10),\n includeServices: z.boolean().default(false),\n proxyMode: z.enum(['location', 'configured', 'none']).default(DEFAULT_MAPS_PROXY_MODE),\n proxyZip: z.string().regex(/^\\d{5}$/).optional(),\n serpRedirect: z.boolean().optional(),\n forceDirectEgress: z.boolean().optional(),\n debug: z.boolean().default(false),\n kernelApiKey: z.string().optional(),\n kernelProxyId: z.string().optional(),\n kernelProxyResolution: z.unknown().optional(),\n headless: z.boolean().default(true),\n})\n\nexport const RawPAAItemSchema = z.object({\n question: z.string().min(1),\n answer: z.string().optional(),\n sourceTitle: z.string().optional(),\n sourceSite: z.string().optional(),\n sourceCite: z.string().optional(),\n})\n\nexport const RawMapsOverviewSchema = z.object({\n name: z.string().nullable(),\n rating: z.string().nullable(),\n reviewCount: z.string().nullable(),\n category: z.string().nullable(),\n address: z.string().nullable(),\n hoursSummary: z.string().nullable(),\n phone: z.string().nullable(),\n phoneDisplay: z.string().nullable(),\n website: z.string().nullable(),\n plusCode: z.string().nullable(),\n bookingUrl: z.string().nullable(),\n})\n\nexport const RawMapsHoursRowSchema = z.object({\n day: z.string(),\n hours: z.string(),\n})\n\nexport const RawMapsReviewStatsSchema = z.object({\n reviewHistogram: z.array(z.object({\n stars: z.number(),\n count: z.string(),\n })),\n reviewTopics: z.array(z.object({\n label: z.string(),\n count: z.string(),\n })),\n})\n\nexport const RawMapsReviewCardSchema = z.object({\n reviewId: z.string(),\n author: z.string().nullable(),\n stars: z.string().nullable(),\n date: z.string().nullable(),\n text: z.string().nullable(),\n ownerResponse: z.string().nullable(),\n})\n\nexport const RawMapsAboutAttributeSchema = z.object({\n section: z.string(),\n attribute: z.string(),\n})\n"],"mappings":";AAAA,SAAS,SAAS;AAEX,IAAM,qBAAqB;AAC3B,IAAM,0BAA0B;AAEhC,IAAM,uBAAuB,EAAE,OAAO;AAAA,EAC3C,OAAc,EAAE,OAAO,EAAE,IAAI,CAAC;AAAA,EAC9B,UAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,IAAc,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,QAAQ,IAAI;AAAA,EAC/C,IAAc,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,QAAQ,IAAI;AAAA,EAC/C,QAAc,EAAE,KAAK,CAAC,WAAW,QAAQ,CAAC,EAAE,QAAQ,SAAS;AAAA,EAC7D,WAAc,EAAE,KAAK,CAAC,YAAY,cAAc,MAAM,CAAC,EAAE,QAAQ,kBAAkB;AAAA,EACnF,UAAc,EAAE,OAAO,EAAE,MAAM,SAAS,EAAE,SAAS;AAAA,EACnD,kBAAkB,EAAE,QAAQ,EAAE,SAAS;AAAA,EACvC,iBAAiB,EAAE,QAAQ,EAAE,SAAS;AAAA,EACtC,aAAc,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,EAAE,EAAE,SAAS;AAAA,EACvD,OAAc,EAAE,QAAQ,EAAE,QAAQ,KAAK;AAAA,EACvC,OAAc,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,EAAE,EAAE,QAAQ,CAAC;AAAA,EACvD,cAAc,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,GAAI,EAAE,QAAQ,GAAG;AAAA,EAC3D,UAAc,EAAE,QAAQ,EAAE,QAAQ,KAAK;AAAA,EACvC,YAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,OAAc,EAAE,OAAO,EAAE,IAAI,EAAE,SAAS;AAAA,EACxC,cAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,eAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