recker 1.0.26 → 1.0.27
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/dist/browser/browser/cache.d.ts +40 -0
- package/dist/browser/browser/cache.js +199 -0
- package/dist/browser/browser/crypto.d.ts +24 -0
- package/dist/browser/browser/crypto.js +80 -0
- package/dist/browser/browser/index.d.ts +31 -0
- package/dist/browser/browser/index.js +31 -0
- package/dist/browser/browser/recker.d.ts +26 -0
- package/dist/browser/browser/recker.js +61 -0
- package/dist/browser/cache/basic-file-storage.d.ts +12 -0
- package/dist/browser/cache/basic-file-storage.js +50 -0
- package/dist/browser/cache/memory-limits.d.ts +20 -0
- package/dist/browser/cache/memory-limits.js +96 -0
- package/dist/browser/cache/memory-storage.d.ts +132 -0
- package/dist/browser/cache/memory-storage.js +454 -0
- package/dist/browser/cache.d.ts +40 -0
- package/dist/browser/cache.js +199 -0
- package/dist/browser/constants/http-status.d.ts +73 -0
- package/dist/browser/constants/http-status.js +156 -0
- package/dist/browser/cookies/memory-cookie-jar.d.ts +30 -0
- package/dist/browser/cookies/memory-cookie-jar.js +210 -0
- package/dist/browser/core/client.d.ts +118 -0
- package/dist/browser/core/client.js +667 -0
- package/dist/browser/core/errors.d.ts +142 -0
- package/dist/browser/core/errors.js +308 -0
- package/dist/browser/core/index.d.ts +5 -0
- package/dist/browser/core/index.js +5 -0
- package/dist/browser/core/request-promise.d.ts +23 -0
- package/dist/browser/core/request-promise.js +82 -0
- package/dist/browser/core/request.d.ts +20 -0
- package/dist/browser/core/request.js +76 -0
- package/dist/browser/core/response.d.ts +34 -0
- package/dist/browser/core/response.js +178 -0
- package/dist/browser/crypto.d.ts +24 -0
- package/dist/browser/crypto.js +80 -0
- package/dist/browser/index.d.ts +31 -0
- package/dist/browser/index.js +31 -0
- package/dist/browser/plugins/auth/api-key.d.ts +8 -0
- package/dist/browser/plugins/auth/api-key.js +27 -0
- package/dist/browser/plugins/auth/auth0.d.ts +33 -0
- package/dist/browser/plugins/auth/auth0.js +94 -0
- package/dist/browser/plugins/auth/aws-sigv4.d.ts +10 -0
- package/dist/browser/plugins/auth/aws-sigv4.js +88 -0
- package/dist/browser/plugins/auth/azure-ad.d.ts +48 -0
- package/dist/browser/plugins/auth/azure-ad.js +152 -0
- package/dist/browser/plugins/auth/basic.d.ts +7 -0
- package/dist/browser/plugins/auth/basic.js +13 -0
- package/dist/browser/plugins/auth/bearer.d.ts +8 -0
- package/dist/browser/plugins/auth/bearer.js +17 -0
- package/dist/browser/plugins/auth/cognito.d.ts +45 -0
- package/dist/browser/plugins/auth/cognito.js +208 -0
- package/dist/browser/plugins/auth/digest.d.ts +8 -0
- package/dist/browser/plugins/auth/digest.js +100 -0
- package/dist/browser/plugins/auth/firebase.d.ts +32 -0
- package/dist/browser/plugins/auth/firebase.js +195 -0
- package/dist/browser/plugins/auth/github-app.d.ts +36 -0
- package/dist/browser/plugins/auth/github-app.js +170 -0
- package/dist/browser/plugins/auth/google-service-account.d.ts +49 -0
- package/dist/browser/plugins/auth/google-service-account.js +172 -0
- package/dist/browser/plugins/auth/index.d.ts +15 -0
- package/dist/browser/plugins/auth/index.js +15 -0
- package/dist/browser/plugins/auth/mtls.d.ts +37 -0
- package/dist/browser/plugins/auth/mtls.js +140 -0
- package/dist/browser/plugins/auth/oauth2.d.ts +8 -0
- package/dist/browser/plugins/auth/oauth2.js +26 -0
- package/dist/browser/plugins/auth/oidc.d.ts +55 -0
- package/dist/browser/plugins/auth/oidc.js +222 -0
- package/dist/browser/plugins/auth/okta.d.ts +47 -0
- package/dist/browser/plugins/auth/okta.js +157 -0
- package/dist/browser/plugins/auth.d.ts +1 -0
- package/dist/browser/plugins/auth.js +1 -0
- package/dist/browser/plugins/cache.d.ts +15 -0
- package/dist/browser/plugins/cache.js +486 -0
- package/dist/browser/plugins/circuit-breaker.d.ts +13 -0
- package/dist/browser/plugins/circuit-breaker.js +100 -0
- package/dist/browser/plugins/compression.d.ts +4 -0
- package/dist/browser/plugins/compression.js +130 -0
- package/dist/browser/plugins/cookie-jar.d.ts +5 -0
