diff options
Diffstat (limited to 'python/src/_cdec.pyx')
| -rw-r--r-- | python/src/_cdec.pyx | 31 |
1 files changed, 18 insertions, 13 deletions
diff --git a/python/src/_cdec.pyx b/python/src/_cdec.pyx index e93474fe..71e083e4 100644 --- a/python/src/_cdec.pyx +++ b/python/src/_cdec.pyx @@ -3,7 +3,7 @@ from libcpp.vector cimport vector from utils cimport * cimport decoder -cdef char* as_str(data, char* error_msg='Cannot convert type %s to str'): +cdef bytes as_str(data, char* error_msg='Cannot convert type %s to str'): cdef bytes ret if isinstance(data, unicode): ret = data.encode('utf8') @@ -25,6 +25,10 @@ decoder.register_feature_functions() class InvalidConfig(Exception): pass class ParseFailed(Exception): pass +def set_silent(yn): + """set_silent(bool): Configure the verbosity of cdec.""" + SetSilent(yn) + def _make_config(config): for key, value in config.items(): if isinstance(value, dict): @@ -34,18 +38,16 @@ def _make_config(config): for name in value: yield key, name else: - yield key, bytes(value) + yield key, str(value) cdef class Decoder: cdef decoder.Decoder* dec cdef DenseVector weights - def __cinit__(self, config_str=None, **config): - """ Configuration can be given as a string: - Decoder('formalism = scfg') - or using keyword arguments: - Decoder(formalism='scfg') - """ + def __init__(self, config_str=None, **config): + """Decoder('formalism = scfg') -> initialize from configuration string + Decoder(formalism='scfg') -> initialize from named parameters + Create a decoder using a given configuration. Formalism is required.""" if config_str is None: formalism = config.get('formalism', None) if formalism not in ('scfg', 'fst', 'lextrans', 'pb', @@ -82,9 +84,10 @@ cdef class Decoder: property formalism: def __get__(self): cdef variables_map* conf = &self.dec.GetConf() - return conf[0]['formalism'].as_str() + return str(conf[0]['formalism'].as_str().c_str()) def read_weights(self, weights): + """decoder.read_weights(filename): Read decoder weights from a file.""" with open(weights) as fp: for line in fp: if line.strip().startswith('#'): continue @@ -92,20 +95,22 @@ cdef class Decoder: self.weights[fname.strip()] = float(value) def translate(self, sentence, grammar=None): + """decoder.translate(sentence, grammar=None) -> Hypergraph + Translate a sentence (string/Lattice) with a grammar (string/list of rules).""" cdef bytes input_str - if isinstance(sentence, unicode) or isinstance(sentence, str): + if isinstance(sentence, basestring): input_str = as_str(sentence.strip()) elif isinstance(sentence, Lattice): input_str = str(sentence) # PLF format else: raise TypeError('Cannot translate input type %s' % type(sentence)) if grammar: - if isinstance(grammar, str) or isinstance(grammar, unicode): - self.dec.AddSupplementalGrammarFromString(string(as_str(grammar))) + if isinstance(grammar, basestring): + self.dec.AddSupplementalGrammarFromString(as_str(grammar)) else: self.dec.AddSupplementalGrammar(TextGrammar(grammar).grammar[0]) cdef decoder.BasicObserver observer = decoder.BasicObserver() - self.dec.Decode(string(input_str), &observer) + self.dec.Decode(input_str, &observer) if observer.hypergraph == NULL: raise ParseFailed() cdef Hypergraph hg = Hypergraph() |