def submit(self, job, test=False):
""" Submit the job to the device. If the device is currently running
a job it will be queued and run when this is finished.
This handles iteration over the path model defined by the job and
sending commands to the actual device using roughly the procedure is
as follows:
device.connect()
model = device.init(job)
for cmd in device.process(model):
device.handle(cmd)
device.finish()
device.disconnect()
Subclasses provided by your own DeviceDriver may reimplement this
to handle path interpolation however needed. The return value is
ignored.
The live plot view will update whenever the device.position object
is updated. On devices with lower cpu/gpu capabilities this should
be updated sparingly (ie the raspberry pi).
Parameters
-----------
job: Instance of `inkcut.job.models.Job`
The job to execute on the device
test: bool
Do a test run. This specifies whether the commands should be
sent to the actual device or not. If True, the connection will
be replaced with a virtual connection that captures all the
command output.
"""
log.debug("device | submit {}".format(job))
try:
#: Only allow one job at a time
if self.busy:
queue = self.queue[:]
queue.append(job)
self.queue = queue #: Copy and reassign so the UI updates
log.info("Job {} put in device queue".format(job))
return
with self.device_busy():
#: Set the current the job
self.job = job
self.status = "Initializing job"
#: Get the time to sleep based for each unit of movement
config = self.config
#: Rate px/ms
if config.custom_rate >= 0:
rate = config.custom_rate
elif self.connection.always_spools or config.spooled:
rate = 0
elif config.interpolate:
if config.step_time > 0:
rate = config.step_size / float(config.step_time)
else:
rate = 0 # Undefined
else:
rate = from_unit(
config.speed, # in/s or cm/s
config.speed_units.split("/")[0]) / 1000.0
# Device model is updated in real time
model = yield defer.maybeDeferred(self.init, job)
#: Local references are faster
info = job.info
#: Determine the length for tracking progress
whole_path = QtGui.QPainterPath()
#: Some versions of Qt seem to require a value in
#: toSubpathPolygons
m = QtGui.QTransform.fromScale(1, 1)
for path in model.toSubpathPolygons(m):
for i, p in enumerate(path):
whole_path.lineTo(p)
total_length = whole_path.length()
total_moved = 0
log.debug("device | Path length: {}".format(total_length))
#: So a estimate of the duration can be determined
info.length = total_length
info.speed = rate * 1000 #: Convert to px/s
#: Waiting for approval
info.status = 'waiting'
#: If marked for auto approve start now
if info.auto_approve:
info.status = 'approved'
else:
#: Check for approval before starting
yield defer.maybeDeferred(info.request_approval)
if info.status != 'approved':
self.status = "Job cancelled"
return
#: Update stats
info.status = 'running'
info.started = datetime.now()
self.status = "Connecting to device"
with self.device_connection(test
or config.test_mode) as connection:
self.status = "Processing job"
try:
yield defer.maybeDeferred(self.connect)
#: Write startup command
if config.commands_before:
yield defer.maybeDeferred(connection.write,
config.commands_before)
self.status = "Working..."
#: For point in the path
for (d, cmd, args, kwargs) in self.process(model):
#: Check if we paused
if info.paused:
self.status = "Job paused"
#: Sleep until resumed, cancelled, or the
#: connection drops
while (info.paused and not info.cancelled
and connection.connected):
yield async_sleep(300) # ms
#: Check for cancel for non interpolated jobs
if info.cancelled:
self.status = "Job cancelled"
info.status = 'cancelled'
break
elif not connection.connected:
self.status = "connection error"
info.status = 'error'
break
#: Invoke the command
#: If you want to let the device handle more complex
#: commands such as curves do it in process and handle
yield defer.maybeDeferred(cmd, *args, **kwargs)
total_moved += d
#: d should be the device must move in px
#: so wait a proportional amount of time for the device
#: to catch up. This avoids buffer errors from dumping
#: everything at once.
#: Since sending is way faster than cutting
#: we must delay (without blocking the UI) before
#: sending the next command or the device's buffer
#: quickly gets filled and crappy china piece cutters
#: get all jacked up. If the transport sends to a spooled
#: output (such as a printer) this can be set to 0
if rate > 0:
# log.debug("d={}, delay={} t={}".format(
# d, delay, d/delay
# ))
yield async_sleep(d / rate)
#: TODO: Check if we need to update the ui
#: Set the job progress based on how far we've gone
if total_length > 0:
info.progress = int(
max(
0,
min(100,
100 * total_moved / total_length)))
if info.status != 'error':
#: We're done, send any finalization commands
yield defer.maybeDeferred(self.finish)
#: Write finalize command
if config.commands_after:
yield defer.maybeDeferred(connection.write,
config.commands_after)
#: Update stats
info.ended = datetime.now()
#: If not cancelled or errored
if info.status == 'running':
info.done = True
info.status = 'complete'
except Exception as e:
log.error(traceback.format_exc())
raise
finally:
if connection.connected:
yield defer.maybeDeferred(self.disconnect)
#: Set the origin
if job.feed_to_end and job.info.status == 'complete':
self.origin = self.position
#: If the user didn't cancel, set the origin and
#: Process any jobs that entered the queue while this was running
if self.queue and not job.info.cancelled:
queue = self.queue[:]
job = queue.pop(0) #: Pull the first job off the queue
log.info("Rescheduling {} from queue".format(job))
self.queue = queue #: Copy and reassign so the UI updates
#: Call a minute later
timed_call(60000, self.submit, job)
except Exception as e:
log.error(' device | Execution error {}'.format(
traceback.format_exc()))
raise
def move(self, x, y, z, absolute=True):
log.debug("protocol.move({x},{y},{z})".format(x=x, y=y, z=z))
#: Wait some time before we get there
return async_sleep(0.1)
def submit(self, job, test=False):
""" Submit the job to the device. If the device is currently running
a job it will be queued and run when this is finished.
