def test_pipeline(request, test_data_dir):
filename = str(request.fspath.dirpath('test_dedup_chunk_counts.yaml'))
with yakonfig.defaulted_config([streamcorpus_pipeline], filename=filename):
## config says read from stdin, so make that have what we want
stdin = sys.stdin
sys.stdin = StringIO(get_test_chunk_path(test_data_dir))
## run the pipeline
stages = PipelineStages()
pf = PipelineFactory(stages)
p = pf(yakonfig.get_global_config('streamcorpus_pipeline'))
from streamcorpus_pipeline.run import SimpleWorkUnit
work_unit = SimpleWorkUnit('long string indicating source of text')
work_unit.data['start_chunk_time'] = time.time()
work_unit.data['start_count'] = 0
g = gevent.spawn(p._process_task, work_unit)
gevent.sleep(5)
with pytest.raises(SystemExit): # pylint: disable=E1101
p.shutdown(sig=signal.SIGTERM)
logger.debug('now joining...')
timeout = gevent.Timeout(1)
g.join(timeout=timeout)
def test_dedup_chunk_counts(request, test_data_dir, tmpdir):
filename = str(request.fspath.dirpath('test_dedup_chunk_counts.yaml'))
with yakonfig.defaulted_config([streamcorpus_pipeline],
filename=filename,
config={'tmp_dir_path':
str(tmpdir)}) as config:
## run the pipeline
pf = PipelineFactory(PipelineStages())
p = pf(config['streamcorpus_pipeline'])
p.run(get_test_chunk_path(test_data_dir))
def test_spinn3r_pipeline(filename, urls, pipeline_config, output_file):
"""minimal end-to-end test, with a fixed pipeline"""
with yakonfig.defaulted_config([streamcorpus_pipeline],
config=pipeline_config):
stages = PipelineStages()
factory = PipelineFactory(stages)
pipeline = factory(yakonfig.get_global_config('streamcorpus_pipeline'))
work_unit = SimpleWorkUnit(filename)
work_unit.data['start_chunk_time'] = 0
work_unit.data['start_count'] = 0
pipeline._process_task(work_unit)
with Chunk(path=output_file, mode='rb') as chunk:
assert [si.abs_url for si in chunk] == urls
def test_spinn3r_pipeline_bogus_prefetched(filename, pipeline_config):
"""supply known-bad prefetched data"""
with yakonfig.defaulted_config([streamcorpus_pipeline],
config=pipeline_config):
stages = PipelineStages()
factory = PipelineFactory(stages)
pipeline = factory(yakonfig.get_global_config('streamcorpus_pipeline'))
key = filename
from_spinn3r_feed._prefetched[key] = 'bogus data, dude!'
work_unit = SimpleWorkUnit(key)
work_unit.data['start_chunk_time'] = 0
work_unit.data['start_count'] = 0
with pytest.raises(DecodeError):
pipeline._process_task(work_unit)
def main():
import argparse
parser = argparse.ArgumentParser(
description='process a sequence of stream items',
usage='streamcorpus_pipeline --config config.yaml --input file.in')
parser.add_argument('-i', '--input', action='append',
help='file paths to input instead of reading from stdin')
parser.add_argument('--in-glob', action='append', default=[], help='path glob specifying input files')
parser.add_argument('--third-dir-path', help='path to third-party tools directory')
parser.add_argument('--tmp-dir-path', help='path to temporary directory for scratch files, can be large')
modules = [yakonfig, kvlayer, dblogger, streamcorpus_pipeline]
args = yakonfig.parse_args(parser, modules)
config = yakonfig.get_global_config()
## this modifies the global config, passed by reference
instantiate_config(config)
input_paths = []
if args.in_glob:
for pattern in args.in_glob:
input_paths.extend(glob.glob(pattern))
if args.input:
if '-' in args.input:
if args.in_glob:
sys.exit('cannot use "-i -" and --in-glob together')
if len(args.input) > 1:
sys.exit('cannot use "-i -" with multiple inputs')
input_paths = sys.stdin
else:
input_paths.extend(args.input)
scp_config = config['streamcorpus_pipeline']
stages = PipelineStages()
if 'external_stages_path' in scp_config:
stages.load_external_stages(scp_config['external_stages_path'])
if 'external_stages_modules' in scp_config:
for mod in scp_config['external_stages_modules']:
stages.load_module_stages(mod)
factory = PipelineFactory(stages)
pipeline = factory(scp_config)
for i_str in input_paths:
work_unit = SimpleWorkUnit(i_str.strip())
work_unit.data['start_chunk_time'] = time.time()
work_unit.data['start_count'] = 0
pipeline._process_task(work_unit)
## explicitly call cleanup, which is idempotent
pipeline.cleanup()
def test_spinn3r_pipeline_unprefetched(urls, pipeline_config):
"""minimal end-to-end test, missing prefetched data"""
