Merge remote-tracking branch 'origin/encounters-i18n'

[zzz-pokedex.git] / pokedex / db / __init__.py

diff --git a/pokedex/db/__init__.py b/pokedex/db/__init__.py
index 9e1b38c..ae06c1b 100644 (file)
--- a/pokedex/db/__init__.py
+++ b/pokedex/db/__init__.py
@@ -1,10 +1,14 @@
-import pkg_resources
+# encoding: utf-8
+import re
 
-from sqlalchemy import MetaData, Table, create_engine, orm
+from sqlalchemy import engine_from_config, orm
 
-from .tables import metadata
+from ..defaults import get_default_db_uri
+from .tables import Language, metadata
+from .multilang import MultilangSession, MultilangScopedSession
 
-def connect(uri=None, session_args={}, engine_args={}):
+
+def connect(uri=None, session_args={}, engine_args={}, engine_prefix=''):
     """Connects to the requested URI.  Returns a session object.
 
     With the URI omitted, attempts to connect to a default SQLite database
@@ -13,14 +17,14 @@ def connect(uri=None, session_args={}, engine_args={}):
     Calling this function also binds the metadata object to the created engine.
     """
 
-    # Default to a URI within the package, which was hopefully created at some point
-    if not uri:
-        sqlite_path = pkg_resources.resource_filename('pokedex',
-                                                      'data/pokedex.sqlite')
-        uri = 'sqlite:///' + sqlite_path
+    # If we didn't get a uri, fall back to the default
+    if uri is None:
+        uri = engine_args.get(engine_prefix + 'url', None)
+    if uri is None:
+        uri = get_default_db_uri()
 
     ### Do some fixery for MySQL
-    if uri[0:5] == 'mysql':
+    if uri.startswith('mysql:'):
         # MySQL uses latin1 for connections by default even if the server is
         # otherwise oozing with utf8; charset fixes this
         if 'charset' not in uri:
@@ -33,13 +37,50 @@ def connect(uri=None, session_args={}, engine_args={}):
             table.kwargs['mysql_charset'] = 'utf8'
 
     ### Connect
-    engine = create_engine(uri, **engine_args)
+    engine_args[engine_prefix + 'url'] = uri
+    engine = engine_from_config(engine_args, prefix=engine_prefix)
     conn = engine.connect()
     metadata.bind = engine
 
     all_session_args = dict(autoflush=True, autocommit=False, bind=engine)
     all_session_args.update(session_args)
-    sm = orm.sessionmaker(**all_session_args)
-    session = orm.scoped_session(sm)
+    sm = orm.sessionmaker(class_=MultilangSession, language_class=Language,
+        **all_session_args)
+    session = MultilangScopedSession(sm)
+
+    # Default to English.  Warning, magic constant, blah blah.  Trying to fetch
+    # English here would kinda break on new databases.  TODO still not an ideal
+    # solution, I guess.
+    session._default_language_id = 9
 
     return session
+
+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