kinokero

Complete Google CloudPrint Proxy client. Includes seperate classes to handle the GCP server protocol (Cloudprint), the GCP Jingle notification protocol (Jingle), and a class for interacting with CUPS devices on a linux system (Printer). The Cloudprint class uses a faraday-based http client for interacting with Google CloudPrint Services (GCPS). Persistence is expected to be handled by whatever is invoking Kinokero. Each of these major classes can (more or less) function in a stand-alone manner for low-level cloudprint primitives.

About this document: I've tried to capture anything and everything relevant to using the gem. This information is not necessarily well organized (yet). Consider it a first draft. If this gem is simply used as a proxy connector (recomended), then usage is much simpler. If instead you only want the GCPS interfaces ( Class Cloudprint ), then you will need more detailed information found in this document.

Kinokero is swahili for an impala (the animal, not the Chevy model).

Kinokero Status

  • The gem is currently at Beta, v0.0.5.
  • All GCP protocol interactions are working as a GCP2.0 printer
  • jingle notifications are working.
  • Files can be cloudprinted remotely.
  • Verbose mode has extensive logging of everything from a protocol standpoint
  • I have a makeshift manual command interpreter ('console') which functions as a debug-level persistence calling module for overall testing kinokero. Please see discussion below.
  • Kinokero uses Ruby threads when polling is required.
  • Documentation draft completed.
  • Unit tests completed (for now).

Discussion

GCP documentation is confusingly bad at best and inaccurate and incomplete at worst. There is nothing approaching correct and accurate protocol documentation with complete examples of actual data requests and responses. There are no state diagrams.

The Chromium Project has a GCP proxy, but the C++ code is the worst code I have ever seen and was almost useless as a reference. Needless to say, the code has no documentation. From a Ruby perspective, the code is horrific and frightening, a Halloween nightmare visited upon the safely slumbering programmer. Is Ruby the only world with packaged interfaces for doing common things like encryption, http client, xmpp client, etc? Compare the code in cloudprint.rb to the spaghetti tangle in the chromium proxy connector.

Libjingle was touted as necessary; but again, there is no API documentation and numerous sources mentioned the impossibility of getting it working on a linux system.

The GCP documents do have some Python code, but they are old and outdated and don't match the current protocol. Lately, Google has added numerous disclaimers to the examples (which are not complete GCP examples!) about the code being outdated and to use it as reference only.

During development, I had to trial-&-error hack my way to get the protocols working.

I used popular working Ruby gems to make this package DRY:

  • faraday as the http client (and needed faraday middleware)
  • xmpp4r as an excellent XMPP module (made skipping libjingle easy)
  • cups as an interface to the linux CUPS system.
  • beefcake as the protocol buffer handler & compiler

Installation

Add this line to your application's Gemfile:

    gem 'kinokero'

And then execute:

    $ bundle install

Or install it yourself as:

    $ gem install kinokero

Console

During development, I needed a convenient setup and testing structure to individually trigger GCP primitives and see traces of the request and result. I'm calling that the "console" and it has no function inherent to the gem, other than a convenient setup and testing apparatus. Rather than having it vanish, I have made it part of the gem superstructure and it can be run indepently, as though it were an application, for calling and testing the gem.

It is rather crude and no further improvements will be made to it. Nor will I be particularly receptive to issues published against it. I will, however, potentially request its usage to isolate situations for issues made against the core gem itself.

The console has a simple means of persistence (a seed yaml file) for any printers which are registered. The seed yaml requires, at the least for initial startup, a section of data for the (required) test printer.

Some of commands have to be issued in a particular order to be useful. This is largely related to the GCP protocol itself.

Invocation

Before invoking, make sure you cd into the console directory and run bundler to install all the required gems.

  $ cd console
  $ bundle install

You will need to also set up your Google developers API for your proxy and have that information in the following environmental variables (I do it in .bashrc). The samples shown below have been sterilized for security.

export GCP_PROXY_API_PROJECT_NBR=407407407407
export GCP_PROXY_API_CLIENT_EMAIL="[email protected]"
export GCP_PROXY_CLIENT_ID="407407407407-abcd1abcd2abcd3abcd4abcd5abcd5ef.apps.googleusercontent.com"
export GCP_PROXY_CLIENT_SECRET="someSECRETencryptedHEREo"

The console can be invoked in two ways: the normal way runs the console and then simply exits to the OS when finished. Or it can be run to exit into irb, still maintaining all instantiated objects, similar to the Rails console.

