X-Git-Url: https://git.openstreetmap.org/nominatim.git/blobdiff_plain/bce93d60bd9480e4b1f6c9a899ee6b8d60485352..f2b2b2b92ca87fdf6ae8fcd7062db76ff4c42f51:/nominatim/tokenizer/icu_tokenizer.py diff --git a/nominatim/tokenizer/icu_tokenizer.py b/nominatim/tokenizer/icu_tokenizer.py index 171d4392..9032d71b 100644 --- a/nominatim/tokenizer/icu_tokenizer.py +++ b/nominatim/tokenizer/icu_tokenizer.py @@ -8,41 +8,53 @@ Tokenizer implementing normalisation as used before Nominatim 4 but using libICU instead of the PostgreSQL module. """ +from typing import Optional, Sequence, List, Tuple, Mapping, Any, cast, \ + Dict, Set, Iterable import itertools import json import logging +from pathlib import Path from textwrap import dedent -from nominatim.db.connection import connect +from nominatim.db.connection import connect, Connection, Cursor +from nominatim.config import Configuration from nominatim.db.utils import CopyBuffer from nominatim.db.sql_preprocessor import SQLPreprocessor from nominatim.data.place_info import PlaceInfo from nominatim.tokenizer.icu_rule_loader import ICURuleLoader +from nominatim.tokenizer.place_sanitizer import PlaceSanitizer +from nominatim.data.place_name import PlaceName +from nominatim.tokenizer.icu_token_analysis import ICUTokenAnalysis from nominatim.tokenizer.base import AbstractAnalyzer, AbstractTokenizer DBCFG_TERM_NORMALIZATION = "tokenizer_term_normalization" LOG = logging.getLogger() -def create(dsn, data_dir): +WORD_TYPES =(('country_names', 'C'), + ('postcodes', 'P'), + ('full_word', 'W'), + ('housenumbers', 'H')) + +def create(dsn: str, data_dir: Path) -> 'ICUTokenizer': """ Create a new instance of the tokenizer provided by this module. """ - return LegacyICUTokenizer(dsn, data_dir) + return ICUTokenizer(dsn, data_dir) -class LegacyICUTokenizer(AbstractTokenizer): - """ This tokenizer uses libICU to covert names and queries to ASCII. +class ICUTokenizer(AbstractTokenizer): + """ This tokenizer uses libICU to convert names and queries to ASCII. Otherwise it uses the same algorithms and data structures as the normalization routines in Nominatim 3. """ - def __init__(self, dsn, data_dir): + def __init__(self, dsn: str, data_dir: Path) -> None: self.dsn = dsn self.data_dir = data_dir - self.loader = None + self.loader: Optional[ICURuleLoader] = None - def init_new_db(self, config, init_db=True): + def init_new_db(self, config: Configuration, init_db: bool = True) -> None: """ Set up a new tokenizer for the database. This copies all necessary data in the project directory to make @@ -55,10 +67,11 @@ class LegacyICUTokenizer(AbstractTokenizer): if init_db: self.update_sql_functions(config) - self._init_db_tables(config) + self._setup_db_tables(config) + self._create_base_indices(config, 'word') - def init_from_project(self, config): + def init_from_project(self, config: Configuration) -> None: """ Initialise the tokenizer from the project directory. """ self.loader = ICURuleLoader(config) @@ -69,16 +82,14 @@ class LegacyICUTokenizer(AbstractTokenizer): self._install_php(config.lib_dir.php, overwrite=False) - def finalize_import(self, config): + def finalize_import(self, config: Configuration) -> None: """ Do any required postprocessing to make the tokenizer data ready for use. """ - with connect(self.dsn) as conn: - sqlp = SQLPreprocessor(conn, config) - sqlp.run_sql_file(conn, 'tokenizer/legacy_tokenizer_indices.sql') + self._create_lookup_indices(config, 'word') - def update_sql_functions(self, config): + def update_sql_functions(self, config: Configuration) -> None: """ Reimport the SQL functions for this tokenizer. """ with connect(self.dsn) as conn: @@ -86,34 +97,119 @@ class LegacyICUTokenizer(AbstractTokenizer): sqlp.run_sql_file(conn, 'tokenizer/icu_tokenizer.sql') - def check_database(self, config): + def check_database(self, config: Configuration) -> None: """ Check that the tokenizer is set up correctly. """ # Will throw an error if there is an issue. self.init_from_project(config) - def update_statistics(self): + def update_statistics(self, config: Configuration, threads: int = 2) -> None: """ Recompute frequencies