# encoding: utf8
-from collections import namedtuple
import os, os.path
-import pkg_resources
import random
import re
import shutil
import whoosh.scoring
import whoosh.spelling
+from pokedex.util import namedtuple
+
from pokedex.db import connect
import pokedex.db.tables as tables
from pokedex.roomaji import romanize
+from pokedex.defaults import get_default_index_dir
__all__ = ['PokedexLookup']
LookupResult = namedtuple('LookupResult',
['object', 'indexed_name', 'name', 'language', 'iso3166', 'exact'])
+class UninitializedIndex(object):
+ class UninitializedIndexError(Exception):
+ pass
+
+ def __nonzero__(self):
+ """Dummy object should identify itself as False."""
+ return False
+
+ def __bool__(self):
+ """Python 3000 version of the above. Future-proofing rules!"""
+ return False
+
+ def __getattr__(self, *args, **kwargs):
+ raise self.UninitializedIndexError(
+ "The lookup index does not exist. Please use `pokedex setup` "
+ "or lookup.rebuild_index() to create it."
+ )
+
class LanguageWeighting(whoosh.scoring.Weighting):
"""A scoring class that forces otherwise-equal English results to come
before foreign results.
)
- def __init__(self, directory=None, session=None, recreate=False):
+ def __init__(self, directory=None, session=None):
"""Opens the whoosh index stored in the named directory. If the index
doesn't already exist, it will be created.
`pokedex` egg directory.
`session`
- If the index needs to be created, this database session will be
- used. Defaults to an attempt to connect to the default SQLite
- database installed by `pokedex setup`.
-
- `recreate`
- If set to True, the whoosh index will be created even if it already
- exists.
+ Used for creating the index and retrieving objects. Defaults to an
+ attempt to connect to the default SQLite database installed by
+ `pokedex setup`.
"""
# By the time this returns, self.index, self.speller, and self.session
# must be set
- # Defaults
- if not directory:
- directory = pkg_resources.resource_filename('pokedex',
- 'data/whoosh-index')
+ # If a directory was not given, use the default
+ if directory is None:
+ directory = get_default_index_dir()
+
+ self.directory = directory
if session:
self.session = session
self.session = connect()
# Attempt to open or create the index
- directory_exists = os.path.exists(directory)
- if directory_exists and not recreate:
- # Already exists; should be an index! Bam, done.
- try:
- self.index = whoosh.index.open_dir(directory, indexname='MAIN')
- spell_store = whoosh.filedb.filestore.FileStorage(directory)
- self.speller = whoosh.spelling.SpellChecker(spell_store)
- return
- except whoosh.index.EmptyIndexError as e:
- # Apparently not a real index. Fall out and create it
- pass
-
- # Delete and start over if we're going to bail anyway.
- if directory_exists and recreate:
- # Be safe and only delete if it looks like a whoosh index, i.e.,
- # everything starts with _
- if all(f[0] == '_' for f in os.listdir(directory)):
- shutil.rmtree(directory)
- directory_exists = False
-
- if not directory_exists:
- os.mkdir(directory)
-
-
- ### Create index
+ if not os.path.exists(directory) or not os.listdir(directory):
+ # Directory doesn't exist OR is empty; caller needs to use
+ # rebuild_index before doing anything. Provide a dummy object that
+ # complains when used
+ self.index = UninitializedIndex()
+ self.speller = UninitializedIndex()
+ return
+
+ # Otherwise, already exists; should be an index! Bam, done.
+ # Note that this will explode if the directory exists but doesn't
+ # contain an index; that's a feature
+ try:
+ self.index = whoosh.index.open_dir(directory, indexname='MAIN')
+ except whoosh.index.EmptyIndexError:
+ raise IOError(
+ "The index directory already contains files. "
+ "Please use a dedicated directory for the lookup index."
+ )
+
+ # Create speller, and done
+ spell_store = whoosh.filedb.filestore.FileStorage(directory)
+ self.speller = whoosh.spelling.SpellChecker(spell_store)
+
+
+ def rebuild_index(self):
+ """Creates the index from scratch."""
+
schema = whoosh.fields.Schema(
name=whoosh.fields.ID(stored=True),
table=whoosh.fields.ID(stored=True),
display_name=whoosh.fields.STORED, # non-lowercased name
)
- self.index = whoosh.index.create_in(directory, schema=schema,
- indexname='MAIN')
+ if not os.path.exists(self.directory):
+ os.mkdir(self.directory)
+
+ self.index = whoosh.index.create_in(self.directory, schema=schema,
+ indexname='MAIN')
writer = self.index.writer()
# Index every name in all our tables of interest
Returns None for a bogus name.
"""
+ # Table object
if hasattr(name, '__tablename__'):
return getattr(name, '__tablename__')
- elif name in self.indexed_tables:
- return name
- elif name + 's' in self.indexed_tables:
- return name + 's'
- else:
- # Bogus. Be nice and return dummy
- return None
+
+ # Table name
+ for table in self.indexed_tables.values():
+ if name in (table.__tablename__, table.__singlename__):
+ return table.__tablename__
+
+ # Bogus. Be nice and return dummy
+ return None
def _whoosh_records_to_results(self, records, exact=True):
"""Converts a list of whoosh's indexed records to LookupResult tuples
# Do different things depending what the query looks like
# Note: Term objects do an exact match, so we don't have to worry about
# a query parser tripping on weird characters in the input
+ try:
+ # Let Python try to convert to a number, so 0xff works
+ name_as_number = int(name, base=0)
+ except ValueError:
+ # Oh well
+ name_as_number = None
+
if '*' in name or '?' in name:
exact_only = True
query = whoosh.query.Wildcard(u'name', name)
- elif rx_is_number.match(name):
+ elif name_as_number is not None:
# Don't spell-check numbers!
exact_only = True
- query = whoosh.query.Term(u'row_id', name)
+ query = whoosh.query.Term(u'row_id', unicode(name_as_number))
else:
# Not an integer
query = whoosh.query.Term(u'name', name)
projects / zzz-pokedex.git / blobdiff