(PHP 5, PHP 7)
curl_multi_select — 等待所有cURL批处理中的活动连接
$mh
[, float $timeout = 1.0
] ) : int阻塞直到cURL批处理连接中有活动连接。
成功时返回描述符集合中描述符的数量。失败时，select失败时返回-1，否则返回超时(从底层的select系统调用).
Before PHP 7.1.23 and 7.2.11, curl_multi_select() would return -1 if cURL had no open file descriptors. Since then, assuming libcurl 7.28 or higher, curl_multi_select() will return 0 if cURL has no open file descriptors and -1 on error.
See https://bugs.php.net/bug.php?id=76480
This is a BC break. See https://bugs.php.net/bug.php?id=77030
Even after so many distro releases, in PHP v5.5.9 and libcurl v7.35.0, curl_multi_select still returns -1. The only work around here is to wait and proceed like nothing ever happened. You have to determine the "wait" required here.
In my application a very small interval like usleep(1) worked. For example:
<?php
// While we're still active, execute curl
$active = null;
do {
$mrc = curl_multi_exec($multi, $active);
} while ($mrc == CURLM_CALL_MULTI_PERFORM);
while ($active && $mrc == CURLM_OK) {
// Wait for activity on any curl-connection
if (curl_multi_select($multi) == -1) {
usleep(1);
}
// Continue to exec until curl is ready to
// give us more data
do {
$mrc = curl_multi_exec($multi, $active);
} while ($mrc == CURLM_CALL_MULTI_PERFORM);
}
?>
Internally php curl_multi_select uses libcurl curl_multi_fdset function and its libcurl documentation says :
http://curl.haxx.se/libcurl/c/curl_multi_fdset.html
"When libcurl returns -1 in max_fd, it is because libcurl currently does something that isn't possible for your application to monitor with a socket and unfortunately you can then not know exactly when the current action is completed using select(). When max_fd returns with -1, you need to wait a while and then proceed and call curl_multi_perform anyway. How long to wait? I would suggest 100 milliseconds at least, but you may want to test it out in your own particular conditions to find a suitable value.
When doing select(), you should use curl_multi_timeout to figure out how long to wait for action."
Untill PHP implements curl_multi_timeout() we have to play with our application and determine the "wait".
According to https://bugs.php.net/bug.php?id=61141:
On Windows setups using libcurl version 7.24 or later (which seems to correspond to PHP 5.3.10 or later), you may find that this always returns -1. This is, apparently, not strictly a bug: according to the libcurl documentation, you should add your own sleep if curl_multi_select returns -1.
For example:
<?php
/* setup $mh */
$active = null;
do {
$mrc = curl_multi_exec($mh, $active);
} while ($mrc == CURLM_CALL_MULTI_PERFORM);
while ($active && $mrc == CURLM_OK) {
if (curl_multi_select($mh) == -1) {
usleep(100);
}
do {
$mrc = curl_multi_exec($mh, $active);
} while ($mrc == CURLM_CALL_MULTI_PERFORM);
}
/* close $mh */
?>
In 5.3.9+, curl_multi_select always returns -1. If this is your case, just wait a bit and then proceed like nothing ever happened:
<?php
do {
$mrc = curl_multi_exec($multi, $active);
} while ($mrc == CURLM_CALL_MULTI_PERFORM);
while ($active && $mrc == CURLM_OK) {
//check for results and execute until everything is done
if (curl_multi_select($multi) == -1) {
//if it returns -1, wait a bit, but go forward anyways!
usleep(100);
}
//do something with the return values
while(($info = curl_multi_info_read($multi)) !== false){
if ($info["result"] == CURLE_OK){
$content = curl_multi_getcontent($info["handle"]);
do_something($content);
}
}
do {
$mrc = curl_multi_exec($multi, $active);
} while ($mrc == CURLM_CALL_MULTI_PERFORM);
}
?>
On php 5.3.18+ be aware that curl_multi_select() may return -1 forever until you call curl_multi_exec().
See https://bugs.php.net/bug.php?id=63411 for more information.
curl_multi_select($mh, $timeout) simply blocks for $timeout seconds while curl_multi_exec() returns CURLM_CALL_MULTI_PERFORM. Otherwise, it works as intended, and blocks until at least one connection has completed or $timeout seconds, whatever happens first.
For that reason, curl_multi_exec() should always be wrapped:
<?php
function full_curl_multi_exec($mh, &$still_running) {
do {
$rv = curl_multi_exec($mh, $still_running);
} while ($rv == CURLM_CALL_MULTI_PERFORM);
return $rv;
}
?>
With that, the core of "multi" processing becomes (ignoring error handling for brevity):
<?php
full_curl_multi_exec($mh, $still_running); // start requests
do { // "wait for completion"-loop
curl_multi_select($mh); // non-busy (!) wait for state change
full_curl_multi_exec($mh, $still_running); // get new state
while ($info = curl_multi_info_read($mh)) {
// process completed request (e.g. curl_multi_getcontent($info['handle']))
}
} while ($still_running);
?>
Note that after starting requests, retrieval is done in the background - one of the better shots at parallel processing in PHP.
Since the docs are still lacking, here's an example of how to use the function. The following code will keep checking all active threads until one of them returns the HTTP 200 Ok status code, or simply end. On success, it will return the URL that worked.
<?php
$running=null;
do {
curl_multi_exec($mh,$running);
$ready=curl_multi_select($mh); // this will pause the loop
if($ready>0){
while($info=curl_multi_info_read($mh)){
$status=curl_getinfo($info['handle'],CURLINFO_HTTP_CODE);
if($status==200){
$successUrl=curl_getinfo($info['handle'],CURLINFO_EFFECTIVE_URL);
break 2;
}
}
}
} while ($running>0 && $ready!=-1);
?>
The question for the $ready variable is, if it will return the value before or after the timeout has occurred. From my tests it appears that it will return the value immediately, and only then pause the execution. That's because it's always zero the first time through the loop, even if the time limit is as high as 10sec. I've expected it to wait and only then return the value so that came unexpected to me.
As stated by someone else, it also doesn't seem to return the overall count of threads in the handle, but only that of the currently active ones.
This function blocks the calling process until there is activity on any of the connections opened by the curl_multi interface, or until the timeout period has expired.
In other words, it waits for data to be received in the opened connections.
Internally it fetches socket pointers with "curl_multi_fdset()" and runs the "select()" C function.
It returns in 3 cases:
1. Activity is detected on any socket;
2. Timeout has ended (second parameter);
3. Process received any signal (#man kill).
The function returns an integer:
* In case of activity it returns a number, usually 1.
I suppose, it returns the number of connections with activity detected.
* If timeout expires it returns 0
* In case of error it returns -1
Thanks for attention, hope this helps.