Currently, both of these run the console with the '-m' switch, which means manual connection for each printer. Without that switch, active printers will be automatically connected (jingle connection) to the GCP servers.

normal invocation

  $ ./console

exit-to-irb invocation

  $ ./irb_console

Command syntax

  <command> [<printer item>]

example: 

  register wild

will register a printer called "gcp_wild_printer" and persist the data in the "wild" item in the seed yaml file.

A single word command (listed below) is required. An optional item keyword to identify which printer. If missing, 'test' is assumed as the default printer. Most commands are specific to a printer.

Console commands

commands which don't require a printer item

  • help - the list of commands
  • quit - exit the console
  • exit - exit the console
  • save - writes the internal seed information to the seed file
  • devices - lists Proxy's my_devices hash for all Kinokero devices (lengthy)
  • cups - switches to submode for querying CUPS primitives

commands which do require a printer item

  • list - returns GCP response listing GCP parameters for the given printer if registered.
  • fetch - returns GCP job fetch list for given printer
  • register - anonymous register the given printer. if the given printer
  • ready - queries CUPS for the printer and performs a GCP /update command for its status
  • refresh - refreshes the OAUTH2 token for the given printer
  • delete - deletes the given printer from GCP registered list
  • connect - starts a persistent jingle connection for the given printer; tells GCP that printer is now 'on-line' and ready to receive print requests. this is only necessary if the console has been started in manual mode.
  • time - time when OAUTH2 token expires next for given printer
  • gcp - dump of Cloudprint gcp_control hash for given printer

CUPS sub-mode commands

The CUPS sub-mode is for querying the local CUPS system and displaying the results.

  • help - lists sub-mode commands
  • quit - exit the cups sub-mode

  • exit - exit the cups sub-mode

  • printers - displays CUPS printers by name

  • default - displays the CUPS default printer name

  • jobs - displays the last CUPS printer job status structure

  • options - displays the options (attributes) for the default printer

  • print - prints a test page to the default printer

  • scan - displays a polled scan of printer status while printer prints a test page

Seed file

As a minimum, the "test" printer item must be defined. Each printer item can be mapped to the same, or different, CUPS printers on the local system. The file is: console/config/gcp_seed.yml

Variables (you supply the correct information) are indicated enclosed in angular brackets: .

test:
  printer_id: 0
  item: test
  cups_alias: <actual CUPS printer name>
  is_active: false
  virgin_access: false
  gcp_printer_name: gcp_test_printer
  capability_ppd: /etc/cups/ppd/<actual CUPS printer name>.ppd
  capability_cdd: /etc/cups/cdd/<actual CUPS printer name>.cdd
  gcp_uuid: <printer serial number>
  gcp_manufacturer: <printer manufacturer name>
  gcp_model: <printer model name>
  gcp_setup_url: <url to a page for setting up the printer>
  gcp_support_url: <url to a page for supporting the printer>
  gcp_update_url: <url to a page for updating the printer>
  gcp_firmware: '<firmware version number>'

For legacy purposes, the PPD file name (complete path) is required. From that, please use the Google CDD converter to convert the PPD to CDD (Cloud Device Description) format and save that on your local system. Provide the complete path to that file. The web tool for this is at: https://www.google.com/cloudprint/tools/cdd/cdd.html

Gem Structure

The kinokero overall structure parallels the different levels for the cloudprint protocol.

Proxy - is the overall, high-level appliance that satisfies the basic functionality of a cloudprint device. This is the expected entry point for an application which also maintains persistence of printer information. The kinokero console (above), for example, functions as a crude persistent application.

Cloudprint - is the primary interace for issuing commands to Google Cloudprint servers via HTTP POST commands. Proxy functions use these primitives to do their work. The main data structure used here is based off of an options hash termed gcp_control (described below).

Jingle - is the interface for the gtalk jingle (XMPP) protocol required for asynchronously receiving notification of pending print jobs (uses a callback mechanism). Class Cloudprint directly accesses Jingle, so you probably won't have to be too concerned about it.

