@agentsean/crawler 1.0.0
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/LICENSE +661 -0
- package/dist/adaptive.d.ts +9 -0
- package/dist/adaptive.d.ts.map +1 -0
- package/dist/adaptive.js +43 -0
- package/dist/adaptive.js.map +1 -0
- package/dist/crawl.d.ts +4 -0
- package/dist/crawl.d.ts.map +1 -0
- package/dist/crawl.js +348 -0
- package/dist/crawl.js.map +1 -0
- package/dist/decode.d.ts +7 -0
- package/dist/decode.d.ts.map +1 -0
- package/dist/decode.js +53 -0
- package/dist/decode.js.map +1 -0
- package/dist/extract.d.ts +4 -0
- package/dist/extract.d.ts.map +1 -0
- package/dist/extract.js +268 -0
- package/dist/extract.js.map +1 -0
- package/dist/hash.d.ts +4 -0
- package/dist/hash.d.ts.map +1 -0
- package/dist/hash.js +11 -0
- package/dist/hash.js.map +1 -0
- package/dist/http.d.ts +38 -0
- package/dist/http.d.ts.map +1 -0
- package/dist/http.js +207 -0
- package/dist/http.js.map +1 -0
- package/dist/index.d.ts +14 -0
- package/dist/index.d.ts.map +1 -0
- package/dist/index.js +13 -0
- package/dist/index.js.map +1 -0
- package/dist/persist.d.ts +20 -0
- package/dist/persist.d.ts.map +1 -0
- package/dist/persist.js +186 -0
- package/dist/persist.js.map +1 -0
- package/dist/probe.d.ts +4 -0
- package/dist/probe.d.ts.map +1 -0
- package/dist/probe.js +132 -0
- package/dist/probe.js.map +1 -0
- package/dist/readability.d.ts +9 -0
- package/dist/readability.d.ts.map +1 -0
- package/dist/readability.js +60 -0
- package/dist/readability.js.map +1 -0
- package/dist/render.d.ts +15 -0
- package/dist/render.d.ts.map +1 -0
- package/dist/render.js +87 -0
- package/dist/render.js.map +1 -0
- package/dist/robots.d.ts +13 -0
- package/dist/robots.d.ts.map +1 -0
- package/dist/robots.js +187 -0
- package/dist/robots.js.map +1 -0
- package/dist/simhash.d.ts +9 -0
- package/dist/simhash.d.ts.map +1 -0
- package/dist/simhash.js +58 -0
- package/dist/simhash.js.map +1 -0
- package/dist/sitemap.d.ts +3 -0
- package/dist/sitemap.d.ts.map +1 -0
- package/dist/sitemap.js +165 -0
- package/dist/sitemap.js.map +1 -0
- package/dist/types.d.ts +251 -0
- package/dist/types.d.ts.map +1 -0
- package/dist/types.js +2 -0
- package/dist/types.js.map +1 -0
- package/dist/ua.d.ts +3 -0
- package/dist/ua.d.ts.map +1 -0
- package/dist/ua.js +3 -0
- package/dist/ua.js.map +1 -0
- package/dist/url.d.ts +11 -0
- package/dist/url.d.ts.map +1 -0
- package/dist/url.js +114 -0
- package/dist/url.js.map +1 -0
- package/package.json +60 -0
package/dist/adaptive.js
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
1
|
+
import { jsDependencyScore } from "./render.js";
|
|
2
|
+
const DETECTION_RATIO = 0.1;
|
|
3
|
+
const FINGERPRINT_N = 50;
|
|
4
|
+
export function decideRenderPolicy(pages, page, index) {
|
|
5
|
+
if (shouldAlwaysRender(page))
|
|
6
|
+
return "always";
|
|
7
|
+
if (index < FINGERPRINT_N)
|
|
8
|
+
return "sample";
|
|
9
|
+
const same = pages.filter((p) => p.templateKey === page.templateKey);
|
|
10
|
+
if (same.length < 3)
|
|
11
|
+
return "sample";
|
|
12
|
+
const scores = same.map((p) => p.jsDependencyScore).toSorted((a, b) => a - b);
|
|
13
|
+
const median = scores[Math.floor(scores.length / 2)] ?? 0;
|
|
14
|
+
const p95 = scores[Math.floor(scores.length * 0.95)] ?? median;
|
|
15
|
+
if (median > 0.35)
|
|
16
|
+
return "always";
|
|
17
|
+
if (p95 < 0.05)
|
|
18
|
+
return "never";
|
|
19
|
+
return Math.random() < DETECTION_RATIO ? "sample" : "never";
|
|
20
|
+
}
|
|
21
|
+
export function shouldAlwaysRender(page) {
|
|
22
|
+
try {
|
|
23
|
+
const u = new URL(page.url);
|
|
24
|
+
if (u.pathname === "/" || u.pathname === "")
|
|
25
|
+
return true;
|
|
26
|
+
}
|
|
27
|
+
catch {
|
|
28
|
+
/* ignore */
|
|
29
|
+
}
|
|
30
|
+
if (!page.extract)
|
|
31
|
+
return false;
|
|
32
|
+
if (page.extract.spaRootEmpty)
|
|
33
|
+
return true;
|
|
34
|
+
if (page.extract.hasNoscriptJsWarning)
|
|
35
|
+
return true;
|
|
36
|
+
if (page.extract.mainWordCount < 200 && page.extract.spaRootEmpty)
|
|
37
|
+
return true;
|
|
38
|
+
return false;
|
|
39
|
+
}
|
|
40
|
+
export function scoreFromHtml(html, extract) {
|
|
41
|
+
return jsDependencyScore(html, extract);
|
|
42
|
+
}
|
|
43
|
+