for all name words. """ with connect(self.dsn) as conn: - if conn.table_exists('search_name'): - with conn.cursor() as cur: - cur.drop_table("word_frequencies") - LOG.info("Computing word frequencies") + if not conn.table_exists('search_name'): + return + + with conn.cursor() as cur: + cur.execute('ANALYSE search_name') + if threads > 1: + cur.execute('SET max_parallel_workers_per_gather TO %s', + (min(threads, 6),)) + + if conn.server_version_tuple() < (12, 0): + LOG.info('Computing word frequencies') + cur.drop_table('word_frequencies') + cur.drop_table('addressword_frequencies') cur.execute("""CREATE TEMP TABLE word_frequencies AS SELECT unnest(name_vector) as id, count(*) FROM search_name GROUP BY id""") - cur.execute("CREATE INDEX ON word_frequencies(id)") - LOG.info("Update word table with recomputed frequencies") - cur.execute("""UPDATE word - SET info = info || jsonb_build_object('count', count) - FROM word_frequencies WHERE word_id = id""") - cur.drop_table("word_frequencies") + cur.execute('CREATE INDEX ON word_frequencies(id)') + cur.execute("""CREATE TEMP TABLE addressword_frequencies AS + SELECT unnest(nameaddress_vector) as id, count(*) + FROM search_name GROUP BY id""") + cur.execute('CREATE INDEX ON addressword_frequencies(id)') + cur.execute("""CREATE OR REPLACE FUNCTION word_freq_update(wid INTEGER, + INOUT info JSONB) + AS $$ + DECLARE rec RECORD; + BEGIN + IF info is null THEN + info = '{}'::jsonb; + END IF; + FOR rec IN SELECT count FROM word_frequencies WHERE id = wid + LOOP + info = info || jsonb_build_object('count', rec.count); + END LOOP; + FOR rec IN SELECT count FROM addressword_frequencies WHERE id = wid + LOOP + info = info || jsonb_build_object('addr_count', rec.count); + END LOOP; + IF info = '{}'::jsonb THEN + info = null; + END IF; + END; + $$ LANGUAGE plpgsql IMMUTABLE; + """) + LOG.info('Update word table with recomputed frequencies') + cur.drop_table('tmp_word') + cur.execute("""CREATE TABLE tmp_word AS + SELECT word_id, word_token, type, word, + word_freq_update(word_id, info) as info + FROM word + """) + cur.drop_table('word_frequencies') + cur.drop_table('addressword_frequencies') + else: + LOG.info('Computing word frequencies') + cur.drop_table('word_frequencies') + cur.execute('ANALYSE search_name') + cur.execute('ANALYSE word') + cur.execute(""" + CREATE TEMP TABLE word_frequencies AS + WITH word_freq AS MATERIALIZED ( + SELECT unnest(name_vector) as id, count(*) + FROM search_name GROUP BY id), + addr_freq AS MATERIALIZED ( + SELECT unnest(nameaddress_vector) as id, count(*) + FROM search_name GROUP BY id) + SELECT coalesce(a.id, w.id) as id, + (CASE WHEN w.count is null THEN '{}'::JSONB + ELSE jsonb_build_object('count', w.count) END + || + CASE WHEN a.count is null THEN '{}'::JSONB + ELSE jsonb_build_object('addr_count', a.count) END) as info + FROM word_freq w FULL JOIN addr_freq a ON a.id = w.id; + """) + cur.execute('CREATE UNIQUE INDEX ON word_frequencies(id) INCLUDE(info)') + cur.execute('ANALYSE word_frequencies') + LOG.info('Update word table with recomputed frequencies') + cur.drop_table('tmp_word') + cur.execute("""CREATE TABLE tmp_word AS + SELECT word_id, word_token, type, word, + (CASE WHEN wf.info is null THEN word.info + ELSE coalesce(word.info, '{}'::jsonb) || wf.info + END) as info + FROM word LEFT JOIN word_frequencies wf + ON word.word_id = wf.id + ORDER BY word_id + """) + cur.drop_table('word_frequencies') + + with conn.cursor() as cur: + cur.execute('SET max_parallel_workers_per_gather TO 0') + + sqlp = SQLPreprocessor(conn, config) + sqlp.run_string(conn, + 'GRANT SELECT ON tmp_word TO "{{config.DATABASE_WEBUSER}}"') conn.commit() + self._create_base_indices(config, 'tmp_word') + self._create_lookup_indices(config, 'tmp_word') + self._move_temporary_word_table('tmp_word') + - def _cleanup_housenumbers(self): + def _cleanup_housenumbers(self) -> None: """ Remove unused house numbers. """ with connect(self.dsn) as conn: @@ -148,7 +244,7 @@ class LegacyICUTokenizer(AbstractTokenizer): - def update_word_tokens(self): + def update_word_tokens(self) -> None: """ Remove unused