def output_top_square_per_token(query):
csv_writer_coords = csv.writer(open('out/couch_coords.csv', 'wb'))
csv_writer_labels = UnicodeWriter(open('out/couch_labels.csv', 'wb'))
rows = query.all_tokens()
for row in rows:
(x_coord,y_coord,token) = query.top_square_for_token(row[1][0])
print (x_coord,y_coord,token)
csv_writer_coords.writerow([x_coord,y_coord])
csv_writer_labels.writerow([token])
def output_top_word_per_square(query):
csv_writer_coords = csv.writer(open('out/couch_coords.csv', 'wb'))
csv_writer_labels = UnicodeWriter(open('out/couch_labels.csv', 'wb'))
for x_coord in range(-200,200):
for y_coord in range(-200,200):
row = query.top_token_for_square(x_coord, y_coord)
print row
if (len(row)>2):
csv_writer_coords.writerow([x_coord,y_coord])
csv_writer_labels.writerow([row[2]])
def write_user_counts(cfg):
couch = Server(cfg.raw.couch_server_url)
db = couch[cfg.raw.couch_db]
#get counts from view
view = db.view('fake_txt/count', group=True)
user_word_vector = []
words = {}
users = {}
word_count = 0
user_count = 0
csv_word_counts = csv.writer(open(cfg.raw.csv_data, 'wb'))
csv_word_counts.writerow(('user_id', 'word_id','count'))
for row in view:
word = row.key[1]
user = row.key[0]
count = row.value
word_id = words.get(word, word_count)
if (word_id==word_count):
word_count = word_count+1 #used default in above, so must be a new word
words.setdefault(word,word_id)
user_id = users.get(user, user_count)
if (user_id==user_count):
user_count = user_count+1 #used default in above, so must be a new user
users.setdefault(user,user_id)
print user_id, word_id, count, "::", user, word, count, user_count, word_count
csv_word_counts.writerow((user_id, word_id,count))
csv_words = UnicodeWriter(open(cfg.raw.words_file, 'wb'))
csv_words.writerow(('word_id', 'word'))
for row in sorted(words.items(), key=lambda x: x[1]): #uses a great magic lambda that i do not fully understand MKT - courtesy http://stackoverflow.com/questions/613183/python-sort-a-dictionary-by-value
csv_words.writerow(("%i"%row[1],row[0]))
csv_users = csv.writer(open(cfg.raw.users_file, 'wb'))
csv_users.writerow(('user_id','screen'))
for row in sorted(users.items(), key=lambda x: x[1]):
csv_users.writerow((row[1],row[0]))
csv_user_labels = csv.writer(open(cfg.output.labels_file, 'wb'))
for row in sorted(users.items(), key=lambda x: x[1]):
csv_user_labels.writerow(([row[0]]))
print "User Count (put this into config.. input.number_of_examples field) :",user_count
print "Word Count (put this into config.. shape.input_vector_length field):",word_count