This handles iteration over the path model defined by the job and
sending commands to the actual device using roughly the procedure is
as follows:
device.connect()
model = device.init(job)
for cmd in device.process(model):
device.handle(cmd)
device.finish()
device.disconnect()
Subclasses provided by your own DeviceDriver may reimplement this
to handle path interpolation however needed. The return value is
ignored.
The live plot view will update whenever the device.position object
is updated. On devices with lower cpu/gpu capabilities this should
be updated sparingly (ie the raspberry pi).
Parameters
-----------
job: Instance of `inkcut.job.models.Job`
The job to execute on the device
test: bool
Do a test run. This specifies whether the commands should be
sent to the actual device or not. If True, the connection will
be replaced with a virtual connection that captures all the
command output.
"""
log.debug("device | submit {}".format(job))
try:
#: Only allow one job at a time
if self.busy:
queue = self.queue[:]
queue.append(job)
self.queue = queue #: Copy and reassign so the UI updates
log.info("Job {} put in device queue".format(job))
return
with self.device_busy():
#: Set the current the job
self.job = job
self.status = "Initializing job"
#: Get the time to sleep based for each unit of movement
config = self.config
#: Rate px/ms
if config.custom_rate >= 0:
rate = config.custom_rate
elif self.connection.always_spools or config.spooled:
rate = 0
elif config.interpolate:
if config.step_time > 0:
rate = config.step_size/float(config.step_time)
else:
rate = 0 # Undefined
else:
rate = from_unit(
config.speed, # in/s or cm/s
config.speed_units.split("/")[0])/1000.0
# Device model is updated in real time
model = yield defer.maybeDeferred(self.init, job)
#: Local references are faster
info = job.info
#: Determine the length for tracking progress
whole_path = QtGui.QPainterPath()
#: Some versions of Qt seem to require a value in
#: toSubpathPolygons
m = QtGui.QTransform.fromScale(1, 1)
for path in model.toSubpathPolygons(m):
for i, p in enumerate(path):
whole_path.lineTo(p)
total_length = whole_path.length()
total_moved = 0
log.debug("device | Path length: {}".format(total_length))
#: So a estimate of the duration can be determined
info.length = total_length
info.speed = rate*1000 #: Convert to px/s
#: Waiting for approval
info.status = 'waiting'
#: If marked for auto approve start now
if info.auto_approve:
info.status = 'approved'
else:
#: Check for approval before starting
yield defer.maybeDeferred(info.request_approval)
if info.status != 'approved':
self.status = "Job cancelled"
return
#: Update stats
info.status = 'running'
info.started = datetime.now()
self.status = "Connecting to device"
with self.device_connection(
test or config.test_mode) as connection:
self.status = "Processing job"
try:
yield defer.maybeDeferred(self.connect)
#: Write startup command
if config.commands_before:
yield defer.maybeDeferred(connection.write,
config.commands_before)
self.status = "Working..."
#: For point in the path
for (d, cmd, args, kwargs) in self.process(model):
#: Check if we paused
if info.paused:
self.status = "Job paused"
#: Sleep until resumed, cancelled, or the
#: connection drops
while (info.paused and not info.cancelled
and connection.connected):
yield async_sleep(300) # ms
#: Check for cancel for non interpolated jobs
if info.cancelled:
self.status = "Job cancelled"
info.status = 'cancelled'
break
elif not connection.connected:
self.status = "connection error"
info.status = 'error'
break
#: Invoke the command
#: If you want to let the device handle more complex
#: commands such as curves do it in process and handle
yield defer.maybeDeferred(cmd, *args, **kwargs)
total_moved += d
#: d should be the device must move in px
#: so wait a proportional amount of time for the device
#: to catch up. This avoids buffer errors from dumping
#: everything at once.
#: Since sending is way faster than cutting
#: we must delay (without blocking the UI) before
#: sending the next command or the device's buffer
#: quickly gets filled and crappy china piece cutters
#: get all jacked up. If the transport sends to a spooled
#: output (such as a printer) this can be set to 0
if rate > 0:
# log.debug("d={}, delay={} t={}".format(
# d, delay, d/delay
# ))
yield async_sleep(d/rate)
#: TODO: Check if we need to update the ui
#: Set the job progress based on how far we've gone
if total_length > 0:
info.progress = int(max(0, min(100,
100*total_moved/total_length)))
if info.status != 'error':
#: We're done, send any finalization commands
yield defer.maybeDeferred(self.finish)
#: Write finalize command
if config.commands_after:
yield defer.maybeDeferred(connection.write,
config.commands_after)
#: Update stats
info.ended = datetime.now()
#: If not cancelled or errored
if info.status == 'running':
info.done = True
info.status = 'complete'
except Exception as e:
log.error(traceback.format_exc())
raise
finally:
if connection.connected:
yield defer.maybeDeferred(self.disconnect)
#: Set the origin
if job.feed_to_end and job.info.status == 'complete':
self.origin = self.position
#: If the user didn't cancel, set the origin and
#: Process any jobs that entered the queue while this was running
if self.queue and not job.info.cancelled:
queue = self.queue[:]
job = queue.pop(0) #: Pull the first job off the queue
log.info("Rescheduling {} from queue".format(job))
self.queue = queue #: Copy and reassign so the UI updates
#: Call a minute later
timed_call(60000, self.submit, job)
except Exception as e:
log.error(' device | Execution error {}'.format(
traceback.format_exc()))
raise