X-Git-Url: http://git.veekun.com/zzz-pokedex.git/blobdiff_plain/b5fde0da1952ae91f8addce211b3e967841fffd7..4a32d3d0331bfb9cfb3be8d7016859e860bf453e:/pokedex/lookup.py?ds=sidebyside diff --git a/pokedex/lookup.py b/pokedex/lookup.py index 16c4e5e..cb6d5f6 100644 --- a/pokedex/lookup.py +++ b/pokedex/lookup.py @@ -3,6 +3,7 @@ from collections import namedtuple import os, os.path import pkg_resources import re +import shutil from sqlalchemy.sql import func import whoosh @@ -53,7 +54,7 @@ def open_index(directory=None, session=None, recreate=False): # Defaults if not directory: directory = pkg_resources.resource_filename('pokedex', - 'data/whoosh_index') + 'data/whoosh-index') if not session: session = connect() @@ -71,17 +72,26 @@ def open_index(directory=None, session=None, recreate=False): # Apparently not a real index. Fall out of the if 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 + ### Create index schema = whoosh.fields.Schema( name=whoosh.fields.ID(stored=True), - table=whoosh.fields.STORED, + table=whoosh.fields.ID(stored=True), row_id=whoosh.fields.ID(stored=True), language=whoosh.fields.STORED, display_name=whoosh.fields.STORED, # non-lowercased name + forme_name=whoosh.fields.ID, ) index = whoosh.index.create_in(directory, schema=schema, indexname='MAIN') @@ -96,18 +106,25 @@ def open_index(directory=None, session=None, recreate=False): for cls in indexed_tables.values(): q = session.query(cls) - # Only index base Pokémon formes - if hasattr(cls, 'forme_base_pokemon_id'): - q = q.filter_by(forme_base_pokemon_id=None) - for row in q.yield_per(5): - row_key = dict(table=cls.__tablename__, row_id=unicode(row.id)) + # XXX need to give forme_name a dummy value because I can't search + # for explicitly empty fields. boo. + row_key = dict(table=unicode(cls.__tablename__), + row_id=unicode(row.id), + forme_name=u'XXX') + + def add(name, language, score): + writer.add_document(name=name.lower(), display_name=name, + language=language, + **row_key) + speller_entries.append((name.lower(), score)) + + # If this is a form, mark it as such + if getattr(row, 'forme_base_pokemon_id', None): + row_key['forme_name'] = row.forme_name name = row.name - writer.add_document(name=name.lower(), - display_name=name, - **row_key) - speller_entries.append((name.lower(), 1)) + add(name, None, 1) # Pokemon also get other languages for foreign_name in getattr(row, 'foreign_names', []): @@ -117,21 +134,12 @@ def open_index(directory=None, session=None, recreate=False): # no point and it makes spell results confusing continue - writer.add_document(name=moonspeak.lower(), - language=foreign_name.language.name, - display_name=moonspeak, - **row_key) - speller_entries.append((moonspeak.lower(), 3)) + add(moonspeak, foreign_name.language.name, 3) # Add Roomaji too if foreign_name.language.name == 'Japanese': roomaji = romanize(foreign_name.name) - writer.add_document(name=roomaji.lower(), - language='Roomaji', - display_name=roomaji, - **row_key) - speller_entries.append((roomaji.lower(), 8)) - + add(roomaji, u'Roomaji', 8) writer.commit() @@ -165,7 +173,7 @@ rx_is_number = re.compile('^\d+$') LookupResult = namedtuple('LookupResult', ['object', 'name', 'language', 'exact']) -def lookup(input, session=None, indices=None, exact_only=False): +def lookup(input, valid_types=[], session=None, indices=None, exact_only=False): """Attempts to find some sort of object, given a database session and name. Returns a list of named (object, name, language, exact) tuples. `object` @@ -182,12 +190,22 @@ def lookup(input, session=None, indices=None, exact_only=False): - Names: "Eevee", "Surf", "Run Away", "Payapa Berry", etc. - Foreign names: "Iibui", "Eivui" - Fuzzy names in whatever language: "Evee", "Ibui" - - IDs: "pokemon 133", "move 192", "item 250" - - Dex numbers: "sinnoh 55", "133", "johto 180" + - IDs: "133", "192", "250" + Also: + - Type restrictions. "type:psychic" will only return the type. This is + how to make ID lookup useful. Multiple type specs can be entered with + commas, as "move,item:1". If `valid_types` are provided, any type prefix + will be ignored. + - Alternate formes can be specified merely like "wash rotom". `input` Name of the thing to look for. + `valid_types` + A list of table objects or names, e.g., `['pokemon', 'moves']`. If + this is provided, only results in one of the given tables will be + returned. + `session` A database session to use for retrieving objects. As with get_index, if this is not provided, a connection to the default database will be @@ -213,6 +231,16 @@ def lookup(input, session=None, indices=None, exact_only=False): name = unicode(input).lower() exact = True + form = None + + # Remove any type prefix (pokemon:133) before constructing a query + if ':' in name: + prefix_chunk, name = name.split(':', 2) + prefixes = prefix_chunk.split(',') + if not valid_types: + # Only use types from the query string if none were explicitly + # provided + valid_types = prefixes # If the input provided is a number, match it as an id. Otherwise, name. # Term objects do an exact match, so we don't have to worry about a query @@ -223,7 +251,34 @@ def lookup(input, session=None, indices=None, exact_only=False): query = whoosh.query.Term(u'row_id', name) else: # Not an integer - query = whoosh.query.Term(u'name', name) + query = whoosh.query.Term(u'name', name) \ + & whoosh.query.Term(u'forme_name', u'XXX') + + # If there's a space in the input, this might be a form + if ' ' in name: + form, formless_name = name.split(' ', 2) + form_query = whoosh.query.Term(u'name', formless_name) \ + & whoosh.query.Term(u'forme_name', form) + query = query | form_query + + ### Filter by type of object + type_terms = [] + for valid_type in valid_types: + if hasattr(valid_type, '__tablename__'): + table_name = getattr(valid_type, '__tablename__') + elif valid_type in indexed_tables: + table_name = valid_type + elif valid_type + 's' in indexed_tables: + table_name = valid_type + 's' + else: + # Bogus. Be nice and ignore it + continue + + type_terms.append(whoosh.query.Term(u'table', table_name)) + + if type_terms: + query = query & whoosh.query.Or(type_terms) + ### Actual searching searcher = index.searcher() @@ -253,6 +308,7 @@ def lookup(input, session=None, indices=None, exact_only=False): cls = indexed_tables[result['table']] obj = session.query(cls).get(result['row_id']) + objects.append(LookupResult(object=obj, name=result['display_name'], language=result['language'],