Printer - is the interface to the local OS CUPS system and does any actual printing. It also does device state polling.

Threads

The gem uses several threads to handle asynchronous polling tasks:

  • polling for user interaction when registering a printer
  • polling device status
  • jingle connection
  • jingle callback

Note: at this point I haven't seen the need for mutex semaphore control over any of the control structures. It might be required in the future.

Usage

Note that the primary example for setting up and using this gem is the console code: console/twiga.rb and all config, setup folders.

Terms: the cloudprint proxy (or connector) orchestrates the overall GCP interactions to do the higher level tasks such as register or remove a printer, intercept print notifications and print jobs, etc. So the "user" of kinokero is whoever is using the proxy to register and control local printers. It is not the end user trying to print something on a cloud printer.

Printer Names, ids, and aliases

This can potentially be a point of confusion to first time kinokero users, so please pay close attention. Kinokero has been set up so that you can map several cloudprint (logical) devices to a single CUPS (physical) device. The reason you might want to do this is because each logical device can be defined with its own CDD (Cloud Device Description) parameters: such as color vs monochrome, dual-sided vs single-sided, etc. For economic purposes, an organization might want to limit access to the color-capabilities of a laser printer and give out broader access to the monochrome-capability of that same printer. With kinokero, this is theoretically possible to do.

So let's work backward through the different names and described them and their purpose. The gcp_control hash (later section) maps these all together.

item: This is the hash key into the @proxy.my_devices hash of all gcp_control hashes being managed by kinokero.

gcp_cups_alias: This is the printer name as recognized by cups when used to issue a print command.

  lp -d <gcp_cups_alias>

gcp_printer_name: This is the name by which a cloudprint user sees the printer when it shows up in their cloudprint managed printers list. For example: "Brother MFC-9340CDW".

gcp_printerid: This is a string issued by GCPS for use in GCPS calls. For example: "d0510370-f36b-356f-1a82-f93c0756a5d9". This also appears in the "advanced details" section of DETAILS for a cloudprint printer in the Manage Printers dashboard.

printer_id: This is some type of unique id (string or integer) which the user's program (the one which invokes kinokero) uses to access a persistence record for the given logical printer. It could, for example, be a database record number for the given persistence.

Examples

  item: color
  gcp_cups_alias: laserjet_1020w
  gcp_printer_name: color laser printer
  gcp_printerid: d0510370-f36b-356f-1a82-f93c0756a5d9
  printer_id:  509

  item: b&w
  gcp_cups_alias: laserjet_1020w
  gcp_printer_name: monochrome laser printer
  gcp_printerid: cda248c1-e1f5-8066-1e10-3efe078cface
  printer_id:  510

  item: test
  gcp_cups_alias: lp_null
  gcp_printer_name: null_printer
  gcp_printerid: 8231517c-716d-4b0a-f721-83fdbe52a05d
  printer_id:  508

In these examples we see three logical cloudprint printers defined. The first two map into the same physical CUPS device. The last one maps into a CUPS device, but has actually been defined as follows, so no actual physical device exists.

  $ lpadmin -p lp_null -E -v file:///dev/null

setting up and initializing the gem

This section will describe the basic first steps of prepping a call to each of the major classes of the kinokero gem: Proxy, Printer, and Cloudprint. Proxy.new is the highest level and in turn invokes Printer.new and Cloudprint.new. So if you're choosing to work at the Proxy level (recommended!), you won't need to worry about the other two.

For references, see the sections below dealing with the primary gcp_control hash, kinokero global parameters, pre-defined gem constants, and template file for Rails config/initializers usage.

Class Proxy

Proxy encapsulates everything needed to function as a Google Cloudprint Proxy Connector Client for any number of printer devices which might be connected to the client machine.

The Proxy has to function correctly even after recovering from power loss or restarting. Thus there has to be a persistence mechanism in place. This mechanism is beyond the scope of the gem and is the invoking program's responsibility.

At its simpliest, the invoking program accesses Proxy in three ways:

  • Kinokero::Proxy.new - instantiates a Proxy object which consists of a (possible) list of all actively managed cloudprint devices (from the persistence). All active printers are automatically logically connected to GCPS thus showing as "on-line" and ready for printing.
  • Proxy#do_register - registers a new printer with GCPS and creates an entry in the managed devices list.
  • Proxy#do_delete - removes a managed printer from GCPS control.

