import sys
import os
from subprocess import Popen, PIPE
from uuid import uuid1
from lap.store import get_lapstore
from pymongo import MongoClient

i = str(uuid1())
in_file = sys.argv[1]

growth = []

cred = get_lapstore()
db = MongoClient(cred['host'], cred['port'])['lapstore']
db.authenticate(cred['user'], cred['password'])

used = int(db.command("dbstats")['storageSize'])
growth.append(('0', 0))
current = used

p = Popen(("/projects/lap/tree/trunk/python/lap/python /projects/lap/tools/trunk/segmentation/nltk_segmenter.py %s /tmp/nltk_segmenter.emanuel.ws940701.%s.rpt" % (in_file, i)).split(), stdout=PIPE, stderr=PIPE, stdin=PIPE)
print p.stdout.read()

used = int(db.command("dbstats")['storageSize'])
growth.append(('sentences', used - current + growth[-1][1]))
current = used

p = Popen(("/projects/lap/tree/trunk/python/lap/python /projects/lap/tools/trunk/segmentation/repp_tokenizer.py /tmp/nltk_segmenter.emanuel.ws940701.%s.rpt /tmp/repp_tokenizer.emanuel.ws940701.%s.rpt nltk_punkt ptb" % (i, i)).split(), stdout=PIPE, stderr=PIPE, stdin=PIPE)
print p.stdout.read()

used = int(db.command("dbstats")['storageSize'])
growth.append(('tokens', used - current + growth[-1][1]))
current = used

p = Popen(("/projects/lap/tree/trunk/python/lap/python /projects/lap/tools/trunk/tagging/hunpos.py /tmp/repp_tokenizer.emanuel.ws940701.%s.rpt /tmp/hunpos.emanuel.ws940701.%s.rpt eng_wsj.model nltk_punkt repp" % (i, i)).split(), stdout=PIPE, stderr=PIPE, stdin=PIPE)
print p.stdout.read()

used = int(db.command("dbstats")['storageSize'])
growth.append(('pos-tags', used - current + growth[-1][1]))
current = used

p = Popen(("/projects/lap/tree/trunk/python/lap/python /projects/lap/tools/trunk/parsing/maltparser.py /tmp/hunpos.emanuel.ws940701.%s.rpt /tmp/maltparser.emanuel.ws940701.%s.rpt nltk_punkt repp hunpos" % (i, i)).split(), stdout=PIPE, stderr=PIPE, stdin=PIPE)
print p.stdout.read()

used = int(db.command("dbstats")['storageSize'])
growth.append(('dependencies', used - current + growth[-1][1]))

print "------------------------"
print "/var/lib/mongodb growth:"
print "------------------------"

for job, used in growth:
    print job, used