def extract_features(pid,job):
def log_callback(msg):
logging.warn('FEATURES %s' % msg)
client.heartbeat(pid,message=msg)
parsed_pid = parse_pid(pid)
bin_lid = parsed_pid[LID]
bin_pid = ''.join([parsed_pid[NAMESPACE], parsed_pid[LID]])
binzip_url = ''.join([bin_pid,'_binzip.zip'])
blob_url = ''.join([bin_pid,'_blob.zip'])
features_url = ''.join([bin_pid,'_features.csv'])
multiblob_url = ''.join([bin_pid,'_multiblob.csv'])
if exists(features_url):
log_callback('skipping %s - features exist' % pid)
return
log_callback('computing features for %s' % pid)
with safe_tempdir() as binzip_dir:
# download bin zip
binzip_path = os.path.join(binzip_dir, '%s.zip' % bin_lid)
log_callback('downloading %s to %s' % (binzip_url, binzip_path))
download(binzip_url, binzip_path)
# download blob zip
blob_path = os.path.join(binzip_dir, '%s_blob.zip' % bin_lid)
log_callback('downloading %s to %s' % (blob_url, blob_path))
download(blob_url, blob_path)
# compute features
with safe_tempdir() as job_dir:
# output of matlab job
feature_csv = os.path.join(job_dir, csvname(bin_pid))
multiblob_csv = os.path.join(job_dir, 'multiblob', multiblobname(bin_pid))
# params for matlab job
namespace = os.path.dirname(binzip_path) + '/'
lid = os.path.basename(binzip_path)
matlab = Matlab(MATLAB_EXEC_PATH, MATLAB_PATH, output_callback=log_callback)
cmd = 'bin_features(\'%s\',\'%s\',\'%s\',\'chatty\')' % (namespace, lid, job_dir + '/')
log_callback('running %s' % cmd)
matlab.run(cmd)
log_callback('matlab exited')
if os.path.exists(feature_csv):
log_callback('features found at %s' % feature_csv)
else:
raise Exception('no features found')
log_callback('uploading %s' % features_url)
upload(feature_csv, features_url)
if os.path.exists(multiblob_csv):
log_callback('multiblob found at %s' % multiblob_csv)
log_callback('uploading %s' % multiblob_url)
upload(multiblob_csv, multiblob_url)
log_callback('complete')
client.wakeup()
def extract_blobs(pid,job):
def log_callback(msg):
logging.warn('BLOBS %s' % msg)
client.heartbeat(pid,message=msg)
parsed_pid = parse_pid(pid)
bin_lid = parsed_pid[LID]
bin_pid = ''.join([parsed_pid[NAMESPACE], parsed_pid[LID]])
binzip_url = ''.join([bin_pid,'_binzip.zip'])
binzip_file = os.path.basename(binzip_url)
deposit_url = '%s_blobs.zip' % bin_pid
if exists(deposit_url):
log_callback('skipping %s - blobs exist' % pid)
return
log_callback('computing blobs for %s' % pid)
with safe_tempdir() as binzip_dir:
# first, copy the zipfile to a temp dir
binzip_path = os.path.join(binzip_dir, '%s.zip' % bin_lid)
log_callback('downloading %s to %s' % (binzip_url, binzip_path))
download(binzip_url, binzip_path)
# now run bin_blobs
with safe_tempdir() as job_dir:
# configure matlab
matlab = Matlab(MATLAB_EXEC_PATH, MATLAB_PATH, output_callback=log_callback)
# run command
blobs_file = os.path.join(job_dir, blob_zip_name(bin_pid))
cmd = 'bin_blobs(\'%s\',\'%s\',\'%s\')' % (bin_pid, binzip_path, job_dir)
log_callback('running %s' % cmd)
matlab.run(cmd)
log_callback('MATLAB done, checking for %s' % blobs_file)
if not os.path.exists(blobs_file):
raise Exception('missing output file')
log_callback('depositing %s' % blobs_file)
upload(blobs_file, deposit_url)
log_callback('deposited %s' % blobs_file)
log_callback('completed %s' % bin_pid)
client.wakeup()
