Test2::Tools::HTTP - Test HTTP / PSGI
version 0.12
use Test2::V0;
use Test2::Tools::HTTP;
use HTTP::Request::Common;
psgi_add_app sub { [ 200, [ 'Content-Type' => 'text/plain;charset=utf-8' ], [ "Test Document\n" ] ] };
# Internally test the app from within the .t file itself
http_request(
# if no host/port/protocol is given then
# the default PSGI app above is assumed
GET('/'),
http_response {
http_code 200;
# http_response {} is a subclass of object {}
# for HTTP::Response objects only, so you can
# also use object {} style comparisons:
call code => 200;
http_content_type match qr/^text\/(html|plain)$/;
http_content_type_charset 'UTF-8';
http_content match qr/Test/;
}
);
use Test2::Tools::JSON::Pointer;
# test an external website
http_request(
# you can also test against a real HTTP server
GET('http://example.test'),
http_response {
http_is_success;
# JSON pointer { "key":"val" }
http_content json '/key' => 'val';
}
);
done_testing;
with short names:
use Test2::Tools::HTTP ':short';
use HTTP::Request::Common;
app_add { [ 200, [ 'Content-Type => 'text/plain' ], [ "Test Document\n" ] ] };
req (
GET('/'),
res {
code 200;
message 'OK';
content_type 'text/plain';
content match qr/Test/;
},
);
done_testing;
This module provides an interface for testing websites and PSGI based apps with a Test2 style comparisons interface. You can specify a PSGI app with a URL and responses from that URL will automatically be routed to that app, without having to actually need a separate server process. Requests to URLs that haven't been registered will be made against the actual networks servers as appropriate. You can also use the user agent returned from http_ua to make requests against PSGI apps. LWP::UserAgent is the user agent used by default, but it is possible to use others assuming an appropriate user agent wrapper class is available (Test2::Tools::HTTP::UA).
By default it uses long function names with either a http_ or psgi_app_ prefix. The intent is to make the module usable when you are importing lots of symbols from lots of different testing tools while reducing the chance of name collisions. You can instead import :short which will give you the most commonly used tools with short names. The short names are indicated below in square brackets, and were chosen to not conflict with Test2::V0.
http_request($request); http_request($request, $check); http_request($request, $check, $message); http_request([$request, %options], ... );
Make a HTTP request. If there is a client level error then it will fail immediately. Otherwise you can use a object {} or http_request comparison check to inspect the HTTP response and ensure that it matches what you expect. By default only one request is made. If the response is a forward (has a Location header) you can use the http_tx-location> method to make the next request.
Options:
This allows the user agent to follow redirects.
my $check = http_response {
... # object or http checks
};
This is a comparison check specific to HTTP::Response objects. You may include these subchecks:
http_response {
http_code $check;
};
The HTTP status code should match the given check.
http_response {
http_message $check;
};
The HTTP status message ('OK' for 200, 'Not Found' for 404, etc) should match the given check.
http_response {
http_content $check;
};
The response body content. Attempt to decode using the HTTP::Message method decoded_content, otherwise use the raw response body. If you want specifically the decoded content or the raw content you can use call to specifically check against them:
http_response {
call content => $check1;
call decoded_content => $check2;
};
http_response {
http_is_info;
http_is_success;
http_is_redirect;
http_is_error;
http_is_client_error;
http_is_server_error;
};
Checks that the response is of the specified type. See HTTP::Status for the meaning of each of these.
http_response {
http_isnt_info;
http_isnt_success;
http_isnt_redirect;
http_isnt_error;
http_isnt_client_error;
http_isnt_server_error;
};
Checks that the response is NOT of the specified type. See HTTP::Status for the meaning of each of these.
http_response {
http_headers $check;
};
Check the HTTP headers as converted into a Perl hash. If the same header appears twice, then the values are joined together using the , character. Example:
http_request(
GET('http://example.test'),
http_response {
http_headers hash {
field 'Content-Type' => 'text/plain;charset=utf-8';
etc;
};
},
);
http_response {
http_header $name, $check;
};
Check an HTTP header against the given check. Can be used with either scalar or array checks. In scalar mode, any list values will be joined with , character. Example:
http_request(
GET('http://example.test'),
http_response {
# single value
http_header 'X-Foo', 'Bar';
# list as scalar, will match either:
# X-Foo: A
# X-Foo: B
# or
# X-Foo: A,B
http_header 'X-Foo', 'A,B';
# list mode, with an array ref:
http_header 'X-Foo', ['A','B'];
# list mode, with an array check:
http_header 'X-Foo', array { item 'A'; item 'B' };
},
);
http_response {
http_content_type $check;
http_content_type_charset $check;
};
Check that the Content-Type header matches the given checks. http_content_type checks just the content type, not the character set, and http_content_type_charset matches just the character set. Hence:
http_response {
http_content_type 'text/html';
http_content_type_charset 'UTF-8';
};
http_response {
http_content_length $check;
};
Check that the Content-Length header matches the given check.
http_response {
http_content_length_ok;
};
Checks that the Content-Length header matches the actual length of the content.
http_response {
http_location $check;
http_location_uri $check;
};
Check the Location HTTP header. The http_location_uri variant converts Location to a URI using the base URL of the response so that it can be tested with Test2::Tools::URL.
my $req = http_tx->req; my $res = http_tx->res; my $bool = http_tx->ok; my $string = http_tx->connection_error; my $url = http_tx->location; http_tx->note; http_tx->diag;
This returns the most recent transaction object, which you can use to get the last request, response and status information related to the most recent http_request.
