HTTP::Tiny - A small, simple, correct HTTP/1.1 client
- VERSION
- SYNOPSIS
- DESCRIPTION
- METHODS
- TLS/SSL SUPPORT
- PROXY SUPPORT
- LIMITATIONS
- SEE ALSO
- SUPPORT
- AUTHORS
- CONTRIBUTORS
- COPYRIGHT AND LICENSE
VERSION
version 0.086
SYNOPSIS
use HTTP::Tiny;
my $response = HTTP::Tiny->new->get('http://example.com/');
die "Failed!\n" unless $response->{success};
print "$response->{status} $response->{reason}\n";
while (my ($k, $v) = each %{$response->{headers}}) {
for (ref $v eq 'ARRAY' ? @$v : $v) {
print "$k: $_\n";
}
}
print $response->{content} if length $response->{content};
DESCRIPTION
This is a very simple HTTP/1.1 client, designed for doing simple requests without the overhead of a large framework like LWP::UserAgent.
It is more correct and more complete than HTTP::Lite. It supports proxies and redirection. It also correctly resumes after EINTR.
If IO::Socket::IP 0.25 or later is installed, HTTP::Tiny will use it instead of IO::Socket::INET for transparent support for both IPv4 and IPv6.
Cookie support requires HTTP::CookieJar or an equivalent class.
METHODS
new
$http = HTTP::Tiny->new( %attributes );
This constructor returns a new HTTP::Tiny object. Valid attributes include:
agent
— A user-agent string (defaults to 'HTTP-Tiny/$VERSION'). Ifagent
— ends in a space character, the default user-agent string is appended.cookie_jar
— An instance of HTTP::CookieJar — or equivalent class that supports theadd
andcookie_header
methodsdefault_headers
— A hashref of default headers to apply to requestslocal_address
— The local IP address to bind tokeep_alive
— Whether to reuse the last connection (if for the same scheme, host and port) (defaults to 1)max_redirect
— Maximum number of redirects allowed (defaults to 5)max_size
— Maximum response size in bytes (only when not using a data callback). If defined, requests with responses larger than this will return a 599 status code.http_proxy
— URL of a proxy server to use for HTTP connections (default is$ENV{http_proxy}
— if set)https_proxy
— URL of a proxy server to use for HTTPS connections (default is$ENV{https_proxy}
— if set)proxy
— URL of a generic proxy server for both HTTP and HTTPS connections (default is$ENV{all_proxy}
— if set)no_proxy
— List of domain suffixes that should not be proxied. Must be a comma-separated string or an array reference. (default is$ENV{no_proxy}
—)timeout
— Request timeout in seconds (default is 60) If a socket open, read or write takes longer than the timeout, the request response status code will be 599.verify_SSL
— A boolean that indicates whether to validate the TLS/SSL certificate of anhttps
— connection (default is true). Changed from false to true in version 0.083.SSL_options
— A hashref ofSSL_*
— options to pass through to IO::Socket::SSL$ENV{PERL_HTTP_TINY_SSL_INSECURE_BY_DEFAULT}
- Changes the default certificate verification behavior to not check server identity if set to 1. Only effective ifverify_SSL
is not set. Added in version 0.083.
An accessor/mutator method exists for each attribute.
Passing an explicit undef
for proxy
, http_proxy
or https_proxy
will prevent getting the corresponding proxies from the environment.
Errors during request execution will result in a pseudo-HTTP status code of 599 and a reason of "Internal Exception". The content field in the response will contain the text of the error.
The keep_alive
parameter enables a persistent connection, but only to a single destination scheme, host and port. If any connection-relevant attributes are modified via accessor, or if the process ID or thread ID change, the persistent connection will be dropped. If you want persistent connections across multiple destinations, use multiple HTTP::Tiny objects.
See "SSL SUPPORT" for more on the verify_SSL
and SSL_options
attributes.
get|head|put|post|patch|delete
$response = $http->get($url);
$response = $http->get($url, \%options);
$response = $http->head($url);
These methods are shorthand for calling request()
for the given method. The URL must have unsafe characters escaped and international domain names encoded. See request()
for valid options and a description of the response.
The success
field of the response will be true if the status code is 2XX.
post_form
$response = $http->post_form($url, $form_data);
$response = $http->post_form($url, $form_data, \%options);
This method executes a POST
request and sends the key/value pairs from a form data hash or array reference to the given URL with a content-type
of application/x-www-form-urlencoded
. If data is provided as an array reference, the order is preserved; if provided as a hash reference, the terms are sorted on key and value for consistency. See documentation for the www_form_urlencode
method for details on the encoding.
