Class: GRPC::ActiveCall

Inherits:
Object
  • Object
show all
Extended by:
Forwardable
Includes:
Core::StatusCodes, Core::CallOps, Core::TimeConsts
Defined in:
lib/grpc/generic/active_call.rb

Overview

The ActiveCall class provides simple methods for sending marshallable data to a call

Constant Summary

Constants included from Core::CallOps

Core::CallOps::RECV_CLOSE_ON_SERVER, Core::CallOps::RECV_INITIAL_METADATA, Core::CallOps::RECV_MESSAGE, Core::CallOps::RECV_STATUS_ON_CLIENT, Core::CallOps::SEND_CLOSE_FROM_CLIENT, Core::CallOps::SEND_INITIAL_METADATA, Core::CallOps::SEND_MESSAGE, Core::CallOps::SEND_STATUS_FROM_SERVER

Instance Attribute Summary collapse

Class Method Summary collapse

Instance Method Summary collapse

Methods included from Core::TimeConsts

from_relative_time

Constructor Details

#initialize(call, q, marshal, unmarshal, deadline, started: true, metadata_tag: nil) ⇒ ActiveCall

Creates an ActiveCall.

ActiveCall should only be created after a call is accepted. That means different things on a client and a server. On the client, the call is accepted after calling call.invoke. On the server, this is after call.accept.

#initialize cannot determine if the call is accepted or not; so if a call that’s not accepted is used here, the error won’t be visible until the ActiveCall methods are called.

deadline is the absolute deadline for the call.

Parameters:

  • call (Call)

    the call used by the ActiveCall

  • q (CompletionQueue)

    the completion queue used to accept the call

  • marshal (Function)

    f(obj)->string that marshal requests

  • unmarshal (Function)

    f(string)->obj that unmarshals responses

  • deadline (Fixnum)

    the deadline for the call to complete

  • metadata_tag (Object) (defaults to: nil)

    the object use obtain metadata for clients

  • started (true|false) (defaults to: true)

    indicates if the call has begun



110
111
112
113
114
115
116
117
118
119
120
121
122
123
# File 'lib/grpc/generic/active_call.rb', line 110

def initialize(call, q, marshal, unmarshal, deadline, started: true,
               metadata_tag: nil)
  fail(TypeError, '!Core::Call') unless call.is_a? Core::Call
  unless q.is_a? Core::CompletionQueue
    fail(TypeError, '!Core::CompletionQueue')
  end
  @call = call
  @cq = q
  @deadline = deadline
  @marshal = marshal
  @started = started
  @unmarshal = unmarshal
  @metadata_tag = 
end

Instance Attribute Details

#deadlineObject (readonly)

Returns the value of attribute deadline.



61
62
63
# File 'lib/grpc/generic/active_call.rb', line 61

def deadline
  @deadline
end

Class Method Details

.client_invoke(call, q, _deadline, **kw) ⇒ Object

client_invoke begins a client invocation.

Flow Control note: this blocks until flow control accepts that client request can go ahead.

deadline is the absolute deadline for the call.

Keyword Arguments ==

any keyword arguments are treated as metadata to be sent to the server if a keyword value is a list, multiple metadata for it’s key are sent

Parameters:

  • call (Call)

    a call on which to start and invocation

  • q (CompletionQueue)

    the completion queue

  • deadline (Fixnum, TimeSpec)

    the deadline



78
79
80
81
82
83
84
85
86
87
# File 'lib/grpc/generic/active_call.rb', line 78

def self.client_invoke(call, q, _deadline, **kw)
  fail(TypeError, '!Core::Call') unless call.is_a? Core::Call
  unless q.is_a? Core::CompletionQueue
    fail(TypeError, '!Core::CompletionQueue')
  end
   = Object.new
  call.run_batch(q, , INFINITE_FUTURE,
                 SEND_INITIAL_METADATA => kw)
  
end

Instance Method Details

#bidi_streamer(requests, **kw, &blk) ⇒ Enumerator?

bidi_streamer sends a stream of requests to the GRPC server, and yields a stream of responses.

