问题
I'm trying to transfer samples from thread A ("Acquisition") to thread B ("P300") using queue but I can't read any data in thread B, although samples are being allocated in thread A. Judging by my output, I think my thread B is rushing and testing things before my thread A starts to put data in.
See an approximation of my code structure bellow:
import threading
import queue
from queue import Empty
import numpy as np
import warnings
warnings.filterwarnings("error")
class AcqThread(threading.Thread):
def __init__(self, dataOutQ1, dataOutQ2, stopQ1, stopQ2, saveQ):
threading.Thread.__init__(self)
self.stopQ2 = stopQ2
self.stopQ1 = stopQ1
self.dataOutQ2 = dataOutQ2
self.dataOutQ1 = dataOutQ1
self.saveQ = saveQ
def run(self):
Acquisition(inlet, self.dataOutQ1, self.dataOutQ2, self.stopQ1, self.stopQ2, self.saveQ)
class P300Thread(threading.Thread):
def __init__(self, dataInQ, featureQ, stopQ):
threading.Thread.__init__(self)
self.dataInQ = dataInQ
self.featureQ = featureQ
self.stopQ = stopQ
def run(self):
P300fun(self.dataInQ, self.featureQ, self.stopQ)
threadLock = threading.Lock()
SaveQ = queue.Queue()
DataOutQ1 = queue.Queue()
DataOutQ2 = queue.Queue()
StopQ1 = queue.Queue()
StopQ2 = queue.Queue()
FeatQ1 = queue.Queue()
StopQ1.put(0)
StopQ2.put(0)
#
AcqTh = AcqThread(DataOutQ1, DataOutQ2, StopQ1, StopQ2, SaveQ)
P300Th = P300Thread(DataOutQ1, FeatQ1, StopQ1)
def Acquisition(inlet, dataOutQ1, dataOutQ2, stopQ1, stopQ2, saveQ):
i = 0
print('Starting...')
while i<1250: #i is the number of samples
sample, timestamp = inlet.pull_sample() #samples coming in @ 250Hz
##Normalization, filtering##
threadLock.acquire()
dataOutQ1.put([filtsamples[:,-250:], rawtimestamps[-250:]]) #I only need the last 250 samples
threadLock.release()
i += 1
def P300fun(dataInQ, featureQ, stopQ):
p300sample = []
p300timestamp = []
print(f"Is DataInQ size true? {DataOutQ1.qsize()}")
print("Is dataInQ emtpy?", DataOutQ1.empty())
while dataInQ.qsize(): #or while not dataqueue.empty():
try:
print("DataInQ has data")
ss, ts = dataInQ.get(0)
print('<>P300\n>>Samples [', ss, ']\nTimestamp [', ts, ']')
except Empty:
return
print('Thread Finished')
if __name__ == '__main__':
print('Looking for an EEG stream...')
streams = resolve_stream('type', 'EEG')
inlet = StreamInlet(streams[0])
print('Connected!\n')
AcqTh.start()
P300Th.start()
AcqTh.join()
P300Th.join()
print("\n\n>>>DONE<<<\n\n")
And output:
Looking for an EEG stream...
Connected!
Is DataInQ size true? 0
Starting...
Is dataInQ emtpy? True
Thread Finished
>>>DONE<<<
In my research, question 1 appeared to present a similar problem, but it seems that the problem was in the image processing part (and they use the multiprocessing package). Question 2 seems to have a concurrency problem, which might be my problem, but I'm not sure how to translate it to my problemlet me know if I'm wrong, tho). Question 3 had just a problem with the order of arguments, so not applicable here, I think.
How should I go about this? Should I recurrently call thread B from within thread A?? Do I need a loop or a delay on thread B? Is there some issue with the .join() part maybe? I'll need to add more threads in a near future, so it would be good to figure out how to work with only two first...
All help is appreaciated!
回答1:
Being a noob can be tricky... So I'll answer my own question to help other beginners that may come across this issue too.
Well, first things first: no, it's not possible to call a thread from within a thread recurrently, because each thread can only be called once.
But there is a way to prevent the thread from ending, making them wait for triggers that will allow them to continue. After some more research, I came across this question that showed me there is a way to create events for threads. The documentation can be found here. And it's quite straight forward: the event objects behave like flags and can be set() (indicating True) or clear() (indicating False, which is the original value). To test an event, one can use the is_set() method for boolean problems or use the wait() method instead of a timer. In my case, it saved me some queues I was going to use:
import threading
import queue
from queue import Empty
import numpy as np
class AcqThread(threading.Thread):
def __init__(self, dataOutQ1, dataOutQ2, saveQ):
threading.Thread.__init__(self)
self.dataOutQ2 = dataOutQ2
self.dataOutQ1 = dataOutQ1
self.saveQ = saveQ
def run(self):
Acquisition(inlet, self.dataOutQ1, self.dataOutQ2, self.saveQ)
class P300Thread(threading.Thread):
def __init__(self, dataInQ, featureQ):
threading.Thread.__init__(self)
self.dataInQ = dataInQ
self.featureQ = featureQ
def run(self):
P300fun(self.dataInQ, self.featureQ)
threadLock = threading.Lock()
SaveQ = queue.Queue()
DataOutQ1 = queue.Queue()
DataOutQ2 = queue.Queue()
FeatQ1 = queue.Queue()
FeatQ2 = queue.Queue()
#NEW:: initializes Events
E = threading.Event()
EP300 = threading.Event()
#
AcqTh = AcqThread(DataOutQ1, DataOutQ2, SaveQ)
P300Th = P300Thread(DataOutQ1, FeatQ1)
And it allows me to "call" thread B "recurrently", as it keeps my first while active (because of event E) and enters the processing part only when the event EP300 is set. Then, EP300 is cleared after the process is done:
def Acquisition(inlet, dataOutQ1, dataOutQ2 saveQ):
i = 0
print('Starting...')
while i<1250:
sample, timestamp = inlet.pull_sample()
##Normalization, filtering##
if _condition_:
threadLock.acquire()
dataOutQ1.put([filtsamples[:,-250:], rawtimestamps[-250:]])
threadLock.release()
EP300.set() #NEW:: allows the P300 function to collect data from queue
i += 1
E.set() #NEW:: flaggs end data collection
def P300fun(dataInQ, featureQ):
p300sample = []
p300timestamp = []
while not E.is_set(): #NEW:: loop until collection is ended
if EP300.is_set(): #NEW:: activated when Event is triggered
while dataInQ.qsize():
try:
print("DataInQ has data")
ss, ts = dataInQ.get(0)
print('<>P300\n>>Samples [', ss, ']\nTimestamp [', ts, ']')
except Empty:
return
if not E.is_set(): #NEW:: Event is cleared in case data collection is not over, waiting for a new set()
EP300.clear()
print('Thread Finished')
if __name__ == '__main__':
print('Looking for an EEG stream...')
streams = resolve_stream('type', 'EEG')
inlet = StreamInlet(streams[0])
print('Connected!\n')
AcqTh.start()
P300Th.start()
AcqTh.join()
P300Th.join()
print("\n\n>>>DONE<<<\n\n")
来源：https://stackoverflow.com/questions/64932804/can-i-call-a-thread-recurrently-from-within-a-thread