The URL must have unsafe characters escaped and international domain names encoded. See request()
for valid options and a description of the response. Any content-type
header or content in the options hashref will be ignored.
The success
field of the response will be true if the status code is 2XX.
mirror
$response = $http->mirror($url, $file, \%options)
if ( $response->{success} ) {
print "$file is up to date\n";
}
Executes a GET
request for the URL and saves the response body to the file name provided. The URL must have unsafe characters escaped and international domain names encoded. If the file already exists, the request will include an If-Modified-Since
header with the modification timestamp of the file. You may specify a different If-Modified-Since
header yourself in the $options->{headers}
hash.
The success
field of the response will be true if the status code is 2XX or if the status code is 304 (unmodified).
If the file was modified and the server response includes a properly formatted Last-Modified
header, the file modification time will be updated accordingly.
request
$response = $http->request($method, $url);
$response = $http->request($method, $url, \%options);
Executes an HTTP request of the given method type ('GET', 'HEAD', 'POST', 'PUT', etc.) on the given URL. The URL must have unsafe characters escaped and international domain names encoded.
NOTE: Method names are case-sensitive per the HTTP/1.1 specification. Don't use get
when you really want GET
. See LIMITATIONS for how this applies to redirection.
If the URL includes a "user:password" stanza, they will be used for Basic-style authorization headers. (Authorization headers will not be included in a redirected request.) For example:
$http->request('GET', 'http://Aladdin:open [email protected]/');
If the "user:password" stanza contains reserved characters, they must be percent-escaped:
$http->request('GET', 'http://john%40example.com:[email protected]/');
A hashref of options may be appended to modify the request.
Valid options are:
headers
— A hashref containing headers to include with the request. If the value for a header is an array reference, the header will be output multiple times with each value in the array. These headers over-write any default headers.content
— A scalar to include as the body of the request OR a code reference that will be called iteratively to produce the body of the requesttrailer_callback
— A code reference that will be called if it exists to provide a hashref of trailing headers (only used with chunked transfer-encoding)data_callback
— A code reference that will be called for each chunks of the response body received.peer
— Override host resolution and force all connections to go only to a specific peer address, regardless of the URL of the request. This will include any redirections! This options should be used with extreme caution (e.g. debugging or very special circumstances). It can be given as either a scalar or a code reference that will receive the hostname and whose response will be taken as the address.
The Host
header is generated from the URL in accordance with RFC 2616. It is a fatal error to specify Host
in the headers
option. Other headers may be ignored or overwritten if necessary for transport compliance.
If the content
option is a code reference, it will be called iteratively to provide the content body of the request. It should return the empty string or undef when the iterator is exhausted.
If the content
option is the empty string, no content-type
or content-length
headers will be generated.
If the data_callback
option is provided, it will be called iteratively until the entire response body is received. The first argument will be a string containing a chunk of the response body, the second argument will be the in-progress response hash reference, as described below. (This allows customizing the action of the callback based on the status
or headers
received prior to the content body.)
Content data in the request/response is handled as "raw bytes". Any encoding/decoding (with associated headers) are the responsibility of the caller.
The request
method returns a hashref containing the response. The hashref will have the following keys:
success
— Boolean indicating whether the operation returned a 2XX status codeurl
— URL that provided the response. This is the URL of the request unless there were redirections, in which case it is the last URL queried in a redirection chainstatus
— The HTTP status code of the responsereason
— The response phrase returned by the servercontent
— The body of the response. If the response does not have any content or if a data callback is provided to consume the response body, this will be the empty stringheaders
— A hashref of header fields. All header field names will be normalized to be lower case. If a header is repeated, the value will be an arrayref; it will otherwise be a scalar string containing the valueprotocol
- If this field exists, it is the protocol of the response such as HTTP/1.0 or HTTP/1.1redirects
If this field exists, it is an arrayref of response hash references from redirects in the same order that redirections occurred. If it does not exist, then no redirections occurred.
On an error during the execution of the request, the status
field will contain 599, and the content
field will contain the text of the error.
www_form_urlencode
$params = $http->www_form_urlencode( $data );
$response = $http->get("http://example.com/query?$params");
This method converts the key/value pairs from a data hash or array reference into a x-www-form-urlencoded
string. The keys and values from the data reference will be UTF-8 encoded and escaped per RFC 3986. If a value is an array reference, the key will be repeated with each of the values of the array reference. If data is provided as a hash reference, the key/value pairs in the resulting string will be sorted by key and value for consistent ordering.
can_ssl
$ok = HTTP::Tiny->can_ssl;
($ok, $why) = HTTP::Tiny->can_ssl;
($ok, $why) = $http->can_ssl;
Indicates if SSL support is available. When called as a class object, it checks for the correct version of Net::SSLeay and IO::Socket::SSL. When called as an object methods, if SSL_verify
is true or if SSL_verify_mode
is set in SSL_options
, it checks that a CA file is available.