//# sourceMappingURL=adaptive.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"adaptive.js","sourceRoot":"","sources":["../src/adaptive.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,iBAAiB,EAAE,MAAM,aAAa,CAAC;AAKhD,MAAM,eAAe,GAAG,GAAG,CAAC;AAC5B,MAAM,aAAa,GAAG,EAAE,CAAC;AAEzB,MAAM,UAAU,kBAAkB,CAChC,KAAuE,EACvE,IAAgF,EAChF,KAAa;IAEb,IAAI,kBAAkB,CAAC,IAAI,CAAC;QAAE,OAAO,QAAQ,CAAC;IAC9C,IAAI,KAAK,GAAG,aAAa;QAAE,OAAO,QAAQ,CAAC;IAC3C,MAAM,IAAI,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,WAAW,KAAK,IAAI,CAAC,WAAW,CAAC,CAAC;IACrE,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC;QAAE,OAAO,QAAQ,CAAC;IACrC,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,iBAAiB,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;IAC9E,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;IAC1D,MAAM,GAAG,GAAG,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,MAAM,GAAG,IAAI,CAAC,CAAC,IAAI,MAAM,CAAC;IAC/D,IAAI,MAAM,GAAG,IAAI;QAAE,OAAO,QAAQ,CAAC;IACnC,IAAI,GAAG,GAAG,IAAI;QAAE,OAAO,OAAO,CAAC;IAC/B,OAAO,IAAI,CAAC,MAAM,EAAE,GAAG,eAAe,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC;AAC9D,CAAC;AAED,MAAM,UAAU,kBAAkB,CAAC,IAGlC;IACC,IAAI,CAAC;QACH,MAAM,CAAC,GAAG,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QAC5B,IAAI,CAAC,CAAC,QAAQ,KAAK,GAAG,IAAI,CAAC,CAAC,QAAQ,KAAK,EAAE;YAAE,OAAO,IAAI,CAAC;IAC3D,CAAC;IAAC,MAAM,CAAC;QACP,YAAY;IACd,CAAC;IACD,IAAI,CAAC,IAAI,CAAC,OAAO;QAAE,OAAO,KAAK,CAAC;IAChC,IAAI,IAAI,CAAC,OAAO,CAAC,YAAY;QAAE,OAAO,IAAI,CAAC;IAC3C,IAAI,IAAI,CAAC,OAAO,CAAC,oBAAoB;QAAE,OAAO,IAAI,CAAC;IACnD,IAAI,IAAI,CAAC,OAAO,CAAC,aAAa,GAAG,GAAG,IAAI,IAAI,CAAC,OAAO,CAAC,YAAY;QAAE,OAAO,IAAI,CAAC;IAC/E,OAAO,KAAK,CAAC;AACf,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,IAAY,EAAE,OAAoB;IAC9D,OAAO,iBAAiB,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;AAC1C,CAAC"}
|
package/dist/crawl.d.ts
ADDED
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"crawl.d.ts","sourceRoot":"","sources":["../src/crawl.ts"],"names":[],"mappings":"AAaA,OAAO,KAAK,EAGV,YAAY,EACZ,WAAW,EAGZ,MAAM,YAAY,CAAC;AAEpB,wBAAgB,cAAc,IAAI,MAAM,CAKvC;AAED,wBAAsB,SAAS,CAAC,OAAO,EAAE,YAAY,GAAG,OAAO,CAAC,WAAW,CAAC,CAuK3E"}
|
package/dist/crawl.js
ADDED
|
@@ -0,0 +1,348 @@
|
|
|
1
|
+
import os from "node:os";
|
|
2
|
+
import { createCrawlerAgent, fetchUrl, ROBOTS_MAX_REDIRECTS } from "./http.js";
|
|
3
|
+
import { extractPage, parseLinkHeaderCanonicals } from "./extract.js";
|
|
4
|
+
import { parseSitemapXml } from "./sitemap.js";
|
|
5
|
+
import { isAllowedByRobots, robotsFromFetch } from "./robots.js";
|
|
6
|
+
import { contentHash } from "./hash.js";
|
|
7
|
+
import { simhashHex } from "./simhash.js";
|
|
8
|
+
import { originOf, templateKey, isInternalUrl, normalizeUrl } from "./url.js";
|
|
9
|
+
import { decideRenderPolicy, scoreFromHtml } from "./adaptive.js";
|
|
10
|
+
import { renderPage } from "./render.js";
|
|
11
|
+
import { probeOrigin } from "./probe.js";
|
|
12
|
+
import { SEAN_UA } from "./ua.js";
|
|
13
|
+
export function ramConcurrency() {
|
|
14
|
+
const freeMb = os.freemem() / (1024 * 1024);
|
|
15
|
+
const cores = Math.max(1, os.cpus().length);
|
|
16
|
+
const fromRam = Math.floor((freeMb - 400) / 50);
|
|
17
|
+
return Math.max(2, Math.min(fromRam || cores, cores * 2, 16));
|
|
18
|
+
}
|
|
19
|
+
export async function crawlSite(options) {
|
|
20
|
+
const startUrl = normalizeUrl(options.startUrl);
|
|
21
|
+
const origin = originOf(startUrl);
|
|
22
|
+
const maxPages = options.maxPages ?? 5000;
|
|
23
|
+
const concurrency = options.concurrency ?? ramConcurrency();
|
|
24
|
+
const rps = options.rps ?? 4;
|
|
25
|
+
const render = options.render ?? true;
|
|
26
|
+
const userAgent = options.userAgent ?? SEAN_UA;
|
|
27
|
+
const startedAt = new Date().toISOString();
|
|
28
|
+
const agent = createCrawlerAgent(2);
|
|
29
|
+
const robots = await fetchRobots(origin, agent, userAgent);
|
|
30
|
+
const sitemaps = await fetchSitemaps(origin, robots, agent, userAgent);
|
|
31
|
+
const seen = new Set(options.resume?.seen ?? []);
|
|
32
|
+
const queue = options.resume?.queue
|
|
33
|
+
? options.resume.queue.map((q) => ({ url: q.url, depth: q.depth }))
|
|
34
|
+
: [];
|
|
35
|
+
const enqueue = (url, depth) => {
|
|
36
|
+
let abs;
|
|
37
|
+
try {
|
|
38
|
+
abs = normalizeUrl(url);
|
|
39
|
+
}
|
|
40
|
+
catch {
|
|
41
|
+
return;
|
|
42
|
+
}
|
|
43
|
+
if (!isInternalUrl(abs, origin) && !options.followExternal)
|
|
44
|
+
return;
|
|
45
|
+
if (seen.has(abs))
|
|
46
|
+
return;
|
|
47
|
+
if (!isAllowedByRobots(robots, abs, userAgent)) {
|
|
48
|
+
seen.add(abs);
|
|
49
|
+
blocked.push(abs);
|
|
50
|
+
return;
|
|
51
|
+
}
|
|
52
|
+
seen.add(abs);
|
|
53
|
+
queue.push({ url: abs, depth });
|
|
54
|
+
};
|
|
55
|
+
const blocked = [];
|
|
56
|
+
if (!options.resume) {
|
|
57
|
+
enqueue(startUrl, 0);
|
|
58
|
+
for (const sm of sitemaps) {
|
|
59
|
+
for (const u of sm.urls)
|
|
60
|
+
enqueue(u.loc, 1);
|
|
61
|
+
}
|
|
62
|
+
}
|
|
63
|
+
const sitemapLocs = new Set(sitemaps.flatMap((s) => s.urls.map((u) => normalizeUrlSafe(u.loc))));
|
|
64
|
+
const pages = [];
|
|
65
|
+
const inlinks = new Map();
|
|
66
|
+
const followInlinks = new Map();
|
|
67
|
+
const minInterval = 1000 / Math.max(0.1, rps);
|
|
68
|
+
let lastFetch = 0;
|
|
69
|
+
let nextIndex = 0;
|
|
70
|
+
let active = 0;
|
|
71
|
+
const snapshot = () => ({
|
|
72
|
+
seen: [...seen],
|
|
73
|
+
queue: queue.map((q) => ({ url: q.url, depth: q.depth })),
|
|
74
|
+
pagesSeen: pages.length,
|
|
75
|
+
});
|
|
76