- package/dist/browser/plugins/cookie-jar.js +72 -0
- package/dist/browser/plugins/dedup.d.ts +5 -0
- package/dist/browser/plugins/dedup.js +35 -0
- package/dist/browser/plugins/graphql.d.ts +13 -0
- package/dist/browser/plugins/graphql.js +58 -0
- package/dist/browser/plugins/grpc-web.d.ts +79 -0
- package/dist/browser/plugins/grpc-web.js +261 -0
- package/dist/browser/plugins/hls.d.ts +105 -0
- package/dist/browser/plugins/hls.js +395 -0
- package/dist/browser/plugins/jsonrpc.d.ts +75 -0
- package/dist/browser/plugins/jsonrpc.js +143 -0
- package/dist/browser/plugins/logger.d.ts +13 -0
- package/dist/browser/plugins/logger.js +108 -0
- package/dist/browser/plugins/odata.d.ts +181 -0
- package/dist/browser/plugins/odata.js +564 -0
- package/dist/browser/plugins/pagination.d.ts +16 -0
- package/dist/browser/plugins/pagination.js +105 -0
- package/dist/browser/plugins/rate-limit.d.ts +15 -0
- package/dist/browser/plugins/rate-limit.js +162 -0
- package/dist/browser/plugins/retry.d.ts +14 -0
- package/dist/browser/plugins/retry.js +116 -0
- package/dist/browser/plugins/scrape.d.ts +21 -0
- package/dist/browser/plugins/scrape.js +82 -0
- package/dist/browser/plugins/server-timing.d.ts +7 -0
- package/dist/browser/plugins/server-timing.js +24 -0
- package/dist/browser/plugins/soap.d.ts +72 -0
- package/dist/browser/plugins/soap.js +347 -0
- package/dist/browser/plugins/xml.d.ts +9 -0
- package/dist/browser/plugins/xml.js +194 -0
- package/dist/browser/plugins/xsrf.d.ts +9 -0
- package/dist/browser/plugins/xsrf.js +48 -0
- package/dist/browser/recker.d.ts +26 -0
- package/dist/browser/recker.js +61 -0
- package/dist/browser/runner/request-runner.d.ts +46 -0
- package/dist/browser/runner/request-runner.js +89 -0
- package/dist/browser/scrape/document.d.ts +44 -0
- package/dist/browser/scrape/document.js +210 -0
- package/dist/browser/scrape/element.d.ts +49 -0
- package/dist/browser/scrape/element.js +176 -0
- package/dist/browser/scrape/extractors.d.ts +16 -0
- package/dist/browser/scrape/extractors.js +356 -0
- package/dist/browser/scrape/types.d.ts +107 -0
- package/dist/browser/scrape/types.js +1 -0
- package/dist/browser/transport/fetch.d.ts +11 -0
- package/dist/browser/transport/fetch.js +143 -0
- package/dist/browser/transport/undici.d.ts +38 -0
- package/dist/browser/transport/undici.js +897 -0
- package/dist/browser/types/ai.d.ts +267 -0
- package/dist/browser/types/ai.js +1 -0
- package/dist/browser/types/index.d.ts +351 -0
- package/dist/browser/types/index.js +1 -0
- package/dist/browser/types/logger.d.ts +16 -0
- package/dist/browser/types/logger.js +66 -0
- package/dist/browser/types/udp.d.ts +138 -0
- package/dist/browser/types/udp.js +1 -0
- package/dist/browser/utils/agent-manager.d.ts +29 -0
- package/dist/browser/utils/agent-manager.js +160 -0
- package/dist/browser/utils/body.d.ts +10 -0
- package/dist/browser/utils/body.js +148 -0
- package/dist/browser/utils/charset.d.ts +15 -0
- package/dist/browser/utils/charset.js +169 -0
- package/dist/browser/utils/concurrency.d.ts +20 -0
- package/dist/browser/utils/concurrency.js +120 -0
- package/dist/browser/utils/dns.d.ts +6 -0
- package/dist/browser/utils/dns.js +26 -0
- package/dist/browser/utils/header-parser.d.ts +94 -0
- package/dist/browser/utils/header-parser.js +617 -0
- package/dist/browser/utils/html-cleaner.d.ts +1 -0
- package/dist/browser/utils/html-cleaner.js +21 -0
- package/dist/browser/utils/link-header.d.ts +69 -0
- package/dist/browser/utils/link-header.js +190 -0
- package/dist/browser/utils/optional-require.d.ts +19 -0
- package/dist/browser/utils/optional-require.js +105 -0
- package/dist/browser/utils/progress.d.ts +8 -0
- package/dist/browser/utils/progress.js +82 -0
- package/dist/browser/utils/request-pool.d.ts +22 -0
- package/dist/browser/utils/request-pool.js +101 -0
- package/dist/browser/utils/sse.d.ts +7 -0
- package/dist/browser/utils/sse.js +67 -0
- package/dist/browser/utils/streaming.d.ts +17 -0
- package/dist/browser/utils/streaming.js +84 -0
- package/dist/browser/utils/try-fn.d.ts +3 -0
- package/dist/browser/utils/try-fn.js +59 -0
- package/dist/browser/utils/user-agent.d.ts +44 -0