tokens. """ LOG.warning("Cleaning up housenumber tokens.") @@ -156,7 +252,7 @@ class LegacyICUTokenizer(AbstractTokenizer): LOG.warning("Tokenizer house-keeping done.") - def name_analyzer(self): + def name_analyzer(self) -> 'ICUNameAnalyzer': """ Create a new analyzer for tokenizing names and queries using this tokinzer. Analyzers are context managers and should be used accordingly: @@ -171,51 +267,136 @@ class LegacyICUTokenizer(AbstractTokenizer): Analyzers are not thread-safe. You need to instantiate one per thread. """ - return LegacyICUNameAnalyzer(self.dsn, self.loader.make_sanitizer(), - self.loader.make_token_analysis()) + assert self.loader is not None + return ICUNameAnalyzer(self.dsn, self.loader.make_sanitizer(), + self.loader.make_token_analysis()) + + + def most_frequent_words(self, conn: Connection, num: int) -> List[str]: + """ Return a list of the `num` most frequent full words + in the database. + """ + with conn.cursor() as cur: + cur.execute("""SELECT word, sum((info->>'count')::int) as count + FROM word WHERE type = 'W' + GROUP BY word + ORDER BY count DESC LIMIT %s""", (num,)) + return list(s[0].split('@')[0] for s in cur) - def _install_php(self, phpdir, overwrite=True): + def _install_php(self, phpdir: Optional[Path], overwrite: bool = True) -> None: """ Install the php script for the tokenizer. """ - php_file = self.data_dir / "tokenizer.php" + if phpdir is not None: + assert self.loader is not None + php_file = self.data_dir / "tokenizer.php" - if not php_file.exists() or overwrite: - php_file.write_text(dedent(f"""\ - None: """ Save the configuration that needs to remain stable for the given database as database properties. """ + assert self.loader is not None with connect(self.dsn) as conn: self.loader.save_config_to_db(conn) - def _init_db_tables(self, config): + def _setup_db_tables(self, config: Configuration) -> None: + """ Set up the word table and fill it with pre-computed word + frequencies. + """ + with connect(self.dsn) as conn: + with conn.cursor() as cur: + cur.drop_table('word') + sqlp = SQLPreprocessor(conn, config) + sqlp.run_string(conn, """ + CREATE TABLE word ( + word_id INTEGER, + word_token text NOT NULL, + type text NOT NULL, + word text, + info jsonb + ) {{db.tablespace.search_data}}; + GRANT SELECT ON word TO "{{config.DATABASE_WEBUSER}}"; + + DROP SEQUENCE IF EXISTS seq_word; + CREATE SEQUENCE seq_word start 1; + GRANT SELECT ON seq_word to "{{config.DATABASE_WEBUSER}}"; + """) + conn.commit() + + + def _create_base_indices(self, config: Configuration, table_name: str) -> None: """ Set up the word table and fill it with pre-computed word frequencies. """ with connect(self.dsn) as conn: sqlp = SQLPreprocessor(conn, config) - sqlp.run_sql_file(conn, 'tokenizer/icu_tokenizer_tables.sql') + sqlp.run_string(conn, + """CREATE INDEX idx_{{table_name}}_word_token ON {{table_name}} + USING BTREE (word_token) {{db.tablespace.search_index}}""", + table_name=table_name) + for name, ctype in WORD_TYPES: + sqlp.run_string(conn, + """CREATE INDEX idx_{{table_name}}_{{idx_name}} ON {{table_name}} + USING BTREE (word) {{db.tablespace.address_index}} + WHERE type = '{{column_type}}' + """, + table_name=table_name, idx_name=name, + column_type=ctype) + conn.commit() + + + def _create_lookup_indices(self, config: Configuration, table_name: str) -> None: + """ Create additional indexes used when running the API. + """ + with connect(self.dsn) as conn: + sqlp = SQLPreprocessor(conn, config) + # Index required for details lookup. + sqlp.run_string(conn, """ + CREATE INDEX IF NOT EXISTS idx_{{table_name}}_word_id + ON {{table_name}} USING BTREE (word_id) {{db.tablespace.search_index}} + """, + table_name=table_name) + conn.commit() + + + def _move_temporary_word_table(self, old: str) -> None: + """ Rename all tables and indexes used by the tokenizer. + """ + with connect(self.dsn) as conn: + with conn.cursor() as cur: + cur.drop_table('word') + cur.execute(f"ALTER TABLE {old} RENAME TO word") + for idx in ('word_token', 'word_id'): + cur.execute(f"""ALTER INDEX idx_{old}_{idx} + RENAME TO idx_word_{idx}""") + for name, _ in WORD_TYPES: + cur.execute(f"""ALTER INDEX idx_{old}_{name} + RENAME TO idx_word_{name}""") conn.commit() -class LegacyICUNameAnalyzer(AbstractAnalyzer): - """ The legacy analyzer uses the ICU library for splitting names. + + +class ICUNameAnalyzer(AbstractAnalyzer): + """ The ICU analyzer uses the ICU library for splitting names. Each instance opens a connection to the database to request the normalization. """ - def __init__(self, dsn, sanitizer, token_analysis): - self.conn = connect(dsn).connection + def __init__(self, dsn: str, sanitizer: PlaceSanitizer, + token_analysis: ICUTokenAnalysis) -> None: + self.conn: Optional[Connection] = connect(dsn).connection self.conn.autocommit = True self.sanitizer = sanitizer self.token_analysis = token_analysis @@ -223,7 +404,7 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): self._cache = _TokenCache() - def close(self): + def close(self) -> None: """ Free all resources used by the analyzer. """ if self.conn: @@ -231,20 +412,20 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): self.conn = None - def _search_normalized(self, name): + def _search_normalized(self, name: str) -> str: """ Return the search token transliteration of the given name. """ - return self.token_analysis.search.transliterate(name).strip() + return cast(str, self.token_analysis.search.transliterate(name)).strip() - def _normalized(self, name): + def _normalized(self, name: str) -> str: """ Return the normalized version of the given name with all non-relevant information removed. """ - return self.token_analysis.normalizer.transliterate(name).strip() + return cast(str, self.token_analysis.normalizer.transliterate(name)).strip() - def get_word_token_info(self, words): + def get_word_token_info(self, words: Sequence[str]) -> List[Tuple[str, str, int]]: """ Return token information for the given list of words. If a word starts with # it is assumed to be a full name otherwise is a partial name. @@ -255,6 +436,7 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): The function is used for testing and debugging only and not necessarily efficient. """ + assert self.conn is not None full_tokens = {} partial_tokens = {} for word in words: @@ -277,7 +459,7 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): + [(k, v, part_ids.get(v, None)) for k, v in partial_tokens.items()] - def normalize_postcode(self, postcode): + def normalize_postcode(self, postcode: str) -> str: """ Convert the postcode to a standardized form. This function must yield exactly the same result as the SQL function @@ -286,10 +468,11 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): return postcode.strip().upper() - def update_postcodes_from_db(self): + def update_postcodes_from_db(self) -> None: """ Update postcode tokens in the word table from the location_postcode table. """ + assert self.conn is not None analyzer = self.token_analysis.analysis.get('@postcode') with self.conn.cursor() as cur: @@ -311,7 +494,7 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): postcode_name = place.name.strip().upper() variant_base = None else: - postcode_name = analyzer.normalize(place.name) + postcode_name = analyzer.get_canonical_id(place) variant_base = place.get_attr("variant") if variant_base: @@ -324,13 +507,15 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): self._delete_unused_postcode_words(word_entries - needed_entries) self._add_missing_postcode_words(needed_entries - word_entries) - def _delete_unused_postcode_words(self, tokens): + def _delete_unused_postcode_words(self, tokens: Iterable[str]) -> None: + assert self.conn is not None if tokens: with self.conn.cursor() as cur: cur.execute("DELETE FROM word WHERE type = 'P' and word = any(%s)", (list(tokens), )) - def _add_missing_postcode_words(self, tokens): + def _add_missing_postcode_words(self, tokens: Iterable[str]) -> None: + assert self.conn is not None if not tokens: return @@ -341,10 +526,12 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): if '@' in postcode_name: term, variant = postcode_name.split('@', 2) term = self._search_normalized(term) - variants = {term} - if