def test_spark_producer(app):
producer_output = []
@app.consumer()
def callback_func(a):
producer_output.append(a)
@app.pipeline()
def spark_simple_pipeline(pipeline, a, b):
return a.subscribe_consumer(callback_func)
@app.spark_producer(spark_simple_pipeline)
def spark_producer():
"""
Simple producer which yields data to `main` pipeline.
More information here: http://stairspy.com/#producer
"""
spark = SparkSession \
.builder \
.getOrCreate()
f = sc.textFile("test.row", 10)
df = spark.read.json(f)
return df
spark_producer.flush()
spark_producer()
spark_simple_pipeline.compile()
run_pipelines(app)
assert len(producer_output) == 20
def test_connected_pipelines_multiple(app):
t = GlobalTestData()
@app.pipeline()
def p_builder(worker, sentence, use_lower):
data = concatenate(sentence=sentence, use_lower=use_lower)
data_with_name = data\
.subscribe_flow(NameExtractionOneWayFlow())
return data_with_name
@app.pipeline()
def p_builder_general(worker, sentence, use_lower):
data = concatenate(sentence=sentence, use_lower=use_lower)
v1 = data.subscribe_pipeline(p_builder)\
.subscribe_flow(NameExtractionFlowMultiple(use_lower=use_lower))
v2 = data.subscribe_pipeline(p_builder)\
.subscribe_flow(NameExtractionFlowMultiple(use_lower=use_lower))
return concatenate(v1=v1, v2=v2).subscribe_func(t.save_one_item)
p_builder.compile()
p_builder_general.compile()
p_builder_general(sentence="Oleg", use_lower=True)
run_pipelines(app)
result = t.get_result()
assert result['v1']['names'][0] == "oleg"
assert result['v2']['names'][0] == "oleg"
def test_multiple_pipelines(app):
t = GlobalTestData()
@app.pipeline()
def callback_func(pipeline, x):
return x.subscribe_func(t.save_multiple_items)
@app.pipeline()
def callback_func2(pipeline, x):
return x.subscribe_func(t.save_multiple_items)
@app.producer(callback_func, callback_func2)
def simple_producer():
for i in range(100):
yield dict(x=1)
callback_func.compile()
callback_func2.compile()
simple_producer.flush()
simple_producer.run()
run_pipelines(app)
assert len(t.get_result()) == 200
def test_make_add_value_with_flow(app):
t = GlobalTestData()
@app.pipeline(config=dict(path='123'))
def p_builder3(worker, sentence2):
return sentence2\
.rename(sentence=sentence2)\
.add_value(path=worker.config['path'])\
.subscribe_flow(NameExtractionOneWayFlow(use_lower=True))\
.subscribe_func(t.save_one_item)
@app.pipeline()
def p_builder_general3(worker, sentence2: DataFrame):
return sentence2.subscribe_pipeline(p_builder3,
config=dict(path='321'))
p_builder3.compile()
p_builder_general3.compile()
p_builder_general3(sentence2="Oleg")
run_pipelines(app)
result = t.get_result()
assert check_keys(result.keys(), ['sentence', 'path', 'names'])
assert result['names'][0] == "oleg"
assert result['path'] == '321'
def test_worker_producer(app):
t = GlobalTestData()
@app.pipeline()
def simple_pipeline(pipeline, x):
return x.subscribe_func(t.save_multiple_items)
@app.producer(simple_pipeline)
def batch_handler(x):
yield dict(x=x)
@app.batch_producer(batch_handler)
def worker_producer():
for i in range(10):
# here we should yield batch of data
yield dict(x=1)
simple_pipeline.compile()
worker_producer()
try:
worker_producer()
run_jobs_processor(app.project,
[batch_handler],
die_when_empty=True)
except SystemExit:
pass
run_pipelines(app)
assert len(t.get_result()) == 20 # 10 batches with 2 jobs
def test_connected_pipelines(app):
@app.consumer_iter()
def save_globaly(func1, func2, **kwargs):
return dict(func1=func1, func2=func2)
@app.pipeline()
def p_builder(worker, sentence):
return sentence \
.subscribe_func(lambda sentence: dict(func1="ok"), name='root1') \
.subscribe_func_as_producer(
lambda sentence: [dict(func2="ok"), dict(func2="ok")], name='root2') \
.subscribe_consumer(save_globaly)
p_builder.compile()
p_builder(sentence="Oleg")
run_pipelines(app)
it = save_globaly.iter(die_when_empty=True)
result = next(it)
assert len(result) == 2
def test_connected_pipelines(app):
@app.pipeline()
def p_builder(worker, sentence, use_lower):
data = concatenate(sentence=sentence, use_lower=use_lower)
data_with_name = data \
.subscribe_flow(NameExtractionOneWayFlow(),
as_worker=False)
return data_with_name
@app.pipeline()
def p_builder_general(worker, sentence, use_lower):
data = concatenate(sentence=sentence, use_lower=use_lower)
return data.subscribe_pipeline(p_builder)\
.subscribe_flow(NameExtractionFlowMultiple(),
as_worker=False)
p_builder.compile()
p_builder_general.compile()
p_builder_general(sentence="Oleg", use_lower=True)
run_pipelines(app)
def test_deep_tree_functions(app):
t = GlobalTestData()
@app.consumer()
def save_globaly(name, result, **kwargs):
t.save_multiple_items(**{name: result})
@app.pipeline()
def p_builder(worker, sentence):
root_branch = sentence \
.subscribe_func(lambda sentence: dict(func1="ok"), name='root1') \
.subscribe_func(lambda sentence: dict(func2="ok"), name='root2')
return concatenate(branch_1=root_branch.subscribe_pipeline(branch_1),
branch_2=root_branch.subscribe_pipeline(branch_2),
branch_3=root_branch.subscribe_pipeline(branch_3))
@app.pipeline()
def branch_1(work, func1, func2):
root_branch = concatenate(func1=func1, func2=func2)
return root_branch \
.add_value(name='branch_1') \
.subscribe_func(lambda func1, func2: dict(func1_1="ok"),
name='branch_1_1') \
.subscribe_func(lambda func1, func2: dict(result="branch_1"),
name='branch_1_2')\
.subscribe_consumer(save_globaly)
@app.pipeline()
def branch_2(work, func1, func2):
root_branch = concatenate(func1=func1, func2=func2)
return root_branch \
.add_value(name='branch_2') \
.subscribe_func(lambda func1, func2: dict(func1_1="ok"),
name='branch_2_1') \
.subscribe_func(lambda func1, func2: dict(result="branch_2"),
name='branch_2_2')\
.subscribe_consumer(save_globaly)
@app.pipeline()
def branch_3(work, func1, func2):
root_branch = concatenate(func1=func1, func2=func2)
return root_branch \
.add_value(name='branch_3') \
.subscribe_func(lambda func1, func2: dict(func1_1="ok"),
name='branch_3_1') \
.subscribe_func(lambda func1, func2: dict(result="branch_3"),
name='branch_3_2')\
.subscribe_consumer(save_globaly)
p_builder.compile()
branch_1.compile()
branch_2.compile()
branch_3.compile()
p_builder(sentence="Oleg", use_lower=True)
run_pipelines(app)
result = t.get_result()
assert len(result) == 3
assert list(result[0].keys()) == list(result[0].values())
assert list(result[1].keys()) == list(result[1].values())
assert list(result[2].keys()) == list(result[2].values())