,uBAAuB,EAAE,QAAQ,EAAE,SAAS;AAAA,EAC5C,WAAc,EAAE,OAAO,EAAE,QAAQ,cAAc;AAAA,EAC/C,QAAc,EAAE,KAAK,CAAC,QAAQ,OAAO,MAAM,CAAC,EAAE,QAAQ,MAAM;AAAA,EAC5D,UAAc,EAAE,QAAQ,EAAE,QAAQ,KAAK;AAAA,EACvC,OAAc,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,CAAC,EAAE,QAAQ,CAAC;AAAA,EACtD,SAAc,EAAE,KAAK,CAAC,OAAO,QAAQ,SAAS,MAAM,CAAC,EAAE,SAAS;AAAA,EAChE,gBAAgB,EAAE,OAAO,EAAE,SAAS;AACtC,CAAC;AAEM,SAAS,aAAa,SAA4E;AACvG,UAAQ,SAAS;AAAA,IACf,KAAK;AAAO,aAAO;AAAA,IACnB,KAAK;AAAQ,aAAO;AAAA,IACpB,KAAK;AAAS,aAAO;AAAA,IACrB,KAAK;AAAQ,aAAO;AAAA,IACpB;AAAS,aAAO;AAAA,EAClB;AACF;AAEO,IAAM,yBAAyB,EAAE,OAAO;AAAA,EAC7C,cAAe,EAAE,OAAO,EAAE,IAAI,CAAC;AAAA,EAC/B,UAAe,EAAE,OAAO,EAAE,IAAI,CAAC;AAAA,EAC/B,IAAe,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,QAAQ,IAAI;AAAA,EAChD,IAAe,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,QAAQ,IAAI;AAAA,EAChD,gBAAgB,EAAE,QAAQ,EAAE,QAAQ,KAAK;AAAA,EACzC,YAAe,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,GAAG,EAAE,QAAQ,EAAE;AAAA,EAC1D,iBAAiB,EAAE,QAAQ,EAAE,QAAQ,KAAK;AAAA,EAC1C,cAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,eAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,UAAe,EAAE,QAAQ,EAAE,QAAQ,IAAI;AACzC,CAAC;AAEM,IAAM,0BAA0B,EAAE,OAAO;AAAA,EAC9C,OAAe,EAAE,OAAO,EAAE,IAAI,CAAC;AAAA,EAC/B,UAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,IAAe,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,QAAQ,IAAI;AAAA,EAChD,IAAe,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,QAAQ,IAAI;AAAA,EAChD,YAAe,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE;AAAA,EACzD,iBAAiB,EAAE,QAAQ,EAAE,QAAQ,KAAK;AAAA,EAC1C,WAAe,EAAE,KAAK,CAAC,YAAY,cAAc,MAAM,CAAC,EAAE,QAAQ,uBAAuB;AAAA,EACzF,UAAe,EAAE,OAAO,EAAE,MAAM,SAAS,EAAE,SAAS;AAAA,EACpD,cAAe,EAAE,QAAQ,EAAE,SAAS;AAAA,EACpC,mBAAmB,EAAE,QAAQ,EAAE,SAAS;AAAA,EACxC,OAAe,EAAE,QAAQ,EAAE,QAAQ,KAAK;AAAA,EACxC,cAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,eAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,uBAAuB,EAAE,QAAQ,EAAE,SAAS;AAAA,EAC5C,UAAe,EAAE,QAAQ,EAAE,QAAQ,IAAI;AACzC,CAAC;AAEM,IAAM,mBAAmB,EAAE,OAAO;AAAA,EACvC,UAAa,EAAE,OAAO,EAAE,IAAI,CAAC;AAAA,EAC7B,QAAa,EAAE,OAAO,EAAE,SAAS;AAAA,EACjC,aAAa,EAAE,OAAO,EAAE,SAAS;AAAA,EACjC,YAAa,EAAE,OAAO,EAAE,SAAS;AAAA,EACjC,YAAa,EAAE,OAAO,EAAE,SAAS;AACnC,CAAC;AAEM,IAAM,wBAAwB,EAAE,OAAO;AAAA,EAC5C,MAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,QAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,aAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,UAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,SAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,cAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,OAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,cAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,SAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,UAAc,EAAE,OAAO,EAAE,SAAS;AAAA,EAClC,YAAc,EAAE,OAAO,EAAE,SAAS;AACpC,CAAC;AAEM,IAAM,wBAAwB,EAAE,OAAO;AAAA,EAC5C,KAAO,EAAE,OAAO;AAAA,EAChB,OAAO,EAAE,OAAO;AAClB,CAAC;AAEM,IAAM,2BAA2B,EAAE,OAAO;AAAA,EAC/C,iBAAiB,EAAE,MAAM,EAAE,OAAO;AAAA,IAChC,OAAO,EAAE