|
+
const emitCheckpoint = () => {
|
|
77
|
+
const every = options.checkpointEvery ?? 10;
|
|
78
|
+
if (pages.length % every !== 0)
|
|
79
|
+
return;
|
|
80
|
+
options.onCheckpoint?.(snapshot());
|
|
81
|
+
};
|
|
82
|
+
let aborted = false;
|
|
83
|
+
const workers = Array.from({ length: Math.min(concurrency, maxPages) }, async () => {
|
|
84
|
+
while (pages.length < maxPages) {
|
|
85
|
+
if (options.signal?.aborted) {
|
|
86
|
+
aborted = true;
|
|
87
|
+
return;
|
|
88
|
+
}
|
|
89
|
+
const item = queue.shift();
|
|
90
|
+
if (!item) {
|
|
91
|
+
if (active === 0)
|
|
92
|
+
return;
|
|
93
|
+
await sleep(20);
|
|
94
|
+
continue;
|
|
95
|
+
}
|
|
96
|
+
active++;
|
|
97
|
+
try {
|
|
98
|
+
const wait = lastFetch + minInterval - Date.now();
|
|
99
|
+
if (wait > 0)
|
|
100
|
+
await sleep(wait);
|
|
101
|
+
lastFetch = Date.now();
|
|
102
|
+
const page = await crawlOne({
|
|
103
|
+
url: item.url,
|
|
104
|
+
depth: item.depth,
|
|
105
|
+
origin,
|
|
106
|
+
agent,
|
|
107
|
+
userAgent,
|
|
108
|
+
timeoutMs: options.timeoutMs ?? 20_000,
|
|
109
|
+
robots,
|
|
110
|
+
sitemapLocs,
|
|
111
|
+
render,
|
|
112
|
+
index: nextIndex++,
|
|
113
|
+
prior: pages,
|
|
114
|
+
check304: options.check304 ?? true,
|
|
115
|
+
});
|
|
116
|
+
pages.push(page);
|
|
117
|
+
emitCheckpoint();
|
|
118
|
+
if (page.extract) {
|
|
119
|
+
for (const link of page.extract.links) {
|
|
120
|
+
if (!link.isInternal)
|
|
121
|
+
continue;
|
|
122
|
+
inlinks.set(link.absUrl, (inlinks.get(link.absUrl) ?? 0) + 1);
|
|
123
|
+
if (!link.isNofollow) {
|
|
124
|
+
followInlinks.set(link.absUrl, (followInlinks.get(link.absUrl) ?? 0) + 1);
|
|
125
|
+
}
|
|
126
|
+
if (pages.length + queue.length < maxPages * 2) {
|
|
127
|
+
enqueue(link.absUrl, item.depth + 1);
|
|
128
|
+
}
|
|
129
|
+
}
|
|
130
|
+
}
|
|
131
|
+
}
|
|
132
|
+
finally {
|
|
133
|
+
active--;
|
|
134
|
+
}
|
|
135
|
+
}
|
|
136
|
+
});
|
|
137
|
+
await Promise.all(workers);
|
|
138
|
+
for (const p of pages) {
|
|
139
|
+
p.inlinkCount = inlinks.get(p.url) ?? 0;
|
|
140
|
+
p.followInlinkCount = followInlinks.get(p.url) ?? 0;
|
|
141
|
+
}
|
|
142
|
+
let originProbe;
|
|
143
|
+
try {
|
|
144
|
+
originProbe = await probeOrigin(startUrl, agent);
|
|
145
|
+
}
|
|
146
|
+
catch {
|
|
147
|
+
originProbe = {
|
|
148
|
+
https: startUrl.startsWith("https:"),
|
|
149
|
+
httpRedirectsToHttps: null,
|
|
150
|
+
wwwSplit: null,
|
|
151
|
+
wwwPreferred: null,
|
|
152
|
+
trailingSlashSplit: null,
|
|
153
|
+
randomSoft404: false,
|
|
154
|
+
randomSoft404Url: null,
|
|
155
|
+
randomSoft404Hash: null,
|
|
156
|
+
certValidTo: null,
|
|
157
|
+
certDaysRemaining: null,
|
|
158
|
+
certError: null,
|
|
159
|
+
alpn: null,
|
|
160
|
+
hsts: null,
|
|
161
|
+
};
|
|
162
|
+
}
|
|
163
|
+
await agent.close();
|
|
164
|
+
const checkpoint = snapshot();
|
|
165
|
+
options.onCheckpoint?.(checkpoint);
|
|
166
|
+
return {
|
|
167
|
+
origin,
|
|
168
|
+
startUrl,
|
|
169
|
+
startedAt,
|
|
170
|
+
finishedAt: new Date().toISOString(),
|
|
171
|
+
pages,
|
|
172
|
+
robots,
|
|
173
|
+
sitemaps,
|
|
174
|
+
originProbe,
|
|
175
|
+
pagesSeen: pages.length,
|
|
176
|
+
pagesChanged: pages.filter((p) => !p.notModified).length,
|
|
177
|
+
maxPages,
|
|
178
|
+
truncated: queue.length > 0 || seen.size > pages.length,
|
|
179
|
+
aborted,
|
|
180
|
+
checkpoint,
|
|
181
|
+
};
|
|
182
|
+
}
|
|
183
|
+
async function crawlOne(opts) {
|
|
184
|
+
const fetched = await fetchUrl(opts.url, {
|
|
185
|
+
agent: opts.agent,
|
|
186
|
+
userAgent: opts.userAgent,
|
|
187
|
+
timeoutMs: opts.timeoutMs,
|
|
188
|
+
});
|
|
189
|
+
const isHtml = /html|xml|xhtml/i.test(fetched.contentType) || looksLikeHtml(fetched.decoded);
|
|
190
|
+
const html = isHtml ? fetched.decoded.toString("utf8") : null;
|
|
191
|
+
const headerCanonical = parseLinkHeaderCanonicals(fetched.headers.link);
|
|
192
|
+
const xRobots = fetched.headers["x-robots-tag"] ?? null;
|
|
193
|
+
const extract = html
|
|
194
|
+
? extractPage(html, fetched.finalUrl, headerCanonical, xRobots)
|
|
195
|
+
: null;
|
|
196
|
+
const hash = fetched.decoded.length ? contentHash(fetched.decoded) : null;
|
|
197
|
+
const sh = extract ? simhashHex(extract.mainText || extract.allText) : null;
|
|
198
|
+
const jsScore = html && extract ? scoreFromHtml(html, extract) : 0;
|
|
199
|
+
const tmpl = templateKey(opts.url);
|
|
200
|
+
let supports304 = null;
|
|
201
|
+
if (opts.check304 &&
|
|
202
|
+
fetched.statusCode === 200 &&
|
|
203
|
+
(fetched.etag || fetched.lastModified)) {
|
|
204
|
+
const again = await fetchUrl(opts.url, {
|
|
205
|
+
agent: opts.agent,
|
|
206
|
+
userAgent: opts.userAgent,
|
|
207
|
+
timeoutMs: opts.timeoutMs,
|
|
208
|
+
etag: fetched.etag ?? undefined,
|
|
209
|
+
lastModified: fetched.lastModified ?? undefined,
|
|
210
|
+
