NAME

WebService::GoogleAPI::Client - Google API Discovery and SDK

VERSION

version 0.24_001

SYNOPSIS

Access Google API Services using an OAUTH2 User Agent.

Includes Discovery, validation, authentication and API Access.

By default assumes gapi.json configuration in working directory with scoped Google project credentials and user authorization created by goauth.

    use WebService::GoogleAPI::Client;
    
    my $gapi_client = WebService::GoogleAPI::Client->new( debug => 1, gapi_json => 'gapi.json', user=> 'peter@pscott.com.au' );
    
    say $gapi_client->list_of_available_google_api_ids();

    my @gmail_endpoint_list =      $gapi_client->methods_available_for_google_api_id('gmail')

    if $gapi_agent->has_scope_to_access_api_endpoint( 'gmail.users.settings.sendAs.get' ) {
      say 'User has Access to GMail Method End-Point gmail.users.settings.sendAs.get';
    }

Package includes goauth CLI Script to collect initial end-user authorisation to scoped services.

Note to intrepid hackers: Any method that isn't documented is considered private, and subject to change in breaking ways without notice. (Although I'm a pretty nice guy, and probably will leave a warning or something).

EXAMPLES

AUTOMATIC API REQUEST CONSTRUCTION - SEND EMAL

    ## using dotted API Endpoint id to invoke helper validation and default value interpolations etc to send email to self
    use Email::Simple;    ## RFC2822 formatted messages
    use MIME::Base64;
    my $my_email_address = 'peter@shotgundriver.com'

    my $raw_email_payload = encode_base64(
      Email::Simple->create(
        header => [
          To      => $my_email_address,
          From    => $my_email_address,
          Subject => "Test email from '$my_email_address' ",
        ],
        body => "This is the body of email to '$my_email_address'",
      )->as_string
    );

    $gapi_client->api_query(
      api_endpoint_id => 'gmail.users.messages.send',
      options         => { raw => $raw_email_payload },
    );

MANUAL API REQUEST CONSTRUCTION - GET CALENDAR LIST

    ## Completely manually constructed API End-Point Request to obtain Perl Data Structure converted from JSON response.
    my $res = $gapi_client->api_query(
      method => 'get',
      path => 'https://www.googleapis.com/calendar/users/me/calendarList',
    )->json;

METHODS

new

  WebService::GoogleAPI::Client->new(
     user => 'peter@pscott.com.au', gapi_json => '/fullpath/gapi.json' );

General parameters

debug

if truthy then diagnostics are send to STDERR - default false. Crank it up to 11 for maximal debug output

chi

an instance to a CHI persistent storage case object - if none provided FILE is used

Login Parameters

You can use either gapi_json, which is the file you get from using the bundled goauth tool, or service_account which is the json file you can download from https://console.cloud.google.com/iam-admin/serviceaccounts.

service_account and gapi_json are mutually exclusive, and gapi_json takes precedence.

If nothing is passed, then we check the GOOGLE_APPLICATION_CREDENTIALS env variable for the location of a service account file. This matches the functionality of the Google Cloud libraries from other languages (well, somewhat. I haven't fully implemented ADC yet - see Google's Docs for some details. PRs are welcome!)

If that doesn't exist, then we default to gapi.json in the current directory.

Be wary! This default is subject to change as more storage backends are implemented. A deprecation warning will be emmitted when this is likely to start happening.

For more advanced usage, you can supply your own auth storage instance, which is a consumer of the WebService::GoogleAPI::Client::AuthStorage role. See the POD for that module for more information.

user

the email address that requests will be made for

gapi_json

Location of end user credentials

service_account

Location of service account credentials

auth_storage

An instance of a class consuming WebService::GoogleAPI::Client::AuthStorage, already set up for returning access tokens (barring the ua).

