package C4::Service; # # Copyright 2008 LibLime # # This file is part of Koha. # # Koha is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 3 of the License, or # (at your option) any later version. # # Koha is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with Koha; if not, see . =head1 NAME C4::Service - functions for JSON webservices. =head1 SYNOPSIS my ( $query, $response) = C4::Service->init( { circulate => 1 } ); my ( $borrowernumber) = C4::Service->require_params( 'borrowernumber' ); C4::Service->return_error( 'internal', 'Frobnication failed', frobnicator => 'foo' ); $response->param( frobnicated => 'You' ); C4::Service->return_success( $response ); =head1 DESCRIPTION This module packages several useful functions for JSON webservices. =cut use strict; use warnings; use CGI qw ( -utf8 ); use C4::Auth qw( check_api_auth ); use C4::Output qw( :ajax ); use C4::Output::JSONStream; use JSON; our $debug; BEGIN { $debug = $ENV{DEBUG} || 0; } our ( $query, $cookie ); sub _output { my ( $response, $status ) = @_; binmode STDOUT, ':encoding(UTF-8)'; if ( $query->param( 'callback' ) ) { output_with_http_headers $query, $cookie, $query->param( 'callback' ) . '(' . $response->output . ');', 'js'; } else { output_with_http_headers $query, $cookie, $response->output, 'json', $status; } } =head1 METHODS =head2 init our ( $query, $response ) = C4::Service->init( %needed_flags ); Initialize the service and check for the permissions in C<%needed_flags>. Also, check that the user is authorized and has a current session, and return an 'auth' error if not. init() returns a C object and a C. The latter can be used for both flat scripts and those that use dispatch(), and should be passed to C. =cut sub init { my ( $class, %needed_flags ) = @_; our $query = CGI->new; my ( $status, $cookie_, $sessionID ) = check_api_auth( $query, \%needed_flags ); our $cookie = $cookie_; # I have no desire to offend the Perl scoping gods $class->return_error( 'auth', $status ) if ( $status ne 'ok' ); return ( $query, C4::Output::JSONStream->new ); } =head2 return_error C4::Service->return_error( $type, $error, %flags ); Exit the script with HTTP status 400, and return a JSON error object. C<$type> should be a short, lower case code for the generic type of error (such as 'auth' or 'input'). C<$error> should be a more specific code giving information on the error. If multiple errors of the same type occurred, they should be joined by '|'; i.e., 'expired|different_ip'. Information in C<$error> does not need to be human-readable, as its formatting should be handled by the client. Any additional information to be given in the response should be passed as param => value pairs. =cut sub return_error { my ( $class, $type, $error, %flags ) = @_; my $response = C4::Output::JSONStream->new; $response->param( message => $error ) if ( $error ); $response->param( type => $type, %flags ); _output( $response, '400 Bad Request' ); exit; } =head2 return_multi C4::Service->return_multi( \@responses, %flags ); return_multi is similar to return_success or return_error, but allows you to return different statuses for several requests sent at once (using HTTP status "207 Multi-Status", much like WebDAV). The toplevel hashref (turned into the JSON response) looks something like this: { multi => JSON::true, responses => \@responses, %flags } Each element of @responses should be either a plain hashref or an arrayref. If it is a hashref, it is sent to the browser as-is. If it is an arrayref, it is assumed to be in the same form as the arguments to return_error, and is turned into an error structure. All key-value pairs %flags are, as stated above, put into the returned JSON structure verbatim. =cut sub return_multi { my ( $class, $responses, @flags ) = @_; my $response = C4::Output::JSONStream->new; if ( !@$responses ) { $class->return_success( $response ); } else { my @responses_formatted; foreach my $response ( @$responses ) { if ( ref( $response ) eq 'ARRAY' ) { my ($type, $error, @error_flags) = @$response; push @responses_formatted, { is_error => JSON::true, type => $type, message => $error, @error_flags }; } else { push @responses_formatted, $response; } } $response->param( 'multi' => JSON::true, responses => \@responses_formatted, @flags ); _output( $response, '207 Multi-Status' ); } exit; } =head2 return_success C4::Service->return_success( $response ); Print out the information in the C C<$response>, then exit with HTTP status 200. =cut sub return_success { my ( $class, $response ) = @_; _output( $response ); } =head2 require_params my @values = C4::Service->require_params( @params ); Check that each of of the parameters specified in @params was sent in the request, then return their values in that order. If a required parameter is not found, send a 'param' error to the browser. =cut sub require_params { my ( $class, @params ) = @_; my @values; for my $param ( @params ) { $class->return_error( 'params', "Missing '$param'" ) if ( !defined( $query->param( $param ) ) ); push @values, scalar $query->param( $param ); # will we ever need multi_param here? } return @values; } =head2 dispatch C4::Service->dispatch( [ $path_regex, \@required_params, \&handler ], ... ); dispatch takes several array-refs, each one describing a 'route', to use the Rails terminology. $path_regex should be a string in regex-form, describing which methods and paths this route handles. Each route is tested in order, from the top down, so put more specific handlers first. Also, the regex is tested on the request method, plus the path. For instance, you might use the route [ 'POST /', ... ] to handle POST requests to your service. Each named parameter in @required_params is tested for to make sure the route matches, but does not raise an error if one is missing; it simply tests the next route. If you would prefer to raise an error, instead use Crequire_params> inside your handler. \&handler is called with each matched group in $path_regex in its arguments. For example, if your service is accessed at the path /blah/123, and you call C with the route [ 'GET /blah/(\\d+)', ... ], your handler will be called with the argument '123'. =cut sub dispatch { my $class = shift; my $path_info = $query->path_info || '/'; ROUTE: foreach my $route ( @_ ) { my ( $path, $params, $handler ) = @$route; next unless ( my @match = ( ($query->request_method . ' ' . $path_info) =~ m,^$path$, ) ); for my $param ( @$params ) { next ROUTE if ( !defined( $query->param ( $param ) ) ); } $debug and warn "Using $path"; $handler->( @match ); return; } $class->return_error( 'no_handler', '' ); } 1; __END__ =head1 AUTHORS Koha Development Team Jesse Weaver