def write():
import loacore.load.file_load as file_load
import loacore.utils.file_writer as file_writer
files = file_load.load_database(load_deptrees=False)
for file in files:
directory = re.findall(r'../../data/raw/(.+)', file.file_name)
filename = re.findall(r'.+/(.+\.txt)', file.file_name)
words = []
for reviews in file.reviews:
for sentence in reviews.sentences:
for word in sentence.words:
words.append(word)
words = sorted(words, key=lambda w: w.word)
text = 'ID_Word\tID_Synset\tSynset\n'
for word in words:
if word.synset is not None:
text += (str(word.id_word) + '\t' +
str(word.synset.id_synset) + '\t' +
word.synset.synset_name + '\t# ' + word.lemma + '\n')
else:
text += (str(word.id_word) + '\t' + str(None) + '\t' +
str(None) + '\t# ' + word.lemma + '\n')
file_writer.write(
text, os.path.join('../../data/disambiguated/', directory[0]),
filename[0])
def adj_patterns_table():
import os
from loacore.conf import RESULT_PATH
import loacore.utils.file_writer as file_writer
import loacore.load.file_load as file_load
import loacore.analysis.pattern_recognition as pattern_recognition
# English files
ids = file_load.get_id_files_by_file_paths([r'.*/uci/.+'])
files = file_load.load_database(id_files=ids)
for file in files:
table = pattern_recognition.adj_pattern_table(file.sentence_list())
directory = os.path.join(RESULT_PATH, 'context_tables', 'adj',
file.get_directory_name())
file_writer.write(table, directory, file.get_filename())
# Spanish files
ids = file_load.get_id_files_by_file_paths([r'.*/corrected/.+'])
files = file_load.load_database(id_files=ids)
for file in files:
table = pattern_recognition.adj_pattern_table(file.sentence_list(),
lang='es')
directory = os.path.join(RESULT_PATH, 'context_tables', 'adj',
file.get_directory_name())
file_writer.write(table, directory, file.get_filename())
def verb_patterns_tables():
import os
from loacore.conf import RESULT_PATH
import loacore.utils.file_writer as file_writer
import loacore.load.file_load as file_load
import loacore.analysis.pattern_recognition as pattern_recognition
ids = file_load.get_id_files_by_file_paths(
[r'.*/uci/.+', r'.*/corrected/.+'])
files = file_load.load_database(id_files=ids, load_deptrees=False)
for file in files:
table = pattern_recognition.verb_context_table(file.sentence_list())
directory = os.path.join(RESULT_PATH, 'context_tables', 'verbs',
file.get_directory_name())
file_writer.write(table, directory, file.get_filename())
def write():
import loacore.load.file_load as file_load
import loacore.utils.file_writer as file_writer
files = file_load.load_database(load_deptrees=False)
for file in files:
directory = re.findall(r'../../data/raw/(.+)', file.file_name)
filename = re.findall(r'.+/(.+\.txt)', file.file_name)
text = ''
for review in file.reviews:
text += (str(review.file_index) + '\t')
text += review.review
text += "\n"
file_writer.write(text,
os.path.join('../../data/normalized/', directory[0]),
filename[0])
def write_frequencies(files_frequencies,
file_path=os.path.join(RESULT_PATH, 'frequencies', 'tables')):
"""
Write frequencies to a txt file in file_path folder. Original raw file names are used.
:param files_frequencies: Dictionary that maps file names to frequencies.
:type files_frequencies: :obj:`dict` of :obj:`str` : :obj:`dict` of label : :obj:`float` .