maxRedirects: 0,
|
|
211
|
+
});
|
|
212
|
+
supports304 = again.statusCode === 304;
|
|
213
|
+
}
|
|
214
|
+
const draft = {
|
|
215
|
+
url: opts.url,
|
|
216
|
+
finalUrl: fetched.finalUrl,
|
|
217
|
+
statusCode: fetched.statusCode,
|
|
218
|
+
error: fetched.error,
|
|
219
|
+
contentType: fetched.contentType,
|
|
220
|
+
headers: fetched.headers,
|
|
221
|
+
ttfbMs: fetched.ttfbMs,
|
|
222
|
+
totalMs: fetched.totalMs,
|
|
223
|
+
wireBytes: fetched.wireBytes,
|
|
224
|
+
decodedBytes: fetched.decodedBytes,
|
|
225
|
+
httpVersion: null,
|
|
226
|
+
redirectChain: fetched.redirectChain,
|
|
227
|
+
redirectLoop: fetched.redirectLoop,
|
|
228
|
+
exceedsGoogleRedirectLimit: fetched.exceedsGoogleRedirectLimit,
|
|
229
|
+
html,
|
|
230
|
+
renderedHtml: null,
|
|
231
|
+
extract,
|
|
232
|
+
renderedExtract: null,
|
|
233
|
+
contentHash: hash,
|
|
234
|
+
simhash: sh,
|
|
235
|
+
etag: fetched.etag,
|
|
236
|
+
lastModified: fetched.lastModified,
|
|
237
|
+
notModified: fetched.notModified,
|
|
238
|
+
supports304,
|
|
239
|
+
inSitemap: opts.sitemapLocs.has(opts.url) || opts.sitemapLocs.has(fetched.finalUrl),
|
|
240
|
+
robotsAllowed: isAllowedByRobots(opts.robots, opts.url, opts.userAgent),
|
|
241
|
+
depth: opts.depth,
|
|
242
|
+
inlinkCount: 0,
|
|
243
|
+
outlinkCount: extract?.links.filter((l) => l.isInternal).length ?? 0,
|
|
244
|
+
followInlinkCount: 0,
|
|
245
|
+
isInternal: isInternalUrl(opts.url, opts.origin),
|
|
246
|
+
fetchedAt: new Date().toISOString(),
|
|
247
|
+
renderPolicy: "skipped",
|
|
248
|
+
rendered: false,
|
|
249
|
+
jsDependencyScore: jsScore,
|
|
250
|
+
templateKey: tmpl,
|
|
251
|
+
};
|
|
252
|
+
if (opts.render && html && extract && fetched.statusCode === 200) {
|
|
253
|
+
const policy = decideRenderPolicy(opts.prior, draft, opts.index);
|
|
254
|
+
draft.renderPolicy = policy === "never" ? "never" : policy;
|
|
255
|
+
if (policy === "always" || policy === "sample") {
|
|
256
|
+
const rendered = await renderPage(opts.url, html);
|
|
257
|
+
if (rendered && !rendered.errors.includes("playwright_unavailable")) {
|
|
258
|
+
draft.renderedHtml = rendered.html;
|
|
259
|
+
draft.renderedExtract = rendered.extract;
|
|
260
|
+
draft.rendered = true;
|
|
261
|
+
if (rendered.timedOut)
|
|
262
|
+
draft.error = draft.error ?? "render_timeout";
|
|
263
|
+
}
|
|
264
|
+
else {
|
|
265
|
+
draft.renderPolicy = "skipped";
|
|
266
|
+
}
|
|
267
|
+
}
|
|
268
|
+
}
|
|
269
|
+
return draft;
|
|
270
|
+
}
|
|
271
|
+
async function fetchRobots(origin, agent, userAgent) {
|
|
272
|
+
const url = `${origin}/robots.txt`;
|
|
273
|
+
const res = await fetchUrl(url, {
|
|
274
|
+
agent,
|
|
275
|
+
userAgent,
|
|
276
|
+
timeoutMs: 10_000,
|
|
277
|
+
maxRedirects: ROBOTS_MAX_REDIRECTS,
|
|
278
|
+
accept: "text/plain,*/*;q=0.1",
|
|
279
|
+
});
|
|
280
|
+
return robotsFromFetch({
|
|
281
|
+
statusCode: res.statusCode,
|
|
282
|
+
raw: res.decoded.toString("utf8"),
|
|
283
|
+
contentType: res.contentType,
|
|
284
|
+
bytes: res.decodedBytes,
|
|
285
|
+
redirectHops: Math.max(0, res.redirectChain.length - 1),
|
|
286
|
+
error: res.error,
|
|
287
|
+
});
|
|
288
|
+
}
|
|
289
|
+
async function fetchSitemaps(origin, robots, agent, userAgent) {
|
|
290
|
+
const candidates = [
|
|
291
|
+
...robots.sitemaps,
|
|
292
|
+
`${origin}/sitemap.xml`,
|
|
293
|
+
`${origin}/sitemap_index.xml`,
|
|
294
|
+
];
|
|
295
|
+
const seen = new Set();
|
|
296
|
+
const out = [];
|
|
297
|
+
for (const loc of candidates) {
|
|
298
|
+
let url;
|
|
299
|
+
try {
|
|
300
|
+
url = new URL(loc, origin).href;
|
|
301
|
+
}
|
|
302
|
+
catch {
|
|
303
|
+
continue;
|
|
304
|
+
}
|
|
305
|
+
if (seen.has(url))
|
|
306
|
+
continue;
|
|
307
|
+
seen.add(url);
|
|
308
|
+
const res = await fetchUrl(url, {
|
|
309
|
+
agent,
|
|
310
|
+
userAgent,
|
|
311
|
+
timeoutMs: 15_000,
|
|
312
|
+
accept: "application/xml,text/xml,*/*;q=0.1",
|
|
313
|
+
});
|
|
314
|
+
if (res.statusCode === 404 || res.statusCode === null) {
|
|
315
|
+
if (!robots.sitemaps.includes(loc) && !robots.sitemaps.includes(url))
|
|
316
|
+
continue;
|
|
317
|
+
}
|
|
318
|
+
const parsed = await parseSitemapXml(res.decoded, url, url.endsWith(".gz") || (res.headers["content-type"] ?? "").includes("gzip"));
|
|
319
|
+
const sm = {
|
|
320
|
+
...parsed,
|
|
321
|
+
statusCode: res.statusCode,
|
|
322
|
+
contentType: res.contentType,
|
|
323
|
+
error: parsed.error ?? res.error,
|
|
324
|
+
};
|
|
325
|
+
out.push(sm);
|
|
326
|
+
for (const child of sm.childSitemaps) {
|
|
327
|
+
if (!seen.has(child) && out.length < 50)
|
|
328
|
+
candidates.push(child);
|
|
329
|
+
}
|
|
330
|
+
}
|
|
331
|
+
return out;
|
|
332
|
+
}
|
|
333
|
+
function looksLikeHtml(buf) {
|
|
334
|
+
const head = buf.slice(0, 256).toString("utf8").toLowerCase();
|
|
335
|
+