The sections below show scaffold coding for preparing those calls.

Note: Any methods not described in the API are not expected to be publicly accessible; in future versions they may become inaccessible. Please do not use.

Class Proxy initialization


  # build a hash of options for all active & inactive printers
  # from the persistence memory (in this case a YAML file)
  # for details on the gcp_control hash, see later sections in this README
  #
  def build_device_list()

    load_gcp_seed()    # load the YAML seed data

      # prep to build up a hash of gcp_control hashes
    gcp_control_hash = {}

    @seed_data.each_key do |item|
      gcp_control_hash[ item ] = @seed_data[ item ].symbolize_keys  # strip item into hash with keys
    end   # convert each seed to a device object

    return gcp_control_hash

  end   # convert each seed to a device object

  gem_options = {
        # true if automatically jingle connect active printers from list
      auto_connect:  true,  
        # true if instance-level verbose log trace of all GCPS and jingle requests
      verbose:       true, 
        # true if truncate long responses
      log_truncate:  true, 
        # true if log trace all responses from GCPS
      log_response:  true 
  }

  @proxy = Kinokero::Proxy.new( build_device_list(), gem_options )

Class Proxy automatic running

Proxy is the higher level way to do all cloudprint control. If your gem_options have auto_connect set to true, then everything else is automatic for all currently registered printers. As job print notifications are received, the files will be downloaded, printed, deleted, and the job status updated to DONE.

Class Proxy register cloudprint printers

This shows the required preparation and invocation to register a new printer as a cloudprint printer managed by the kinokero proxy. This is the only way to register a printer from the proxy. It uses GCPS' anonymous printer registration method (meaning that the owner google account is not known at the time when registration is invoked. The owner must 'claim' the printer using a GCPS-issued token. In the process of claiming the printer, the owner might have to sign in to a Google account, at which point authentication is established and then the proxy can proceed to get OAUTH2 authorization tokens).


def register_printer_scaffold( any_parameters )
  new_request = {
      item:  'name for this item',
      printer_id:   0,  # will be cue to create new record
      gcp_printer_name: "gcp_#{item}_printer",
      capability_ppd: 'pathname for the PPD file',  # legacy GCP 1.0
      capability_cdd: 'pathname for the CDD file',  # required GCP 2.0
      cups_alias: 'cups printer name',
      gcp_uuid:        string, # see gcp_control hash below 
      gcp_manufacturer:string, # see gcp_control hash below 
      gcp_model:       string, # see gcp_control hash below 
      gcp_setup_url:   string, # see gcp_control hash below 
      gcp_support_url: string, # see gcp_control hash below 
      gcp_update_url:  string, # see gcp_control hash below 
      gcp_firmware:    string, # see gcp_control hash below 
  }

  response = @proxy.do_register( new_request ) do |gcp_control|

      # item_persistence is user-defined means to persist the
      # GCPS-issued information for a printer
      # which must be supplied again to the proxy whenever
      # rebooting
      # remember to set up a gcp_control[:printer_id] as the
    # database record number for this new item for any
    # future need to update persistence (such as on 
      # refresh token
    item_persistence( gcp_control )

  end   # do persist new printer information

  unless response[:success]
    puts "printer registration failed: #{response[:message]}"
  end

end

Class Proxy remove cloudprint printers

This shows the required set up for removing a printer from the cloudprint management. Cloudprint spec requires the proxy to offer the user a way to remove a printer from cloudprint management.

def remove_printer_scaffold( item )

   @proxy.do_delete( item )  # removes printer associated with item

      # deactivate_item_persistence is user-defined means to
      # do any persistence cleanup associated with a deactivated item,
      # including removal from the my_devices hash, if desired.
   deactivate_item_persistence( @proxy.my_devices, item )
end

Class Proxy other API methods

These normally won't be needed. They were originally made high-level so that the debugging console could individually trigger an action and the resulting log trace be viewed.

Format for discussion will be instance method name & invocation, description. Typically returns the body of a json-parsed GCP response hash (they all have response['success'] set to true if no errors). See the GCP documentation for more detail.