pipeline_config['streamcorpus_pipeline']['from_spinn3r_feed'] = {
'use_prefetched': True
}
with yakonfig.defaulted_config([streamcorpus_pipeline],
config=pipeline_config):
stages = PipelineStages()
factory = PipelineFactory(stages)
pipeline = factory(yakonfig.get_global_config('streamcorpus_pipeline'))
key = 'test_file.bin'
work_unit = SimpleWorkUnit(key)
work_unit.data['start_chunk_time'] = 0
work_unit.data['start_count'] = 0
with pytest.raises(ConfigurationError):
pipeline._process_task(work_unit)
def test_spinn3r_pipeline_filter_no_matches(filename, pipeline_config,
output_file):
"""set a publisher_type filter that matches nothing in the feed"""
pipeline_config['streamcorpus_pipeline']['from_spinn3r_feed'] = {
'publisher_type': 'MICROBLOG'
}
with yakonfig.defaulted_config([streamcorpus_pipeline],
config=pipeline_config):
stages = PipelineStages()
factory = PipelineFactory(stages)
pipeline = factory(yakonfig.get_global_config('streamcorpus_pipeline'))
work_unit = SimpleWorkUnit(filename)
work_unit.data['start_chunk_time'] = 0
work_unit.data['start_count'] = 0
pipeline._process_task(work_unit)
# no chunks means the output file won't actually get written
assert not os.path.exists(output_file)
def test_spinn3r_pipeline_filter_matches(filename, urls, pipeline_config,
output_file):
"""set a publisher_type filter that matches everything in the feed"""
pipeline_config['streamcorpus_pipeline']['from_spinn3r_feed'] = {
'publisher_type': 'WEBLOG'
}
with yakonfig.defaulted_config([streamcorpus_pipeline],
config=pipeline_config):
stages = PipelineStages()
factory = PipelineFactory(stages)
pipeline = factory(yakonfig.get_global_config('streamcorpus_pipeline'))
work_unit = SimpleWorkUnit(filename)
work_unit.data['start_chunk_time'] = 0
work_unit.data['start_count'] = 0
pipeline._process_task(work_unit)
with Chunk(path=output_file, mode='rb') as chunk:
assert [si.abs_url for si in chunk] == urls
def test_spinn3r_pipeline_ignore_prefetched(filename, urls, pipeline_config,
output_file):
"""configuration explicitly ignores bad prefetched data"""
pipeline_config['streamcorpus_pipeline']['from_spinn3r_feed'] = {
'use_prefetched': False
}
with yakonfig.defaulted_config([streamcorpus_pipeline],
config=pipeline_config):
stages = PipelineStages()
factory = PipelineFactory(stages)
pipeline = factory(yakonfig.get_global_config('streamcorpus_pipeline'))
key = filename
from_spinn3r_feed._prefetched[key] = 'bogus data, dude!'
work_unit = SimpleWorkUnit(key)
work_unit.data['start_chunk_time'] = 0
work_unit.data['start_count'] = 0
pipeline._process_task(work_unit)
del from_spinn3r_feed._prefetched[key]
with Chunk(path=output_file, mode='rb') as chunk:
assert [si.abs_url for si in chunk] == urls
def rejester_run_function(work_unit):
with yakonfig.defaulted_config([kvlayer, streamcorpus_pipeline],
config=work_unit.spec.get('config', {})):
scp_config = yakonfig.get_global_config('streamcorpus_pipeline')
stages = PipelineStages()
if 'external_stages_path' in scp_config:
stages.load_external_stages(scp_config['external_stages_path'])
if 'external_stages_modules' in scp_config:
for mod in scp_config['external_stages_modules']:
stages.load_module_stages(mod)
factory = PipelineFactory(stages)
pipeline = factory(scp_config)
work_unit.data['start_chunk_time'] = time.time()
work_unit.data['start_count'] = 0
pipeline._process_task(work_unit)
## explicitly call cleanup, which is idempotent and might not
## get called by atexit if we are running under
## multiprocessing
pipeline.cleanup()
def test_spinn3r_pipeline_prefetched(filename, urls, pipeline_config,
output_file):
"""minimal end-to-end test, preloading data in the loader"""
pipeline_config['streamcorpus_pipeline']['from_spinn3r_feed'] = {
'use_prefetched': True
}
with yakonfig.defaulted_config([streamcorpus_pipeline],
config=pipeline_config):
stages = PipelineStages()
factory = PipelineFactory(stages)
pipeline = factory(yakonfig.get_global_config('streamcorpus_pipeline'))
key = 'test_file.bin'
with open(filename, 'rb') as f:
from_spinn3r_feed._prefetched[key] = f.read()
work_unit = SimpleWorkUnit(key)
work_unit.data['start_chunk_time'] = 0
work_unit.data['start_count'] = 0
pipeline._process_task(work_unit)
del from_spinn3r_feed._prefetched[key]
with Chunk(path=output_file, mode='rb') as chunk:
assert [si.abs_url for si in chunk] == urls