return (head.includes("<html") || head.includes("<!doctype html") || head.includes("<head"));
|
|
336
|
+
}
|
|
337
|
+
function normalizeUrlSafe(url) {
|
|
338
|
+
try {
|
|
339
|
+
return normalizeUrl(url);
|
|
340
|
+
}
|
|
341
|
+
catch {
|
|
342
|
+
return url;
|
|
343
|
+
}
|
|
344
|
+
}
|
|
345
|
+
function sleep(ms) {
|
|
346
|
+
return new Promise((r) => setTimeout(r, ms));
|
|
347
|
+
}
|
|
348
|
+
//# sourceMappingURL=crawl.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"crawl.js","sourceRoot":"","sources":["../src/crawl.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,SAAS,CAAC;AAEzB,OAAO,EAAE,kBAAkB,EAAE,QAAQ,EAAE,oBAAoB,EAAE,MAAM,WAAW,CAAC;AAC/E,OAAO,EAAE,WAAW,EAAE,yBAAyB,EAAE,MAAM,cAAc,CAAC;AACtE,OAAO,EAAE,eAAe,EAAE,MAAM,cAAc,CAAC;AAC/C,OAAO,EAAE,iBAAiB,EAAE,eAAe,EAAE,MAAM,aAAa,CAAC;AACjE,OAAO,EAAE,WAAW,EAAE,MAAM,WAAW,CAAC;AACxC,OAAO,EAAE,UAAU,EAAE,MAAM,cAAc,CAAC;AAC1C,OAAO,EAAE,QAAQ,EAAE,WAAW,EAAE,aAAa,EAAE,YAAY,EAAE,MAAM,UAAU,CAAC;AAC9E,OAAO,EAAE,kBAAkB,EAAE,aAAa,EAAE,MAAM,eAAe,CAAC;AAClE,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AACzC,OAAO,EAAE,WAAW,EAAE,MAAM,YAAY,CAAC;AACzC,OAAO,EAAE,OAAO,EAAE,MAAM,SAAS,CAAC;AAUlC,MAAM,UAAU,cAAc;IAC5B,MAAM,MAAM,GAAG,EAAE,CAAC,OAAO,EAAE,GAAG,CAAC,IAAI,GAAG,IAAI,CAAC,CAAC;IAC5C,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,EAAE,CAAC,IAAI,EAAE,CAAC,MAAM,CAAC,CAAC;IAC5C,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,MAAM,GAAG,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC;IAChD,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,OAAO,IAAI,KAAK,EAAE,KAAK,GAAG,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC;AAChE,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,SAAS,CAAC,OAAqB;IACnD,MAAM,QAAQ,GAAG,YAAY,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC;IAChD,MAAM,MAAM,GAAG,QAAQ,CAAC,QAAQ,CAAC,CAAC;IAClC,MAAM,QAAQ,GAAG,OAAO,CAAC,QAAQ,IAAI,IAAI,CAAC;IAC1C,MAAM,WAAW,GAAG,OAAO,CAAC,WAAW,IAAI,cAAc,EAAE,CAAC;IAC5D,MAAM,GAAG,GAAG,OAAO,CAAC,GAAG,IAAI,CAAC,CAAC;IAC7B,MAAM,MAAM,GAAG,OAAO,CAAC,MAAM,IAAI,IAAI,CAAC;IACtC,MAAM,SAAS,GAAG,OAAO,CAAC,SAAS,IAAI,OAAO,CAAC;IAC/C,MAAM,SAAS,GAAG,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC;IAC3C,MAAM,KAAK,GAAG,kBAAkB,CAAC,CAAC,CAAC,CAAC;IAEpC,MAAM,MAAM,GAAG,MAAM,WAAW,CAAC,MAAM,EAAE,KAAK,EAAE,SAAS,CAAC,CAAC;IAC3D,MAAM,QAAQ,GAAG,MAAM,aAAa,CAAC,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,SAAS,CAAC,CAAC;IAEvE,MAAM,IAAI,GAAG,IAAI,GAAG,CAAS,OAAO,CAAC,MAAM,EAAE,IAAI,IAAI,EAAE,CAAC,CAAC;IACzD,MAAM,KAAK,GAAqC,OAAO,CAAC,MAAM,EAAE,KAAK;QACnE,CAAC,CAAC,OAAO,CAAC,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,GAAG,EAAE,CAAC,CAAC,GAAG,EAAE,KAAK,EAAE,CAAC,CAAC,KAAK,EAAE,CAAC,CAAC;QACnE,CAAC,CAAC,EAAE,CAAC;IACP,MAAM,OAAO,GAAG,CAAC,GAAW,EAAE,KAAa,EAAE,EAAE;QAC7C,IAAI,GAAW,CAAC;QAChB,IAAI,CAAC;YACH,GAAG,GAAG,YAAY,CAAC,GAAG,CAAC,CAAC;QAC1B,CAAC;QAAC,MAAM,CAAC;YACP,OAAO;QACT,CAAC;QACD,IAAI,CAAC,aAAa,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,cAAc;YAAE,OAAO;QACnE,IAAI,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC;YAAE,OAAO;QAC1B,IAAI,CAAC,iBAAiB,CAAC,MAAM,EAAE,GAAG,EAAE,SAAS,CAAC,EAAE,CAAC;YAC/C,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YACd,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YAClB,OAAO;QACT,CAAC;QACD,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QACd,KAAK,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,GAAG,EAAE,KAAK,EAAE,CAAC,CAAC;IAClC,CAAC,CAAC;IAEF,MAAM,OAAO,GAAa,EAAE,CAAC;IAC7B,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC;QACpB,OAAO,CAAC,QAAQ,EAAE,CAAC,CAAC,CAAC;QACrB,KAAK,MAAM,EAAE,IAAI,QAAQ,EAAE,CAAC;YAC1B,KAAK,MAAM,CAAC,IAAI,EAAE,CAAC,IAAI;gBAAE,OAAO,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC;QAC7C,CAAC;IACH,CAAC;IAED,MAAM,WAAW,GAAG,IAAI,GAAG,CACzB,QAAQ,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,gBAAgB,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CACpE,CAAC;IACF,MAAM,KAAK,GAAkB,EAAE,CAAC;IAChC,MAAM,OAAO,GAAG,IAAI,GAAG,EAAkB,CAAC;IAC1C,MAAM,aAAa,GAAG,IAAI,GAAG,EAAkB,CAAC;IAChD,MAAM,WAAW,GAAG,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC;IAC9C,IAAI,SAAS,GAAG,CAAC,CAAC;IAClB,IAAI,SAAS,GAAG,CAAC,CAAC;IAClB,IAAI,MAAM,GAAG,CAAC,CAAC;IAEf,MAAM,QAAQ,GAAG,GAAoB,EAAE,CAAC,CAAC;QACvC,IAAI,EAAE,CAAC,GAAG,IAAI,CAAC;QACf,KAAK,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,GAAG,EAAE,CAAC,CAAC,GAAG,EAAE,KAAK,EAAE,CAAC,CAAC,KAAK,EAAE,CAAC,CAAC;QACzD,SAAS,EAAE,KAAK,CAAC,MAAM;KACxB,CAAC,CAAC;IACH,MAAM,cAAc,GAAG,GAAG,EAAE;QAC1B,MAAM,KAAK,GAAG,OAAO,CAAC,eAAe,IAAI,EAAE,CAAC;QAC5C,IAAI,KAAK,CAAC,MAAM,GAAG,KAAK,KAAK,CAAC;YAAE,OAAO;QACvC,OAAO,CAAC,YAAY,EAAE,CAAC,QAAQ,EAAE,CAAC,CAAC;IACrC,CAAC,CAAC;IACF,IAAI,OAAO,GAAG,KAAK,CAAC;IAEpB,MAAM,OAAO,GAAG,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,WAAW,EAAE,QAAQ,CAAC,EAAE,EAAE,KAAK,IAAI,EAAE;QACjF,OAAO,KAAK,CAAC,MAAM,GAAG,QAAQ,EAAE,CAAC;YAC/B,IAAI,OAAO,CAAC,MAAM,EAAE,OAAO,EAAE,CAAC;gBAC5B,OAAO,GAAG,IAAI,CAAC;gBACf,OAAO;YACT,CAAC;YACD,MAAM,IAAI,GAAG,KAAK,CAAC,KAAK,EAAE,CAAC;YAC3B,IAAI,CAAC,IAAI,EAAE,CAAC;gBACV,IAAI,MAAM,KAAK,CAAC;oBAAE,OAAO;gBACzB,MAAM,KAAK,CAAC,EAAE,CAAC,CAAC;gBAChB,SAAS;YACX,CAAC;YACD,MAAM,EAAE,CAAC;YACT,IAAI,CAAC;gBACH,MAAM,IAAI,GAAG,SAAS,GAAG,WAAW,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;gBAClD,IAAI,IAAI,GAAG,CAAC;oBAAE,MAAM,KAAK,CAAC,IAAI,CAAC,CAAC;gBAChC,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;gBACvB,MAAM,IAAI,GAAG,MAAM,QAAQ,CAAC;oBAC1B,GAAG,EAAE,IAAI,CAAC,GAAG;oBACb,KAAK,EAAE,IAAI,CAAC,KAAK;oBACjB,MAAM;oBACN,KAAK;oBACL,SAAS;oBACT,SAAS,EAAE,OAAO,CAAC,SAAS,IAAI,MAAM;oBACtC,MAAM;oBACN,WAAW;oBACX,MAAM;oBACN,KAAK,EAAE,SAAS,EAAE;oBAClB,KAAK,EAAE,KAAK;oBACZ,QAAQ,EAAE,OAAO,CAAC,QAAQ,IAAI,IAAI;iBACnC,CAAC,CAAC;gBACH,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;gBACjB,cAAc,EAAE,CAAC;gBACjB,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;oBACjB,KAAK,MAAM,IAAI,IAAI,IAAI,CAAC,OAAO,CAAC,KAAK,EAAE,CAAC;wBACtC,IAAI,CAAC,IAAI,CAAC,UAAU;4BAAE,SAAS;wBAC/B,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;wBAC9D,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE,CAAC;4BACrB,aAAa,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC,aAAa,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;wBAC5E,CAAC;wBACD,IAAI,KAAK,CAAC,MAAM,GAAG,KAAK,CAAC,MAAM,GAAG,QAAQ,GAAG,CAAC,EAAE,CAAC;4BAC/C,OAAO,CAAC,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;wBACvC,CAAC;oBACH,CAAC;gBACH,CAAC;YACH,CAAC;oBAAS,CAAC;gBACT,MAAM,EAAE,CAAC;YACX,CAAC;QACH,CAAC;IACH,CAAC,CAAC,CAAC;IAEH,MAAM,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;IAE3B,KAAK,MAAM,CAAC,IAAI,KAAK,EAAE,CAAC;QACtB,CAAC,CAAC,WAAW,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QACxC,CAAC,CAAC,iBAAiB,GAAG,aAAa,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IACtD,CAAC;IAED,IAAI,WAAW,CAAC;IAChB,IAAI,CAAC;QACH,WAAW,GAAG,MAAM,WAAW,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC;IACnD,CAAC;IAAC,MAAM,CAAC;QACP,WAAW,GAAG;YACZ,KAAK,EAAE,QAAQ,CAAC,UAAU,CAAC,QAAQ,CAAC;YACpC,oBAAoB,EAAE,IAAI;YAC1B,QAAQ,EAAE,IAAI;YACd,YAAY,EAAE,IAAI;YAClB,kBAAkB,EAAE,IAAI;YACxB,aAAa,EAAE,KAAK;YACpB,gBAAgB,EAAE,IAAI;YACtB,iBAAiB,EAAE,IAAI;YACvB,WAAW,EAAE,IAAI;YACjB,iBAAiB,EAAE,IAAI;YACvB,SAAS,EAAE,IAAI;YACf,IAAI,EAAE,IAAI;YACV,IAAI,EAAE,IAAI;SACX,CAAC;IACJ,CAAC;IAED,MAAM,KAAK,CAAC,KAAK,EAAE,CAAC;IAEpB,MAAM,UAAU,GAAG,QAAQ,EAAE,CAAC;IAC9B,OAAO,CAAC,YAAY,EAAE,CAAC,UAAU,CAAC,CAAC;IAEnC,OAAO;QACL,MAAM;QACN,QAAQ;QACR,SAAS;QACT,UAAU,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;QACpC,KAAK;QACL,MAAM;QACN,QAAQ;QACR,WAAW;QACX,SAAS,EAAE,KAAK,CAAC,MAAM;QACvB,YAAY,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC,MAAM;QACxD,QAAQ;QACR,SAAS,EAAE,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,IAAI,CAAC,IAAI,GAAG,KAAK,CAAC,MAAM;QACvD,OAAO;QACP,UAAU;KACX,CAAC;AACJ,CAAC;AAED,KAAK,UAAU,QAAQ,CAAC,IAavB;IACC,MAAM,OAAO,GAAG,MAAM,QAAQ,CAAC,IAAI,CAAC,GAAG,EAAE;QACvC,KAAK,EAAE,IAAI,CAAC,KAAK;QACjB,SAAS,EAAE,IAAI,CAAC,SAAS;QACzB,SAAS,EAAE,IAAI,CAAC,SAAS;KAC1B,CAAC,CAAC;IACH,MAAM,MAAM,GACV,iBAAiB,CAAC,IAAI,CAAC,OAAO,CAAC,WAAW,CAAC,IAAI,aAAa,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC;IAChF,MAAM,IAAI,GAAG,MAAM,CAAC,CAAC,CAAC,OAAO,CAAC,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;IAC9D,MAAM,eAAe,GAAG,yBAAyB,CAAC,OAAO,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;IACxE,MAAM,OAAO,GAAG,OAAO,CAAC,OAAO,CAAC,cAAc,CAAC,IAAI,IAAI,CAAC;IACxD,MAAM,OAAO,GAAG,IAAI;QAClB,CAAC,CAAC,WAAW,CAAC,IAAI,EAAE,OAAO,CAAC,QAAQ,EAAE,eAAe,EAAE,OAAO,CAAC;QAC/D,CAAC,CAAC,IAAI,CAAC;IACT,MAAM,IAAI,GAAG,OAAO,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,WAAW,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;IAC1E,MAAM,EAAE,GAAG,OAAO,CAAC,CAAC,CAAC,UAAU,CAAC,OAAO,CAAC,QAAQ,IAAI,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;IAC5E,MAAM,OAAO,GAAG,IAAI,IAAI,OAAO,CAAC,CAAC,CAAC,aAAa,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACnE,MAAM,IAAI,GAAG,WAAW,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IAEnC,IAAI,WAAW,GAAmB,IAAI,CAAC;IACvC,IACE,IAAI,CAAC,QAAQ;QACb,OAAO,CAAC,UAAU,KAAK,GAAG;QAC1B,CAAC,OAAO,CAAC,IAAI,IAAI,OAAO,CAAC,YAAY,CAAC,EACtC,CAAC;QACD,MAAM,KAAK,GAAG,MAAM,QAAQ,CAAC,IAAI,CAAC,GAAG,EAAE;YACrC,KAAK,EAAE,IAAI,CAAC,KAAK;YACjB,SAAS,EAAE,IAAI,CAAC,SAAS;YACzB,SAAS,EAAE,IAAI,CAAC,SAAS;YACzB,IAAI,EAAE,OAAO,CAAC,IAAI,IAAI,SAAS;YAC/B,YAAY,EAAE,OAAO,CAAC,YAAY,IAAI,SAAS;YAC/C,YAAY,EAAE,CAAC;SAChB,CAAC,CAAC;QACH,WAAW,GAAG,KAAK,CAAC,UAAU,KAAK,GAAG,CAAC;IACzC,CAAC;IAED,MAAM,KAAK,GAAgB;QACzB,GAAG,EAAE,IAAI,CAAC,GAAG;QACb,QAAQ,EAAE,OAAO,CAAC,QAAQ;QAC1B,UAAU,EAAE,OAAO,CAAC,UAAU;QAC9B,KAAK,EAAE,OAAO,CAAC,KAAK;QACpB,WAAW,EAAE,OAAO,CAAC,WAAW;QAChC,OAAO,EAAE,OAAO,CAAC,OAAO;QACxB,MAAM,EAAE,OAAO,CAAC,MAAM;QACtB,OAAO,EAAE,OAAO,CAAC,OAAO;QACxB,SAAS,EAAE,OAAO,CAAC,SAAS;QAC5B,YAAY,EAAE,OAAO,CAAC,YAAY;QAClC,WAAW,EAAE,IAAI;QACjB,aAAa,EAAE,OAAO,CAAC,aAAa;QACpC,YAAY,EAAE,OAAO,CAAC,YAAY;QAClC,0BAA0B,EAAE,OAAO,CAAC,0BAA0B;QAC9D,IAAI;QACJ,YAAY,EAAE,IAAI;QAClB,OAAO;QACP,eAAe,EAAE,IAAI;QACrB,WAAW,EAAE,IAAI;QACjB,OAAO,EAAE,EAAE;QACX,IAAI,EAAE,OAAO,CAAC,IAAI;QAClB,YAAY,EAAE,OAAO,CAAC,YAAY;QAClC,WAAW,EAAE,OAAO,CAAC,WAAW;QAChC,WAAW;QACX,SAAS,EAAE,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,OAAO,CAAC,QAAQ,CAAC;QACnF,aAAa,EAAE,iBAAiB,CAAC,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC,GAAG,EAAE,IAAI,CAAC,SAAS,CAAC;QACvE,KAAK,EAAE,IAAI,CAAC,KAAK;QACjB,WAAW,EAAE,CAAC;QACd,YAAY,EAAE,OAAO,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,MAAM,IAAI,CAAC;QACpE,iBAAiB,EAAE,CAAC;QACpB,UAAU,EAAE,aAAa,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,CAAC,MAAM,CAAC;QAChD,SAAS,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;QACnC,YAAY,EAAE,SAAS;QACvB,QAAQ,EAAE,KAAK;QACf,iBAAiB,EAAE,OAAO;QAC1B,WAAW,EAAE,IAAI;KAClB,CAAC;IAEF,IAAI,IAAI,CAAC,MAAM,IAAI,IAAI,IAAI,OAAO,IAAI,OAAO,CAAC,UAAU,KAAK,GAAG,EAAE,CAAC;QACjE,MAAM,MAAM,GAAG,kBAAkB,CAAC,IAAI,CAAC,KAAK,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC;QACjE,KAAK,CAAC,YAAY,GAAG,MAAM,KAAK,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC;QAC3D,IAAI,MAAM,KAAK,QAAQ,IAAI,MAAM,KAAK,QAAQ,EAAE,CAAC;YAC/C,MAAM,QAAQ,GAAG,MAAM,UAAU,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC;YAClD,IAAI,QAAQ,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,QAAQ,CAAC,wBAAwB,CAAC,EAAE,CAAC;gBACpE,KAAK,CAAC,YAAY,GAAG,QAAQ,CAAC,IAAI,CAAC;gBACnC,KAAK,CAAC,eAAe,GAAG,QAAQ,CAAC,OAAO,CAAC;gBACzC,KAAK,CAAC,QAAQ,GAAG,IAAI,CAAC;gBACtB,IAAI,QAAQ,CAAC,QAAQ;oBAAE,KAAK,CAAC,KAAK,GAAG,KAAK,CAAC,KAAK,IAAI,gBAAgB,CAAC;YACvE,CAAC;iBAAM,CAAC;gBACN,KAAK,CAAC,YAAY,GAAG,SAAS,CAAC;YACjC,CAAC;QACH,CAAC;IACH,CAAC;IAED,OAAO,KAAK,CAAC;AACf,CAAC;AAED,KAAK,UAAU,WAAW,CACxB,MAAc,EACd,KAAY,EACZ,SAAiB;IAEjB,MAAM,GAAG,GAAG,GAAG,MAAM,aAAa,CAAC;IACnC,MAAM,GAAG,GAAG,MAAM,QAAQ,CAAC,GAAG,EAAE;QAC9B,KAAK;QACL,SAAS;QACT,SAAS,EAAE,MAAM;QACjB,YAAY,EAAE,oBAAoB;QAClC,MAAM,EAAE,sBAAsB;KAC/B,CAAC,CAAC;IACH,OAAO,eAAe,CAAC;QACrB,UAAU,EAAE,GAAG,CAAC,UAAU;QAC1B,GAAG,EAAE,GAAG,CAAC,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAC;QACjC,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,KAAK,EAAE,GAAG,CAAC,YAAY;QACvB,YAAY,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,GAAG,CAAC,aAAa,CAAC,MAAM,GAAG,CAAC,CAAC;QACvD,KAAK,EAAE,GAAG,CAAC,KAAK;KACjB,CAAC,CAAC;AACL,CAAC;AAED,KAAK,UAAU,aAAa,CAC1B,MAAc,EACd,MAAqB,EACrB,KAAY,EACZ,SAAiB;IAEjB,MAAM,UAAU,GAAG;QACjB,GAAG,MAAM,CAAC,QAAQ;QAClB,GAAG,MAAM,cAAc;QACvB,GAAG,MAAM,oBAAoB;KAC9B,CAAC;IACF,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAC;IAC/B,MAAM,GAAG,GAAoB,EAAE,CAAC;IAChC,KAAK,MAAM,GAAG,IAAI,UAAU,EAAE,CAAC;QAC7B,IAAI,GAAW,CAAC;QAChB,IAAI,CAAC;YACH,GAAG,GAAG,IAAI,GAAG,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC,IAAI,CAAC;QAClC,CAAC;QAAC,MAAM,CAAC;YACP,SAAS;QACX,CAAC;QACD,IAAI,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC;YAAE,SAAS;QAC5B,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QACd,MAAM,GAAG,GAAG,MAAM,QAAQ,CAAC,GAAG,EAAE;YAC9B,KAAK;YACL,SAAS;YACT,SAAS,EAAE,MAAM;YACjB,MAAM,EAAE,oCAAoC;SAC7C,CAAC,CAAC;QACH,IAAI,GAAG,CAAC,UAAU,KAAK,GAAG,IAAI,GAAG,CAAC,UAAU,KAAK,IAAI,EAAE,CAAC;YACtD,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,QAAQ,CAAC,GAAG,CAAC;gBAAE,SAAS;QACjF,CAAC;QACD,MAAM,MAAM,GAAG,MAAM,eAAe,CAClC,GAAG,CAAC,OAAO,EACX,GAAG,EACH,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,cAAc,CAAC,IAAI,EAAE,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,CAC5E,CAAC;QACF,MAAM,EAAE,GAAkB;YACxB,GAAG,MAAM;YACT,UAAU,EAAE,GAAG,CAAC,UAAU;YAC1B,WAAW,EAAE,GAAG,CAAC,WAAW;YAC5B,KAAK,EAAE,MAAM,CAAC,KAAK,IAAI,GAAG,CAAC,KAAK;SACjC,CAAC;QACF,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QACb,KAAK,MAAM,KAAK,IAAI,EAAE,CAAC,aAAa,EAAE,CAAC;YACrC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,GAAG,CAAC,MAAM,GAAG,EAAE;gBAAE,UAAU,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QAClE,CAAC;IACH,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED,SAAS,aAAa,CAAC,GAAW;IAChC,MAAM,IAAI,GAAG,GAAG,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,WAAW,EAAE,CAAC;IAC9D,OAAO,CACL,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,gBAAgB,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,CACpF,CAAC;AACJ,CAAC;AAED,SAAS,gBAAgB,CAAC,GAAW;IACnC,IAAI,CAAC;QACH,OAAO,YAAY,CAAC,GAAG,CAAC,CAAC;IAC3B,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,GAAG,CAAC;IACb,CAAC;AACH,CAAC;AAED,SAAS,KAAK,CAAC,EAAU;IACvB,OAAO,IAAI,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,UAAU,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC;AAC/C,CAAC"}
|
package/dist/decode.d.ts
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* Decode Content-Encoding. undici.request() does not decompress; we must.
|
|
3
|
+
* Apply chained encodings right-to-left.
