+ properties.set_property(conn, DBCFG_MAXWORDFREQ, config.MAX_WORD_FREQUENCY)
+
+
+class LegacyNameAnalyzer(AbstractAnalyzer):
+ """ The legacy analyzer uses the special Postgresql module for
+ splitting names.
+
+ Each instance opens a connection to the database to request the
+ normalization.
+ """
+
+ def __init__(self, dsn: str, normalizer: Any):
+ self.conn: Optional[Connection] = connect(dsn).connection
+ self.conn.autocommit = True
+ self.normalizer = normalizer
+ psycopg2.extras.register_hstore(self.conn)
+
+ self._cache = _TokenCache(self.conn)
+
+
+ def close(self) -> None:
+ """ Free all resources used by the analyzer.
+ """
+ if self.conn:
+ self.conn.close()
+ self.conn = None
+
+
+ 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.
+
+ The function returns a list of tuples with
+ (original word, word token, word id).
+
+ The function is used for testing and debugging only
+ and not necessarily efficient.
+ """
+ assert self.conn is not None
+ with self.conn.cursor() as cur:
+ cur.execute("""SELECT t.term, word_token, word_id
+ FROM word, (SELECT unnest(%s::TEXT[]) as term) t
+ WHERE word_token = (CASE
+ WHEN left(t.term, 1) = '#' THEN
+ ' ' || make_standard_name(substring(t.term from 2))
+ ELSE
+ make_standard_name(t.term)
+ END)
+ and class is null and country_code is null""",
+ (words, ))
+
+ return [(r[0], r[1], r[2]) for r in cur]
+
+
+ def normalize(self, phrase: str) -> str:
+ """ Normalize the given phrase, i.e. remove all properties that
+ are irrelevant for search.
+ """
+ return cast(str, self.normalizer.transliterate(phrase))
+
+
+ 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
+ 'token_normalized_postcode()'.
+ """
+ return postcode.strip().upper()
+
+
+ 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
+
+ with self.conn.cursor() as cur:
+ # This finds us the rows in location_postcode and word that are
+ # missing in the other table.
+ cur.execute("""SELECT * FROM
+ (SELECT pc, word FROM
+ (SELECT distinct(postcode) as pc FROM location_postcode) p
+ FULL JOIN
+ (SELECT word FROM word
+ WHERE class ='place' and type = 'postcode') w
+ ON pc = word) x
+ WHERE pc is null or word is null""")
+
+ to_delete = []
+ to_add = []
+
+ for postcode, word in cur:
+ if postcode is None:
+ to_delete.append(word)
+ else:
+ to_add.append(postcode)
+
+ if to_delete:
+ cur.execute("""DELETE FROM WORD
+ WHERE class ='place' and type = 'postcode'
+ and word = any(%s)
+ """, (to_delete, ))
+ if to_add:
+ cur.execute("""SELECT count(create_postcode_id(pc))
+ FROM unnest(%s) as pc
+ """, (to_add, ))
+
+
+
+ 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.
+ """
+ assert self.conn is not None
+
+ norm_phrases = set(((self.normalize(p[0]), p[1], p[2], p[3])
+ for p in phrases))
+
+ with self.conn.cursor() as cur:
+ # Get the old phrases.
+ existing_phrases = set()
+ cur.execute("""SELECT word, class, type, operator FROM word
+ WHERE class != 'place'
+ OR (type != 'house' AND type != 'postcode')""")
+ for label, cls, typ, oper in cur:
+ existing_phrases.add((label, cls, typ, oper or '-'))
+
+ to_add = norm_phrases - existing_phrases
+ to_delete = existing_phrases - norm_phrases
+
+ if to_add:
+ cur.execute_values(
+ """ INSERT INTO word (word_id, word_token, word, class, type,
+ search_name_count, operator)
+ (SELECT nextval('seq_word'), ' ' || make_standard_name(name), name,
+ class, type, 0,
+ CASE WHEN op in ('in', 'near') THEN op ELSE null END
+ FROM (VALUES %s) as v(name, class, type, op))""",
+ to_add)
+
+ if to_delete and should_replace:
+ cur.execute_values(
+ """ DELETE FROM word USING (VALUES %s) as v(name, in_class, in_type, op)
+ WHERE word = name and class = in_class and type = in_type
+ and ((op = '-' and operator is null) or op = operator)""",
+ to_delete)
+
+ LOG.info("Total phrases: %s. Added: %s. Deleted: %s",
+ len(norm_phrases), len(to_add), len(to_delete))
+
+
+ def add_country_names(self, country_code: str, names: Mapping[str, str]) -> None:
+ """ Add names for the given country to the search index.
