def _analyze_company_source(self, company_id, source_type, from_date, to_date, write_to_filename):
# Get documents.
documents = getattr(self.dbmodel, 'get_'+source_type)(company_id, from_date, to_date)
# Create a final list.
docs_list = []
for doc in documents:
#print doc['id'],
# Process text.
if source_type == 'articles':
text = TextProcessing.process_article_text(doc['text'])
else:
text = TextProcessing.process_facebook_text(doc['text'])
# Skip empty documents.
if len(text) == 0:
continue
# Get sentiment values of the text.
if source_type == 'articles':
sentiment_number = self.s_analyzer.calculate_vader_sentiment('custom_dict_orig', text, True)
else:
sentiment_number = self.s_analyzer.calculate_vader_sentiment('custom_dict_orig', text, False)
sentiment_polarity = self.s_analyzer.format_sentiment_value(sentiment_number)
# Save data.
doc_date = self._get_doc_date(source_type, doc)
docs_list.append([company_id, doc_date, sentiment_number, sentiment_polarity])
# Write to file.
self.text_writer.write_file(docs_list, write_to_filename, 'csv', ',', 'a')
def _analyze_documents_by_source(self, company_id, source_type, from_date, to_date, write_to_filename=False):
# Get documents
documents = getattr(self.dbmodel, 'get_'+source_type)(company_id, from_date, to_date)
# Create a final list
docs_list = []
for doc in documents:
# Process text
if source_type == 'articles':
text = TextProcessing.process_article_text(doc['text'])
else:
text = TextProcessing.process_facebook_text(doc['text'])
# Skip empty documents
if len(text) == 0:
continue
# Get sentiment values of the text
sent_sum, sent_division = self.s_analyzer.calculate_vader_sentiment_values('vader', text)
# Add this to list
docs_list.append([sent_sum, sent_division, text])
# Prepare header
header = ['sentiment_sum', 'sentiment_division', 'text']
docs_list.insert(0, header)
file_name = '%d_%s' % (company_id, source_type)
self.text_writer.write_file(docs_list, file_name, 'csv', '\t', 'w')
def _process_tweets(self, company_id, examined_date, used_dict_name='vader'):
# Select all Yahoo Finance articles.
tweets = self.dbmodel.get_daily_tweets(company_id, examined_date)
counter = {'pos': 0, 'neu': 0, 'neg': 0}
# Calculate sentiment for all articles.
for tw in tweets:
#print("Tweet: %s") % tw['tw_id'],
tw_text = TextProcessing.process_facebook_text(tw['text'])
if len(tw_text) == 0:
continue # skip empty tweets
sent_value = self.s_analyzer.calculate_vader_sentiment(used_dict_name, tw_text, False)
polarity = self.s_analyzer.format_sentiment_value(sent_value)
counter[polarity] += 1
#print("| %s ... %s") % (str(round(sent_value, 4)), polarity)
# result
return counter
def _process_yahoo(self, company_id, examined_date, used_dict_name='vader'):
# Select all Yahoo Finance articles.
articles = self.dbmodel.get_daily_articles(company_id, examined_date)
counter = {'pos': 0, 'neu': 0, 'neg': 0}
# Calculate sentiment for all articles
for art in articles:
#print("Yahoo article: %s") % art['id'],
art_text = TextProcessing.process_article_text(art['text'])
if len(art_text) == 0:
continue # skip empty articles
sent_value = self.s_analyzer.calculate_vader_sentiment(used_dict_name, art_text, True)
polarity = self.s_analyzer.format_sentiment_value(sent_value)
counter[polarity] += 1
#print("| %s ... %s") % (str(round(sent_value, 4)), polarity)
# result
return counter
def _process_fb_comments(self, company_id, examined_date, used_dict_name='vader'):
# Select all FB comments.
comments = self.dbmodel.get_daily_fb_comments(company_id, examined_date)
counter = {'pos': 0, 'neu': 0, 'neg': 0}
# Calculate sentiment for all posts
for com in comments:
#print("FB comment: %s") % com['id'],
com_text = TextProcessing.process_facebook_text(com['text'])
if len(com_text) == 0:
continue # skip empty comments
sent_value = self.s_analyzer.calculate_vader_sentiment(used_dict_name, com_text, False)
polarity = self.s_analyzer.format_sentiment_value(sent_value)
counter[polarity] += 1
#print("| %s ... %s") % (str(round(sent_value, 4)), polarity)
# result
return counter