If you're using a service account, user represents the user that you're impersonating. Make sure you have domain-wide delegation set up, or else this won't work.

api_query

query Google API with option to validate request before submitting

handles user auth token inclusion in request headers and refreshes token if required and possible

Required params: method, route

Optional params: api_endpoint_id cb_method_discovery_modify, options

$self->get_access_token must return a valid token

  $gapi->api_query({
      method => 'get',
      path => 'https://www.googleapis.com/calendar/users/me/calendarList',
    });

  $gapi->api_query({
      method => 'post',
      path => 'https://www.googleapis.com/calendar/v3/calendars/'.$calendar_id.'/events',
      options => { key => value }
  }

  ## if provide the Google API Endpoint to inform pre-query validation
  say $gapi_agent->api_query(
      api_endpoint_id => 'gmail.users.messages.send',
      options    => 
          { raw => encode_base64( Email::Simple->create( 
                       header => [To => $user, From => $user, 
                                  Subject =>"Test email from $user",], 
                       body   => "This is the body of email from $user to $user", 
                   )->as_string ), 
          },
      )->to_string; ##

  print  $gapi_agent->api_query(
            api_endpoint_id => 'gmail.users.messages.list', 
            ## auto sets method to GET, and the path to 
            ## 'https://www.googleapis.com/gmail/v1/users/me/messages'
          )->to_string;

If the pre-query validation fails then a 418 - I'm a Teapot error response is returned with the body containing the specific description of the errors ( Tea Leaves ;^) ).

Dealing with inconsistencies

NB: If you pass a 'path' parameter this takes precendence over the API Discovery Spec. Any parameters defined in the path of the format {VARNAME} will be filled in with values within the options=>{ VARNAME => 'value '} parameter structure. This is the simplest way of addressing issues where the API discovery spec is inaccurate. ( See dev_sheets_example.pl as at 14/11/18 for illustration ). This particular issue has been since solved, but you never know where else there are problems with the discovery spec.

Sometimes, Google is slightly inconsistent about how to name the parameters. For example, error messages sent back to the user tend to have the param names in snake_case, whereas the discovery document always has them in camelCase. To address this issue, and in the DWIM spirit of perl, parameters may be passed in camelCase or snake_case. That means that

    $gapi_agent->api_query(
        api_endpoint_id => 'gmail.users.messages.list',
        options => { userId => 'foobar' });

and

    $gapi_agent->api_query(
        api_endpoint_id => 'gmail.users.messages.list',
        options => { user_id => 'foobar' });

will produce the same result.

Sometimes a param expects a dynamic part and a static part. The endpoint jobs.projects.jobs.list, for example, has a param called 'parent' which has a format '^projects/[^/]+$'. In cases like this, you can just skip out the constant part, making

  $gapi_agent->api_query( api_endpoint_id => 'jobs.projects.jobs.list',
    options => { parent => 'sner' } );

and

  $gapi_agent->api_query( api_endpoint_id => 'jobs.projects.jobs.list',
    options => { parent => 'projects/sner' } );

the same. How's that for DWIM?

In addition, you can use different names to refer to multi-part parameters. For example, the endpoint jobs.projects.jobs.delete officially expects one parameter, 'name'. The description for the param tells you that you it expects it to contain 'projectsId' and 'jobsId'. For cases like this,

  $gapi_agent->api_query( api_endpoint_id => 'jobs.projects.jobs.delete',
    options => {name => 'projects/sner/jobs/bler'} );

and

  $gapi_agent->api_query( api_endpoint_id => 'jobs.projects.jobs.delete',
    options => {projectsId => 'sner', jobsId => 'bler'} );

will produce the same result. Note that for now, in this case you can't pass the official param name without the constant parts. That may change in the future.

To further fix discrepencies in the Discovery Specification, the cb_method_discovery_modify callback can be used which must accept the method specification as a parameter and must return a (potentially modified) method spec.

eg.

    my $r = $gapi_client->api_query(  
                api_endpoint_id => "sheets:v4.spreadsheets.values.update",  
                options => { 
                   spreadsheetId => '1111111111111111111',
                   valueInputOption => 'RAW',
                   range => 'Sheet1!A1:A2',
                   'values' => [[99],[98]]
                },
                cb_method_discovery_modify => sub { 
                   my  $meth_spec  = shift; 
                   $meth_spec->{parameters}{valueInputOption}{location} = 'path';
                   $meth_spec->{path} .= "?valueInputOption={valueInputOption}";
                   return $meth_spec;
                 }
            );

Again, this specific issue has been fixed.