:param file_path: Directory path
:type file_path: |path-like-object|
"""
import loacore.utils.file_writer as file_writer
for file_name in files_frequencies.keys():
table = ''
for label in files_frequencies[file_name].keys():
table += str(label) + '\t' + str(files_frequencies[file_name][label]) + '\n'
file_writer.write(table, file_path, file_name)
def write():
import loacore.load.file_load as file_load
import loacore.utils.file_writer as file_writer
files = file_load.load_database()
for file in files:
directory = re.findall(r'../../data/raw/(.+)', file.file_name)
filename = re.findall(r'.+/(.+\.txt)', file.file_name)
text = ''
for review in file.reviews:
for sentence in review.sentences:
text += "+-------------------------------------------------------------------------------------------\n"
text += "| File Index : " + str(review.file_index) + "\n"
text += "| Sentence : " + " ".join(
[w.word for w in sentence.words]) + "\n"
text += "+-------------------------------------------------------------------------------------------\n"
text += sentence.dep_tree.dep_tree_str(print_dep_tree=False)
text += "\n\n"
file_writer.write(text,
os.path.join('../../data/dep_trees/', directory[0]),
filename[0])
def write():
import loacore.load.file_load as file_load
import loacore.utils.file_writer as file_writer
files = file_load.load_database(load_deptrees=False)
for file in files:
directory = re.findall(r'../../data/raw/(.+)', file.file_name)
filename = re.findall(r'.+/(.+\.txt)', file.file_name)
words = []
for reviews in file.reviews:
for sentence in reviews.sentences:
for word in sentence.words:
words.append(word)
words = sorted(words, key=lambda w: w.word)
text = 'ID_Word\ttoken'
for word in words:
text += (str(word.id_word) + '\t' + word.word + '\n')
file_writer.write(text,
os.path.join('../../data/tokenized/', directory[0]),
filename[0])
def write_polarity_check(files,
analysis_to_check=("simple", "optimistic",
"pessimistic", "pattern_adj_cc",
"pattern_adj", "pattern_cc"),
ref="label",
select="all",
terminal_print=True,
colored_polarity=True,
directory_path=os.path.join(RESULT_PATH,
'sentiment_analysis',
'check')):
"""
Write polarity in .txt files. If *select* is set to all, each review is written with polarities corresponding to
*ref* and *analysis_to_check*. If *select* is set to *false_positive* or *false_negative*, only the review with a
false positive or false negative for at least one analysis of *analysis_to_check* are written, with the
corresponding(s) polarities.
:param files: Files to process.
:type files: :obj:`list` of |File|
:param analysis_to_check: Analysis to compare with *ref*
:type analysis_to_check: :obj:`sequence` of :obj:`str`
:param ref: Reference analysis
:type ref: string
:param select: Select option
:type select: :obj:`list` of :obj:`str` : {'all', 'false_positive', 'false_negative'}
:param terminal_print: If True, print results in the terminal.
:type terminal_print: boolean
:param colored_polarity:
If True, write words with colored polarity. Notice that colors are not display in most of the .txt editors.
For example, in Linux, use
.. code-block:: console
cat file_name.txt
To show the colored file in terminal.
:param directory_path:
Path of the directory in which to write files.
Default value computed using os.path.join(RESULT_PATH, 'sentiment_analysis', 'check')
:type directory_path: |path-like-object|
"""
import loacore.utils.file_writer as file_writer
for file in files:
if select == "all":
check_str =\
'\n'.join([r.review_str(colored_polarity=colored_polarity, analysis=ref+analysis_to_check)
for r in file.reviews])
else:
false_str = []
for review in file.reviews:
selected_analysis = [ref]
for analysis in analysis_to_check:
if select == "false_positive" and is_false_positive(review, ref, analysis) \
or select == "false_negative" and is_false_negative(review, ref, analysis):
selected_analysis.append(analysis)
if len(selected_analysis) > 1:
false_str.append(
review.review_str(colored_polarity=colored_polarity,
analysis=selected_analysis))
check_str = '\n'.join(false_str)
if terminal_print:
print(check_str)
if colored_polarity:
file_writer.write(check_str, os.path.join(directory_path,
"colored"),
"check_" + select + "_" + file.get_filename())
print("File write : " + str(
os.path.join(os.path.join(directory_path, "colored"),
"check_" + select + "_" + file.get_filename())))
else:
file_writer.write(check_str,
os.path.join(directory_path, "uncolored"),
"check_" + select + "_" + file.get_filename())
print("File write : " + str(
os.path.join(os.path.join(directory_path, "uncolored"),
"check_" + select + "_" + file.get_filename())))