This method takes an Enumerable of requests, and returns and enumerable of responses.

requests ==

requests provides an ‘iterable’ of Requests. I.e. it follows Ruby’s #each enumeration protocol. In the simplest case, requests will be an array of marshallable objects; in typical case it will be an Enumerable that allows dynamic construction of the marshallable objects.

responses ==

This is an enumerator of responses. I.e, its #next method blocks waiting for the next response. Also, if at any point the block needs to consume all the remaining responses, this can be done using #each or #collect. Calling #each or #collect should only be done if the_call#writes_done has been called, otherwise the block will loop forever.

Keyword Arguments ==

any keyword arguments are treated as metadata to be sent to the server if a keyword value is a list, multiple metadata for it’s key are sent

Parameters:

  • requests (Object)

    an Enumerable of requests to send

Returns:

  • (Enumerator, nil)

    a response Enumerator



408
409
410
411
412
413
414
415
# File 'lib/grpc/generic/active_call.rb', line 408

def bidi_streamer(requests, **kw, &blk)
  start_call(**kw) unless @started
  bd = BidiCall.new(@call, @cq, @marshal, @unmarshal, @deadline)
  bd.run_on_client(requests, &blk)
rescue GRPC::Core::CallError => e
  finished  # checks for Cancelled
  raise e
end

#client_streamer(requests, **kw) ⇒ Object

client_streamer sends a stream of requests to a GRPC server, and returns a single response.

requests provides an ‘iterable’ of Requests. I.e. it follows Ruby’s #each enumeration protocol. In the simplest case, requests will be an array of marshallable objects; in typical case it will be an Enumerable that allows dynamic construction of the marshallable objects.

Keyword Arguments ==

any keyword arguments are treated as metadata to be sent to the server if a keyword value is a list, multiple metadata for it’s key are sent

Parameters:

  • requests (Object)

    an Enumerable of requests to send

Returns:

  • (Object)

    the response received from the server



338
339
340
341
342
343
344
345
346
347
348
# File 'lib/grpc/generic/active_call.rb', line 338

def client_streamer(requests, **kw)
  start_call(**kw) unless @started
  requests.each { |r| remote_send(r) }
  writes_done(false)
  response = remote_read
  finished unless response.is_a? Struct::Status
  response
rescue GRPC::Core::CallError => e
  finished  # checks for Cancelled
  raise e
end

#each_remote_readEnumerator

each_remote_read passes each response to the given block or returns an enumerator the responses if no block is given.

Enumerator ==

  • #next blocks until the remote endpoint sends a READ or FINISHED

  • for each read, enumerator#next yields the response

  • on status

    * if it's is OK, enumerator#next raises StopException
    * if is not OK, enumerator#next raises RuntimeException
    

Block ==

  • if provided it is executed for each response

  • the call blocks until no more responses are provided

Returns:

  • (Enumerator)

    if no block was given



261
262
263
264
265
266
267
268
# File 'lib/grpc/generic/active_call.rb', line 261

def each_remote_read
  return enum_for(:each_remote_read) unless block_given?
  loop do
    resp = remote_read
    break if resp.nil?  # the last response was received
    yield resp
  end
end

#each_remote_read_then_finishEnumerator

each_remote_read_then_finish passes each response to the given block or returns an enumerator of the responses if no block is given.

It is like each_remote_read, but it blocks on finishing on detecting the final message.

Enumerator ==

  • #next blocks until the remote endpoint sends a READ or FINISHED

  • for each read, enumerator#next yields the response

  • on status

    * if it's is OK, enumerator#next raises StopException
    * if is not OK, enumerator#next raises RuntimeException
    

Block ==

  • if provided it is executed for each response

  • the call blocks until no more responses are provided

Returns:

  • (Enumerator)

    if no block was given



290
291
292
293
294
295
296
297
298
299
300
301
# File 'lib/grpc/generic/active_call.rb', line 290

def each_remote_read_then_finish
  return enum_for(:each_remote_read_then_finish) unless block_given?
  loop do
    resp = remote_read
    break if resp.is_a? Struct::Status  # is an OK status
    if resp.nil?  # the last response was received, but not finished yet
      finished
      break
    end
    yield resp
  end
end

#finishedObject

finished waits until a client call is completed.

