1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
use std::sync::Arc;
use tokio::task::JoinSet;
use tracing::{error, info};
use crate::{
error::{ErrorPolicy, KumoError},
frontier::{Frontier, memory::MemoryFrontier},
middleware::Middleware,
pipeline::Pipeline,
request::{CrawlRequest, FrontierRequest},
scheduler::{CrawlScheduler, SchedulerPoll},
spider::Spider,
stats::{CrawlStats, domain_key},
};
use super::{
builder::CrawlEngine,
erased::{ErasedSpider, SpiderErased},
setup::{
FetcherArgs, build_http_client, build_raw_fetcher, build_robots_cache, wrap_with_cache,
},
task::{TaskContext, is_cancelled, process_request_once, should_enqueue},
};
impl CrawlEngine {
/// Consume the engine, run the spider, and return crawl statistics.
pub async fn run<S>(self, spider: S) -> Result<CrawlStats, KumoError>
where
S: Spider + 'static,
{
let start = std::time::Instant::now();
let metrics_interval = self.metrics_interval;
let stream_cancelled = self.stream_cancelled.clone();
let spider: Arc<dyn ErasedSpider> = Arc::new(SpiderErased(spider));
let frontier: Arc<dyn Frontier> = self
.frontier
.unwrap_or_else(|| Arc::new(MemoryFrontier::new(self.max_urls)));
let scheduler = CrawlScheduler::from_arc(frontier, self.politeness_policy)
.with_fingerprint_policy(self.fingerprint_policy);
let store = self
.store
.unwrap_or_else(|| Arc::new(crate::store::stdout::StdoutStore));
let middleware: Arc<Vec<Arc<dyn Middleware>>> = Arc::new(self.middleware);
let pipelines: Arc<Vec<Arc<dyn Pipeline>>> = Arc::new(self.pipelines);
// Warn if both AutoThrottle and RateLimiter are registered — they compound delays.
{
let has_throttle = middleware
.iter()
.any(|mw| std::any::type_name_of_val(mw.as_ref()).contains("AutoThrottle"));
let has_limiter = middleware
.iter()
.any(|mw| std::any::type_name_of_val(mw.as_ref()).contains("RateLimiter"));
if has_throttle && has_limiter {
tracing::warn!(
"Both AutoThrottle and RateLimiter are registered. \
They apply delays independently and will compound. \
Consider using only one."
);
}
}
let concurrency = self.concurrency;
let retry_policy = self.retry_policy;
let robots_cache = build_robots_cache(self.respect_robots, self.robots_ttl);
let client =
build_http_client(concurrency, self.request_timeout, self.http_client_builder)?;
let fetcher = build_raw_fetcher(FetcherArgs {
fetcher_override: self.fetcher_override,
client: client.clone(),
concurrency,
#[cfg(feature = "stealth")]
stealth_profile: self.stealth_profile,
#[cfg(feature = "browser")]
browser: self.browser,
})
.await?;
let fetcher = wrap_with_cache(fetcher, self.cache_dir, self.cache_ttl)?;
let mut stats = CrawlStats::default();
spider.open().await?;
let start_urls = spider.start_urls();
info!(
spider = spider.name(),
start_urls = start_urls.len(),
"spider.open"
);
for url in start_urls {
let domain = domain_key(&url);
if scheduler.push_request(CrawlRequest::get(url), 0).await {
stats.record_scheduled(&domain);
} else {
stats.record_deduped(&domain);
}
}
type TaskResult = (
FrontierRequest,
Result<(u64, u64, Vec<(CrawlRequest, usize)>), KumoError>,
);
let mut join_set: JoinSet<TaskResult> = JoinSet::new();
// Spawn periodic metrics logger if configured.
let live_stats = Arc::new(tokio::sync::Mutex::new(CrawlStats::default()));
let _metrics_task = metrics_interval.map(|interval| {
let live = live_stats.clone();
tokio::spawn(async move {
loop {
tokio::time::sleep(interval).await;
let s = live.lock().await;
tracing::info!(
pages = s.pages_crawled,
items = s.items_scraped,
errors = s.errors,
bytes = s.bytes_downloaded,
elapsed_secs = s.duration.as_secs_f64(),
"[kumo metrics]"
);
}
})
});
let shutdown = async {
#[cfg(not(target_arch = "wasm32"))]
{
tokio::signal::ctrl_c().await.ok();
tracing::info!("ctrl-c received — finishing in-flight tasks then exiting");
}
#[cfg(target_arch = "wasm32")]
std::future::pending::<()>().await
};
tokio::pin!(shutdown);
let mut shutting_down = false;
loop {
if is_cancelled(&stream_cancelled) {
shutting_down = true;
stats.interrupted = true;
}
let mut next_scheduler_wait: Option<std::time::Duration> = None;
if !shutting_down {
// Fill up to the concurrency limit.
while join_set.len() < concurrency {
match scheduler.poll_ready().await {
SchedulerPoll::Ready(queued) => {
let queued = *queued;
// Check robots.txt before dispatching.
if let Some(ref cache) = robots_cache
&& !cache.is_allowed(&client, queued.request.url()).await
{
tracing::debug!(url = %queued.request.url(), "blocked by robots.txt, skipping");
stats.record_robots_blocked(&domain_key(queued.request.url()));
scheduler.finish(&queued).await;
continue;
}
let ctx = TaskContext {
spider: spider.clone(),
store: store.clone(),
middleware: middleware.clone(),
pipelines: pipelines.clone(),
fetcher: fetcher.clone(),
stream_cancelled: stream_cancelled.clone(),
};
join_set.spawn(async move {
let result = process_request_once(queued.clone(), ctx).await;
(queued, result)
});
}
SchedulerPoll::Pending(wait) => {
next_scheduler_wait =
Some(next_scheduler_wait.map_or(wait, |current| current.min(wait)));
break;
}
// Frontier currently empty — tasks may still add URLs.