analyzer is not None: - variants.update(analyzer.get_variants_ascii(variant)) - variants = list(variants) + if analyzer is None: + variants = [term] + else: + variants = analyzer.compute_variants(variant) + if term not in variants: + variants.append(term) else: variants = [self._search_normalized(postcode_name)] terms.append((postcode_name, variants)) @@ -358,12 +545,14 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): - def update_special_phrases(self, phrases, should_replace): + def update_special_phrases(self, phrases: Iterable[Tuple[str, str, str, str]], + should_replace: bool) -> None: """ Replace the search index for special phrases with the new phrases. If `should_replace` is True, then the previous set of will be completely replaced. Otherwise the phrases are added to the already existing ones. """ + assert self.conn is not None norm_phrases = set(((self._normalized(p[0]), p[1], p[2], p[3]) for p in phrases)) @@ -386,7 +575,9 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): len(norm_phrases), added, deleted) - def _add_special_phrases(self, cursor, new_phrases, existing_phrases): + def _add_special_phrases(self, cursor: Cursor, + new_phrases: Set[Tuple[str, str, str, str]], + existing_phrases: Set[Tuple[str, str, str, str]]) -> int: """ Add all phrases to the database that are not yet there. """ to_add = new_phrases - existing_phrases @@ -407,9 +598,10 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): return added - @staticmethod - def _remove_special_phrases(cursor, new_phrases, existing_phrases): - """ Remove all phrases from the databse that are no longer in the + def _remove_special_phrases(self, cursor: Cursor, + new_phrases: Set[Tuple[str, str, str, str]], + existing_phrases: Set[Tuple[str, str, str, str]]) -> int: + """ Remove all phrases from the database that are no longer in the new phrase list. """ to_delete = existing_phrases - new_phrases @@ -425,7 +617,7 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): return len(to_delete) - def add_country_names(self, country_code, names): + def add_country_names(self, country_code: str, names: Mapping[str, str]) -> None: """ Add default names for the given country to the search index. """ # Make sure any name preprocessing for country names applies. @@ -437,10 +629,12 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): internal=True) - def _add_country_full_names(self, country_code, names, internal=False): + def _add_country_full_names(self, country_code: str, names: Sequence[PlaceName], + internal: bool = False) -> None: """ Add names for the given country from an already sanitized name list. """ + assert self.conn is not None word_tokens = set() for name in names: norm_name = self._search_normalized(name.name) @@ -453,7 +647,8 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): FROM word WHERE type = 'C' and word = %s""", (country_code, )) - existing_tokens = {True: set(), False: set()} # internal/external names + # internal/external names + existing_tokens: Dict[bool, Set[str]] = {True: set(), False: set()} for word in cur: existing_tokens[word[1]].add(word[0]) @@ -486,7 +681,7 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): cur.execute(sql, (country_code, list(new_tokens))) - def process_place(self, place): + def process_place(self, place: PlaceInfo) -> Mapping[str, Any]: """ Determine tokenizer information about the given place. Returns a JSON-serializable structure that will be handed into @@ -500,6 +695,7 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): token_info.set_names(*self._compute_name_tokens(names)) if place.is_country(): + assert place.country_code is not None self._add_country_full_names(place.country_code, names) if address: @@ -508,7 +704,8 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): return token_info.to_dict() - def _process_place_address(self, token_info, address): + def _process_place_address(self, token_info: '_TokenInfo', + address: Sequence[PlaceName]) -> None: for item in address: if item.kind == 'postcode': token_info.set_postcode(self._add_postcode(item)) @@ -524,12 +721,13 