script/lib/iri_tweet/export_tweet_db.py
author Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
Wed, 27 Jul 2011 00:04:55 +0200
changeset 242 cdd7d3c0549c
parent 122 4c3a15877f80
child 243 9213a63fa34a
permissions -rw-r--r--
Starting 'parallel_twitter' branch
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
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
122
4c3a15877f80 clean php and python scripts
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents: 15
diff changeset
     4
from utils import set_logging_options, set_logging, TwitterProcessor
9
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     5
import logging
122
4c3a15877f80 clean php and python scripts
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents: 15
diff changeset
     6
import sqlite3 #@UnresolvedImport
9
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
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     9
#    'entities': "tweet_entity",     
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    10
#    'user': "tweet_user"
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    11
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    12
def get_option():
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    13
    
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    14
    parser = OptionParser()
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    15
15
5d552b6a0e55 add oauth authentication to tweetstream
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents: 12
diff changeset
    16
    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
    17
                      help="Token file name")
5d552b6a0e55 add oauth authentication to tweetstream
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents: 12
diff changeset
    18
11
54d7f1486ac4 implement get_oauth_token
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents: 9
diff changeset
    19
    set_logging_options(parser)
9
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    20
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    21
    return parser.parse_args()
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    22
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    23
if __name__ == "__main__":
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    24
    
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    25
    (options, args) = get_option()
11
54d7f1486ac4 implement get_oauth_token
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents: 9
diff changeset
    26
        
54d7f1486ac4 implement get_oauth_token
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents: 9
diff changeset
    27
    set_logging(options)
9
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    28
    
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    29
    with sqlite3.connect(args[0]) as conn_in:
11
54d7f1486ac4 implement get_oauth_token
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents: 9
diff changeset
    30
        engine, metadata = 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
    31
        Session = sessionmaker(bind=engine)
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    32
        session = Session()
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    33
        try:
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    34
            curs_in = conn_in.cursor()
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    35
            fields_mapping = {}
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    36
            for i,res in enumerate(curs_in.execute("select json from tweet_tweet;")):
122
4c3a15877f80 clean php and python scripts
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents: 15
diff changeset
    37
                logging.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
    38
                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
    39
                processor.process()
9
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    40
                session.commit()
122
4c3a15877f80 clean php and python scripts
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents: 15
diff changeset
    41
            logging.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
    42
        except Exception, e:
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    43
            session.rollback()
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    44
            raise e
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    45
        finally:
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    46
            session.close()
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    47
            
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    48
    
bb44692e09ee script apres traitement enmi
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    49