The HTTP::Request object.
The HTTP::Response object.
Warning: Depending on the user agent class in use, in the case of a connection error, this may be either a synthetic response or not defined. For example LWP::UserAgent produced a synthetic response, while Mojo::UserAgent does not produce a response in the event of a connection error.
True if the most recent call to http_request passed.
The connection error if any from the most recent http_reequest.
The Location header converted to an absolute URL, if included in the response.
Send the request, response and ok to Test2's "note" output. Note that the message bodies may be decoded, but the headers will not be modified.
Send the request, response and ok to Test2's "diag" output. Note that the message bodies may be decoded, but the headers will not be modified.
http_base_url($url); my $url = http_base_url;
Sets the base URL for all requests made by http_request. This is used if you do not provide a fully qualified URL. For example:
http_base_url 'http://httpbin.org';
http_request(
GET('/status/200') # actually makes a request against http://httpbin.org
);
If you use psgi_add_app without a URL, then this is the URL which will be used to access your app. If you do not specify a base URL, then localhost with a random unused port will be picked.
http_ua(LWP::UserAgent->new); my $ua = http_ua;
Gets/sets the LWP::UserAgent object used to make requests against real web servers. For tests against a PSGI app, this will NOT be used. If not provided, the default LWP::UserAgent will call env_proxy and add an in-memory cookie jar.
psgi_app_add $app; psgi_app_add $url, $app;
Add the given PSGI app to the testing environment. If you provide a URL, then requests to that URL will be intercepted by http_request and routed to the app instead of making a real HTTP request via LWP::UserAgent.
psgi_app_del; psgi_app_del $url;
Remove the app at the given (or default) URL.
my $guard = psgi_app_guard $app; my $guard = psgi_app_guard $url, $app; my $guard = psgi_app_guard $url, $app, ...;
Similar to psgi_app_add except a guard object is returned. When the guard object falls out of scope, the old apps are restored automatically. The intent is for this to be used in subtests or other scoped blocks to temporarily override the internet or other PSGI apps.
psgi_add_add 'http://foo.test' => sub { ... };
subtest 'mysubtest' => sub {
my $guard = psgi_app_guard
'http://foo.test' => sub { ... },
'https://www.google.com' => sub { ... };
http_request
# gets the foo.test for this scope.
GET('http://foo.test'),
http_response {
...
};
http_request
# gets the mock google
GET('https://www.google.com'),
http_response {
...;
};
};
http_request
# gets the original foo.test mock
GET('http://foo.test'),
http_response {
...;
};
http_request
# gets the real google
GET('https://www.google.com'),
http_response {
...;
};
Because calling a function that returns a guard in void context is usually a mistake, this function will throw an exception if you attempt to call it in void context.
This is a very capable web application testing module. Definitely worth checking out, even if you aren't developing a Mojolicious app since it can be used (with Test::Mojo::Role::PSGI) to test any PSGI application.
Also allows you to make HTTP::Request requests against a PSGI app and get the appropriate HTTP::Response response back. Doesn't provide any special tools for interrogating that response. This module in fact uses this one internally.
This is a subclass of LWP::UserAgent that can return responses from a local PSGI app, similar to the way this module instruments an instance of LWP::UserAgent for similar purposes. The limitation to this approach is that it cannot be used with classes which cannot be used with subclasses of LWP::UserAgent. By contrast, this module can instrument an existing LWP::UserAgent object without having to rebless it into another class or other such shenanigans. If you can at least get access to another class's user agent instance, it can be used with Test2::Tools::HTTP's mock website system. Doesn't work with anything that is not an LWP::UserAgent object.
Provides a similar functionality to Test::LWP::UserAgent, but registers apps globally using LWP::Protocol so that you do not need access to a specific LWP::UserAgent object. Does not work with anything that is not an LWP::UserAgent object. Test2::Tools::HTTP::UA provides similar functionality, but is an abstraction layer which can be used with any appropriately adapted user agent class or instance, although we use LWP::UserAgent by default. Support for Mojo::UserAgent and HTTP::AnyUA is available, although not bundled with this distribution. One advantage of this abstraction is that it can be used to instrument either a single instance or all objects belonging to a particular class.
Graham Ollis <plicease@cpan.org>
This software is copyright (c) 2018-2022 by Graham Ollis.
This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.