def generate_docs_file(file_):
from past.model.user import User
uids = User.get_ids(start=0, limit=400)
with open(file_, "aw") as f:
for x in uids:
print 'get user %s text' % x
text = get_all_text_by_user(x)
f.write(text.encode("utf8"))
def get_keywords(user_id=config.MY_USER_ID, count=30):
text = get_all_text_by_user(user_id)
file_ = "/tmp/tag_%s" % randbytes(8)
with open(file_, 'w') as f:
f.write(text.encode("utf8"))
try:
cmd = '%s -I -d %s -c utf8 -t500 -i "%s"|grep -E "^[0-9]+"' \
% (config.SCWS, config.HOT_TERMS_DICT, file_)
r = commands.getoutput(cmd)
except Exception, e:
print e
def get_keywords(user_id=config.MY_USER_ID, count=30):
text = get_all_text_by_user(user_id)
file_ = "/tmp/tag_%s" % randbytes(8)
with open(file_, 'w') as f:
f.write(text.encode("utf8"))
try:
cmd = '%s -I -d %s -c utf8 -t200 -i "%s"|grep -E "^[0-9]+"' \
% (config.SCWS, config.HOT_TERMS_DICT, file_)
r = commands.getoutput(cmd)
except Exception, e:
import traceback
print traceback.format_exc()