It blocks until the remote endpoint acknowledges by sending a status.



170
171
172
173
174
175
176
177
178
179
180
181
# File 'lib/grpc/generic/active_call.rb', line 170

def finished
  batch_result = @call.run_batch(@cq, self, INFINITE_FUTURE,
                                 RECV_STATUS_ON_CLIENT => nil)
  unless batch_result.status.nil?
    if @call..nil?
      @call. = batch_result.status.
    else
      @call..merge!(batch_result.status.)
    end
  end
  batch_result.check_status
end

#multi_req_viewObject

multi_req_view provides a restricted view of this ActiveCall for use in a server client-streaming handler.



133
134
135
# File 'lib/grpc/generic/active_call.rb', line 133

def multi_req_view
  MultiReqView.new(self)
end

#operationObject

operation provides a restricted view of this ActiveCall for use as a Operation.



145
146
147
# File 'lib/grpc/generic/active_call.rb', line 145

def operation
  Operation.new(self)
end

#output_metadataObject

output_metadata are provides access to hash that can be used to save metadata to be sent as trailer



127
128
129
# File 'lib/grpc/generic/active_call.rb', line 127

def 
  @output_metadata ||= {}
end

#remote_readObject

remote_read reads a response from the remote endpoint.

It blocks until the remote endpoint replies with a message or status. On receiving a message, it returns the response after unmarshalling it. On receiving a status, it returns nil if the status is OK, otherwise raising BadStatus



225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
# File 'lib/grpc/generic/active_call.rb', line 225

def remote_read
  ops = { RECV_MESSAGE => nil }
  ops[RECV_INITIAL_METADATA] = nil unless @metadata_tag.nil?
  batch_result = @call.run_batch(@cq, self, INFINITE_FUTURE, ops)
  unless @metadata_tag.nil?
    @call. = batch_result.
    @metadata_tag = nil
  end
  logger.debug("received req: #{batch_result}")
  unless batch_result.nil? || batch_result.message.nil?
    logger.debug("received req.to_s: #{batch_result.message}")
    res = @unmarshal.call(batch_result.message)
    logger.debug("received_req (unmarshalled): #{res.inspect}")
    return res
  end
  logger.debug('found nil; the final response has been sent')
  nil
end

#remote_send(req, marshalled = false) ⇒ Object

remote_send sends a request to the remote endpoint.

It blocks until the remote endpoint accepts the message.

marshalled.

Parameters:

  • req (Object, String)

    the object to send or it’s marshal form.

  • marshalled (false, true) (defaults to: false)

    indicates if the object is already



190
191
192
193
194
195
196
197
198
# File 'lib/grpc/generic/active_call.rb', line 190

def remote_send(req, marshalled = false)
  logger.debug("sending #{req}, marshalled? #{marshalled}")
  if marshalled
    payload = req
  else
    payload = @marshal.call(req)
  end
  @call.run_batch(@cq, self, INFINITE_FUTURE, SEND_MESSAGE => payload)
end

#request_response(req, **kw) ⇒ Object

request_response sends a request to a GRPC server, and returns the response.

Keyword Arguments ==

any keyword arguments are treated as metadata to be sent to the server if a keyword value is a list, multiple metadata for it’s key are sent

Parameters:

  • req (Object)

    the request sent to the server

Returns:

  • (Object)

    the response received from the server



312
313
314
315
316
317
318
319
320
321
322
# File 'lib/grpc/generic/active_call.rb', line 312

def request_response(req, **kw)
  start_call(**kw) unless @started
  remote_send(req)
  writes_done(false)
  response = remote_read
  finished unless response.is_a? Struct::Status
  response
rescue GRPC::Core::CallError => e
  finished  # checks for Cancelled
  raise e
end

#run_server_bidi(gen_each_reply) ⇒ Object

run_server_bidi orchestrates a BiDi stream processing on a server.