In scalar context, returns a boolean indicating if SSL is available. In list context, returns the boolean and a (possibly multi-line) string of errors indicating why SSL isn't available.
connected
$host = $http->connected;
($host, $port) = $http->connected;
Indicates if a connection to a peer is being kept alive, per the keep_alive
option.
In scalar context, returns the peer host and port, joined with a colon, or undef
(if no peer is connected). In list context, returns the peer host and port or an empty list (if no peer is connected).
Note: This method cannot reliably be used to discover whether the remote host has closed its end of the socket.
TLS/SSL SUPPORT
Direct https
connections are supported only if IO::Socket::SSL 1.56 or greater and Net::SSLeay 1.49 or greater are installed. An error will occur if new enough versions of these modules are not installed or if the TLS encryption fails. You can also use HTTP::Tiny::can_ssl()
utility function that returns boolean to see if the required modules are installed.
An https
connection may be made via an http
proxy that supports the CONNECT command (i.e. RFC 2817). You may not proxy https
via a proxy that itself requires https
to communicate.
TLS/SSL provides two distinct capabilities:
Encrypted communication channel
Verification of server identity
By default, HTTP::Tiny verifies server identity.
This was changed in version 0.083 due to security concerns. The previous default behavior can be enabled by setting $ENV{PERL_HTTP_TINY_SSL_INSECURE_BY_DEFAULT}
to 1.
Verification is done by checking that that the TLS/SSL connection has a valid certificate corresponding to the host name of the connection and that the certificate has been verified by a CA. Assuming you trust the CA, this will protect against machine-in-the-middle attacks.
Certificate verification requires a file containing trusted CA certificates.
If the environment variable SSL_CERT_FILE
is present, HTTP::Tiny will try to find a CA certificate file in that location.
If the Mozilla::CA module is installed, HTTP::Tiny will use the CA file included with it as a source of trusted CA's.
If that module is not available, then HTTP::Tiny will search several system-specific default locations for a CA certificate file:
/etc/ssl/certs/ca-certificates.crt
/etc/pki/tls/certs/ca-bundle.crt
/etc/ssl/ca-bundle.pem
/etc/openssl/certs/ca-certificates.crt
/etc/ssl/cert.pem
/usr/local/share/certs/ca-root-nss.crt
/etc/pki/tls/cacert.pem
/etc/certs/ca-certificates.crt
An error will be occur if verify_SSL
is true and no CA certificate file is available.
If you desire complete control over TLS/SSL connections, the SSL_options
attribute lets you provide a hash reference that will be passed through to IO::Socket::SSL::start_SSL()
, overriding any options set by HTTP::Tiny. For example, to provide your own trusted CA file:
SSL_options => {
SSL_ca_file => $file_path,
}
The SSL_options
attribute could also be used for such things as providing a client certificate for authentication to a server or controlling the choice of cipher used for the TLS/SSL connection. See IO::Socket::SSL documentation for details.
PROXY SUPPORT
HTTP::Tiny can proxy both http
and https
requests. Only Basic proxy authorization is supported and it must be provided as part of the proxy URL: http://user:[email protected]/
.
HTTP::Tiny supports the following proxy environment variables:
http_proxy or HTTP_PROXY
https_proxy or HTTPS_PROXY
all_proxy or ALL_PROXY
If the REQUEST_METHOD
environment variable is set, then this might be a CGI process and HTTP_PROXY
would be set from the Proxy:
header, which is a security risk. If REQUEST_METHOD
is set, HTTP_PROXY
(the upper case variant only) is ignored, but CGI_HTTP_PROXY
is considered instead.
Tunnelling https
over an http
proxy using the CONNECT method is supported. If your proxy uses https
itself, you can not tunnel https
over it.
Be warned that proxying an https
connection opens you to the risk of a man-in-the-middle attack by the proxy server.
The no_proxy
environment variable is supported in the format of a comma-separated list of domain extensions proxy should not be used for.
Proxy arguments passed to new
will override their corresponding environment variables.