SchedulerPoll::Empty => break,
}
}
}
if join_set.is_empty() {
if scheduler.is_empty().await {
break;
}
tokio::time::sleep(
next_scheduler_wait.unwrap_or(std::time::Duration::from_millis(10)),
)
.await;
continue;
}
let scheduler_sleep = tokio::time::sleep(
next_scheduler_wait.unwrap_or(std::time::Duration::from_secs(24 * 60 * 60)),
);
tokio::pin!(scheduler_sleep);
tokio::select! {
_ = &mut scheduler_sleep, if next_scheduler_wait.is_some() => {
continue;
}
_ = &mut shutdown, if !shutting_down => {
shutting_down = true;
stats.interrupted = true;
}
result = join_set.join_next() => {
match result {
Some(Ok((queued, Ok((item_count, bytes, follows))))) => {
scheduler.finish(&queued).await;
stats.record_completed(&domain_key(queued.request.url()));
stats.pages_crawled += 1;
stats.items_scraped += item_count;
stats.bytes_downloaded += bytes;
if is_cancelled(&stream_cancelled) {
shutting_down = true;
stats.interrupted = true;
}
// Keep live snapshot up to date for the metrics task.
if metrics_interval.is_some() {
let mut snap = live_stats.lock().await;
*snap = stats.clone();
snap.duration = start.elapsed();
}
if !shutting_down {
for (follow_request, follow_depth) in follows {
if should_enqueue(&follow_request, follow_depth, spider.as_ref()) {
let domain = domain_key(follow_request.url());
if scheduler.push_request(follow_request, follow_depth).await {
stats.record_scheduled(&domain);
} else {
stats.record_deduped(&domain);
}
}
}
}
}
Some(Ok((queued, Err(e)))) => {
scheduler.finish(&queued).await;
let url = queued.request.url().to_string();
// Notify all middleware of the permanent failure.
for mw in middleware.iter() {
mw.on_error(&url, &e).await;
}
if !shutting_down
&& queued.retry_count < retry_policy.max_attempts
&& retry_policy.is_retriable(&e)
{
let delay = retry_policy.delay_for(queued.retry_count);
stats.record_retry(&domain_key(&url));
tracing::warn!(
spider = spider.name(),
url = %url,
attempt = queued.retry_count + 1,
max = retry_policy.max_attempts,
retry_in_ms = delay.as_millis(),
error = %e,
"scheduling retry"
);
scheduler
.push_request_force(
FrontierRequest::new(
queued.request,
queued.depth,
queued.retry_count + 1,
)
.scheduled_after(delay),
)
.await;
continue;
}
stats.errors += 1;
stats.record_failed(&domain_key(&url));
match spider.on_error(&url, &e) {
ErrorPolicy::Abort => {
error!(url = %url, error = %e, "aborting crawl");
return Err(e);
}
ErrorPolicy::Retry(max) if queued.retry_count < max => {
tracing::warn!(
spider = spider.name(),
url = %url,
attempt = queued.retry_count + 1,
max,
error = %e,
"re-queuing failed URL"
);
if !shutting_down {
stats.record_retry(&domain_key(&url));
scheduler.push_request_force(FrontierRequest::new(
queued.request,
queued.depth,
queued.retry_count + 1,
)).await;
}
}
ErrorPolicy::Retry(_) => {
tracing::warn!(spider = spider.name(), url = %url, error = %e, "fetch.skip.retry_exhausted");
}
ErrorPolicy::Skip => {
tracing::warn!(spider = spider.name(), url = %url, error = %e, "fetch.skip");
}
}
}
Some(Err(join_err)) => {
stats.errors += 1;
error!(spider = spider.name(), error = %join_err, "crawl task panicked");
}
None => break,
}
if shutting_down && join_set.is_empty() {
break;
}
}
}
}
scheduler.flush().await?;
store.flush().await?;
stats.duration = start.elapsed();
// close() errors are intentionally not propagated — the crawl and store
// flush completed successfully. Cleanup failures are logged only.
if let Err(e) = spider.close(&stats).await {
tracing::error!(error = %e, "spider::close failed");
}
let rps = if stats.duration.as_secs_f64() > 0.0 {
stats.pages_crawled as f64 / stats.duration.as_secs_f64()
} else {
0.0
};
info!(
pages = stats.pages_crawled,
items = stats.items_scraped,
errors = stats.errors,
bytes = stats.bytes_downloaded,
duration_secs = stats.duration.as_secs_f64(),
pages_per_sec = format!("{rps:.1}"),
interrupted = stats.interrupted,
"crawl complete"
);
Ok(stats)
}
}