From: Eevee Date: Wed, 25 Aug 2010 03:06:40 +0000 (-0700) Subject: Allow restricting lookup by language, with a @ja: prefix. #90 X-Git-Tag: veekun-promotions/2010091201~11 X-Git-Url: http://git.veekun.com/zzz-pokedex.git/commitdiff_plain/d131ce6a550d4d3572584c04984550d151eccd7b?ds=inline Allow restricting lookup by language, with a @ja: prefix. #90 --- diff --git a/pokedex/lookup.py b/pokedex/lookup.py index 8c3e444..b8dc1a1 100644 --- a/pokedex/lookup.py +++ b/pokedex/lookup.py @@ -26,8 +26,9 @@ __all__ = ['PokedexLookup'] rx_is_number = re.compile('^\d+$') -LookupResult = namedtuple('LookupResult', - ['object', 'indexed_name', 'name', 'language', 'iso3166', 'exact']) +LookupResult = namedtuple('LookupResult', [ + 'object', 'indexed_name', 'name', 'language', 'iso639', 'iso3166', 'exact', +]) class UninitializedIndex(object): class UninitializedIndexError(Exception): @@ -168,7 +169,8 @@ class PokedexLookup(object): table=whoosh.fields.ID(stored=True), row_id=whoosh.fields.ID(stored=True), language=whoosh.fields.STORED, - iso3166=whoosh.fields.STORED, + iso639=whoosh.fields.ID(stored=True), + iso3166=whoosh.fields.ID(stored=True), display_name=whoosh.fields.STORED, # non-lowercased name ) @@ -188,12 +190,12 @@ class PokedexLookup(object): row_key = dict(table=unicode(cls.__tablename__), row_id=unicode(row.id)) - def add(name, language, iso3166): + def add(name, language, iso639, iso3166): normalized_name = self.normalize_name(name) writer.add_document( name=normalized_name, display_name=name, - language=language, iso3166=iso3166, + language=language, iso639=iso639, iso3166=iso3166, **row_key ) @@ -204,14 +206,14 @@ class PokedexLookup(object): if cls == tables.Pokemon: # Pokémon need their form name added # XXX kinda kludgy - add(row.full_name, None, u'us') + add(row.full_name, None, u'en', u'us') # If this is a default form, ALSO add the unadorned name, # so 'Deoxys' alone will still do the right thing if row.forme_name and not row.forme_base_pokemon_id: - add(row.name, None, u'us') + add(row.name, None, u'en', u'us') else: - add(row.name, None, u'us') + add(row.name, None, u'en', u'us') # Some things also have other languages' names # XXX other language form names..? @@ -224,12 +226,13 @@ class PokedexLookup(object): continue add(moonspeak, foreign_name.language.name, + foreign_name.language.iso639, foreign_name.language.iso3166) # Add Roomaji too if foreign_name.language.name == 'Japanese': roomaji = romanize(foreign_name.name) - add(roomaji, u'Roomaji', u'jp') + add(roomaji, u'Roomaji', u'ja', u'jp') writer.commit() @@ -310,16 +313,31 @@ class PokedexLookup(object): # Construct the term type_terms = [] + lang_terms = [] final_valid_types = [] for valid_type in combined_valid_types: - table_name = self._parse_table_name(valid_type) - - # Quietly ignore bogus valid_types; more likely to DTRT - if table_name: - final_valid_types.append(valid_type) - type_terms.append(whoosh.query.Term(u'table', table_name)) - - return name, final_valid_types, whoosh.query.Or(type_terms) + if valid_type.startswith(u'@'): + # @foo means: language must be foo. + # Allow for either country or language codes + lang_code = valid_type[1:] + lang_terms.append(whoosh.query.Term(u'iso639', lang_code)) + lang_terms.append(whoosh.query.Term(u'iso3166', lang_code)) + else: + # otherwise, this is a type/table name + table_name = self._parse_table_name(valid_type) + + # Quietly ignore bogus valid_types; more likely to DTRT + if table_name: + type_terms.append(whoosh.query.Term(u'table', table_name)) + + # Combine both kinds of restriction + all_terms = [] + if type_terms: + all_terms.append(whoosh.query.Or(type_terms)) + if lang_terms: + all_terms.append(whoosh.query.Or(lang_terms)) + + return name, combined_valid_types, whoosh.query.And(all_terms) def _parse_table_name(self, name): @@ -362,6 +380,7 @@ class PokedexLookup(object): indexed_name=record['name'], name=record['display_name'], language=record['language'], + iso639=record['iso639'], iso3166=record['iso3166'], exact=exact)) @@ -371,12 +390,11 @@ class