  • do_refresh(item) - refreshes the token authorization; update persistence afterwards.
  • do_list(item) - returns the list of printers as a json-parsed list of hashes in response['printers']
  • do_connect(item) - manually initiates the jingle connection which will show the printer as on-line to the cloudprint user.
  • do_fetch_jobs(item) - returns a json-parsed list of gcp job hashes
  • do_print_jobs(printerid) - fetches, then prints all jobs in queue for given printerid
  • do_ready_state(item) - tells GCPS that the printer is ready for jobs
  • item_from_printerid(printerid) - returns the item key associated with the printerid; raises PrinterNotFound exception if nothing matched.
  • print_gcp_registration_information(response) - prints the necessary GCP-issued information needed for claiming a newly registered printer. Printing occurs on the printer being registered. "response" parameter is the kinokero-generated hash of information required.

Class Printer

This encapsulates all generic CUPS printer interactions. NOTE: if you are using Class Proxy, you will not need to access the cloudprint objects and you will not need the information in this section. You may safely skip this section!

Note: Any methods not described in the API are not expected to be publicly accessible; in future versions they may become inaccessible. Please do not use.

Printer#start_poll_thread

Starts a background thread to periodically poll the printer status and advise GCPS of any changes or error conditions.

Printer#stop_poll_thread

Stops the background thread to periodically poll the printer status.

Printer#print_file(file)

Prints the file (full path filename) to the object's printer (using cups_alias).

Class Cloudprint

This encapsulates all low-level calls to GCPS. NOTE: if you are using Class Proxy, you will not need to access the cloudprint objects and you will not need the information in this section. You may safely skip this section!

Initialization requires the gcp_control hash and the options hash (which controls a few instance settings like verbose, auto_connect, log_truncate, and log_response). A new Cloudprint object is created, a faraday connection is set up, and a Jingle object created if the printer is active.

Any responses that are labelled "Returns a GCP json-parsed response hash" means that the result is a hash per the GCP documentation, which should be consulted for the fields returned.

Note: Any methods not described in the API are not expected to be publicly accessible; in future versions they may become inaccessible. Please do not use.

Cloudprint.regsiter_anonymous_printer

Pass a request hash and a closure block (invoked upon successful registration confirmation). A thread is used to handle the polling of the confirmation status, so execution immediategly returns after starting (but before confirmation) passing back a response hash containing:

  • :success - true if the registration process was successfully begun
  • :message - error message if not successful
  • :printer_id - any persistence id that might have been passed
  • :cups_alias - the cups name of the printer.
  • :gcp_printer - GCPS printer name (echoed from gcp_control hash)
  • :gcp_printer_id - GCPS-issued printer id
  • :gcp_invite_page_url - complete URL for claim printer page
  • :gcp_easy_reg_url - simple one-click URL to claim the printer
  • :gcp_auto_invite_url - similar
  • :gcp_claim_token_url - similar
  • :gcp_printer_reg_token - token itself for claiming printer

GCP documentation states that the invoker should print out instructions to the user on how to claim the printer using the information in the response hash.

All the other class-level methods are invoked from this method and show should not normally be directly accessed by an outside program.

Cloudprint#gtalk_start_connection(&block)

Invoke this to establish a jingle connection with GCPS. This will let GCPS know that the printer is on-line and actively seeking jobs. Jingle sets up an asynch notification thread. When a notification is received, the closure block is invoked receiving the printerid (gcp_issued id) for the printer have queued jobs.

Cloudprint#gcp_get_job_file( file_url )

Uses the file_url to request the file from GCPS and returns it (the entire file!) as the response. If there was an error, then returns nil.

Cloudprint#gcp_refresh_tokens

Requests a refresh of the cloudprint OAUTH2 tokens if expired. Persistence should be updated after returning new info in gcp_control hash.

Cloudprint#gcp_get_printer_fetch( printerid )

Returns a GCP json-parsed response hash containing list of jobs pending for a given printer.

Cloudprint#gcp_job_status( jobid, status, nbr_pages )

Update the job status. Status should be one of the GCP_JOBSTATE_xxxx constants. nbr_pages is the number of pages that have been printed.

