def main():
description = "Get the phonemes from a language"
args = commandline.LanguageAndWordInput.parse_arguments(description)
pronunciations = get_pronunciations(args.language, args.word)
if not pronunciations:
message = "No pronunciations found for word '{}' in language '{}'"
raise RuntimeError(message.format(args.word, args.language))
commandline.output_list(pronunciations)
self.all_data.append(row)
def _get_language_code(self):
return language_codes.Phoibe.map(self.language)
def get_all_phonemes(self):
""" get a set of all phonemes for the language of the phonemeCollector
"""
return set(row['Phoneme'].decode('utf8') for row in self.all_data)
def get_phonemes(language):
"""Main entry point for the module
will return a set of phonemes for the given language
Arguments:
language=language for which to get the phonemes
Returns:
set of phonemes
"""
phoibe_data = resources.phoible_database
phonemes_collector = PhonemesCollector(language)
phonemes_collector.parse_source(phoibe_data)
return phonemes_collector.get_all_phonemes()
if __name__ == '__main__':
description = 'Get the phonemes from a language'
args = commandline.LanguageInput.parse_arguments(description)
phonemes = get_phonemes(args.language)
commandline.output_list(phonemes)
pronunciation_entries = (entry.get(key, []) for entry in wiktionary_entry)
pronunciations = itertools.chain.from_iterable(pronunciation_entries)
return pronunciations
def list_pronunciations(pronunciation_entries):
""" Parses all the pronunciations from the entries
Arguments:
pronunciation_entries = a iteratable of entries to parse
Returns:
set of pronunciations : set(pron1, pron2, ...)
"""
pronunciations = set()
pattern = re.compile('IPA: */(.*?)/')
for entry in pronunciation_entries:
found_pronunciations = pattern.findall(entry)
for pronunciation in found_pronunciations:
pronunciations.add(pronunciation)
return pronunciations
if __name__ == '__main__':
description = 'Get the phonemes from a language'
args = commandline.LanguageAndWordInput.parse_arguments(
description, extra_arguments=['local'])
pronunciations = get_pronunciations(args.language, args.word, args.local)
if not pronunciations:
message = "No pronunciations found for word '{}' in language '{}'"
raise RuntimeError(message.format(args.word, args.language))
commandline.output_list(pronunciations)
text = requests.get(page_path).text
stream = StringIO.StringIO(text)
return stream
def _get_frequency_list_from_file(file_pointer):
"""Take a pointer to a file and get the frequency list from it
"""
with open(file_pointer) as instream:
freq_list = _get_frequency_list_from_filestream(instream)
return freq_list
def _get_frequency_list_from_filestream(instream):
"""Take a file stream and get the frequency list from it
"""
freq_list = []
for line in instream:
if not line:
break
word, freq = line.split()
freq_list.append(word)
return freq_list
if __name__ == '__main__':
description = 'Get the word frequencies for a language'
args = commandline.LanguageInput.parse_arguments(description)
frequency_list = get_frequency_list(args.language)
commandline.output_list(frequency_list[:5])
[resources.hermit_dave_github, language_code, page_name])
text = requests.get(page_path).text
stream = StringIO.StringIO(text)
return stream
class FrequencySources(object):
language_code = staticmethod(language_codes.HermitDave.map)
frequency_filestream = staticmethod(_get_hermitdave_page)
def _frequency_list_from_filestream(filestream, extended_return_value=False):
"""Take a filestream and get the frequency list from it
if extended_return -> list of (word, ranking, occurances)
"""
freq_list = (line.strip().split() for line in filestream if line.strip())
freq_list = [
word if not extended_return_value else (word, i + 1, int(freq))
for i, (word, freq) in enumerate(freq_list)
]
if not freq_list:
raise RuntimeError("No entries found for creating a frequency list")
return freq_list
if __name__ == '__main__':
description = 'Get the word frequencies for a language'
args = commandline.LanguageInput.parse_arguments(description)
frequency_list = get_frequency_list(args.language)
commandline.output_list(frequency_list[:5])