def evaluate_and_save_stats(input_file, file_name, output_dir, helpers,
includeSemanticSimilarity):
measures = defaultdict(dict)
data = ''
with open(input_file) as data_file:
data = json.load(data_file)
stopwords = helpers['stopwords']
spellcheck = helpers['spellcheck']
texts = merge_texts(data)
original_text = texts[0]
obfuscated_text = texts[1]
document_stats_original = DocumentStats(original_text, spellcheck,
stopwords)
document_stats_obfuscation = DocumentStats(obfuscated_text, spellcheck,
stopwords)
measures['safety_sentenceLengthChange'] = measure_rate_change(
document_stats_original.average_sentence_length,
document_stats_obfuscation.average_sentence_length)
measures['safety_nounRateChange'] = measure_rate_change(
document_stats_original.average_noun_rate,
document_stats_obfuscation.average_noun_rate)
measures['safety_verbRateChange'] = measure_rate_change(
document_stats_original.average_verb_rate,
document_stats_obfuscation.average_verb_rate)
measures['safety_adjectivesRateChange'] = measure_rate_change(
document_stats_original.average_adj_rate,
document_stats_obfuscation.average_adj_rate)
measures['safety_adverbsRateChange'] = measure_rate_change(
document_stats_original.average_adv_rate,
document_stats_obfuscation.average_adv_rate)
measures['safety_punctuationRateChange'] = measure_rate_change(
document_stats_original.average_punct_rate,
document_stats_obfuscation.average_punct_rate)
measures['safety_stopWordsRateChange'] = measure_rate_change(
document_stats_original.stop_words_ratio,
document_stats_obfuscation.stop_words_ratio)
measures['safety_uniqueWordsRateChange'] = measure_rate_change(
document_stats_original.unique_words_ratio,
document_stats_obfuscation.unique_words_ratio)
measures['safety_capitalizedWordsRateChange'] = measure_rate_change(
document_stats_original.words_all_capital_letters_ratio,
document_stats_obfuscation.words_all_capital_letters_ratio)
measures['safety_sequenceDistance'] = similarity.simple_distance(
original_text, obfuscated_text)
if includeSemanticSimilarity:
measures['soundness_semanticSimilarity'] = average_semantic_similarity(
data)
for key, value in measures.items():
measure = \
'measure {\n' + \
' key : "' + str(key) + '"\n' + \
' value: "' + str(math.fabs(round(value, 4))) + '"\n' + \
'}\n'
#print(measure)
output_writer.ensure_directory_exists(output_dir + '/' + file_name)
output_writer.write_text_to_file(
measure, output_dir + '/' + file_name + '/evaluation_result.txt')
def save_stats_for_all_files(input_dir, output_dir):
# Make sure the given output directory exists
output_writer.ensure_directory_exists(output_dir)
# Create helper objects in order to not create them for every file
helpers = {'spellcheck': Spellcheck(), 'stopwords': StopWords()}
# Read all files in the given directory and save stats in one file
with open(output_dir + '\stats.csv', 'w', newline='') as csvfile:
csv_writer = csv.writer(csvfile,
delimiter=',',
quotechar='|',
quoting=csv.QUOTE_MINIMAL)
csv_writer.writerow([
'filename', 'attr_name', 'original', 'obfuscated', 'change',
'change_rate', 'average'
])
for file_name in os.listdir(input_dir):
print(file_name)
doc_stats = get_document_stats(input_dir + '/' + file_name,
helpers)
original_stats = doc_stats[0]
obfuscation_stats = doc_stats[1]
for attr_name, average_value in AverageStats.EVALUATION_MEASURES.items(
):
orig = getattr(original_stats, attr_name)
obf = getattr(obfuscation_stats, attr_name)
change = obf - orig
change_rate = change / orig
csv_writer.writerow([
file_name, attr_name,
str(orig),
str(obf),
str(change),
str(change_rate),
str(average_value)
])
def evaluate_and_save_stats_all_files(input_dir, output_dir,
includeSemanticSimilarity):
# Make sure the given output directory exists
output_writer.ensure_directory_exists(output_dir)
# Create helper objects in order to not create them for every file
helpers = {'spellcheck': Spellcheck(), 'stopwords': StopWords()}
for file_name in os.listdir(input_dir):
evaluate_and_save_stats(
input_dir + '/' + file_name + '/obfuscation.json', file_name,
output_dir, helpers, includeSemanticSimilarity)
def obfuscate_and_save_file(input_file_path, dir_name, output_dir, helpers):
print('Obfuscating file ' + input_file_path)
# Read the input text
input_text = input_reader.read_input_file(input_file_path)
# Get array of split parts of the input text
file_parts = text_utils.split_text(input_text)
file_parts = obfuscator.obfuscate_all(input_text, file_parts, helpers)
output_writer.ensure_directory_exists(output_dir + '/' + dir_name)
output_file = output_dir + '/' + dir_name + '/obfuscation.json'
print('Output: ' + output_file)
output_writer.write_output_file(output_file, file_parts)
def obfuscate_and_save_all_files(input_dir, output_dir):
print('Parsing input...')
helpers = {
'stopwords': StopWords(),
'spellcheck': SpellChecker(
), # It seems there is a name clash with another module, so needed to be renamed for this file
'errorCreator': ErrorCreator(),
'punctuation': Punctuation(),
'britishToAmerican': BritishAmericanNormalization(),
'fillerWords': FillerWords(),
'symbolReplacement': SymbolReplacement(),
'paraphraseCorpus': ParaphraseCorpus()
}
# Make sure the given output directory exists
output_writer.ensure_directory_exists(output_dir)
# Read all files in the given directory and for each execute obfuscation
for dir_name in os.listdir(input_dir):
if dir_name[0] != '.': #skip hidden files and folders
print(dir_name)
obfuscate_and_save_file(
input_dir + '/' + dir_name + '/original.txt', dir_name,
output_dir, helpers)