Cloudprint#gcp_job_status_abort( jobid, status, nbr_pages )

Updates the abort status for a job that failed. Use one of the GCP_USER_ACTION_xxxx constants.

Cloudprint#gcp_get_printer_list

Returns a GCP json-parsed response hash containing list of printers. Currently, kinokero only supports one printer per cloudprint object (corresponds to specific OAUTH2 credentials), so the printer list will have only one item.

Class Jingle

This encapsulates all XMPP interactions with Google's jingle server which delivers asynch notifications about the presence of print jobs. Jingle includes Google's extensions to XMPP for this purpose. There really isn't an API here. The jingle.rb code is mainly what's needed to use xmpp4r in a Google Jingle context. Notification callbacks are handled through Ruby's block closure and infer reentreant code.

NOTE: if you are using Class Proxy, you will not need to access the cloudprint objects and you will not need the information in this section. You may safely skip this section!

NOTE: Any methods not described in the API are not expected to be publicly accessible; in future versions they may become inaccessible. Please do not use.

gcp_control hash

This is used throughout the gem, so each attribute will be explained here. Most attributes require persistence, meaning that they have to be supplied to the proxy at initialization for any active printers (this occur at a restart/reboot state).

None of the GCPS-issued items are required when first registering a printer; only the "Items supplied by the proxy user."

The console uses a yml file (in lieu of any other type of persistence mechanism) for storing and retrieving these values between sessions. The file is at: console/config/gcp_seed.yml.

GCPS-issued items which need persistence

  • gcp_xmpp_jid: GCPS-issued id for accessing jingle servers
  • gcp_confirmation_url: GCPS-issued url for confirming the printer registration

  • gcp_owner_email: GCPS-issued owner's email

  • gcp_printerid: GCPS-issued printer id

  • gcp_refresh_token: GCPS-issued used to refresh the OAUTH2 token

  • gcp_access_token: GCPS-issued OAUTH2 token

  • gcp_token_expiry_time: UTC datetime for when the OAUTH2 token expires (thus needing to be refreshed)

  • gcp_token_type: GCPS-issued, used to form the OAUTH2 token

Items supplied by proxy user

  • item: identifier for this printer item, such as: 'test'
  • printer_id: local persistence id (such as for a database record)
  • cups_alias: printer name from the OS' standpoint (such as registered with CUPS)

  • gcp_printer_name: user-determined name which shows up in cloudprint user's managed printer list

  • capability_ppd: complete file pathname for the PPD printer description file (legacy)

  • capability_cdd: complete file pathname for the CCD cloud device description file (gcp v2.0)

  • gcp_uuid: printer serial number, such as: 'VND3R11877'

  • gcp_manufacturer: printer manufacturer, such as: 'Hewlett-Packard'

  • gcp_model: printer model, such as: 'LaserJet P1102w'

  • gcp_setup_url: a url for how-to set up the printer

  • gcp_support_url: a url for getting support

  • gcp_update_url: a url for getting updates

  • gcp_firmware: printer firmware version number, such as: '20130703'

Kinokero-internal usage (persistence required)

  • is_active: set to true after successful registration; false if no longer in use
  • virgin_access: true for initial oauth2 token for a freshly registered printer; false after first refresh

Future API usage

  • message: last error message saved

gem configuration

There are several global parameters which can be set at gem configuration. Default values are provided for each parameter. These are expained below. Many parameters, however, are fixed by Google Cloud Printer documentation demands and should not be changed.

configuration in Rails application

When using kinokero in a Rails application, use the template console/config/kinokero_initializer_template.rb, which shows how to access and change any of these parameters during your rails project initialization. Copy it into your Rails project-name/config/initializers directory, and rename it to: kinokero.rb .

The template file also contains comments describing each parameter.

configuration in non-Rails application

You can change these parameters similar to the example below for changing the verbose setting.