+ """
+ assert self.conn is not None
+
+ with self.conn.cursor() as cur:
+ cur.execute(
+ """INSERT INTO word (word_id, word_token, country_code)
+ (SELECT nextval('seq_word'), lookup_token, %s
+ FROM (SELECT DISTINCT ' ' || make_standard_name(n) as lookup_token
+ FROM unnest(%s)n) y
+ WHERE NOT EXISTS(SELECT * FROM word
+ WHERE word_token = lookup_token and country_code = %s))
+ """, (country_code, list(names.values()), country_code))
+
+
+ def process_place(self, place: PlaceInfo) -> Mapping[str, Any]:
+ """ Determine tokenizer information about the given place.
+
+ Returns a JSON-serialisable structure that will be handed into
+ the database via the token_info field.
+ """
+ assert self.conn is not None
+
+ token_info = _TokenInfo(self._cache)
+
+ names = place.name
+
+ if names:
+ token_info.add_names(self.conn, names)
+
+ if place.is_country():
+ assert place.country_code is not None
+ self.add_country_names(place.country_code, names)
+
+ address = place.address
+ if address:
+ self._process_place_address(token_info, address)
+
+ return token_info.data
+
+
+ def _process_place_address(self, token_info: '_TokenInfo', address: Mapping[str, str]) -> None:
+ assert self.conn is not None
+ hnrs = []
+ addr_terms = []
+
+ for key, value in address.items():
+ if key == 'postcode':
+ # Make sure the normalized postcode is present in the word table.
+ if re.search(r'[:,;]', value) is None:
+ norm_pc = self.normalize_postcode(value)
+ token_info.set_postcode(norm_pc)
+ self._cache.add_postcode(self.conn, norm_pc)
+ elif key in ('housenumber', 'streetnumber', 'conscriptionnumber'):
+ hnrs.append(value)
+ elif key == 'street':
+ token_info.add_street(self.conn, value)
+ elif key == 'place':
+ token_info.add_place(self.conn, value)
+ elif not key.startswith('_') \
+ and key not in ('country', 'full', 'inclusion'):
+ addr_terms.append((key, value))
+
+ if hnrs:
+ token_info.add_housenumbers(self.conn, hnrs)
+
+ if addr_terms:
+ token_info.add_address_terms(self.conn, addr_terms)
+
+
+
+class _TokenInfo:
+ """ Collect token information to be sent back to the database.
+ """
+ def __init__(self, cache: '_TokenCache') -> None:
+ self.cache = cache
+ self.data: Dict[str, Any] = {}
+
+
+ def add_names(self, conn: Connection, names: Mapping[str, str]) -> None:
+ """ Add token information for the names of the place.
+ """
+ with conn.cursor() as cur:
+ # Create the token IDs for all names.
+ self.data['names'] = cur.scalar("SELECT make_keywords(%s)::text",
+ (names, ))
+
+
+ def add_housenumbers(self, conn: Connection, hnrs: Sequence[str]) -> None:
+ """ Extract housenumber information from the address.
+ """
+ if len(hnrs) == 1:
+ token = self.cache.get_housenumber(hnrs[0])
+ if token is not None:
+ self.data['hnr_tokens'] = token
+ self.data['hnr'] = hnrs[0]
+ return
+
+ # split numbers if necessary
+ simple_list: List[str] = []
+ for hnr in hnrs:
+ simple_list.extend((x.strip() for x in re.split(r'[;,]', hnr)))
+
+ if len(simple_list) > 1:
+ simple_list = list(set(simple_list))
+
+ with conn.cursor() as cur:
+ cur.execute("SELECT * FROM create_housenumbers(%s)", (simple_list, ))
+ result = cur.fetchone()
+ assert result is not None
+ self.data['hnr_tokens'], self.data['hnr'] = result
+
+
+ def set_postcode(self, postcode: str) -> None:
+ """ Set or replace the postcode token with the given value.
