2011-04-03 11:49:34 +00:00
|
|
|
|
# encoding: utf-8
|
|
|
|
|
import re
|
|
|
|
|
|
2011-04-06 04:03:41 +00:00
|
|
|
|
from sqlalchemy import engine_from_config, orm
|
2009-02-05 08:05:42 +00:00
|
|
|
|
|
2013-12-18 23:36:43 +00:00
|
|
|
|
from hashlib import md5
|
|
|
|
|
|
2010-05-13 17:33:07 +00:00
|
|
|
|
from ..defaults import get_default_db_uri
|
2011-04-06 04:03:41 +00:00
|
|
|
|
from .tables import Language, metadata
|
2011-03-30 03:15:41 +00:00
|
|
|
|
from .multilang import MultilangSession, MultilangScopedSession
|
2009-02-05 08:05:42 +00:00
|
|
|
|
|
2011-04-10 07:54:14 +00:00
|
|
|
|
ENGLISH_ID = 9
|
|
|
|
|
|
2010-05-13 17:33:07 +00:00
|
|
|
|
|
2011-01-31 06:29:23 +00:00
|
|
|
|
def connect(uri=None, session_args={}, engine_args={}, engine_prefix=''):
|
2009-02-05 08:05:42 +00:00
|
|
|
|
"""Connects to the requested URI. Returns a session object.
|
|
|
|
|
|
2009-08-19 01:02:53 +00:00
|
|
|
|
With the URI omitted, attempts to connect to a default SQLite database
|
|
|
|
|
contained within the package directory.
|
|
|
|
|
|
2009-02-05 08:05:42 +00:00
|
|
|
|
Calling this function also binds the metadata object to the created engine.
|
|
|
|
|
"""
|
|
|
|
|
|
2010-05-13 17:33:07 +00:00
|
|
|
|
# If we didn't get a uri, fall back to the default
|
2011-01-31 06:29:23 +00:00
|
|
|
|
if uri is None:
|
2011-03-12 14:46:04 +00:00
|
|
|
|
uri = engine_args.get(engine_prefix + 'url', None)
|
2010-05-13 17:33:07 +00:00
|
|
|
|
if uri is None:
|
|
|
|
|
uri = get_default_db_uri()
|
2009-08-19 01:02:53 +00:00
|
|
|
|
|
2009-02-05 08:05:42 +00:00
|
|
|
|
### Do some fixery for MySQL
|
2011-04-03 11:26:45 +00:00
|
|
|
|
if uri.startswith('mysql:'):
|
2009-02-05 08:05:42 +00:00
|
|
|
|
# MySQL uses latin1 for connections by default even if the server is
|
|
|
|
|
# otherwise oozing with utf8; charset fixes this
|
|
|
|
|
if 'charset' not in uri:
|
|
|
|
|
uri += '?charset=utf8'
