def test_get_stemmer_positive():
"""Test for the method get_stemmer()."""
stemmer = Stemmer.get_stemmer("LancasterStemmer")
assert isinstance(stemmer, nltk.stem.LancasterStemmer)
stemmer = Stemmer.get_stemmer("PorterStemmer")
assert isinstance(stemmer, nltk.stem.PorterStemmer)
stemmer = Stemmer.get_stemmer("EnglishStemmer")
assert isinstance(stemmer, nltk.stem.snowball.EnglishStemmer)
def test_get_stemmer_negative():
"""Test for the method get_stemmer()."""
with pytest.raises(StemmerNotFoundError):
stemmer = Stemmer.get_stemmer("unknown")
print(stemmer)
with pytest.raises(StemmerNotFoundError):
stemmer = Stemmer.get_stemmer("")
print(stemmer)
with pytest.raises(StemmerNotFoundError):
stemmer = Stemmer.get_stemmer(None)
print(stemmer)
def _prepare_lookup(keywords_file=None, stopwords_file=None, ngram_size=None, lemmatize=False,
stemmer=None):
# pylint: disable=too-many-arguments
"""Prepare resources for keywords lookup.
:param keywords_file: keywords file to be used
:param stopwords_file: stopwords file to be used
:param ngram_size: size of ngrams, if None, ngram size is computed
:param lemmatize: use lemmatizer
:type lemmatize: bool
:param stemmer: stemmer to be used
:type stemmer: str
"""
stemmer_instance = Stemmer.get_stemmer(stemmer) if stemmer is not None else None
lemmatizer_instance = Lemmatizer.get_lemmatizer() if lemmatize else None
chief = KeywordsChief(keywords_file, lemmatizer=lemmatizer_instance, stemmer=stemmer_instance)
computed_ngram_size = chief.compute_ngram_size()
if ngram_size is not None and computed_ngram_size > ngram_size:
_logger.warning("Computed ngram size (%d) does not reflect supplied ngram size (%d), "
"some synonyms will be omitted", chief.compute_ngram_size(), ngram_size)
elif ngram_size is None:
ngram_size = computed_ngram_size
tokenizer = Tokenizer(stopwords_file, ngram_size, lemmatizer=lemmatizer_instance,
stemmer=stemmer_instance)
return ngram_size, tokenizer, chief, CoreParser()
def reckon(keywords_file=None,
stopwords_file=None,
stemmer=None,
lemmatize=False):
"""Compute keywords and stopwords based on stemmer and lemmatizer configuration.
:param keywords_file: keywords file to be used
:param stopwords_file: stopwords file to be used
:param stemmer: stemmer to be used
:param lemmatize: True if lemmatization should be done
:return: computed keywords and stopwords, duplicit entries are not removed
"""
result = dict.fromkeys(('keywords', 'stopwords'))
stemmer_instance = Stemmer.get_stemmer(
stemmer) if stemmer is not None else None
lemmatizer_instance = Lemmatizer.get_lemmatizer() if lemmatize else None
chief = KeywordsChief(keywords_file,
lemmatizer=lemmatizer_instance,
stemmer=stemmer_instance)
tokenizer = Tokenizer(stopwords_file,
lemmatizer=lemmatizer_instance,
stemmer=stemmer_instance)
result['keywords'] = chief.keywords
result['stopwords'] = sorted(tokenizer.raw_stopwords) + sorted(
tokenizer.regexp_stopwords)
return result
def test_get_registered_stemmers():
"""Test for the class method get_registered_stemmers()."""
stemmers = Stemmer.get_registered_stemmers()
assert stemmers
# we expected at least three stemmers to be registered
assert len(stemmers) >= 3
def get_registered_stemmers():
"""Get all stemmers that are available in NLTK."""
return Stemmer.get_registered_stemmers()
def lookup(path,
keywords_file=None,
stopwords_file=None,
ignore_errors=False,
ngram_size=None,
use_progressbar=False,
lemmatize=False,
stemmer=None):
# pylint: disable=too-many-arguments,too-many-locals
"""Perform keywords lookup.
:param path: path of directory tree or file on which the lookup should be done
:param keywords_file: keywords file to be used
:param stopwords_file: stopwords file to be used
:param ignore_errors: True, if errors should be reported but computation shouldn't be stopped
:param ngram_size: size of ngrams, if None, ngram size is computed
:param use_progressbar: True if progressbar should be shown
:param lemmatize: use lemmatizer
:type lemmatize: bool
:param stemmer: stemmer to be used
:type stemmer: str
:return: found keywords, reported per file
"""
ret = {}
stemmer_instance = Stemmer.get_stemmer(
stemmer) if stemmer is not None else None
lemmatizer_instance = Lemmatizer.get_lemmatizer() if lemmatize else None
chief = KeywordsChief(keywords_file,
lemmatizer=lemmatizer_instance,
stemmer=stemmer_instance)
computed_ngram_size = chief.compute_ngram_size()
if ngram_size is not None and computed_ngram_size > ngram_size:
_logger.warning(
"Computed ngram size (%d) does not reflect supplied ngram size (%d), "
"some synonyms will be omitted", chief.compute_ngram_size(),
ngram_size)
elif ngram_size is None:
ngram_size = computed_ngram_size
tokenizer = Tokenizer(stopwords_file,
ngram_size,
lemmatizer=lemmatizer_instance,
stemmer=stemmer_instance)
for file in progressbarize(iter_files(path, ignore_errors),
progress=use_progressbar):
_logger.info("Processing file '%s'", file)
try:
content = CoreParser().parse_file(file)
tokens = tokenizer.tokenize(content)
# We do not perform any analysis on sentences now, so treat all tokens as one array (sentences of tokens).
tokens = chain(*tokens)
keywords = chief.extract_keywords(tokens)
except Exception as exc: # pylint: disable=broad-except
if not ignore_errors:
raise
_logger.exception("Failed to parse content in file '%s': %s", file,
str(exc))
continue
ret[file] = keywords
return ret