N.B. gen_each_reply is a func(Enumerable<Requests>)

It takes an enumerable of requests as an arg, in case there is a relationship between the stream of requests and the stream of replies.

This does not mean that must necessarily be one. E.g, the replies produced by gen_each_reply could ignore the received_msgs

Parameters:

  • gen_each_reply (Proc)

    generates the BiDi stream replies



428
429
430
431
# File 'lib/grpc/generic/active_call.rb', line 428

def run_server_bidi(gen_each_reply)
  bd = BidiCall.new(@call, @cq, @marshal, @unmarshal, @deadline)
  bd.run_on_server(gen_each_reply)
end

#send_status(code = OK, details = '', assert_finished = false, **kw) ⇒ Object

send_status sends a status to the remote endpoint.

FINISHED.

Keyword Arguments ==

any keyword arguments are treated as metadata to be sent to the server if a keyword value is a list, multiple metadata for it’s key are sent

Parameters:

  • code (int) (defaults to: OK)

    the status code to send

  • details (String) (defaults to: '')

    details

  • assert_finished (true, false) (defaults to: false)

    when true(default), waits for



210
211
212
213
214
215
216
217
# File 'lib/grpc/generic/active_call.rb', line 210

def send_status(code = OK, details = '', assert_finished = false, **kw)
  ops = {
    SEND_STATUS_FROM_SERVER => Struct::Status.new(code, details, kw)
  }
  ops[RECV_CLOSE_ON_SERVER] = nil if assert_finished
  @call.run_batch(@cq, self, INFINITE_FUTURE, ops)
  nil
end

#server_streamer(req, **kw) ⇒ Enumerator|nil

server_streamer sends one request to the GRPC server, which yields a stream of responses.

responses provides an enumerator over the streamed responses, i.e. it follows Ruby’s #each iteration protocol. The enumerator blocks while waiting for each response, stops when the server signals that no further responses will be supplied. If the implicit block is provided, it is executed with each response as the argument and no result is returned.

Keyword Arguments ==

any keyword arguments are treated as metadata to be sent to the server if a keyword value is a list, multiple metadata for it’s key are sent any keyword arguments are treated as metadata to be sent to the server.

Parameters:

  • req (Object)

    the request sent to the server

Returns:

  • (Enumerator|nil)

    a response Enumerator



367
368
369
370
371
372
373
374
375
376
377
# File 'lib/grpc/generic/active_call.rb', line 367

def server_streamer(req, **kw)
  start_call(**kw) unless @started
  remote_send(req)
  writes_done(false)
  replies = enum_for(:each_remote_read_then_finish)
  return replies unless block_given?
  replies.each { |r| yield r }
rescue GRPC::Core::CallError => e
  finished  # checks for Cancelled
  raise e
end

#single_req_viewObject

single_req_view provides a restricted view of this ActiveCall for use in a server request-response handler.



139
140
141
# File 'lib/grpc/generic/active_call.rb', line 139

def single_req_view
  SingleReqView.new(self)
end

#writes_done(assert_finished = true) ⇒ Object

writes_done indicates that all writes are completed.

It blocks until the remote endpoint acknowledges with at status unless assert_finished is set to false. Any calls to #remote_send after this call will fail.

FINISHED.

Parameters:

  • assert_finished (true, false) (defaults to: true)

    when true(default), waits for



157
158
159
160
161
162
163
164
165
# File 'lib/grpc/generic/active_call.rb', line 157

def writes_done(assert_finished = true)
  ops = {
    SEND_CLOSE_FROM_CLIENT => nil
  }
  ops[RECV_STATUS_ON_CLIENT] = nil if assert_finished
  batch_result = @call.run_batch(@cq, self, INFINITE_FUTURE, ops)
  return unless assert_finished
  batch_result.check_status
end