Why does an incomplete page come up when using curl + proxy?

Question

PHP

HTTP experts, help with a problem!
I use curl (in PHP) through a proxy (no access to a proxy) to get the content of the site (HTTP proxy). Very often incomplete pages arrive, break off at the semi-tag (for example
...<p class="myclas, the content ends further). There are no regularities when and how the content will break. Moreover, if you do not receive through a proxy, there is no problem. I searched for solutions on the Internet, found the following solutions:
1) I updated the version of curl to 7.36 - it did not help.
2) Sending header "Expect:" - didn't help.
3) Timeout set to 0 - did not help.
4) Set the CURLOPT_HTTP_VERSION option to CURL_HTTP_VERSION_1_0 - if the response comes in HTTP 1.1 format, it still breaks.
5) Received content bypassing CURLOPT_RETURNTRANSFER = 1 via ob_start and so on - didn't help.
6) CURLOPT_HTTP_TRANSFER_DECODING = 0 and CURLOPT_HTTP_CONTENT_DECODING = 0 - did not help.
If you enable VERBOSE mode in curl, then on problem pages it says
...
nread <= 0, server closed connection, bailing
curl: (18) transfer closed with outstanding read data remaining
...
Also, the problem is observed only with Transfer-Encoding: chunked . Seen on many sites.
How do you still get the content completely, or, if this is not possible, how to determine that the content came incomplete?

Answer 1

is your proxy controlled?
determine what is incomplete