X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/5ecae1071397c24c6333e82a842dbb55c5e0bc36..b2ae7156995c7ad461add1f751c59bdf58e75f8f:/nominatim/tools/special_phrases.py diff --git a/nominatim/tools/special_phrases.py b/nominatim/tools/special_phrases.py index 3bb240ab..9d0259dc 100644 --- a/nominatim/tools/special_phrases.py +++ b/nominatim/tools/special_phrases.py @@ -27,11 +27,22 @@ class SpecialPhrasesImporter(): self.black_list, self.white_list = self._load_white_and_black_lists() #Compile the regex here to increase performances. self.occurence_pattern = re.compile( - r'\| ([^\|]+) \|\| ([^\|]+) \|\| ([^\|]+) \|\| ([^\|]+) \|\| ([\-YN])' + r'\| *([^\|]+) *\|\| *([^\|]+) *\|\| *([^\|]+) *\|\| *([^\|]+) *\|\| *([\-YN])' ) self.sanity_check_pattern = re.compile(r'^\w+$') self.transliterator = Transliterator.createFromRules("special-phrases normalizer", self.config.TERM_NORMALIZATION) + #This set will contain all existing phrases from the word table which + #no longer exist on the wiki. + #It contain tuples with the following format: (normalized_word, class, type, operator) + self.words_phrases_to_delete = set() + #This set will contain the phrases which still exist from the wiki. + #It is used to prevent duplicates on the wiki by removing them from + #the word_phrases_to_delete only at the end. + self.words_phrases_still_exist = set() + #This set will contain all existing place_classtype tables which doesn't match any + #special phrases class/type on the wiki. + self.table_phrases_to_delete = set() def import_from_wiki(self, languages=None): """ @@ -41,6 +52,9 @@ class SpecialPhrasesImporter(): if languages is not None and not isinstance(languages, list): raise TypeError('The \'languages\' argument should be of type list.') + self._fetch_existing_words_phrases() + self._fetch_existing_place_classtype_tables() + #Get all languages to process. languages = self._load_languages() if not languages else languages @@ -53,9 +67,46 @@ class SpecialPhrasesImporter(): class_type_pairs.update(self._process_xml_content(wiki_page_xml_content, lang)) self._create_place_classtype_table_and_indexes(class_type_pairs) + self._remove_non_existent_phrases_from_db() self.db_connection.commit() LOG.warning('Import done.') + def _fetch_existing_words_phrases(self): + """ + Fetch existing special phrases from the word table. + Fill the word_phrases_to_delete set of the class. + """ + #Only extract special phrases terms: + #If class=place and type=house then it is a housenumber term. + #If class=place and type=postcode then it is a postcode term. + word_query = """ + SELECT word, class, type, operator FROM word + WHERE class != 'place' OR (type != 'house' AND type != 'postcode') + """ + with self.db_connection.cursor() as db_cursor: + db_cursor.execute(SQL(word_query)) + for row in db_cursor: + row[3] = '-' if row[3] is None else row[3] + self.words_phrases_to_delete.add( + (row[0], row[1], row[2], row[3]) + ) + + def _fetch_existing_place_classtype_tables(self): + """ + Fetch existing place_classtype tables. + Fill the table_phrases_to_delete set of the class. + """ + query = """ + SELECT table_name + FROM information_schema.tables + WHERE table_schema='public' + AND table_name like 'place_classtype_%'; + """ + with self.db_connection.cursor() as db_cursor: + db_cursor.execute(SQL(query)) + for row in db_cursor: + self.table_phrases_to_delete.add(row[0]) + def _load_white_and_black_lists(self): """ Load white and black lists from phrases-settings.json. @@ -102,8 +153,10 @@ class SpecialPhrasesImporter(): class_matchs = self.sanity_check_pattern.findall(phrase_class) if len(class_matchs) < 1 or len(type_matchs) < 1: - raise UsageError("Bad class/type for language {}: {}={}".format( - lang, phrase_class, phrase_type)) + LOG.warning("Bad class/type for language %s: %s=%s. It will not be imported", + lang, phrase_class, phrase_type) + return