- package/dist/browser/utils/user-agent.js +100 -0
- package/dist/browser/utils/whois.d.ts +32 -0
- package/dist/browser/utils/whois.js +246 -0
- package/dist/browser/websocket/client.d.ts +65 -0
- package/dist/browser/websocket/client.js +313 -0
- package/dist/cli/index.d.ts +1 -0
- package/dist/cli/index.js +1 -0
- package/dist/transport/fetch.d.ts +7 -1
- package/dist/transport/fetch.js +58 -76
- package/package.json +34 -2
|
@@ -0,0 +1,617 @@
|
|
|
1
|
+
export function parseCacheInfo(headers) {
|
|
2
|
+
const info = { hit: false };
|
|
3
|
+
const cfCacheStatus = headers.get('cf-cache-status');
|
|
4
|
+
if (cfCacheStatus) {
|
|
5
|
+
info.provider = 'cloudflare';
|
|
6
|
+
info.status = cfCacheStatus.toUpperCase();
|
|
7
|
+
info.hit = cfCacheStatus.toUpperCase() === 'HIT';
|
|
8
|
+
}
|
|
9
|
+
const xCache = headers.get('x-cache');
|
|
10
|
+
if (xCache) {
|
|
11
|
+
const upper = xCache.toUpperCase();
|
|
12
|
+
info.hit = upper.includes('HIT');
|
|
13
|
+
if (upper.includes('FASTLY'))
|
|
14
|
+
info.provider = 'fastly';
|
|
15
|
+
else if (upper.includes('VARNISH'))
|
|
16
|
+
info.provider = 'varnish';
|
|
17
|
+
if (upper.includes('HIT'))
|
|
18
|
+
info.status = 'HIT';
|
|
19
|
+
else if (upper.includes('MISS'))
|
|
20
|
+
info.status = 'MISS';
|
|
21
|
+
else if (upper.includes('EXPIRED'))
|
|
22
|
+
info.status = 'EXPIRED';
|
|
23
|
+
else if (upper.includes('STALE'))
|
|
24
|
+
info.status = 'STALE';
|
|
25
|
+
}
|
|
26
|
+
const akamaiCache = headers.get('x-akamai-cache-status');
|
|
27
|
+
if (akamaiCache) {
|
|
28
|
+
info.provider = 'akamai';
|
|
29
|
+
info.hit = akamaiCache.toUpperCase().includes('HIT');
|
|
30
|
+
}
|
|
31
|
+
const cloudFrontCache = headers.get('x-cache');
|
|
32
|
+
const cloudFrontId = headers.get('x-amz-cf-id');
|
|
33
|
+
if (cloudFrontId) {
|
|
34
|
+
info.provider = 'cloudfront';
|
|
35
|
+
}
|
|
36
|
+
const cacheControl = headers.get('cache-control');
|
|
37
|
+
if (cacheControl) {
|
|
38
|
+
const maxAgeMatch = cacheControl.match(/max-age=(\d+)/);
|
|
39
|
+
if (maxAgeMatch) {
|
|
40
|
+
info.maxAge = parseInt(maxAgeMatch[1], 10);
|
|
41
|
+
}
|
|
42
|
+
}
|
|
43
|
+
const age = headers.get('age');
|
|
44
|
+
if (age) {
|
|
45
|
+
info.age = parseInt(age, 10);
|
|
46
|
+
}
|
|
47
|
+
return info;
|
|
48
|
+
}
|
|
49
|
+
const PLATFORM_DETECTORS = [
|
|
50
|
+
{
|
|
51
|
+
name: 'cloudflare',
|
|
52
|
+
category: 'cdn',
|
|
53
|
+
headers: ['cf-ray', 'cf-cache-status'],
|
|
54
|
+
detect: (headers) => headers.get('cf-ray') !== null,
|
|
55
|
+
extract: (headers) => {
|
|
56
|
+
const ray = headers.get('cf-ray');
|
|
57
|
+
const metadata = {};
|
|
58
|
+
if (ray) {
|
|
59
|
+
metadata.ray = ray;
|
|
60
|
+
const parts = ray.split('-');
|
|
61
|
+
if (parts.length > 1) {
|
|
62
|
+
return { region: parts[1], metadata };
|
|
63
|
+
}
|
|
64
|
+
}
|
|
65
|
+
return { metadata };
|
|
66
|
+
}
|
|
67
|
+
},
|
|
68
|
+
{
|
|
69
|
+
name: 'fastly',
|
|
70
|
+
category: 'cdn',
|
|
71
|
+
headers: ['fastly-debug-digest', 'x-served-by'],
|
|
72
|
+
detect: (headers) => {
|
|
73
|
+
const servedBy = headers.get('x-served-by');
|
|
74
|
+
return servedBy?.includes('cache-') || headers.get('fastly-debug-digest') !== null;
|
|
75
|
+
},
|
|
76
|
+
extract: (headers) => ({
|
|
77
|
+
server: headers.get('x-served-by') || undefined,
|
|
78
|
+
metadata: {
|
|
79
|
+
digest: headers.get('fastly-debug-digest')
|
|
80
|
+
}
|
|
81
|
+
})
|
|
82
|
+
},
|
|
83
|
+
{
|
|
84
|
+
name: 'akamai',
|
|
85
|
+
category: 'cdn',
|
|
86
|
+
headers: ['x-akamai-request-id', 'x-akamai-cache-status'],
|
|
87
|
+
detect: (headers) => headers.get('x-akamai-request-id') !== null,
|
|
88
|
+
extract: (headers) => ({
|
|
89
|
+
metadata: {
|
|
90
|
+
requestId: headers.get('x-akamai-request-id'),
|
|
91
|
+
cacheId: headers.get('x-akamai-cache-status')
|
|
92
|
+
}
|
|
93
|
+
})
|
|
94
|
+
},
|
|
95
|
+
{
|
|
96
|
+
name: 'aws-cloudfront',
|
|
97
|
+
category: 'cloud',
|
|
98
|
+
headers: ['x-amz-cf-id', 'x-amz-cf-pop'],
|
|
99
|
+
detect: (headers) => headers.get('x-amz-cf-id') !== null,
|
|
100
|
+