,OAAO;AAAA,IAChB,OAAO,EAAE,OAAO;AAAA,EAClB,CAAC,CAAC;AAAA,EACF,cAAc,EAAE,MAAM,EAAE,OAAO;AAAA,IAC7B,OAAO,EAAE,OAAO;AAAA,IAChB,OAAO,EAAE,OAAO;AAAA,EAClB,CAAC,CAAC;AACJ,CAAC;AAEM,IAAM,0BAA0B,EAAE,OAAO;AAAA,EAC9C,UAAe,EAAE,OAAO;AAAA,EACxB,QAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,OAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,MAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,MAAe,EAAE,OAAO,EAAE,SAAS;AAAA,EACnC,eAAe,EAAE,OAAO,EAAE,SAAS;AACrC,CAAC;AAEM,IAAM,8BAA8B,EAAE,OAAO;AAAA,EAClD,SAAW,EAAE,OAAO;AAAA,EACpB,WAAW,EAAE,OAAO;AACtB,CAAC;","names":[]}
|
|
@@ -0,0 +1,194 @@
|
|
|
1
|
+
import {
|
|
2
|
+
CaptchaError,
|
|
3
|
+
LocationMismatchError,
|
|
4
|
+
PUBLIC_SERVICE_UNAVAILABLE_MESSAGE,
|
|
5
|
+
RequestAbortedError,
|
|
6
|
+
publicErrorMessage
|
|
7
|
+
} from "./chunk-5UN33CGU.js";
|
|
8
|
+
import {
|
|
9
|
+
finishHarvestAttempt,
|
|
10
|
+
startHarvestAttempt
|
|
11
|
+
} from "./chunk-IQTT7CAB.js";
|
|
12
|
+
|
|
13
|
+
// src/api/webhook.ts
|
|
14
|
+
async function deliverWebhook(url, payload, retries = 3) {
|
|
15
|
+
for (let attempt = 1; attempt <= retries; attempt++) {
|
|
16
|
+
try {
|
|
17
|
+
const res = await fetch(url, {
|
|
18
|
+
method: "POST",
|
|
19
|
+
headers: { "content-type": "application/json" },
|
|
20
|
+
body: JSON.stringify(payload),
|
|
21
|
+
signal: AbortSignal.timeout(1e4)
|
|
22
|
+
});
|
|
23
|
+
if (res.ok) return;
|
|
24
|
+
console.warn(`[webhook] attempt ${attempt} \u2192 ${res.status} from ${url}`);
|
|
25
|
+
} catch (err) {
|
|
26
|
+
console.warn(`[webhook] attempt ${attempt} failed:`, err instanceof Error ? err.message : err);
|
|
27
|
+
}
|
|
28
|
+
if (attempt < retries) await new Promise((r) => setTimeout(r, 1e3 * attempt * 2));
|
|
29
|
+
}
|
|
30
|
+
console.error(`[webhook] gave up after ${retries} attempts for ${url}`);
|
|
31
|
+
}
|
|
32
|
+
|
|
33
|
+
// src/api/harvest-problems.ts
|
|
34
|
+
function errorMessage(err) {
|
|
35
|
+
return err instanceof Error ? err.message : String(err);
|
|
36
|
+
}
|
|
37
|
+
function looksLikeTimeout(err, message) {
|
|
38
|
+
if (err instanceof DOMException && (err.name === "TimeoutError" || err.name === "AbortError")) return true;
|
|
39
|
+
return /timeout|timed out|Timeout \d+ms exceeded|deadline/i.test(message);
|
|
40
|
+
}
|
|
41
|
+
function looksLikeCaptcha(message) {
|
|
42
|
+
return /captcha|recaptcha|unusual traffic|google\.com\/sorry|blocked/i.test(message);
|
|
43
|
+
}
|
|
44
|
+
function looksLikeProxyTunnelFailure(message) {
|
|
45
|
+
return /ERR_TUNNEL_CONNECTION_FAILED|ERR_PROXY_CONNECTION_FAILED|ERR_SOCKS_CONNECTION_FAILED|tunnel connection failed|proxy connection failed|transport error: proxy/i.test(message);
|
|
46
|
+
}
|
|
47
|
+
function looksLikeProxyUnavailable(message) {
|
|
48
|
+
return /proxy unavailable|proxy_unavailable|connection_test_failed|did not return a proxy id|configured fallback/i.test(message);
|
|
49
|
+
}
|
|
50
|
+
function looksLikeInternalServiceUnavailable(message) {
|
|
51
|
+
