class AnalysisProcess:
""" Provides methods to perform automatic morphological analysis of a set of inputs.
"""
def __init__(self, dictionary, input_manager, output_manager):
""" Initializes a new instance of AnalysisManager.
:param dictionary: A morphological dictionary.
:param input_manager: An instance of mosyn.util.io.Input as input method.
:param output_manager: An instance of mosyn.util.io.Output as output method.
"""
self.on_process = EventHook()
self.input_manager = input_manager
self.output_manager = output_manager
self.analysis = MorphologicalAnalysis(dictionary)
self.on_start = EventHook()
self.on_completed = EventHook()
def start_analysis(self):
""" Starts the analysis of the input instance with the defined dictionary.
"""
self.on_start.fire(self)
with self.input_manager, self.output_manager:
for line in self.input_manager.read():
for result in self.analysis.analyze_text(line):
self.output_manager.write(result)
self.on_completed.fire(self)
class MorphologicalAnalysis:
""" Provides methods to perform a morphological analysis to a text entry.
"""
def __init__(self, dictionary):
""" Initializes a new instance of MorphologicalAnalysis.
:param dictionary: A morphological dictionary.
"""
self.dictionary = dictionary
self.on_process = EventHook()
self._cache_dict = {} # Output Dictionary Cache
self._default_eagles = u'NP00000'
self.on_start = EventHook()
self.on_completed = EventHook()
def analyze_text(self, value):
""" Analyzes a piece of text and returns their morphological information.
:param value: The text to analyse.
:return: An iterable collection of morphological representation of the text.
"""
self.on_start.fire(self)
tokens = nltk.word_tokenize(value)
for token in tokens: # Iterate over all tokens in the line
data = self._cache_dict.get(token)
if not data:
data = self.dictionary.get_word(token.lower(), self._default_eagles)
self._cache_dict[token] = data
yield [token, data]
self.on_completed.fire(self)
def __init__(self, dictionary):
""" Initializes a new instance of MorphologicalAnalysis.
:param dictionary: A morphological dictionary.
"""
self.dictionary = dictionary
self.on_process = EventHook()
self._cache_dict = {} # Output Dictionary Cache
self._default_eagles = u'NP00000'
self.on_start = EventHook()
self.on_completed = EventHook()
def __init__(self, dictionary, input_manager, output_manager):
""" Initializes a new instance of AnalysisManager.
:param dictionary: A morphological dictionary.
:param input_manager: An instance of mosyn.util.io.Input as input method.
:param output_manager: An instance of mosyn.util.io.Output as output method.
"""
self.on_process = EventHook()
self.input_manager = input_manager
self.output_manager = output_manager
self.analysis = MorphologicalAnalysis(dictionary)
self.on_start = EventHook()
self.on_completed = EventHook()
def __init__(self, filename):
""" Initializes a new instance of MorphologicalDictionary.
:param filename: The dictionary filename.
"""
self.filename = filename
self.dictionary = {}
self.on_loaded = EventHook()
class MorphologicalDictionary:
""" Represents a morphological dictionary.
"""
def __init__(self, filename):
""" Initializes a new instance of MorphologicalDictionary.
:param filename: The dictionary filename.
"""
self.filename = filename
self.dictionary = {}
self.on_loaded = EventHook()
def load(self):
""" Loads the dictionary filename.
:return: An instance of mosyn.MorphologicalDictionary.
"""
with codecs.open(self.filename, mode='r', encoding='utf-8') as f:
ix = 0
cache = []
for line in f:
if ix == 0:
ix += 1
continue
temp = self._parse_line(line)
if not cache or temp and temp[0] != cache[0]:
if cache:
self.dictionary[cache[0]] = cache[1]
cache = []
cache = [temp[0], [temp[1:3]]]
elif temp and temp[0] == cache[0]:
cache[1].append(temp[1:3])
if cache:
self.dictionary[cache[0]] = cache[1]
self.on_loaded.fire(self)
return self.dictionary
def get_word(self, value, default=u'NP00000'):
""" Gets the morphological information of the word.
:param value: The word value to be find.
:param default: The eagles default classification if the word is not found.
:return: The morphological representation of the word.
"""
return self.dictionary.get(value, [(value.lower(), default)])
def _parse_line(self, text):
""" Converts a text to an array of elements.
:param text: The value to be converted.
:return: The array representation of the text.
:raise Exception: if the array does not agree with the dictionary format.
"""
initial = True
if initial and text.startswith(u','):
initial = False
temp = text.rstrip().split(',')
items = [',', ',', temp[len(temp)-1]]
else:
items = text.rstrip().split(u',')
if len(items) >= 3:
return items[0], items[1], items[2]
else:
raise Exception('Error on dictionary format')