extract: (headers) => ({
|
|
101
|
+
region: headers.get('x-amz-cf-pop') || undefined,
|
|
102
|
+
metadata: {
|
|
103
|
+
requestId: headers.get('x-amz-cf-id'),
|
|
104
|
+
pop: headers.get('x-amz-cf-pop')
|
|
105
|
+
}
|
|
106
|
+
})
|
|
107
|
+
},
|
|
108
|
+
{
|
|
109
|
+
name: 'gcp',
|
|
110
|
+
category: 'cloud',
|
|
111
|
+
headers: ['x-cloud-trace-context', 'x-goog-request-id'],
|
|
112
|
+
detect: (headers) => headers.get('x-cloud-trace-context') !== null || headers.get('x-goog-request-id') !== null,
|
|
113
|
+
extract: (headers) => {
|
|
114
|
+
const trace = headers.get('x-cloud-trace-context');
|
|
115
|
+
return {
|
|
116
|
+
metadata: {
|
|
117
|
+
traceId: trace ? trace.split('/')[0] : undefined,
|
|
118
|
+
requestId: headers.get('x-goog-request-id')
|
|
119
|
+
}
|
|
120
|
+
};
|
|
121
|
+
}
|
|
122
|
+
},
|
|
123
|
+
{
|
|
124
|
+
name: 'azure',
|
|
125
|
+
category: 'cloud',
|
|
126
|
+
headers: ['x-ms-request-id', 'x-azure-ref'],
|
|
127
|
+
detect: (headers) => headers.get('x-ms-request-id') !== null,
|
|
128
|
+
extract: (headers) => ({
|
|
129
|
+
metadata: {
|
|
130
|
+
requestId: headers.get('x-ms-request-id'),
|
|
131
|
+
ref: headers.get('x-azure-ref')
|
|
132
|
+
}
|
|
133
|
+
})
|
|
134
|
+
},
|
|
135
|
+
{
|
|
136
|
+
name: 'oracle-cloud',
|
|
137
|
+
category: 'cloud',
|
|
138
|
+
headers: ['x-oracle-dms-rid', 'x-oracle-dms-ecid'],
|
|
139
|
+
detect: (headers) => headers.get('x-oracle-dms-rid') !== null,
|
|
140
|
+
extract: (headers) => ({
|
|
141
|
+
metadata: {
|
|
142
|
+
requestId: headers.get('x-oracle-dms-rid'),
|
|
143
|
+
ecid: headers.get('x-oracle-dms-ecid')
|
|
144
|
+
}
|
|
145
|
+
})
|
|
146
|
+
},
|
|
147
|
+
{
|
|
148
|
+
name: 'alibaba-cloud',
|
|
149
|
+
category: 'cloud',
|
|
150
|
+
headers: ['ali-swift-global-savetime', 'eagleeye-traceid'],
|
|
151
|
+
detect: (headers) => headers.get('eagleeye-traceid') !== null,
|
|
152
|
+
extract: (headers) => ({
|
|
153
|
+
metadata: {
|
|
154
|
+
traceId: headers.get('eagleeye-traceid')
|
|
155
|
+
}
|
|
156
|
+
})
|
|
157
|
+
},
|
|
158
|
+
{
|
|
159
|
+
name: 'vercel',
|
|
160
|
+
category: 'hosting',
|
|
161
|
+
headers: ['x-vercel-id', 'x-vercel-cache'],
|
|
162
|
+
detect: (headers) => headers.get('x-vercel-id') !== null,
|
|
163
|
+
extract: (headers) => ({
|
|
164
|
+
metadata: {
|
|
165
|
+
requestId: headers.get('x-vercel-id'),
|
|
166
|
+
cache: headers.get('x-vercel-cache')
|
|
167
|
+
}
|
|
168
|
+
})
|
|
169
|
+
},
|
|
170
|
+
{
|
|
171
|
+
name: 'netlify',
|
|
172
|
+
category: 'hosting',
|
|
173
|
+
headers: ['x-nf-request-id', 'x-nf-trace-id'],
|
|
174
|
+
detect: (headers) => headers.get('x-nf-request-id') !== null,
|
|
175
|
+
extract: (headers) => ({
|
|
176
|
+
metadata: {
|
|
177
|
+
requestId: headers.get('x-nf-request-id'),
|
|
178
|
+
traceId: headers.get('x-nf-trace-id')
|
|
179
|
+
}
|
|
180
|
+
})
|
|
181
|
+
},
|
|
182
|
+
{
|
|
183
|
+
name: 'render',
|
|
184
|
+
category: 'hosting',
|
|
185
|
+
headers: ['x-render-origin-server'],
|
|
186
|
+
detect: (headers) => headers.get('x-render-origin-server') !== null,
|
|
187
|
+
extract: (headers) => ({
|
|
188
|
+
metadata: {
|
|
189
|
+
origin: headers.get('x-render-origin-server')
|
|
190
|
+
}
|
|
191
|
+
})
|
|
192
|
+
},
|
|
193
|
+
{
|
|
194
|
+
name: 'datadome',
|
|
195
|
+
category: 'security',
|
|
196
|
+
headers: ['x-datadome', 'x-datadome-cid'],
|
|
197
|
+
detect: (headers) => headers.get('x-datadome') !== null || headers.get('x-datadome-cid') !== null,
|
|
198
|
+
extract: (headers) => ({
|
|
199
|
+
metadata: { cid: headers.get('x-datadome-cid') }
|
|
200
|
+
})
|
|
201
|
+
},
|
|
202
|
+
{
|
|
203
|
+
name: 'incapsula',
|
|
204
|
+
category: 'security',
|
|
205
|
+
headers: ['x-iinfo', 'x-cdn'],
|
|
206
|
+
detect: (headers) => headers.get('x-iinfo') !== null || (headers.get('x-cdn') || '').includes('Incapsula'),
|
|
207
|
+
extract: (headers) => ({
|
|
208
|
+
metadata: { iinfo: headers.get('x-iinfo') }
|
|
209
|
+
})
|
|
210
|
+
},
|
|
211
|
+
{
|
|
212
|
+
name: 'imperva',
|
|
213
|
+
category: 'security',
|
|
214
|
+
headers: ['x-imperva-uuid'],
|
|
215
|
+