Returns Mojo::Message::Response object

has_scope_to_access_api_endpoint

Given an API Endpoint such as 'gmail.users.settings.sendAs.get' returns 1 iff user has scope to access

    say 'User has Access'  if $gapi_agent->has_scope_to_access_api_endpoint( 'gmail.users.settings.sendAs.get' );

Returns 0 if scope to access is not available to the user.

warns and returns 0 on error ( eg user or config not specified etc )

METHODS DELEGATED TO WebService::GoogleAPI::Client::Discovery

discover_all

  Return details about all Available Google APIs as provided by Google or in CHI Cache

  On Success: Returns HASHREF containing items key => list of hashes describing each API
  On Failure: Warns and returns empty hashref

    my $client = WebService::GoogleAPI::Client->new; ## has discovery member WebService::GoogleAPI::Client::Discovery

    $d = $client->discover_all();
    $d = $client->discover_all(1); ## NB if include a parameter that evaluates to true such as '1' then the cache is flushed with a new version

    ## OR
    $d = $client->discovery-> discover_all();
    $d = WebService::GoogleAPI::Client::Discovery->discover_all();

    print Dumper $d;

      $VAR1 = {
                'items' => [
                            {
                              'preferred' => bless( do{\(my $o = 1)}, 'JSON::PP::Boolean' ),
                              'id' => 'abusiveexperiencereport:v1',
                              'icons' => {
                                            'x32' => 'https://www.gstatic.com/images/branding/product/1x/googleg_32dp.png',
                                            'x16' => 'https://www.gstatic.com/images/branding/product/1x/googleg_16dp.png'
                                          },
                              'version' => 'v1',
                              'documentationLink' => 'https://developers.google.com/abusive-experience-report/',
                              'kind' => 'discovery#directoryItem',
                              'discoveryRestUrl' => 'https://abusiveexperiencereport.googleapis.com/$discovery/rest?version=v1',
                              'title' => 'Abusive Experience Report API',
                              'name' => 'abusiveexperiencereport',
                              'description' => 'Views Abusive Experience Report data, and gets a list of sites that have a significant number of abusive experiences.'
                            }, ...

    ## NB because the structure isn't indexed on the api name it can be convenient to post-process it
    ## 
    
    my $new_hash = {};
    foreach my $api ( @{ %{$client->discover_all()}{items} } )
    {
        # convert JSON::PP::Boolean to true|false strings
        $api->{preferred}  = "$api->{preferred}" if defined $api->{preferred};
        $api->{preferred}  = $api->{preferred} eq '0' ? 'false' : 'true';

        $new_hash->{ $api->{name} } = $api;
    }
    print dump $new_hash->{gmail};

get_api_document

returns the cached version if avaiable in CHI otherwise retrieves discovery data via HTTP, stores in CHI cache and returns as a Perl data structure.

    my $hashref = $self->get_api_document( 'gmail' );
    my $hashref = $self->get_api_document( 'gmail:v3' );

returns the api discovery specification structure ( cached by CHI ) for api id ( eg 'gmail ')

returns the discovery data as a hashref, an empty hashref on certain failing conditions or croaks on critical errors.

methods_available_for_google_api_id

Returns a hashref keyed on the Google service API Endpoint in dotted format. The hashed content contains a structure representing the corresponding discovery specification for that method ( API Endpoint ).

    methods_available_for_google_api_id('gmail')

get_method_details

    $my $api_detail = $gapi->discovery->get_method_details( 'gmail.users.settings' );

returns a hashref representing the discovery specification for the method identified by $tree in dotted API format such as texttospeech.text.synthesize

returns an empty hashref if not found

list_api_ids

Returns an array list of all the available API's described in the API Discovery Resource that is either fetched or cached in CHI locally for 30 days.

    my $r = $agent->list_api_ids();
    print "List of API Services ( comma separated): $r\n";

    my @list = $agent->list_api_ids();

To check for just one service id, use service_exists instead.

FEATURES

  • API Discovery requests cached with CHI ( Default File )

  • OAUTH app and user credentials (client_id, client_secret, scope, users access_token and refresh_tokens) stored in local file (default name = gapi.json)

  • access_token auto-refreshes when expires (if user has refresh_token) saving refreshed token back to json file

  • helper api_query to streamline request composition without preventing manual construction if preferred.

  • CLI tool (goauth) with lightweight Mojo HTTP server to simplify OAuth2 configuration, sccoping, authorization and obtaining access_ and refresh_ tokens from users

AUTHORS

  • Veesh Goldman <veesh@cpan.org>

  • Peter Scott <localshop@cpan.org>

COPYRIGHT AND LICENSE

This software is Copyright (c) 2017-2021 by Peter Scott and others.

This is free software, licensed under:

  The Apache License, Version 2.0, January 2004