print ffs
# <codecell>
report = {}
tasks = ['PegTx','Cutting','Suturing','ClipApply']
%load_ext autoreload
%autoreload
for f in ffs:
print f
uid = f.split('.')[-3]
task = int(f.split('.')[-2])
if not report.get(uid, False): report[uid] = {'Summary':[0,0,0,0], tasks[0]:[], tasks[1]:[], tasks[2]:[], tasks[3]:[]}
data, meta = myS3.getData(bucket, f, labeled=True)
#Compute Summary Metrics
jsonSimscore = vm.summary_metrics(meta, data, conn)
jsonSimscore = vm.data_metrics_append(jsonSimscore, data, f)
jsonSimscore = vm.machine_health_append(jsonSimscore, meta, data)
failtypes = jsonSimscore['FailTypes']
report[uid]['Summary'][task] += 1
report[uid][tasks[task]].append([f, failtypes])
# <codecell>
import json
from openpyxl.workbook import Workbook
from openpyxl.writer.excel import ExcelWriter
import numpy as np
import scoring
import fetch.myS3 as myS3
import fetch.mySQS as mySQS
import scrub
import validity_metrics as vm
from aws import aws_ak, aws_sk
from boto.sqs.message import Message
# <codecell>
conn = boto.connect_s3(aws_ak, aws_sk)
bucket = conn.get_bucket('incoming-simscore-org')
filename = 'edge0/2013/01/10.06.22.05.109.0.txt'
data, meta = myS3.getData(bucket, filename, labeled=True)
jsonSimscore = vm.summary_metrics(meta, data, conn)
jsonSimscore = vm.data_metrics_append(jsonSimscore, data, filename)
jsonSimscore = vm.machine_health_append(jsonSimscore, meta, data)
# <codecell>
print jsonSimscore['IgnoreErrors']
print jsonSimscore['NaNSensors']
# <codecell>
def main():
#Get a file off the SQS stack using 20sec long poll
rs = q.read(wait_time_seconds=20)
#if there's a file in the queue,
if rs:
'''Compute all metrics and send'''
try:
#Pull filename and from queued message
filename = mySQS.get_sqs_filename(rs) #'edge6/2012/11/05.18.46.23.340.0.txt'
bucketname = mySQS.get_sqs_bucket(rs)
logit(log,'{0}\n{1}\nProcessing {2}\nfrom bucket {3}\n'.format('-'*20,datetime.now(),filename, bucketname) )
#Ensure this isn't a Reference block trace
if 'Trace' in filename:
logit(log,'Is a reference block trace.\n');
d = q.delete_message(rs)
logit(log, 'Deleted from queue\n'); print 'Deleted %s from queue' %filename
return rs
#If everything looks good, load the dataaa!
data, meta = myS3.getData(whichBucket(bucketname), filename, labeled=True)
if not data: raise ValueError, "Data file is empty!"
'''Where the magic happens'''
#Compute Summary Metrics
jsonSimscore = vm.summary_metrics(meta, data, conn)
jsonSimscore = vm.data_metrics_append(jsonSimscore, data, filename)
jsonSimscore = vm.machine_health_append(jsonSimscore, meta, data)
jsonSimscore['Bucket'] = bucketname
#Score data
jsonSimscore.update({'Score': 'None'}) #scoring.score_test(data, meta)} )
#cleanup long floats, NaN values
jsonSimscore = vm.round_dict(jsonSimscore,3)
jsonSimscore = vm.nan_replace(jsonSimscore)
logit(log,'Successfully processed.\n'); print 'Successfully processed.'
'''Processing is completed --Add json to new SQS stack for POST'''
receipt = mySQS.append_to_queue(jsonSimscore, shipq, raw=False)
assert receipt, "Could not write to queue"
#If json is DEFINITELY received by new SQS, delete from original Files2Process queue
d = q.delete_message(rs)
assert d, "Could not delete from queue"
logit(log, 'Deleted from queue\n'); print 'Deleted %s from queue' %filename
s = add_file_sdb(sdb_domain, jsonSimscore)
assert s, "SimpleDB not updated"
logit(log, "Updated SimpleDB\n")
except Exception as err:
#make more invisible, print/log exception, email me, then continue
rs.change_visibility(5*60)
if filename == None: filename = 'Unknown'
logit(log,'ERROR: %s - %s\n'%(filename, str(err)) )
#Connect to ses
ses_conn = boto.connect_ses(aws_ak, aws_sk)
send_fail('Error computing {0}. computeSimscore.py error: {1}.'.format(filename, err), ses_conn)
#TODO handle incoming-simscore-org-test bucket requests in better way
if err.message in ["File not found on S3","Data file is empty!"]:
d = q.delete_message(rs)
logit(log, 'Deleted from queue\n'); print 'Deleted %s from queue' %filename
return rs
