def __init__(self):
# to display internal messages for debugging
#~debug = False
# limit of words to vocalize, default value is 1000 words.
self.limit = 1000
# set the option value to enable the Last mark on voaclize
# words in output
# default value is True, can be disabled for debuging porpus
self.enabled_last_mark = True
# set the option to do statistical vocalization based
# on collocations
# default value is True, can be disabled for debuging porpus
#self.enabled_stat_tashkeel = False
self.enabled_stat_tashkeel = True
# set the option to show the collocations marks
# default value is False, can be enabled for debuging porpus
self.enabled_show_collocation_mark = False
# set the option to use scoring teashkeel chosing.
self.select_by_score_enabled = False
# set the option to do syntaxic Analysis
# default value is True, can be disabled for debuging porpus
self.enabled_syntaxic_analysis = True
# set the option to do allow ajusting voaclization result,
# for التقاء الساكنين
# default value is True, can be disabled for debuging porpus
self.enabled_ajust_vocalization = True
# set the option to do Semantic Analysis
# default value is True, can be disabled for debuging porpus
self.enabled_semantic_analysis = True
#~ self.enabled_semantic_analysis = False
# enable the last mark (Harakat Al-I3rab)
self.allow_syntax_last_mark = True
# lexical analyzer
self.analyzer = qalsadi.analex.Analex()
self.analyzer.disable_allow_cache_use()
#~ self.analyzer.enable_allow_cache_use()
# syntaxic analyzer
self.anasynt = aranasyn.anasyn.SyntaxAnalyzer()
# semantic analyzer
self.anasem = asmai.anasem.SemanticAnalyzer()
#set the lexical analzer debugging
self.analyzer.set_debug(debug)
#set the lexical analzer word limit
self.analyzer.set_limit(self.limit)
#collocations dictionary for statistical tashkeel
self.collo = coll.CollocationClass(self.enabled_show_collocation_mark)
# unknown vocalizer for unrecognized words
self.unknown_vocalizer = unknown_tashkeel.UnknownTashkeel()
def __init__(self, mycache_path=False):
# configure logging
logging.basicConfig(level=logging.INFO)
#~ logging.basicConfig(level=logging.DEBUG)
self.logger = logging.getLogger(__name__)
#~ self.logger.info("Cache Path %s"%mycache_path)
# to display internal messages for debugging
#~debug = False
# limit of words to vocalize, default value is 1000 words.
self.limit = 1000
# set the option value to enable the Last mark on voaclize
# words in output
# default value is True, can be disabled for debuging porpus
self.enabled_last_mark = True
# set the option to do statistical vocalization based
# on collocations
# default value is True, can be disabled for debuging porpus
#self.enabled_stat_tashkeel = False
self.enabled_stat_tashkeel = True
# set the option to show the collocations marks
# default value is False, can be enabled for debuging porpus
self.enabled_show_collocation_mark = False
# set the option to use scoring teashkeel chosing.
self.select_by_score_enabled = False
# set the option to do syntaxic Analysis
# default value is True, can be disabled for debuging porpus
self.enabled_syntaxic_analysis = True
# set the option to do allow ajusting voaclization result,
# for التقاء الساكنين
# default value is True, can be disabled for debuging porpus
self.enabled_ajust_vocalization = True
# set the option to do Semantic Analysis
# default value is True, can be disabled for debuging porpus
self.enabled_semantic_analysis = True
#~ self.enabled_semantic_analysis = False
# enable the last mark (Harakat Al-I3rab)
self.allow_syntax_last_mark = True
# lexical analyzer
self.analyzer = qalsadi.analex.Analex(cache_path = mycache_path)
#~ self.logger.info("Cache Path cache %s"%self.analyzer.cache.DB_PATH)
#~ self.logger.info("Cache Path cache %s"%self.analyzer.cache.db.path)
#~ self.analyzer.disable_allow_cache_use()
self.analyzer.enable_allow_cache_use()
# syntaxic analyzer
self.anasynt = aranasyn.anasyn.SyntaxAnalyzer(cache_path = mycache_path)
#~ self.logger.info("Cache Path cache syntax %s"%self.anasynt.cache.db.path)
# to disable the training when do Tashkeel
self.syntax_train_enabled = False
# semantic analyzer
self.anasem = asmai.anasem.SemanticAnalyzer(cache_path = mycache_path)
#~ self.logger.info("Cache Path cache anasem %s"%self.anasem.syncache.db.path)
#set the lexical analzer debugging
self.analyzer.set_debug(debug)
#set the lexical analzer word limit
self.analyzer.set_limit(self.limit)
#collocations dictionary for statistical tashkeel
self.collo = coll.CollocationClass(self.enabled_show_collocation_mark)
# unknown vocalizer for unrecognized words
self.unknown_vocalizer = unknown_tashkeel.UnknownTashkeel()