detect: (headers) => headers.get('x-imperva-uuid') !== null,
|
|
216
|
+
extract: (headers) => ({
|
|
217
|
+
metadata: { uuid: headers.get('x-imperva-uuid') }
|
|
218
|
+
})
|
|
219
|
+
},
|
|
220
|
+
{
|
|
221
|
+
name: 'aws-waf',
|
|
222
|
+
category: 'security',
|
|
223
|
+
headers: ['x-amzn-waf-action'],
|
|
224
|
+
detect: (headers) => headers.get('server') === 'awselb/2.0',
|
|
225
|
+
extract: (headers) => ({})
|
|
226
|
+
},
|
|
227
|
+
{
|
|
228
|
+
name: 'envoy',
|
|
229
|
+
category: 'proxy',
|
|
230
|
+
headers: ['x-envoy-upstream-service-time'],
|
|
231
|
+
detect: (headers) => headers.get('x-envoy-upstream-service-time') !== null || headers.get('server') === 'envoy',
|
|
232
|
+
extract: (headers) => ({})
|
|
233
|
+
},
|
|
234
|
+
{
|
|
235
|
+
name: 'traefik',
|
|
236
|
+
category: 'proxy',
|
|
237
|
+
headers: [],
|
|
238
|
+
detect: (headers) => (headers.get('server') || '').toLowerCase().includes('traefik'),
|
|
239
|
+
extract: (headers) => ({})
|
|
240
|
+
},
|
|
241
|
+
{
|
|
242
|
+
name: 'caddy',
|
|
243
|
+
category: 'server',
|
|
244
|
+
headers: [],
|
|
245
|
+
detect: (headers) => (headers.get('server') || '').toLowerCase().includes('caddy'),
|
|
246
|
+
extract: (headers) => ({})
|
|
247
|
+
},
|
|
248
|
+
{
|
|
249
|
+
name: 'haproxy',
|
|
250
|
+
category: 'proxy',
|
|
251
|
+
headers: [],
|
|
252
|
+
detect: (headers) => (headers.get('server') || '').toLowerCase().includes('haproxy'),
|
|
253
|
+
extract: (headers) => ({})
|
|
254
|
+
},
|
|
255
|
+
{
|
|
256
|
+
name: 'iis',
|
|
257
|
+
category: 'server',
|
|
258
|
+
headers: ['x-powered-by'],
|
|
259
|
+
detect: (headers) => (headers.get('server') || '').toLowerCase().includes('iis'),
|
|
260
|
+
extract: (headers) => ({})
|
|
261
|
+
},
|
|
262
|
+
{
|
|
263
|
+
name: 'kestrel',
|
|
264
|
+
category: 'server',
|
|
265
|
+
headers: [],
|
|
266
|
+
detect: (headers) => (headers.get('server') || '').toLowerCase().includes('kestrel'),
|
|
267
|
+
extract: (headers) => ({})
|
|
268
|
+
},
|
|
269
|
+
{
|
|
270
|
+
name: 'express',
|
|
271
|
+
category: 'framework',
|
|
272
|
+
headers: ['x-powered-by'],
|
|
273
|
+
detect: (headers) => (headers.get('x-powered-by') || '').toLowerCase().includes('express'),
|
|
274
|
+
extract: (headers) => ({})
|
|
275
|
+
},
|
|
276
|
+
{
|
|
277
|
+
name: 'rails',
|
|
278
|
+
category: 'framework',
|
|
279
|
+
headers: ['x-runtime', 'x-request-id'],
|
|
280
|
+
detect: (headers) => headers.get('x-runtime') !== null && !(headers.get('x-powered-by') || '').includes('Express'),
|
|
281
|
+
extract: (headers) => ({
|
|
282
|
+
metadata: { runtime: headers.get('x-runtime') }
|
|
283
|
+
})
|
|
284
|
+
},
|
|
285
|
+
{
|
|
286
|
+
name: 'django',
|
|
287
|
+
category: 'framework',
|
|
288
|
+
headers: [],
|
|
289
|
+
detect: (headers) => (headers.get('server') || '').toLowerCase().includes('wsgi'),
|
|
290
|
+
extract: (headers) => ({})
|
|
291
|
+
},
|
|
292
|
+
{
|
|
293
|
+
name: 'aspnet',
|
|
294
|
+
category: 'framework',
|
|
295
|
+
headers: ['x-aspnet-version', 'x-powered-by'],
|
|
296
|
+
detect: (headers) => headers.get('x-aspnet-version') !== null || (headers.get('x-powered-by') || '').includes('ASP.NET'),
|
|
297
|
+
extract: (headers) => ({
|
|
298
|
+
metadata: { version: headers.get('x-aspnet-version') }
|
|
299
|
+
})
|
|
300
|
+
},
|
|
301
|
+
{
|
|
302
|
+
name: 'php',
|
|
303
|
+
category: 'framework',
|
|
304
|
+
headers: ['x-powered-by'],
|
|
305
|
+
detect: (headers) => (headers.get('x-powered-by') || '').toLowerCase().includes('php'),
|
|
306
|
+
extract: (headers) => ({
|
|
307
|
+
metadata: { version: headers.get('x-powered-by') }
|
|
308
|
+
})
|
|
309
|
+
}
|
|
310
|
+
];
|
|
311
|
+
export function parsePlatformInfo(headers) {
|
|
312
|
+
const info = {
|
|
313
|
+
metadata: {}
|
|
314
|
+
};
|
|
315
|
+
for (const detector of PLATFORM_DETECTORS) {
|
|
316
|
+
if (detector.detect(headers)) {
|
|
317
|
+
info.provider = detector.name;
|
|
318
|
+
info.category = detector.category;
|
|
319
|
+
const extracted = detector.extract(headers);
|
|
320
|
+
if (extracted.region)
|
|
321
|
+
info.region = extracted.region;