LIMITATIONS
HTTP::Tiny is conditionally compliant with the HTTP/1.1 specifications:
"Message Syntax and Routing" [RFC7230]
"Semantics and Content" [RFC7231]
"Conditional Requests" [RFC7232]
"Range Requests" [RFC7233]
"Caching" [RFC7234]
"Authentication" [RFC7235]
It attempts to meet all "MUST" requirements of the specification, but does not implement all "SHOULD" requirements. (Note: it was developed against the earlier RFC 2616 specification and may not yet meet the revised RFC 7230-7235 spec.) Additionally, HTTP::Tiny supports the PATCH
method of RFC 5789.
Some particular limitations of note include:
HTTP::Tiny focuses on correct transport. Users are responsible for ensuring that user-defined headers and content are compliant with the HTTP/1.1 specification.
Users must ensure that URLs are properly escaped for unsafe characters and that international domain names are properly encoded to ASCII. See URI::Escape, URI::_punycode and Net::IDN::Encode.
Redirection is very strict against the specification. Redirection is only automatic for response codes 301, 302, 307 and 308 if the request method is 'GET' or 'HEAD'. Response code 303 is always converted into a 'GET' redirection, as mandated by the specification. There is no automatic support for status 305 ("Use proxy") redirections.
There is no provision for delaying a request body using an
Expect
header. Unexpected1XX
responses are silently ignored as per the specification.Only 'chunked'
Transfer-Encoding
is supported.There is no support for a Request-URI of '*' for the 'OPTIONS' request.
Headers mentioned in the RFCs and some other, well-known headers are generated with their canonical case. Other headers are sent in the case provided by the user. Except for control headers (which are sent first), headers are sent in arbitrary order.
Despite the limitations listed above, HTTP::Tiny is considered feature-complete. New feature requests should be directed to HTTP::Tiny::UA.
SEE ALSO
HTTP::Tiny::UA - Higher level UA features for HTTP::Tiny
HTTP::Thin - HTTP::Tiny wrapper with HTTP::Request/HTTP::Response compatibility
HTTP::Tiny::Mech - Wrap WWW::Mechanize instance in HTTP::Tiny compatible interface
IO::Socket::IP - Required for IPv6 support
IO::Socket::SSL - Required for SSL support
LWP::UserAgent - If HTTP::Tiny isn't enough for you, this is the "standard" way to do things
Mozilla::CA - Required if you want to validate SSL certificates
Net::SSLeay - Required for SSL support
SUPPORT
Bugs / Feature Requests
Please report any bugs or feature requests through the issue tracker at https://github.com/chansen/p5-http-tiny/issues. You will be notified automatically of any progress on your issue.
Source Code
This is open source software. The code repository is available for public review and contribution under the terms of the license.
https://github.com/chansen/p5-http-tiny
git clone https://github.com/chansen/p5-http-tiny.git
AUTHORS
Christian Hansen <[email protected]>
David Golden <[email protected]>
CONTRIBUTORS
Alan Gardner <[email protected]>
Alessandro Ghedini <[email protected]>
A. Sinan Unur <[email protected]>
Brad Gilbert <[email protected]>
brian m. carlson <[email protected]>
Chris Nehren <[email protected]>
Chris Weyl <[email protected]>
Claes Jakobsson <[email protected]>
Clinton Gormley <[email protected]>
Craig A. Berry <[email protected]>
Craig Berry <[email protected]>
David Golden <[email protected]>
David Mitchell <[email protected]>
Dean Pearce <[email protected]>
Edward Zborowski <[email protected]>
Felipe Gasper <[email protected]>
Graham Knop <[email protected]>
Greg Kennedy <[email protected]>
James E Keenan <[email protected]>
James Raspass <[email protected]>
Jeremy Mates <[email protected]>
Jess Robinson <[email protected]>
Karen Etheridge <[email protected]>
Lukas Eklund <[email protected]>
Martin J. Evans <[email protected]>
Martin-Louis Bright <[email protected]>
Matthew Horsfall <[email protected]>
Michael R. Davis <[email protected]>
Mike Doherty <[email protected]>
Nicolas Rochelemagne <[email protected]>
Olaf Alders <[email protected]>
Olivier Mengué <[email protected]>
Petr Písař <[email protected]>
sanjay-cpu <[email protected]>
Serguei Trouchelle <[email protected]>
Shoichi Kaji <[email protected]>
SkyMarshal <[email protected]>
Sören Kornetzki <[email protected]>
Steve Grazzini <[email protected]>
Stig Palmquist <[email protected]>
Syohei YOSHIDA <[email protected]>
Tatsuhiko Miyagawa <[email protected]>
Tom Hukins <[email protected]>
Tony Cook <[email protected]>
Xavier Guimard <[email protected]>
COPYRIGHT AND LICENSE
This software is copyright (c) 2023 by Christian Hansen.
This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.