+ """
+ self.data['postcode'] = postcode
+
+ def add_street(self, conn: Connection, street: str) -> None:
+ """ Add addr:street match terms.
+ """
+ def _get_street(name: str) -> List[int]:
+ with conn.cursor() as cur:
+ return cast(List[int],
+ cur.scalar("SELECT word_ids_from_name(%s)::text", (name, )))
+
+ tokens = self.cache.streets.get(street, _get_street)
+ if tokens:
+ self.data['street'] = tokens
+
+
+ def add_place(self, conn: Connection, place: str) -> None:
+ """ Add addr:place search and match terms.
+ """
+ def _get_place(name: str) -> Tuple[List[int], List[int]]:
+ with conn.cursor() as cur:
+ cur.execute("""SELECT make_keywords(hstore('name' , %s))::text,
+ word_ids_from_name(%s)::text""",
+ (name, name))
+ return cast(Tuple[List[int], List[int]], cur.fetchone())
+
+ self.data['place_search'], self.data['place_match'] = \
+ self.cache.places.get(place, _get_place)
+
+
+ def add_address_terms(self, conn: Connection, terms: Sequence[Tuple[str, str]]) -> None:
+ """ Add additional address terms.
+ """
+ def _get_address_term(name: str) -> Tuple[List[int], List[int]]:
+ with conn.cursor() as cur:
+ cur.execute("""SELECT addr_ids_from_name(%s)::text,
+ word_ids_from_name(%s)::text""",
+ (name, name))
+ return cast(Tuple[List[int], List[int]], cur.fetchone())
+
+ tokens = {}
+ for key, value in terms:
+ items = self.cache.address_terms.get(value, _get_address_term)
+ if items[0] or items[1]:
+ tokens[key] = items
+
+ if tokens:
+ self.data['addr'] = tokens
+
+
+class _LRU:
+ """ Least recently used cache that accepts a generator function to
+ produce the item when there is a cache miss.
+ """
+
+ def __init__(self, maxsize: int = 128):
+ self.data: 'OrderedDict[str, Any]' = OrderedDict()
+ self.maxsize = maxsize
+
+
+ def get(self, key: str, generator: Callable[[str], Any]) -> Any:
+ """ Get the item with the given key from the cache. If nothing
+ is found in the cache, generate the value through the
+ generator function and store it in the cache.
+ """
+ value = self.data.get(key)
+ if value is not None:
+ self.data.move_to_end(key)
+ else:
+ value = generator(key)
+ if len(self.data) >= self.maxsize:
+ self.data.popitem(last=False)
+ self.data[key] = value
+
+ return value
+
+
+class _TokenCache:
+ """ Cache for token information to avoid repeated database queries.
+
+ This cache is not thread-safe and needs to be instantiated per
+ analyzer.
+ """
+ def __init__(self, conn: Connection):
+ # various LRU caches
+ self.streets = _LRU(maxsize=256)
+ self.places = _LRU(maxsize=128)
+ self.address_terms = _LRU(maxsize=1024)
+
+ # Lookup houseunumbers up to 100 and cache them
+ with conn.cursor() as cur:
+ cur.execute("""SELECT i, ARRAY[getorcreate_housenumber_id(i::text)]::text
+ FROM generate_series(1, 100) as i""")
+ self._cached_housenumbers: Dict[str, str] = {str(r[0]): r[1] for r in cur}
+
+ # For postcodes remember the ones that have already been added
+ self.postcodes: Set[str] = set()
+
+ def get_housenumber(self, number: str) -> Optional[str]:
+ """ Get a housenumber token from the cache.
+ """
+ return self._cached_housenumbers.get(number)
+
+
+ def add_postcode(self, conn: Connection, postcode: str) -> None:
+ """ Make sure the given postcode is in the database.
+ """
+ if postcode not in self.postcodes:
+ with conn.cursor() as cur:
+ cur.execute('SELECT create_postcode_id(%s)', (postcode, ))
+ self.postcodes.add(postcode)
projects / nominatim.git / blobdiff