|
|
|
|
|
|
2009-03-08 02:54:01 +00:00
|
|
|
|
# Tables should be InnoDB, in the event that we're creating them, and
|
|
|
|
|
# use UTF-8 goddammit!
|
2009-02-05 08:05:42 +00:00
|
|
|
|
for table in metadata.tables.values():
|
|
|
|
|
table.kwargs['mysql_engine'] = 'InnoDB'
|
2009-03-08 02:54:01 +00:00
|
|
|
|
table.kwargs['mysql_charset'] = 'utf8'
|
2009-02-05 08:05:42 +00:00
|
|
|
|
|
2013-12-18 13:32:13 +00:00
|
|
|
|
### Do some fixery for Oracle
|
|
|
|
|
if uri.startswith('oracle:') or uri.startswith('oracle+cx_oracle:'):
|
|
|
|
|
# Oracle requires auto_setinputsizes=False (or at least a special
|
|
|
|
|
# set of exclusions from it, which I don't know)
|
|
|
|
|
if 'auto_setinputsizes' not in uri:
|
|
|
|
|
uri += '?auto_setinputsizes=FALSE'
|
|
|
|
|
|
|
|
|
|
# Shorten table names, Oracle limits table and column names to 30 chars
|
|
|
|
|
# Easy solution : drop the vowels, differents words are unlikely to
|
|
|
|
|
# end up the same after the vowels are gone
|
|
|
|
|
for table in metadata.tables.values():
|
2013-12-18 22:37:03 +00:00
|
|
|
|
table._orginal_name = table.name[:]
|
2013-12-18 13:32:13 +00:00
|
|
|
|
if len(table.name) > 30:
|
|
|
|
|
for letter in ['a', 'e', 'i', 'o', 'u', 'y']:
|
|
|
|
|
table.name=table.name.replace(letter,'')
|
2013-12-18 23:36:43 +00:00
|
|
|
|
# Aggressive renaming if the length is still too long:
|
|
|
|
|
# Take the initials of the table, add a hash to make a new name
|
|
|
|
|
if len(table.name) > 30:
|
|
|
|
|
hashedname = md5(table._orginal_name).hexdigest()
|
|
|
|
|
shortname = ''.join(word[:1] for word in table.name.split('_'))
|
|
|
|
|
shortname = ''.join([shortname, hashedname])
|
|
|
|
|
table.name = shortname[:30]
|
2013-12-18 13:32:13 +00:00
|
|
|
|
|
2009-02-05 08:05:42 +00:00
|
|
|
|
### Connect
|
2011-01-31 06:29:23 +00:00
|
|
|
|
engine_args[engine_prefix + 'url'] = uri
|
|
|
|
|
engine = engine_from_config(engine_args, prefix=engine_prefix)
|
2009-02-05 08:05:42 +00:00
|
|
|
|
conn = engine.connect()
|
|
|
|
|
metadata.bind = engine
|
|
|
|
|
|
2010-03-17 07:44:19 +00:00
|
|
|
|
all_session_args = dict(autoflush=True, autocommit=False, bind=engine)
|
|
|
|
|
all_session_args.update(session_args)
|
2011-04-10 07:54:14 +00:00
|
|
|
|
sm = orm.sessionmaker(class_=MultilangSession,
|
|
|
|
|
default_language_id=ENGLISH_ID, **all_session_args)
|
2011-03-30 03:15:41 +00:00
|
|
|
|
session = MultilangScopedSession(sm)
|
2009-02-05 08:05:42 +00:00
|
|
|
|
|
|
|
|
|
return session
|
2011-04-03 11:49:34 +00:00
|
|
|
|
|
|
|
|
|
def identifier_from_name(name):
|
|
|
|
|
"""Make a string safe to use as an identifier.
|
|
|
|
|
|
|
|
|
|
Valid characters are lowercase alphanumerics and "-". This function may
|
|
|
|
|
raise ValueError if it can't come up with a suitable identifier.
|
|
|
|
|
|
|
|
|
|
This function is useful for scripts which add things with names.
|
|
|
|
|
"""
|
|
|
|
|
if isinstance(name, str):
|
|
|
|
|
identifier = name.decode('utf-8')
|
|
|
|
|
else:
|
|
|
|
|
identifier = name
|
|
|
|
|
identifier = identifier.lower()
|
|
|
|
|
identifier = identifier.replace(u'+', u' plus ')
|
|
|
|
|
identifier = re.sub(u'[ _–]+', u'-', identifier)
|
|
|
|
|
identifier = re.sub(u"['./;’(),:]", u'', identifier)
|
|
|
|
|
identifier = identifier.replace(u'é', u'e')
|
|
|
|
|
identifier = identifier.replace(u'♀', u'-f')
|
|
|
|
|
identifier = identifier.replace(u'♂', u'-m')
|
|
|
|
|
if identifier in (u'???', u'????'):
|
|
|
|
|
identifier = u'unknown'
|
|
|
|
|
elif identifier == u'!':
|
|
|
|
|
identifier = u'exclamation'
|
|
|
|
|
elif identifier == u'?':
|
|
|
|
|
identifier = u'question'
|
|
|
|
|
|
|
|
|
|
if not identifier.replace(u"-", u"").isalnum():
|
|
|
|
|
raise ValueError(identifier)
|
|
|
|
|
return identifier
|