return /(?:spending|billing|organization|account).{0,80}(?:cap|limit|blocked|disabled)|(?:quota|capacity).{0,40}(?:reached|exceeded)|https?:\/\/\S*dashboard/i.test(message);
|
|
52
|
+
}
|
|
53
|
+
function classifyHarvestProblem(err) {
|
|
54
|
+
const message = errorMessage(err);
|
|
55
|
+
if (err instanceof RequestAbortedError) {
|
|
56
|
+
return {
|
|
57
|
+
error_code: "request_aborted",
|
|
58
|
+
error_type: "request_aborted",
|
|
59
|
+
message: publicErrorMessage("request_aborted"),
|
|
60
|
+
retryable: true,
|
|
61
|
+
httpStatus: 408,
|
|
62
|
+
terminalStatus: "cancelled"
|
|
63
|
+
};
|
|
64
|
+
}
|
|
65
|
+
if (err instanceof CaptchaError || looksLikeCaptcha(message)) {
|
|
66
|
+
return {
|
|
67
|
+
error_code: "captcha_exhausted",
|
|
68
|
+
error_type: "captcha",
|
|
69
|
+
message: publicErrorMessage("captcha_exhausted"),
|
|
70
|
+
retryable: true,
|
|
71
|
+
httpStatus: 503,
|
|
72
|
+
terminalStatus: "failed"
|
|
73
|
+
};
|
|
74
|
+
}
|
|
75
|
+
if (err instanceof LocationMismatchError) {
|
|
76
|
+
return {
|
|
77
|
+
error_code: "location_mismatch",
|
|
78
|
+
error_type: "location_mismatch",
|
|
79
|
+
message: publicErrorMessage("location_mismatch"),
|
|
80
|
+
retryable: true,
|
|
81
|
+
httpStatus: 503,
|
|
82
|
+
terminalStatus: "failed"
|
|
83
|
+
};
|
|
84
|
+
}
|
|
85
|
+
if (looksLikeProxyTunnelFailure(message)) {
|
|
86
|
+
return {
|
|
87
|
+
error_code: "proxy_tunnel_failed",
|
|
88
|
+
error_type: "proxy_tunnel_failed",
|
|
89
|
+
message: publicErrorMessage("proxy_tunnel_failed"),
|
|
90
|
+
retryable: true,
|
|
91
|
+
httpStatus: 503,
|
|
92
|
+
terminalStatus: "failed"
|
|
93
|
+
};
|
|
94
|
+
}
|
|
95
|
+
if (looksLikeProxyUnavailable(message)) {
|
|
96
|
+
return {
|
|
97
|
+
error_code: "proxy_unavailable",
|
|
98
|
+
error_type: "proxy_unavailable",
|
|
99
|
+
message: publicErrorMessage("proxy_unavailable"),
|
|
100
|
+
retryable: true,
|
|
101
|
+
httpStatus: 503,
|
|
102
|
+
terminalStatus: "failed"
|
|
103
|
+
};
|
|
104
|
+
}
|
|
105
|
+
if (looksLikeTimeout(err, message)) {
|
|
106
|
+
return {
|
|
107
|
+
error_code: "harvest_timeout",
|
|
108
|
+
error_type: "timeout",
|
|
109
|
+
message: publicErrorMessage("harvest_timeout"),
|
|
110
|
+
retryable: true,
|
|
111
|
+
httpStatus: 504,
|
|
112
|
+
terminalStatus: "failed"
|
|
113
|
+
};
|
|
114
|
+
}
|
|
115
|
+
if (looksLikeInternalServiceUnavailable(message)) {
|
|
116
|
+
return {
|
|
117
|
+
error_code: "service_unavailable",
|
|
118
|
+
error_type: "service_unavailable",
|
|
119
|
+
message: PUBLIC_SERVICE_UNAVAILABLE_MESSAGE,
|
|
120
|
+
retryable: true,
|
|
121
|
+
httpStatus: 503,
|
|
122
|
+
terminalStatus: "failed"
|
|
123
|
+
};
|
|
124
|
+
}
|
|
125
|
+
return {
|
|
126
|
+
error_code: "extraction_failed",
|
|
127
|
+
error_type: "extraction",
|
|
128
|
+
message: PUBLIC_SERVICE_UNAVAILABLE_MESSAGE,
|
|
129
|
+
retryable: false,
|
|
130
|
+
httpStatus: 500,
|
|
131
|
+
terminalStatus: "failed"
|
|
132
|
+
};
|
|
133
|
+
}
|
|
134
|
+