PokedexLookup(object): def lookup(self, input, valid_types=[], exact_only=False): """Attempts to find some sort of object, given a name. - Returns a list of named (object, name, language, iso3166, exact) - tuples. `object` is a database object, `name` is the name under which - the object was found, `language` and `iso3166` are the name and country - code of the language in which the name was found, and `exact` is True - iff this was an - exact match. + Returns a list of named (object, name, language, iso639, iso3166, + exact) tuples. `object` is a database object, `name` is the name under + which the object was found, `language` and the two isos are the name + and country codes of the language in which the name was found, and + `exact` is True iff this was an exact match. This function currently ONLY does fuzzy matching if there are no exact matches. @@ -394,17 +412,19 @@ class PokedexLookup(object): 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. + with commas, as "move,item:1". + - Language restrictions. "@fr:charge" will only return Tackle, which + is called "Charge" in French. These can be combined with type + restrictions, e.g., "@fr,move:charge". - 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. + A list of type or language restrictions, e.g., `['pokemon', + '@ja']`. If this is provided, only results in one of the given + tables will be returned. `exact_only` If True, only exact matches are returned. If set to False (the @@ -503,17 +523,18 @@ class PokedexLookup(object): `valid_types`. """ - tables = [] + table_names = [] for valid_type in valid_types: table_name = self._parse_table_name(valid_type) + # Skip anything not recognized. Could be, say, a language code if table_name: - tables.append(self.indexed_tables[table_name]) + table_names.append(table_name) - if not tables: + if not table_names: # n.b.: It's possible we got a list of valid_types and none of them # were valid, but this function is guaranteed to return - # *something*, so it politely selects from the entire index isntead - tables = self.indexed_tables.values() + # *something*, so it politely selects from the entire index instead + table_names = self.indexed_tables.keys() # Rather than create an array of many hundred items and pick randomly # from it, just pick a number up to the total number of potential @@ -523,10 +544,10 @@ class PokedexLookup(object): # XXX ought to cache this (in the index?) if possible total = 0 partitions = [] - for table in tables: - count = self.session.query(table).count() + for table_name in table_names: + count = self.session.query(self.indexed_tables[table_name]).count() total += count - partitions.append((table, count)) + partitions.append((table_name, count)) n = random.randint(1, total) while n > partitions[0][1]: diff --git a/pokedex/tests/test_lookup.py b/pokedex/tests/test_lookup.py index 96c76cb..094e630 100644 --- a/pokedex/tests/test_lookup.py +++ b/pokedex/tests/test_lookup.py @@ -71,6 +71,26 @@ def test_type_lookup(): results = lookup.lookup(u'1', valid_types=['pokemon']) assert_equal(results[0].name, u'Bulbasaur', u'valid_types works as well as type: prefix') +def test_language_lookup(): + # There are two objects named "charge": the move Charge, and the move + # Tackle, which is called "Charge" in French. + results = lookup.lookup(u'charge') + assert_true(len(results) > 1, u'There are multiple "charge"s') + + results = lookup.lookup(u'@fr:charge') + assert_equal(results[0].iso639, u'fr', u'Language restriction works correctly') + assert_equal(len(results), 1, u'Only one "charge" result when language is specified') + assert_equal(results[0].object.name, u'Tackle', + u'Language + vague name returns the right result') + + results = lookup.lookup(u'charge', valid_types=['@fr']) + assert_equal(results[0].object.name, u'Tackle', + u'valid_types works as well as @lang: prefix') + + results = lookup.lookup(u'@fr,move:charge') + assert_equal(results[0].object.name, u'Tackle', + u'Languages and types both work together') + def test_fuzzy_lookup(): tests = [ # Regular English names