| author | Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com> |
| Tue, 21 Feb 2012 10:47:06 +0100 | |
| changeset 529 | 99215db3da25 |
| parent 289 | script/lib/iri_tweet/export_tweet_db.py@a5eff8f2b81d |
| child 891 | 8628c590f608 |
| permissions | -rw-r--r-- |
|
122
4c3a15877f80
clean php and python scripts
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
15
diff
changeset
|
1 |
from models import setup_database |
|
4c3a15877f80
clean php and python scripts
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
15
diff
changeset
|
2 |
from optparse import OptionParser #@UnresolvedImport |
|
9
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
3 |
from sqlalchemy.orm import sessionmaker |
|
243
9213a63fa34a
- debug multithread (still database lock problem)
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
242
diff
changeset
|
4 |
from utils import set_logging_options, set_logging, TwitterProcessor, logger |
|
122
4c3a15877f80
clean php and python scripts
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
15
diff
changeset
|
5 |
import sqlite3 #@UnresolvedImport |
|
9
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
6 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
7 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
8 |
# 'entities': "tweet_entity", |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
9 |
# 'user': "tweet_user" |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
10 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
11 |
def get_option(): |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
12 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
13 |
parser = OptionParser() |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
14 |
|
|
15
5d552b6a0e55
add oauth authentication to tweetstream
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
12
diff
changeset
|
15 |
parser.add_option("-t", dest="token_filename", metavar="TOKEN_FILENAME", default=".oauth_token", |
|
5d552b6a0e55
add oauth authentication to tweetstream
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
12
diff
changeset
|
16 |
help="Token file name") |
|
5d552b6a0e55
add oauth authentication to tweetstream
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
12
diff
changeset
|
17 |
|
|
11
54d7f1486ac4
implement get_oauth_token
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
9
diff
changeset
|
18 |
set_logging_options(parser) |
|
9
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
19 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
20 |
return parser.parse_args() |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
21 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
22 |
if __name__ == "__main__": |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
23 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
24 |
(options, args) = get_option() |
|
11
54d7f1486ac4
implement get_oauth_token
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
9
diff
changeset
|
25 |
|
|
54d7f1486ac4
implement get_oauth_token
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
9
diff
changeset
|
26 |
set_logging(options) |
|
9
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
27 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
28 |
with sqlite3.connect(args[0]) as conn_in: |
|
289
a5eff8f2b81d
improve session maker creation + models version + add model version in db
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
243
diff
changeset
|
29 |
engine, metadata, Session = setup_database('sqlite:///'+args[1], echo=((options.verbose-options.quiet)>0)) |
|
9
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
30 |
session = Session() |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
31 |
try: |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
32 |
curs_in = conn_in.cursor() |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
33 |
fields_mapping = {} |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
34 |
for i,res in enumerate(curs_in.execute("select json from tweet_tweet;")): |
|
243
9213a63fa34a
- debug multithread (still database lock problem)
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
242
diff
changeset
|
35 |
logger.debug("main loop %d : %s" % (i, res[0])) #@UndefinedVariable |
|
242
cdd7d3c0549c
Starting 'parallel_twitter' branch
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
122
diff
changeset
|
36 |
processor = TwitterProcessor(eval(res[0]), res[0], None, session, options.token_filename) |
|
11
54d7f1486ac4
implement get_oauth_token
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
9
diff
changeset
|
37 |
processor.process() |
|
9
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
38 |
session.commit() |
|
243
9213a63fa34a
- debug multithread (still database lock problem)
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
242
diff
changeset
|
39 |
logger.debug("main : %d tweet processed" % (i+1)) #@UndefinedVariable |
|
9
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
40 |
except Exception, e: |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
41 |
session.rollback() |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
42 |
raise e |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
43 |
finally: |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
44 |
session.close() |
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
45 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
46 |
|
|
bb44692e09ee
script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
47 |