function serializeHarvestProblem(problem) {
|
|
135
|
+
return JSON.stringify({
|
|
136
|
+
error_code: problem.error_code,
|
|
137
|
+
error_type: problem.error_type,
|
|
138
|
+
message: problem.message,
|
|
139
|
+
retryable: problem.retryable
|
|
140
|
+
});
|
|
141
|
+
}
|
|
142
|
+
function harvestProblemResponse(problem) {
|
|
143
|
+
return {
|
|
144
|
+
error: problem.message,
|
|
145
|
+
error_code: problem.error_code,
|
|
146
|
+
error_type: problem.error_type,
|
|
147
|
+
retryable: problem.retryable
|
|
148
|
+
};
|
|
149
|
+
}
|
|
150
|
+
|
|
151
|
+
// src/api/harvest-attempt-events.ts
|
|
152
|
+
function createHarvestAttemptRecorder(jobId, userId) {
|
|
153
|
+
return async (event) => {
|
|
154
|
+
if (event.type === "started") {
|
|
155
|
+
await startHarvestAttempt({
|
|
156
|
+
jobId,
|
|
157
|
+
userId,
|
|
158
|
+
attemptNumber: event.attemptNumber,
|
|
159
|
+
maxAttempts: event.maxAttempts,
|
|
160
|
+
query: event.query,
|
|
161
|
+
location: event.location,
|
|
162
|
+
maxQuestions: event.maxQuestions,
|
|
163
|
+
startedAt: event.startedAt
|
|
164
|
+
});
|
|
165
|
+
return;
|
|
166
|
+
}
|
|
167
|
+
await finishHarvestAttempt({
|
|
168
|
+
jobId,
|
|
169
|
+
attemptNumber: event.attemptNumber,
|
|
170
|
+
outcome: event.outcome,
|
|
171
|
+
kernelSessionId: event.kernelSessionId,
|
|
172
|
+
questionCount: event.questionCount,
|
|
173
|
+
durationMs: event.durationMs,
|
|
174
|
+
error: event.error,
|
|
175
|
+
willRetry: event.willRetry,
|
|
176
|
+
kernelDeleteStarted: event.cleanup.kernelDeleteStarted,
|
|
177
|
+
kernelDeleteSucceeded: event.cleanup.kernelDeleteSucceeded,
|
|
178
|
+
kernelDeleteError: event.cleanup.kernelDeleteError,
|
|
179
|
+
browserCloseSucceeded: event.cleanup.browserCloseSucceeded,
|
|
180
|
+
browserCloseError: event.cleanup.browserCloseError,
|
|
181
|
+
debug: event.debug,
|
|
182
|
+
completedAt: event.completedAt
|
|
183
|
+
});
|
|
184
|
+
};
|
|
185
|
+
}
|
|
186
|
+
|
|
187
|
+
export {
|
|
188
|
+
deliverWebhook,
|
|
189
|
+
classifyHarvestProblem,
|
|
190
|
+
serializeHarvestProblem,
|
|
191
|
+
harvestProblemResponse,
|
|
192
|
+
createHarvestAttemptRecorder
|
|
193
|
+
};
|
|
194
|
+
//# sourceMappingURL=chunk-CS4HE6QY.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"sources":["../src/api/webhook.ts","../src/api/harvest-problems.ts","../src/api/harvest-attempt-events.ts"],"sourcesContent":["export async function deliverWebhook(url: string, payload: object, retries = 3): Promise<void> {\n for (let attempt = 1; attempt <= retries; attempt++) {\n try {\n const res = await fetch(url, {\n method: 'POST',\n headers: { 'content-type': 'application/json' },\n body: JSON.stringify(payload),\n signal: AbortSignal.timeout(10_000),\n })\n if (res.ok) return\n console.warn(`[webhook] attempt ${attempt} → ${res.status} from ${url}`)\n } catch (err) {\n console.warn(`[webhook] attempt ${attempt} failed:`, err instanceof Error ? err.message : err)\n }\n if (attempt < retries) await new Promise((r) => setTimeout(r, 1000 * attempt * 2))\n }\n console.error(`[webhook] gave up after ${retries} attempts for ${url}`)\n}\n","import {\n CaptchaError,\n LocationMismatchError,\n PUBLIC_SERVICE_UNAVAILABLE_MESSAGE,\n RequestAbortedError,\n publicErrorMessage,\n} from '../errors.js'\n\nexport type HarvestProblemCode =\n | 'request_aborted'\n | 'captcha_exhausted'\n | 'location_mismatch'\n | 'proxy_tunnel_failed'\n | 'proxy_unavailable'\n | 'harvest_timeout'\n | 'service_unavailable'\n | 'extraction_failed'\n\nexport interface HarvestProblem {\n error_code: HarvestProblemCode\n error_type: string\n message: string\n retryable: boolean\n httpStatus: number\n terminalStatus: 'cancelled' | 'failed'\n}\n\nfunction errorMessage(err: unknown): string {\n return err instanceof Error ? err.message : String(err)\n}\n\nfunction looksLikeTimeout(err: unknown, message: string): boolean {\n if (err instanceof DOMException && (err.name === 'TimeoutError' || err.name === 'AbortError')) return true\n return /timeout|timed out|Timeout \\d+ms exceeded|deadline/i.test(message)\n}\n\nfunction looksLikeCaptcha(message: string): boolean {\n return /captcha|recaptcha|unusual traffic|google\\.com\\/sorry|blocked/i.test(message)\n}\n\nfunction looksLikeProxyTunnelFailure(message: string): boolean {\n return /ERR_TUNNEL_CONNECTION_FAILED|ERR_PROXY_CONNECTION_FAILED|ERR_SOCKS_CONNECTION_FAILED|tunnel connection failed|proxy connection failed|transport error: proxy/i.test(message)\n}\n\nfunction looksLikeProxyUnavailable(message: string): boolean {\n return /proxy unavailable|proxy_unavailable|connection_test_failed|did not return a proxy id|configured fallback/i.test(message)\n}\n\nfunction looksLikeInternalServiceUnavailable(message: string): boolean {\n return /(?:spending|billing|organization|account).{0,80}(?:cap|limit|blocked|disabled)|(?:quota|capacity).{0,40}(?:reached|exceeded)|https?:\\/\\/\\S*dashboard/i.test(message)\n}\n\nexport function classifyHarvestProblem(err: unknown): HarvestProblem {\n const message = errorMessage(err)\n\n if (err instanceof RequestAbortedError) {\n return {\n error_code: 'request_aborted',\n error_type: 'request_aborted',\n message: publicErrorMessage('request_aborted'),\n retryable: true,\n httpStatus: 408,\n terminalStatus: 'cancelled',\n }\n }\n\n if (err instanceof CaptchaError || looksLikeCaptcha(message)) {\n return {\n error_code: 'captcha_exhausted',\n error_type: 'captcha',\n message: publicErrorMessage('captcha_exhausted'),\n retryable: true,\n httpStatus: 503,\n terminalStatus: 'failed',\n }\n }\n\n if (err instanceof LocationMismatchError) {\n return {\n error_code: 'location_mismatch',\n error_type: 'location_mismatch',\n message: publicErrorMessage('location_mismatch'),\n retryable: true,\n httpStatus: 503,\n terminalStatus: 'failed',\n }\n }\n\n if (looksLikeProxyTunnelFailure(message)) {\n return {\n error_code: 'proxy_tunnel_failed',\n error_type: 'proxy_tunnel_failed',\n message: publicErrorMessage('proxy_tunnel_failed'),\n retryable: true,\n httpStatus: 503,\n