def main(args):
from argparse import ArgumentParser
from simplelogsetter import SimpleLogSetter
parser = ArgumentParser(description=__doc__)
parser.add_argument(
"-v",
"--verbose",
dest="verbosity",
default=0,
action="count",
help="Verbosity. Invoke many times for higher verbosity")
parser.add_argument("-d",
"--dictionary",
dest="dictionary",
required=True,
help="Dictionary file")
parser.add_argument("-m",
"--model",
dest="model",
required=True,
help="Model file")
parser.add_argument("-s",
"--show-examples",
dest="showExamples",
type=int,
default=None,
help="Show n examples that rank high in each topic")
parser.add_argument(
"-e",
"--english-only",
dest="filterEnglishOnly",
default=False,
action="store_true",
help="Filter corpus and only let through english lyrics")
parser.add_argument("--elastic-search",
dest="elasticSearch",
default=False,
action="store_true",
help="Get data from elastic search instead of a file")
parser.add_argument("lyrics", nargs="?", help="File to load lyrics from")
parameters = parser.parse_args(args)
logger = SimpleLogSetter(verbosity=parameters.verbosity)
logger.startLogging()
dictionaryPath = os.path.expanduser(parameters.dictionary)
logging.info("Loading dictionary from %s" % dictionaryPath)
dictionary = corpora.Dictionary.load(dictionaryPath)
modelPath = os.path.expanduser(parameters.model)
logging.info("Loading model from %s" % modelPath)
model = models.ldamodel.LdaModel.load(modelPath)
if parameters.elasticSearch:
lyricsGenerator = getLyricsFromElasticSearch()
else:
lyricsGenerator = getLyricsFromJson(
os.path.expanduser(parameters.lyrics))
for index, songLyrics in enumerate(lyricsGenerator):
frequencied = dictionary.doc2bow(cleanLyrics(songLyrics),
allow_update=False)
if not parameters.filterEnglishOnly or isItEnglish(
frequencied, dictionary):
distribution = model[frequencied]
print distribution
print songLyrics
if index > 30:
break
lyricRatings = {}
return 0
def main(args):
from argparse import ArgumentParser
from simplelogsetter import SimpleLogSetter
parser = ArgumentParser(description=__doc__)
parser.add_argument(
"-v",
"--verbose",
dest="verbosity",
default=0,
action="count",
help="Verbosity. Invoke many times for higher verbosity")
parser.add_argument("-c",
"--corpus",
dest="corpus",
required=True,
help="Corpus file to save to/load from")
parser.add_argument("-d",
"--dictionary",
dest="dictionary",
required=True,
help="Dictionary file to save to/load from")
parser.add_argument("-m",
"--model",
dest="model",
required=True,
help="Model file to save to/load from")
parser.add_argument("-t",
"--topics",
dest="topics",
required=True,
type=int,
help="Number of topics")
parser.add_argument(
"-p",
"--passes",
dest="passes",
type=int,
default=1,
help="How many passes of the data (default: %(default)s)")
parser.add_argument(
"-e",
"--english-only",
dest="filterEnglishOnly",
default=False,
action="store_true",
help="Filter corpus and only let through english lyrics")
parser.add_argument("--save-filtered",
dest="filteredCorpus",
default=None,
help="Save the filtered corpus to this filename")
parser.add_argument(
"-s",
"--filter-stop",
dest="filterStopWords",
default=False,
action="store_true",
help=
"Filter stop words before modelling (only implemented for english at the moment)"
)
parser.add_argument("--elastic-search",
dest="elasticSearch",
default=False,
action="store_true",
help="Get data from elastic search instead of a file")
parser.add_argument("lyrics", nargs="?", help="File to load lyrics from")
parameters = parser.parse_args(args)
logger = SimpleLogSetter(verbosity=parameters.verbosity)
logger.startLogging()
dictionaryPath = os.path.expanduser(parameters.dictionary)
if os.path.exists(dictionaryPath):
logging.info("Loading dictionary from %s" % dictionaryPath)
dictionary = corpora.Dictionary.load(dictionaryPath)
allowUpdate = False
else:
dictionary = corpora.Dictionary()
allowUpdate = True
corpusPath = os.path.expanduser(parameters.corpus)
if len(dictionary) > 0 and os.path.exists(corpusPath):
logging.info("Loading corpus from %s" % corpusPath)
allLyrics = corpora.MmCorpus(corpusPath)
else:
if parameters.elasticSearch:
lyricsGenerator = getLyricsFromElasticSearch()
else:
lyricsGenerator = getLyricsFromJson(
os.path.expanduser(parameters.lyrics))
allLyrics = []
for songLyrics in lyricsGenerator:
allLyrics.append(
dictionary.doc2bow(cleanLyrics(songLyrics),
allow_update=allowUpdate))
corpora.MmCorpus.serialize(corpusPath, allLyrics)
if allowUpdate:
logging.info("Saving dictionary to %s" % dictionaryPath)
dictionary.save(dictionaryPath)
if parameters.filterEnglishOnly:
logging.info("Filtering for english only")
allLyrics = englishOnly(allLyrics, dictionary)
if parameters.filteredCorpus:
corpora.MmCorpus.serialize(
os.path.expanduser(parameters.filteredCorpus), allLyrics)
if parameters.filterStopWords:
logging.info("Filtering out stop words")
allLyrics = filterFrequentWords(allLyrics, dictionary)
model = models.ldamodel.LdaModel(corpus=allLyrics,
id2word=dictionary,
num_topics=parameters.topics,
passes=parameters.passes)
model.save(os.path.expanduser(parameters.model))
return 0