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): token_info.add_address_term(item.kind, self._compute_partial_tokens(item.name)) - def _compute_housenumber_token(self, hnr): + def _compute_housenumber_token(self, hnr: PlaceName) -> Tuple[Optional[int], Optional[str]]: """ Normalize the housenumber and return the word token and the canonical form. """ + assert self.conn is not None analyzer = self.token_analysis.analysis.get('@housenumber') - result = None, None + result: Tuple[Optional[int], Optional[str]] = (None, None) if analyzer is None: # When no custom analyzer is set, simply normalize and transliterate @@ -538,32 +736,34 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): result = self._cache.housenumbers.get(norm_name, result) if result[0] is None: with self.conn.cursor() as cur: - cur.execute("SELECT getorcreate_hnr_id(%s)", (norm_name, )) - result = cur.fetchone()[0], norm_name + hid = cur.scalar("SELECT getorcreate_hnr_id(%s)", (norm_name, )) + + result = hid, norm_name self._cache.housenumbers[norm_name] = result else: # Otherwise use the analyzer to determine the canonical name. # Per convention we use the first variant as the 'lookup name', the # name that gets saved in the housenumber field of the place. - norm_name = analyzer.normalize(hnr.name) - if norm_name: - result = self._cache.housenumbers.get(norm_name, result) + word_id = analyzer.get_canonical_id(hnr) + if word_id: + result = self._cache.housenumbers.get(word_id, result) if result[0] is None: - variants = analyzer.get_variants_ascii(norm_name) + variants = analyzer.compute_variants(word_id) if variants: with self.conn.cursor() as cur: - cur.execute("SELECT create_analyzed_hnr_id(%s, %s)", - (norm_name, list(variants))) - result = cur.fetchone()[0], variants[0] - self._cache.housenumbers[norm_name] = result + hid = cur.scalar("SELECT create_analyzed_hnr_id(%s, %s)", + (word_id, list(variants))) + result = hid, variants[0] + self._cache.housenumbers[word_id] = result return result - def _compute_partial_tokens(self, name): + def _compute_partial_tokens(self, name: str) -> List[int]: """ Normalize the given term, split it into partial words and return then token list for them. """ + assert self.conn is not None norm_name = self._search_normalized(name) tokens = [] @@ -582,16 +782,18 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): (need_lookup, )) for partial, token in cur: + assert token is not None tokens.append(token) self._cache.partials[partial] = token return tokens - def _retrieve_full_tokens(self, name): + def _retrieve_full_tokens(self, name: str) -> List[int]: """ Get the full name token for the given name, if it exists. - The name is only retrived for the standard analyser. + The name is only retrieved for the standard analyser. """ + assert self.conn is not None norm_name = self._search_normalized(name) # return cached if possible @@ -608,51 +810,55 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): return full - def _compute_name_tokens(self, names): + def _compute_name_tokens(self, names: Sequence[PlaceName]) -> Tuple[Set[int], Set[int]]: """ Computes the full name and partial name tokens for the given dictionary of names. """ - full_tokens = set() - partial_tokens = set() + assert self.conn is not None + full_tokens: Set[int] = set() + partial_tokens: Set[int] = set() for name in names: analyzer_id = name.get_attr('analyzer') analyzer = self.token_analysis.get_analyzer(analyzer_id) - norm_name = analyzer.normalize(name.name) + word_id = analyzer.get_canonical_id(name) if analyzer_id is None: - token_id = norm_name + token_id = word_id else: - token_id = f'{norm_name}@{analyzer_id}' + token_id = f'{word_id}@{analyzer_id}' full, part = self._cache.names.get(token_id, (None, None)) if full is None: - variants = analyzer.get_variants_ascii(norm_name) + variants = analyzer.compute_variants(word_id) if not variants: continue with self.conn.cursor() as cur: cur.execute("SELECT * FROM getorcreate_full_word(%s, %s)", (token_id, variants)) - full, part = cur.fetchone() + full, part = cast(Tuple[int, List[int]], cur.fetchone()) self._cache.names[token_id] = (full, part) + assert