|
|
322
|
+
if (extracted.server)
|
|
323
|
+
info.server = extracted.server;
|
|
324
|
+
if (extracted.metadata) {
|
|
325
|
+
Object.assign(info.metadata, extracted.metadata);
|
|
326
|
+
}
|
|
327
|
+
break;
|
|
328
|
+
}
|
|
329
|
+
}
|
|
330
|
+
if (!info.provider) {
|
|
331
|
+
const server = headers.get('server');
|
|
332
|
+
if (server) {
|
|
333
|
+
info.server = server;
|
|
334
|
+
const serverLower = server.toLowerCase();
|
|
335
|
+
if (serverLower.includes('cloudflare')) {
|
|
336
|
+
info.provider = 'cloudflare';
|
|
337
|
+
info.category = 'cdn';
|
|
338
|
+
}
|
|
339
|
+
else if (serverLower.includes('cloudfront')) {
|
|
340
|
+
info.provider = 'aws-cloudfront';
|
|
341
|
+
info.category = 'cloud';
|
|
342
|
+
}
|
|
343
|
+
else if (serverLower.includes('fastly')) {
|
|
344
|
+
info.provider = 'fastly';
|
|
345
|
+
info.category = 'cdn';
|
|
346
|
+
}
|
|
347
|
+
else if (serverLower.includes('akamai')) {
|
|
348
|
+
info.provider = 'akamai';
|
|
349
|
+
info.category = 'cdn';
|
|
350
|
+
}
|
|
351
|
+
else if (serverLower.includes('nginx')) {
|
|
352
|
+
info.provider = 'nginx';
|
|
353
|
+
info.category = 'server';
|
|
354
|
+
}
|
|
355
|
+
else if (serverLower.includes('apache')) {
|
|
356
|
+
info.provider = 'apache';
|
|
357
|
+
info.category = 'server';
|
|
358
|
+
}
|
|
359
|
+
else if (serverLower.includes('varnish')) {
|
|
360
|
+
info.provider = 'varnish';
|
|
361
|
+
info.category = 'proxy';
|
|
362
|
+
}
|
|
363
|
+
else {
|
|
364
|
+
info.provider = 'unknown';
|
|
365
|
+
info.category = 'unknown';
|
|
366
|
+
}
|
|
367
|
+
}
|
|
368
|
+
}
|
|
369
|
+
if (!info.provider) {
|
|
370
|
+
const via = headers.get('via');
|
|
371
|
+
if (via) {
|
|
372
|
+
const viaLower = via.toLowerCase();
|
|
373
|
+
if (viaLower.includes('cloudflare')) {
|
|
374
|
+
info.provider = 'cloudflare';
|
|
375
|
+
info.category = 'cdn';
|
|
376
|
+
}
|
|
377
|
+
else if (viaLower.includes('akamai')) {
|
|
378
|
+
info.provider = 'akamai';
|
|
379
|
+
info.category = 'cdn';
|
|
380
|
+
}
|
|
381
|
+
else if (viaLower.includes('varnish')) {
|
|
382
|
+
info.provider = 'varnish';
|
|
383
|
+
info.category = 'proxy';
|
|
384
|
+
}
|
|
385
|
+
else if (viaLower.includes('squid')) {
|
|
386
|
+
info.provider = 'squid';
|
|
387
|
+
info.category = 'proxy';
|
|
388
|
+
}
|
|
389
|
+
}
|
|
390
|
+
}
|
|
391
|
+
if (Object.keys(info.metadata || {}).length === 0) {
|
|
392
|
+
delete info.metadata;
|
|
393
|
+
}
|
|
394
|
+
return info;
|
|
395
|
+
}
|
|
396
|
+
export function parseRateLimitInfo(headers, status) {
|
|
397
|
+
const info = { limited: status === 429 };
|
|
398
|
+
const limit = headers.get('x-ratelimit-limit') || headers.get('ratelimit-limit');
|
|
399
|
+
const remaining = headers.get('x-ratelimit-remaining') || headers.get('ratelimit-remaining');
|
|
400
|
+
const reset = headers.get('x-ratelimit-reset') || headers.get('ratelimit-reset');
|
|
401
|
+
if (limit)
|
|
402
|
+
info.limit = parseInt(limit, 10);
|
|
403
|
+
if (remaining)
|
|
404
|
+
info.remaining = parseInt(remaining, 10);
|
|
405
|
+
if (reset) {
|
|
406
|
+
const resetValue = parseInt(reset, 10);
|
|
407
|
+
if (resetValue > 1000000000) {
|
|
408
|
+
info.reset = new Date(resetValue * 1000);
|
|
409
|
+
}
|
|
410
|
+
else {
|
|
411
|
+
info.reset = new Date(Date.now() + resetValue * 1000);
|
|
412
|
+
}
|
|
413
|
+
}
|
|
414
|
+
const retryAfter = headers.get('retry-after');
|
|
415
|
+
if (retryAfter) {
|
|
416
|
+
const seconds = parseInt(retryAfter, 10);
|
|
417
|
+
if (!isNaN(seconds)) {
|
|
418
|
+
info.retryAfter = seconds;
|
|
419
|
+
}
|
|
420
|
+
else {
|
|
421
|
+
const retryDate = new Date(retryAfter);
|
|
422
|
+
if (!isNaN(retryDate.getTime())) {
|
|
423
|
+
info.retryAfter = Math.ceil((retryDate.getTime() - Date.now()) / 1000);
|
|
424
|
+
}
|
|
425
|
+
}
|
|
426
|
+
}
|
|
427
|
+
const policy = headers.get('ratelimit-policy');
|
|
428
|
+
if (policy) {
|
|
429
|
+
info.policy = policy;
|
|
430
|
+
}
|
|
431
|
+
if (info.remaining !== undefined && info.remaining === 0) {
|
|
432
|
+
info.limited = true;
|
|
433
|
+
}
|
|
434
|
+
return info;