False + return True def _process_xml_content(self, xml_content, lang): """ @@ -122,12 +175,11 @@ class SpecialPhrasesImporter(): phrase_class = match[1].strip() phrase_type = match[2].strip() phrase_operator = match[3].strip() + #Needed if some operator in the wiki are not written in english + phrase_operator = '-' if phrase_operator not in ('near', 'in') else phrase_operator #hack around a bug where building=yes was imported with quotes into the wiki phrase_type = re.sub(r'\"|"', '', phrase_type) - #sanity check, in case somebody added garbage in the wiki - self._check_sanity(lang, phrase_class, phrase_type) - #blacklisting: disallow certain class/type combinations if ( phrase_class in self.black_list.keys() and @@ -141,7 +193,23 @@ class SpecialPhrasesImporter(): ): continue - #add class/type to the pairs dict + #Check if the phrase already exists in the database. + if ( + (normalized_label, phrase_class, phrase_type, phrase_operator) + in self.words_phrases_to_delete + ): + #Remove this phrase from the ones to delete as it still exist on the wiki. + self.words_phrases_still_exist.add( + (normalized_label, phrase_class, phrase_type, phrase_operator) + ) + class_type_pairs.add((phrase_class, phrase_type)) + #Dont need to add this phrase as it already exists in the word table. + continue + + #sanity check, in case somebody added garbage in the wiki + if not self._check_sanity(lang, phrase_class, phrase_type): + continue + class_type_pairs.add((phrase_class, phrase_type)) self._process_amenity( @@ -191,6 +259,15 @@ class SpecialPhrasesImporter(): phrase_class = pair[0] phrase_type = pair[1] + table_name = 'place_classtype_{}_{}'.format(phrase_class, phrase_type) + + if table_name in self.table_phrases_to_delete: + #Remove this table from the ones to delete as it match a class/type + #still existing on the special phrases of the wiki. + self.table_phrases_to_delete.remove(table_name) + #So dont need to create the table and indexes. + continue + #Table creation self._create_place_classtype_table(sql_tablespace, phrase_class, phrase_type) @@ -251,6 +328,41 @@ class SpecialPhrasesImporter(): .format(Identifier(table_name), Identifier(self.config.DATABASE_WEBUSER))) + def _remove_non_existent_phrases_from_db(self): + """ + Remove special phrases which doesn't exist on the wiki anymore. + Delete from the word table and delete the place_classtype tables. + """ + LOG.warning('Cleaning database...') + self.words_phrases_to_delete = self.words_phrases_to_delete - self.words_phrases_still_exist + #Array containing all queries to execute. Contain tuples of format (query, parameters) + queries_parameters = [] + + #Delete phrases from the word table which are not on the wiki anymore. + for phrase_to_delete in self.words_phrases_to_delete: + if phrase_to_delete[3] == '-': + query = """ + DELETE FROM word WHERE word = %s AND class = %s AND type = %s AND operator IS null + """ + parameters = (phrase_to_delete[0], phrase_to_delete[1], phrase_to_delete[2], ) + queries_parameters.append((query, parameters)) + else: + query = """ + DELETE FROM word WHERE word = %s AND class = %s AND type = %s AND operator = %s + """ + parameters = (phrase_to_delete[0], phrase_to_delete[1], + phrase_to_delete[2], phrase_to_delete[3], ) + queries_parameters.append((query, parameters)) + + #Delete place_classtype tables corresponding to class/type which are not on the wiki anymore + for table in self.table_phrases_to_delete: + query = SQL('DROP TABLE IF EXISTS {}').format(Identifier(table)) + queries_parameters.append((query, ())) + + with self.db_connection.cursor() as db_cursor: + for query, parameters in queries_parameters: + db_cursor.execute(query, parameters) + def _convert_php_settings_if_needed(self, file_path): """ Convert php settings file of special phrases to json file if it is still in php format.