part is not None + full_tokens.add(full) partial_tokens.update(part) return full_tokens, partial_tokens - def _add_postcode(self, item): + def _add_postcode(self, item: PlaceName) -> Optional[str]: """ Make sure the normalized postcode is present in the word table. """ + assert self.conn is not None analyzer = self.token_analysis.analysis.get('@postcode') if analyzer is None: postcode_name = item.name.strip().upper() variant_base = None else: - postcode_name = analyzer.normalize(item.name) + postcode_name = analyzer.get_canonical_id(item) variant_base = item.get_attr("variant") if variant_base: @@ -667,7 +873,7 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): variants = {term} if analyzer is not None and variant_base: - variants.update(analyzer.get_variants_ascii(variant_base)) + variants.update(analyzer.compute_variants(variant_base)) with self.conn.cursor() as cur: cur.execute("SELECT create_postcode_word(%s, %s)", @@ -680,25 +886,24 @@ class LegacyICUNameAnalyzer(AbstractAnalyzer): class _TokenInfo: """ Collect token information to be sent back to the database. """ - def __init__(self): - self.names = None - self.housenumbers = set() - self.housenumber_tokens = set() - self.street_tokens = set() - self.place_tokens = set() - self.address_tokens = {} - self.postcode = None - - - @staticmethod - def _mk_array(tokens): + def __init__(self) -> None: + self.names: Optional[str] = None + self.housenumbers: Set[str] = set() + self.housenumber_tokens: Set[int] = set() + self.street_tokens: Optional[Set[int]] = None + self.place_tokens: Set[int] = set() + self.address_tokens: Dict[str, str] = {} + self.postcode: Optional[str] = None + + + def _mk_array(self, tokens: Iterable[Any]) -> str: return f"{{{','.join((str(s) for s in tokens))}}}" - def to_dict(self): + def to_dict(self) -> Dict[str, Any]: """ Return the token information in database importable format. """ - out = {} + out: Dict[str, Any] = {} if self.names: out['names'] = self.names @@ -707,7 +912,7 @@ class _TokenInfo: out['hnr'] = ';'.join(self.housenumbers) out['hnr_tokens'] = self._mk_array(self.housenumber_tokens) - if self.street_tokens: + if self.street_tokens is not None: out['street'] = self._mk_array(self.street_tokens) if self.place_tokens: @@ -722,40 +927,43 @@ class _TokenInfo: return out - def set_names(self, fulls, partials): + def set_names(self, fulls: Iterable[int], partials: Iterable[int]) -> None: """ Adds token information for the normalised names. """ self.names = self._mk_array(itertools.chain(fulls, partials)) - def add_housenumber(self, token, hnr): + def add_housenumber(self, token: Optional[int], hnr: Optional[str]) -> None: """ Extract housenumber information from a list of normalised housenumbers. """ if token: + assert hnr is not None self.housenumbers.add(hnr) self.housenumber_tokens.add(token) - def add_street(self, tokens): + def add_street(self, tokens: Iterable[int]) -> None: """ Add addr:street match terms. """ + if self.street_tokens is None: + self.street_tokens = set() self.street_tokens.update(tokens) - def add_place(self, tokens): + def add_place(self, tokens: Iterable[int]) -> None: """ Add addr:place search and match terms. """ self.place_tokens.update(tokens) - def add_address_term(self, key, partials): + def add_address_term(self, key: str, partials: Iterable[int]) -> None: """ Add additional address terms. """ if partials: self.address_tokens[key] = self._mk_array(partials) - def set_postcode(self, postcode): + def set_postcode(self, postcode: Optional[str]) -> None: """ Set the postcode to the given one. """ self.postcode = postcode @@ -767,9 +975,9 @@ class _TokenCache: This cache is not thread-safe and needs to be instantiated per analyzer. """ - def __init__(self): - self.names = {} - self.partials = {} - self.fulls = {} - self.postcodes = set() - self.housenumbers = {} + def __init__(self) -> None: + self.names: Dict[str, Tuple[int, List[int]]] = {} + self.partials: Dict[str, int] = {} + self.fulls: Dict[str, List[int]] = {} + self.postcodes: Set[str] = set() + self.housenumbers: Dict[str, Tuple[Optional[int], Optional[str]]] = {}