|
|
4
|
+
*/
|
|
5
|
+
export declare function decodeBody(buf: Buffer, contentEncoding: string | undefined): Promise<Buffer>;
|
|
6
|
+
export declare function acceptEncodingHeader(): string;
|
|
7
|
+
//# sourceMappingURL=decode.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"decode.d.ts","sourceRoot":"","sources":["../src/decode.ts"],"names":[],"mappings":"AAYA;;;GAGG;AACH,wBAAsB,UAAU,CAC9B,GAAG,EAAE,MAAM,EACX,eAAe,EAAE,MAAM,GAAG,SAAS,GAClC,OAAO,CAAC,MAAM,CAAC,CA2BjB;AAED,wBAAgB,oBAAoB,IAAI,MAAM,CAI7C"}
|
package/dist/decode.js
ADDED
|
@@ -0,0 +1,53 @@
|
|
|
1
|
+
import { promisify } from "node:util";
|
|
2
|
+
import zlib from "node:zlib";
|
|
3
|
+
const gunzip = promisify(zlib.gunzip);
|
|
4
|
+
const inflate = promisify(zlib.inflate);
|
|
5
|
+
const inflateRaw = promisify(zlib.inflateRaw);
|
|
6
|
+
const brotli = promisify(zlib.brotliDecompress);
|
|
7
|
+
const zstd = "zstdDecompress" in zlib && typeof zlib.zstdDecompress === "function"
|
|
8
|
+
? promisify(zlib.zstdDecompress)
|
|
9
|
+
: null;
|
|
10
|
+
/**
|
|
11
|
+
* Decode Content-Encoding. undici.request() does not decompress; we must.
|
|
12
|
+
* Apply chained encodings right-to-left.
|
|
13
|
+
*/
|
|
14
|
+
export async function decodeBody(buf, contentEncoding) {
|
|
15
|
+
if (!contentEncoding)
|
|
16
|
+
return buf;
|
|
17
|
+
const encodings = contentEncoding
|
|
18
|
+
.split(",")
|
|
19
|
+
.map((s) => s.trim().toLowerCase())
|
|
20
|
+
.filter((s) => s && s !== "identity");
|
|
21
|
+
let out = buf;
|
|
22
|
+
for (let i = encodings.length - 1; i >= 0; i--) {
|
|
23
|
+
const enc = encodings[i];
|
|
24
|
+
if (enc === "gzip" || enc === "x-gzip") {
|
|
25
|
+
out = Buffer.from(await gunzip(out));
|
|
26
|
+
}
|
|
27
|
+
else if (enc === "deflate") {
|
|
28
|
+
try {
|
|
29
|
+
out = Buffer.from(await inflate(out));
|
|
30
|
+
}
|
|
31
|
+
catch {
|
|
32
|
+
out = Buffer.from(await inflateRaw(out));
|
|
33
|
+
}
|
|
34
|
+
}
|
|
35
|
+
else if (enc === "br") {
|
|
36
|
+
out = Buffer.from(await brotli(out));
|
|
37
|
+
}
|
|
38
|
+
else if (enc === "zstd" || enc === "zst") {
|
|
39
|
+
if (!zstd) {
|
|
40
|
+
throw new Error("zstd Content-Encoding requires Node >= 22.15");
|
|
41
|
+
}
|
|
42
|
+
out = Buffer.from(await zstd(out));
|
|
43
|
+
}
|
|
44
|
+
}
|
|
45
|
+
return out;
|
|
46
|
+
}
|
|
47
|
+
export function acceptEncodingHeader() {
|
|
48
|
+
const parts = ["gzip", "deflate", "br"];
|
|
49
|
+
if (zstd)
|
|
50
|
+
parts.push("zstd");
|
|
51
|
+
return parts.join(", ");
|
|
52
|
+
}
|
|
53
|
+
//# sourceMappingURL=decode.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"decode.js","sourceRoot":"","sources":["../src/decode.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,SAAS,EAAE,MAAM,WAAW,CAAC;AACtC,OAAO,IAAI,MAAM,WAAW,CAAC;AAE7B,MAAM,MAAM,GAAG,SAAS,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;AACtC,MAAM,OAAO,GAAG,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;AACxC,MAAM,UAAU,GAAG,SAAS,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;AAC9C,MAAM,MAAM,GAAG,SAAS,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAC;AAChD,MAAM,IAAI,GACR,gBAAgB,IAAI,IAAI,IAAI,OAAO,IAAI,CAAC,cAAc,KAAK,UAAU;IACnE,CAAC,CAAC,SAAS,CAAC,IAAI,CAAC,cAAc,CAAC;IAChC,CAAC,CAAC,IAAI,CAAC;AAEX;;;GAGG;AACH,MAAM,CAAC,KAAK,UAAU,UAAU,CAC9B,GAAW,EACX,eAAmC;IAEnC,IAAI,CAAC,eAAe;QAAE,OAAO,GAAG,CAAC;IACjC,MAAM,SAAS,GAAG,eAAe;SAC9B,KAAK,CAAC,GAAG,CAAC;SACV,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC;SAClC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,KAAK,UAAU,CAAC,CAAC;IACxC,IAAI,GAAG,GAAW,GAAG,CAAC;IACtB,KAAK,IAAI,CAAC,GAAG,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;QAC/C,MAAM,GAAG,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC;QACzB,IAAI,GAAG,KAAK,MAAM,IAAI,GAAG,KAAK,QAAQ,EAAE,CAAC;YACvC,GAAG,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC;QACvC,CAAC;aAAM,IAAI,GAAG,KAAK,SAAS,EAAE,CAAC;YAC7B,IAAI,CAAC;gBACH,GAAG,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC;YACxC,CAAC;YAAC,MAAM,CAAC;gBACP,GAAG,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC;YAC3C,CAAC;QACH,CAAC;aAAM,IAAI,GAAG,KAAK,IAAI,EAAE,CAAC;YACxB,GAAG,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC;QACvC,CAAC;aAAM,IAAI,GAAG,KAAK,MAAM,IAAI,GAAG,KAAK,KAAK,EAAE,CAAC;YAC3C,IAAI,CAAC,IAAI,EAAE,CAAC;gBACV,MAAM,IAAI,KAAK,CAAC,8CAA8C,CAAC,CAAC;YAClE,CAAC;YACD,GAAG,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;QACrC,CAAC;IACH,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED,MAAM,UAAU,oBAAoB;IAClC,MAAM,KAAK,GAAG,CAAC,MAAM,EAAE,SAAS,EAAE,IAAI,CAAC,CAAC;IACxC,IAAI,IAAI;QAAE,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAC7B,OAAO,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AAC1B,CAAC"}
|
|
@@ -0,0 +1,4 @@
|
|
|
1
|
+
import type { PageExtract } from "./types.js";
|
|
2
|
+
export declare function extractPage(html: string, pageUrl: string, headerCanonical?: string[], xRobots?: string | null): PageExtract;
|
|
3
|
+
export declare function parseLinkHeaderCanonicals(header: string | undefined): string[];
|
|
4
|
+
//# sourceMappingURL=extract.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"extract.d.ts","sourceRoot":"","sources":["../src/extract.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAKV,WAAW,EACZ,MAAM,YAAY,CAAC;AAKpB,wBAAgB,WAAW,CACzB,IAAI,EAAE,MAAM,EACZ,OAAO,EAAE,MAAM,EACf,eAAe,GAAE,MAAM,EAAO,EAC9B,OAAO,GAAE,MAAM,GAAG,IAAW,GAC5B,WAAW,CA4Mb;AAiED,wBAAgB,yBAAyB,CAAC,MAAM,EAAE,MAAM,GAAG,SAAS,GAAG,MAAM,EAAE,CAQ9E"}
|