terminalStatus: 'failed',\n }\n }\n\n if (looksLikeProxyUnavailable(message)) {\n return {\n error_code: 'proxy_unavailable',\n error_type: 'proxy_unavailable',\n message: publicErrorMessage('proxy_unavailable'),\n retryable: true,\n httpStatus: 503,\n terminalStatus: 'failed',\n }\n }\n\n if (looksLikeTimeout(err, message)) {\n return {\n error_code: 'harvest_timeout',\n error_type: 'timeout',\n message: publicErrorMessage('harvest_timeout'),\n retryable: true,\n httpStatus: 504,\n terminalStatus: 'failed',\n }\n }\n\n if (looksLikeInternalServiceUnavailable(message)) {\n return {\n error_code: 'service_unavailable',\n error_type: 'service_unavailable',\n message: PUBLIC_SERVICE_UNAVAILABLE_MESSAGE,\n retryable: true,\n httpStatus: 503,\n terminalStatus: 'failed',\n }\n }\n\n return {\n error_code: 'extraction_failed',\n error_type: 'extraction',\n message: PUBLIC_SERVICE_UNAVAILABLE_MESSAGE,\n retryable: false,\n httpStatus: 500,\n terminalStatus: 'failed',\n }\n}\n\nexport function serializeHarvestProblem(problem: HarvestProblem): string {\n return JSON.stringify({\n error_code: problem.error_code,\n error_type: problem.error_type,\n message: problem.message,\n retryable: problem.retryable,\n })\n}\n\nexport function harvestProblemResponse(problem: HarvestProblem): {\n error: string\n error_code: HarvestProblemCode\n error_type: string\n retryable: boolean\n} {\n return {\n error: problem.message,\n error_code: problem.error_code,\n error_type: problem.error_type,\n retryable: problem.retryable,\n }\n}\n","import type { HarvestAttemptLogEvent } from '../harvest.js'\nimport { finishHarvestAttempt, startHarvestAttempt } from './db.js'\n\nexport function createHarvestAttemptRecorder(jobId: string, userId: number | bigint) {\n return async (event: HarvestAttemptLogEvent): Promise<void> => {\n if (event.type === 'started') {\n await startHarvestAttempt({\n jobId,\n userId,\n attemptNumber: event.attemptNumber,\n maxAttempts: event.maxAttempts,\n query: event.query,\n location: event.location,\n maxQuestions: event.maxQuestions,\n startedAt: event.startedAt,\n })\n return\n }\n\n await finishHarvestAttempt({\n jobId,\n attemptNumber: event.attemptNumber,\n outcome: event.outcome,\n kernelSessionId: event.kernelSessionId,\n questionCount: event.questionCount,\n durationMs: event.durationMs,\n error: event.error,\n willRetry: event.willRetry,\n kernelDeleteStarted: event.cleanup.kernelDeleteStarted,\n kernelDeleteSucceeded: event.cleanup.kernelDeleteSucceeded,\n kernelDeleteError: event.cleanup.kernelDeleteError,\n browserCloseSucceeded: event.cleanup.browserCloseSucceeded,\n browserCloseError: event.cleanup.browserCloseError,\n debug: event.debug,\n completedAt: event.completedAt,\n })\n }\n}\n"],"mappings":";;;;;;;;;;;;;AAAA,eAAsB,eAAe,KAAa,SAAiB,UAAU,GAAkB;AAC7F,WAAS,UAAU,GAAG,WAAW,SAAS,WAAW;AACnD,QAAI;AACF,YAAM,MAAM,MAAM,MAAM,KAAK;AAAA,QAC3B,QAAQ;AAAA,QACR,SAAS,EAAE,gBAAgB,mBAAmB;AAAA,QAC9C,MAAM,KAAK,UAAU,OAAO;AAAA,QAC5B,QAAQ,YAAY,QAAQ,GAAM;AAAA,MACpC,CAAC;AACD,UAAI,IAAI,GAAI;AACZ,cAAQ,KAAK,qBAAqB,OAAO,WAAM,IAAI,MAAM,SAAS,GAAG,EAAE;AAAA,IACzE,SAAS,KAAK;AACZ,cAAQ,KAAK,qBAAqB,OAAO,YAAY,eAAe,QAAQ,IAAI,UAAU,GAAG;AAAA,IAC/F;AACA,QAAI,UAAU,QAAS,OAAM,IAAI,QAAQ,CAAC,MAAM,WAAW,GAAG,MAAO,UAAU,CAAC,CAAC;AAAA,EACnF;AACA,UAAQ,MAAM,2BAA2B,OAAO,iBAAiB,GAAG,EAAE;AACxE;;;ACUA,SAAS,aAAa,KAAsB;AAC1C,SAAO,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;AACxD;AAEA,SAAS,iBAAiB,KAAc,SAA0B;AAChE,MAAI,eAAe,iBAAiB,IAAI,SAAS,kBAAkB,IAAI,SAAS,cAAe,QAAO;AACtG,SAAO,qDAAqD,KAAK,OAAO;AAC1E;AAEA,SAAS,iBAAiB,SAA0B;AAClD,SAAO,gEAAgE,KAAK,OAAO;AACrF;AAEA,SAAS,4BAA4B,SAA0B;AAC7D,SAAO,gKAAgK,KAAK,OAAO;AACrL;AAEA,SAAS,0BAA0B,SAA0B;AAC3D,SAAO,4GAA4G,KAAK,OAAO;AACjI;AAEA,SAAS,oCAAoC,SAA0B;AACrE,SAAO,wJAAwJ,KAAK,OAAO;AAC7K;AAEO,SAAS,uBAAuB,KAA8B;AACnE,QAAM,UAAU,aAAa,GAAG;AAEhC,MAAI,eAAe,qBAAqB;AACtC,WAAO;AAAA,MACL,YAAY;AAAA,MACZ,YAAY;AAAA,MACZ,SAAS,mBAAmB,iBAAiB;AAAA,MAC7C,WAAW;AAAA,MACX,YAAY;AAAA,MACZ,gBAAgB;AAAA,IAClB;AAAA,EACF;AAEA,MAAI,eAAe,gBAAgB,iBAAiB,OAAO,GAAG;AAC5D,WAAO;AAAA,MACL,YAAY;AAAA,MACZ,YAAY;AAAA,MACZ,SAAS,mBAAmB,mBAAmB;AAAA,MAC/C,WAAW;AAAA,MACX,YAAY;AAAA,MACZ,gBAAgB;AAAA,IAClB;AAAA,EACF;AAEA,MAAI,eAAe,uBAAuB;AACxC,WAAO;AAAA,MACL,YAAY;AAAA,MACZ,YAAY;AAAA,MACZ,SAAS,mBAAmB,mBAAmB;AAAA,MAC/C,WAAW;AAAA,MACX,YAAY;AAAA,MACZ,gBAAgB;AAAA,IAClB;AAAA,EACF;AAEA,MAAI,4BAA4B,OAAO,GAAG;AACxC,WAAO;AAAA,MACL,YAAY;AAAA,MACZ,YAAY;AAAA,MACZ,SAAS,mBAAmB,qBAAqB;AAAA,MACjD,WAAW;AAAA,MACX,YAAY;AAAA,MACZ,gBAAgB;AAAA,IAClB;AAAA,EACF;AAEA,MAAI,0BAA0B,OAAO,GAAG;AACtC,WAAO;AAAA,MACL,YAAY;AAAA,MACZ,YAAY;AAAA,MACZ,SAAS,mBAAmB,mBAAmB;AAAA,MAC/C,WAAW;AAAA,MACX,YAAY;AAAA,MACZ,gBAAgB;AAAA,IAClB;AAAA,EACF;AAEA,MAAI,iBAAiB,KAAK,OAAO,GAAG;AAClC,WAAO;AAAA,MACL,YAAY;AAAA,MACZ,YAAY;AAAA,MACZ,SAAS,mBAAmB,iBAAiB;AAAA,MAC7C,WAAW;AAAA,MACX,YAAY;AAAA,MACZ,gBAAgB;AAAA,IAClB;AAAA,EACF;AAEA,MAAI,oCAAoC,OAAO,GAAG;AAChD,WAAO;AAAA,MACL,YAAY;AAAA,MACZ,YAAY;AAAA,MACZ,SAAS;AAAA,MACT,WAAW;AAAA,MACX,YAAY;AAAA,MACZ,gBAAgB;AAAA,IAClB;AAAA,EACF;AAEA,SAAO;AAAA,IACL,YAAY;AAAA,IACZ,YAAY;AAAA,IACZ,SAAS;AAAA,IACT,WAAW;AAAA,IACX,YAAY;AAAA,IACZ,gBAAgB;AAAA,EAClB;AACF;AAEO,SAAS,wBAAwB,SAAiC;AACvE,SAAO,KAAK,UAAU;AAAA,IACpB,YAAY,QAAQ;AAAA,IACpB,YAAY,QAAQ;AAAA,IACpB,SAAS,QAAQ;AAAA,IACjB,WAAW,QAAQ;AAAA,EACrB,CAAC;AACH;AAEO,SAAS,uBAAuB,SAKrC;AACA,SAAO;AAAA,IACL,OAAO,QAAQ;AAAA,IACf,YAAY,QAAQ;AAAA,IACpB,YAAY,QAAQ;AAAA,IACpB,WAAW,QAAQ;AAAA,EACrB;AACF;;;AChKO,SAAS,6BAA6B,OAAe,QAAyB;AACnF,SAAO,OAAO,UAAiD;AAC7D,QAAI,MAAM,SAAS,WAAW;AAC5B,YAAM,oBAAoB;AAAA,QACxB;AAAA,QACA;AAAA,QACA,eAAe,MAAM;AAAA,QACrB,aAAa,MAAM;AAAA,QACnB,OAAO,MAAM;AAAA,QACb,UAAU,MAAM;AAAA,QAChB,cAAc,MAAM;AAAA,QACpB,WAAW,MAAM;AAAA,MACnB,CAAC;AACD;AAAA,IACF;AAEA,UAAM,qBAAqB;AAAA,MACzB;AAAA,MACA,eAAe,MAAM;AAAA,MACrB,SAAS,MAAM;AAAA,MACf,iBAAiB,MAAM;AAAA,MACvB,eAAe,MAAM;AAAA,MACrB,YAAY,MAAM;AAAA,MAClB,OAAO,MAAM;AAAA,MACb,WAAW,MAAM;AAAA,MACjB,qBAAqB,MAAM,QAAQ;AAAA,MACnC,uBAAuB,MAAM,QAAQ;AAAA,MACrC,mBAAmB,MAAM,QAAQ;AAAA,MACjC,uBAAuB,MAAM,QAAQ;AAAA,MACrC,mBAAmB,MAAM,QAAQ;AAAA,MACjC,OAAO,MAAM;AAAA,MACb,aAAa,MAAM;AAAA,IACrB,CAAC;AAAA,EACH;AACF;","names":[]}
|