|
|
435
|
+
}
|
|
436
|
+
export function parseCompressionInfo(headers) {
|
|
437
|
+
const info = {};
|
|
438
|
+
const encoding = headers.get('content-encoding');
|
|
439
|
+
if (encoding) {
|
|
440
|
+
info.encoding = encoding.toLowerCase();
|
|
441
|
+
}
|
|
442
|
+
const xOriginalSize = headers.get('x-original-size') || headers.get('x-uncompressed-size');
|
|
443
|
+
if (xOriginalSize) {
|
|
444
|
+
info.originalSize = parseInt(xOriginalSize, 10);
|
|
445
|
+
}
|
|
446
|
+
const contentLength = headers.get('content-length');
|
|
447
|
+
if (contentLength) {
|
|
448
|
+
info.compressedSize = parseInt(contentLength, 10);
|
|
449
|
+
}
|
|
450
|
+
if (info.originalSize && info.compressedSize) {
|
|
451
|
+
info.ratio = info.originalSize / info.compressedSize;
|
|
452
|
+
}
|
|
453
|
+
return info;
|
|
454
|
+
}
|
|
455
|
+
export function parseCSPInfo(headers) {
|
|
456
|
+
const info = {
|
|
457
|
+
directives: {},
|
|
458
|
+
reportOnly: false
|
|
459
|
+
};
|
|
460
|
+
let cspHeader = headers.get('content-security-policy');
|
|
461
|
+
if (!cspHeader) {
|
|
462
|
+
cspHeader = headers.get('content-security-policy-report-only');
|
|
463
|
+
if (cspHeader) {
|
|
464
|
+
info.reportOnly = true;
|
|
465
|
+
}
|
|
466
|
+
}
|
|
467
|
+
if (cspHeader) {
|
|
468
|
+
info.policy = cspHeader;
|
|
469
|
+
const directives = cspHeader.split(';').map(d => d.trim()).filter(Boolean);
|
|
470
|
+
for (const directive of directives) {
|
|
471
|
+
const [name, ...values] = directive.split(/\s+/);
|
|
472
|
+
if (name) {
|
|
473
|
+
info.directives[name] = values;
|
|
474
|
+
}
|
|
475
|
+
}
|
|
476
|
+
}
|
|
477
|
+
return info;
|
|
478
|
+
}
|
|
479
|
+
export function parseContentType(headers) {
|
|
480
|
+
const info = {};
|
|
481
|
+
const contentType = headers.get('content-type');
|
|
482
|
+
if (!contentType)
|
|
483
|
+
return info;
|
|
484
|
+
const parts = contentType.split(';').map(p => p.trim());
|
|
485
|
+
if (parts[0]) {
|
|
486
|
+
info.mediaType = parts[0];
|
|
487
|
+
const [type, subtype] = parts[0].split('/');
|
|
488
|
+
if (type)
|
|
489
|
+
info.type = type;
|
|
490
|
+
if (subtype)
|
|
491
|
+
info.subtype = subtype;
|
|
492
|
+
}
|
|
493
|
+
for (let i = 1; i < parts.length; i++) {
|
|
494
|
+
const param = parts[i];
|
|
495
|
+
const [key, value] = param.split('=').map(s => s.trim());
|
|
496
|
+
if (key === 'charset' && value) {
|
|
497
|
+
info.charset = value.replace(/['"]/g, '');
|
|
498
|
+
}
|
|
499
|
+
else if (key === 'boundary' && value) {
|
|
500
|
+
info.boundary = value.replace(/['"]/g, '');
|
|
501
|
+
}
|
|
502
|
+
}
|
|
503
|
+
return info;
|
|
504
|
+
}
|
|
505
|
+
export function parseAcceptInfo(headers) {
|
|
506
|
+
const info = {
|
|
507
|
+
types: [],
|
|
508
|
+
encodings: [],
|
|
509
|
+
languages: []
|
|
510
|
+
};
|
|
511
|
+
const accept = headers.get('accept');
|
|
512
|
+
if (accept) {
|
|
513
|
+
const types = accept.split(',').map(t => t.trim());
|
|
514
|
+
for (const typeStr of types) {
|
|
515
|
+
const [mediaType, ...params] = typeStr.split(';').map(s => s.trim());
|
|
516
|
+
let q = 1.0;
|
|
517
|
+
for (const param of params) {
|
|
518
|
+
if (param.startsWith('q=')) {
|
|
519
|
+
q = parseFloat(param.substring(2)) || 1.0;
|
|
520
|
+
}
|
|
521
|
+
}
|
|
522
|
+
const [type, subtype] = mediaType.split('/');
|
|
523
|
+
info.types.push({
|
|
524
|
+
mediaType,
|
|
525
|
+
q,
|
|
526
|
+
type: type || undefined,
|
|
527
|
+
subtype: subtype || undefined
|
|
528
|
+
});
|
|
529
|
+
}
|
|
530
|
+
info.types.sort((a, b) => b.q - a.q);
|
|
531
|
+
}
|
|
532
|
+
const acceptEncoding = headers.get('accept-encoding');
|
|
533
|
+
if (acceptEncoding) {
|
|
534
|
+
const encodings = acceptEncoding.split(',').map(e => e.trim());
|
|
535
|
+
for (const encodingStr of encodings) {
|
|
536
|
+
const [encoding, ...params] = encodingStr.split(';').map(s => s.trim());
|
|
537
|
+
let q = 1.0;
|
|
538
|
+
for (const param of params) {
|
|
539
|
+
if (param.startsWith('q=')) {
|
|
540
|
+
q = parseFloat(param.substring(2)) || 1.0;
|
|
541
|
+
}
|
|
542
|
+
}
|
|
543
|
+
info.encodings.push({ encoding, q });