  Kinokero.verbose = true

listing of all configuration parameters and defaults

class Proxy required

  • my_proxy_id: = MY_PROXY_ID # unique name for this running of the GCP connector client
  • proxy_client_id: = ENV["GCP_PROXY_CLIENT_ID"] || 'missing'
  • proxy_client_secret: = ENV["GCP_PROXY_CLIENT_SECRET"] || 'missing'
  • proxy_serial_nbr: = ENV["GCP_PROXY_SERIAL_NBR"] || 'missing'
  • verbose: = false # for any class-level decisions

class Cloudprint required

  • mimetype_oauth: = MIMETYPE_OAUTH # how to encoade oauth files
  • mimetype_ppd: = MIMETYPE_PPD # how to encode PPD files
  • mimetype_cdd: = MIMETYPE_CDD # how to encode CDD files
  • polling_secs: = POLLING_SECS # secs to sleep before register polling again
  • truncate_log: = TRUNCATE_LOG # number of characters to truncate response logs
  • followup_host: = FOLLOWUP_HOST #
  • followup_uri: = FOLLOWUP_URI #
  • gaia_host: = GAIA_HOST #
  • login_uri: = LOGIN_URI #
  • login_url: = LOGIN_URL #
  • gcp_url: = GCP_URL #
  • gcp_service: = GCP_SERVICE #

  • ssl_ca_path: = SSL_CERT_PATH # SSL certificates path for this machine

  • authorization_scope: = AUTHORIZATION_SCOPE #

  • authorization_redirect_uri: = AUTHORIZATION_REDIRECT_URI #

  • oauth2_token_endpoint: = OAUTH2_TOKEN_ENDPOINT #

class Jingle required

  • xmpp_server: = XMPP_SERVER #
  • ns_google_push: = NS_GOOGLE_PUSH #
  • gcp_channel: = GCP_CHANNEL #

cups testpage file path

  • cups_testpage_file: = CUPS_TESTPAGE_FILE

printer device/cups related

  • printer_poll_cycle: = PRINTER_POLL_CYCLE

kinokero global constants used for defaults

Current constant defaults are defined as follows, with the actual definitions in lib/kinokero.rb . 

# mimetype for how to encode CDD files
MIMETYPE_JSON      = 'application/json'
MIMETYPE_PROTOBUF  = 'application/protobuf'
MIMETYPE_GENERAL   = 'application/octet-stream'
MIMETYPE_CDD       =  MIMETYPE_GENERAL

# mimetype for how to encode PPD files
MIMETYPE_PPD     = 'application/vnd.cups.ppd'

# number of secs to sleep before polling again
POLLING_SECS = 30     

# number of characters before truncate response logs
TRUNCATE_LOG = 1000    

# authentication function constants
FOLLOWUP_HOST = 'www.google.com/cloudprint'
FOLLOWUP_URI = 'select%2Fgaiaauth'
GAIA_HOST = 'www.google.com'
LOGIN_URI = '/accounts/ServiceLoginAuth'
LOGIN_URL = 'https://www.google.com/accounts/ClientLogin'

# GCP documentation constants
AUTHORIZATION_SCOPE = "https://www.googleapis.com/auth/cloudprint"
CLIENT_REDIRECT_URI = "urn:ietf:wg:oauth:2.0:oob"
AUTHORIZATION_REDIRECT_URI = 'oob'
OAUTH2_TOKEN_ENDPOINT = "https://accounts.google.com/o/oauth2/token"
MIMETYPE_OAUTH =  "application/x-pkcs12"

# The GCP URL path is composed of URL + SERVICE + ACTION
# below three are used when testing locally
#   GCP_URL = 'http://0.0.0.0:3000'
#   GCP_SERVICE = '/'
#   GCP_REGISTER = ''
GCP_URL = 'https://www.google.com/'
GCP_SERVICE = 'cloudprint'

# jingle constants required

XMPP_SERVER     = "talk.google.com" 
NS_GOOGLE_PUSH  = "google:push"
GCP_CHANNEL     = "cloudprint.google.com"

# MY_PROXY_ID is a unique name for this running of the GCP connector client
# formed with gem name + machine-node name (expected to be  unique)
# TODO: make sure machine nodename is unique
MY_PROXY_ID = "kinokero::"+`uname -n`.chop

# SSL certificates path for this machine
SSL_CERT_PATH = "/usr/lib/ssl/certs"

# CUPS system default testpage file
CUPS_TESTPAGE_FILE = "/usr/share/cups/data/default-testpage.pdf"

# printer device status polling cycle (float secs to sleep)
PRINTER_POLL_CYCLE = 15    # wait fifteen seconds before recheck status

Testing

Unit Testing is somewhat problematic because the kinokero gem is also a wrapper for interactions with the Google Cloud Print Server (GCPS). GCPS, unfortunately, are not really set up to handle unit testing. There is no sandbox; everything is live.