|
|
544
|
+
}
|
|
545
|
+
info.encodings.sort((a, b) => b.q - a.q);
|
|
546
|
+
}
|
|
547
|
+
const acceptLanguage = headers.get('accept-language');
|
|
548
|
+
if (acceptLanguage) {
|
|
549
|
+
const languages = acceptLanguage.split(',').map(l => l.trim());
|
|
550
|
+
for (const languageStr of languages) {
|
|
551
|
+
const [language, ...params] = languageStr.split(';').map(s => s.trim());
|
|
552
|
+
let q = 1.0;
|
|
553
|
+
for (const param of params) {
|
|
554
|
+
if (param.startsWith('q=')) {
|
|
555
|
+
q = parseFloat(param.substring(2)) || 1.0;
|
|
556
|
+
}
|
|
557
|
+
}
|
|
558
|
+
info.languages.push({ language, q });
|
|
559
|
+
}
|
|
560
|
+
info.languages.sort((a, b) => b.q - a.q);
|
|
561
|
+
}
|
|
562
|
+
return info;
|
|
563
|
+
}
|
|
564
|
+
export function parseAuthInfo(headers) {
|
|
565
|
+
const info = { methods: [] };
|
|
566
|
+
const wwwAuth = headers.get('www-authenticate');
|
|
567
|
+
if (wwwAuth) {
|
|
568
|
+
if (wwwAuth.toLowerCase().includes('basic'))
|
|
569
|
+
info.methods.push('Basic');
|
|
570
|
+
if (wwwAuth.toLowerCase().includes('bearer'))
|
|
571
|
+
info.methods.push('Bearer');
|
|
572
|
+
if (wwwAuth.toLowerCase().includes('digest'))
|
|
573
|
+
info.methods.push('Digest');
|
|
574
|
+
if (wwwAuth.toLowerCase().includes('negotiate'))
|
|
575
|
+
info.methods.push('Negotiate');
|
|
576
|
+
if (wwwAuth.toLowerCase().includes('aws4-hmac-sha256'))
|
|
577
|
+
info.methods.push('AWS4');
|
|
578
|
+
const realmMatch = wwwAuth.match(/realm="([^"]+)"/);
|
|
579
|
+
if (realmMatch)
|
|
580
|
+
info.realm = realmMatch[1];
|
|
581
|
+
const errorMatch = wwwAuth.match(/error="([^"]+)"/);
|
|
582
|
+
if (errorMatch)
|
|
583
|
+
info.error = errorMatch[1];
|
|
584
|
+
const descMatch = wwwAuth.match(/error_description="([^"]+)"/);
|
|
585
|
+
if (descMatch)
|
|
586
|
+
info.errorDescription = descMatch[1];
|
|
587
|
+
}
|
|
588
|
+
const xAuthError = headers.get('x-auth-error') || headers.get('x-authentication-error');
|
|
589
|
+
if (xAuthError) {
|
|
590
|
+
if (!info.error)
|
|
591
|
+
info.error = xAuthError;
|
|
592
|
+
}
|
|
593
|
+
return info;
|
|
594
|
+
}
|
|
595
|
+
export function parseClockSkew(headers) {
|
|
596
|
+
const dateHeader = headers.get('date');
|
|
597
|
+
if (!dateHeader)
|
|
598
|
+
return {};
|
|
599
|
+
const serverTime = new Date(dateHeader);
|
|
600
|
+
if (isNaN(serverTime.getTime()))
|
|
601
|
+
return {};
|
|
602
|
+
const skewMs = serverTime.getTime() - Date.now();
|
|
603
|
+
return { serverTime, skewMs };
|
|
604
|
+
}
|
|
605
|
+
export function parseHeaders(headers, status) {
|
|
606
|
+
return {
|
|
607
|
+
cache: parseCacheInfo(headers),
|
|
608
|
+
platform: parsePlatformInfo(headers),
|
|
609
|
+
rateLimit: parseRateLimitInfo(headers, status),
|
|
610
|
+
compression: parseCompressionInfo(headers),
|
|
611
|
+
csp: parseCSPInfo(headers),
|
|
612
|
+
contentType: parseContentType(headers),
|
|
613
|
+
accept: parseAcceptInfo(headers),
|
|
614
|
+
auth: parseAuthInfo(headers),
|
|
615
|
+
clockSkew: parseClockSkew(headers)
|
|
616
|
+
};
|
|
617
|
+
}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
export declare function cleanHtml(html: string): string;
|
|
@@ -0,0 +1,21 @@
|
|
|
1
|
+
export function cleanHtml(html) {
|
|
2
|
+
if (!html)
|
|
3
|
+
return '';
|
|
4
|
+
let text = html;
|
|
5
|
+
text = text.replace(/<script\b[^>]*>[\s\S]*?<\/script>/gi, '');
|
|
6
|
+
text = text.replace(/<style\b[^>]*>[\s\S]*?<\/style>/gi, '');
|
|
7
|
+
text = text.replace(/<!--[\s\S]*?-->/g, '');
|
|
8
|
+
text = text.replace(/<\/(div|p|h[1-6]|li|ul|ol|tr|table|section|article|main|header|footer|nav)>/gi, '\n');
|
|
9
|
+
text = text.replace(/<(br|hr)\s*\/?>/gi, '\n');
|
|
10
|
+
text = text.replace(/<[^>]+>/g, ' ');
|
|
11
|
+
text = text
|
|
12
|
+
.replace(/ /g, ' ')
|
|
13
|
+
.replace(/&/g, '&')
|
|
14
|
+
.replace(/</g, '<')
|
|
15
|
+
.replace(/>/g, '>')
|
|
16
|
+
.replace(/"/g, '"')
|
|
17
|
+
.replace(/'/g, "'");
|
|
18
|
+
text = text.replace(/[ \t]+/g, ' ');
|
|
19
|
+
text = text.replace(/\n\s*\n\s*\n+/g, '\n\n');
|
|
20
|
+
return text.trim();
|
|
21
|
+
}
|