There are several issues associated with this.

To use unit testing, you'll need to register yourself into the Google API and obtains client id and secret. You'll then have to set these up in your local environment variables. The unit testing for Cloudprint will verify whether these exist or not and fail the unit tests at the outset if not.

export GCP_PROXY_CLIENT_ID="407123456789-321ngkabcdefghijklmnooppqrstuvwx.apps.googleusercontent.com"
export GCP_PROXY_CLIENT_SECRET="iAMsecretiAMsecretiAMsec"

Next, you'll have to create and register a test printer and claim it at some Google account. You can use the console to do that. Then copy and paste the gcp_seed.yml information into the test item area of test/test/fixtures/gcp_seed.yml.

The next problem is the manually-intensive nature of registering a printer. This isn't conducive to automated tests, so the actual registration of a printer is not currently in the cloudprint_test.rb.

And finally, a number of GCPS commands require an active on-line printer to fully test. But it is possible to test the failure of various GCPS commands by issueing them against non-existent print jobs and files. At the very least, we can test the command set up, invocation against GCPS, and an unsuccessful GCPS response. This has been done for many of the GCPS actions. These tests are seperated in the test file and a comment marks the start of the section for those tests. That does mean that there will be an OAUTH2 token fetch fail message printed to the log (below) when running the test. This is good and should cause no alarm, since it means the code is working correctly!

E, [2014-09-20T12:29:37.479671 #10941] ERROR -- oauth2 token fetch fail: **********************************

If you do encounter errors or failures when running the cloudprint tests, you may want to turn on logging by switching the verbose setting. You can change this in the test/test/test_kinokero.rb test helper, line 206: set it to true for a verbose logging; otherwise false for brevity. When logging is verbose (enabled) then ALL requests AND responses to/from GCPS are logged, as well as all Jingle interactions.

      full_verbose = false    # ok to change this setting

Available & planned unit tests

The list below has the status of all kinokero unit tests. All these unit tests are completed and working.

  • Cloudprint - test/models/cloudprint_test.rb this tests the primary GCPS wrapped interface for all GCPS actions.
  • Jingle - test/models/jingle_test.rb
  • Printer - test/models/printer_test.rb
  • Proxy - test/models/proxy_test.rb

Running a unit test

There is a mini Rails application for the testing environment. From the kinokero gem directory:

  $cd test   # gets you to the mini Rails app
  $ruby -I test test/models/cloudprint_test.rb

or substitute different unit test filenames for cloudprint_test.rb

references

These references are for Google Cloudprint documentation.

future stuff

  • use a robot account; A gcp developer claims: As long as the same robot account is used for all printers handled by the proxy, you can use just one XMPP connection for notifications for all of the printers. This is mentioned briefly in the documentation at https://developers.google.com/cloud-print/docs/devguide#connectorregistration - just use the robot account from the first registered printer as the 'owner' for all subsequent registrations. In this special case, we make the user who created that robot account the owner of the new printer as well, and use the same robot account credentials for it. This will also resolve the /list issue you mentioned, as calling /list authenticated with the robot account will allow you to see all of the associated printers.
  • autotest register printer; same developer claims, but I couldn't figure it out: We don't have a special server to test registration against, but you can automate the claim flow for unit tests. It isn't documented externally, but if you use your browser's devtools to look at the requests we make on the claim page, you can see the flow your tests would need to follow to automatically 'claim' a test printer (you can do this with a test Gmail account you register, or with a user's credentials - same for generating the XMPP JID).

Contributing

GCP interaction is, shall we say, delicate. Any changes to Cloudprint or Jingle need to be carefully considered and tested under various conditions.

  1. Fork it
  2. Create your feature branch (git checkout -b my-new-feature)
  3. Commit your changes (git commit -am 'Add some feature')
  4. Push to the branch (git push origin my-new-feature)
  5. Run the unit tests and make sure nothing is broken. Supply new unit tests for any added features.
  6. Create new Pull Request