def testSpawnThreads(self):
self.jobmanager.loadJob(Job(1, "*/2 * * * *", "echo haha", 5))
self.jobmanager.indexJobs()
self.assertEqual(1, len(self.jobmanager.hour_index[15]))
threads = self.jobmanager.spwanTasks(DateUtil.datetime("2014-02-18 15:30:00"), DateUtil.datetime("2014-02-18 16:00:00"))
self.assertEqual(15, len(threads))
threads = self.jobmanager.spwanTasks(DateUtil.datetime("2014-02-18 15:30:23"), DateUtil.datetime("2014-02-18 16:00:00"))
self.assertEqual(14, len(threads))
def testSpawnThreads2(self):
self.jobmanager.loadJob(Job(1, "* * * * *", "echo haha", 5))
self.jobmanager.indexJobs()
threads = self.jobmanager.spwanTasks(DateUtil.datetime("2014-02-19 15:00:00"), DateUtil.datetime("2014-02-19 16:00:00"))
self.assertEqual(60, len(threads))
print threads[0].args[0].get_exc_time()
def transfer_texts_2_tfRecord(self, start_date, end_date, raw_feature_file_path, tf_record_file_path,
column_names,
label_name, need_feature_cols, negative_ratio=None, var_length_cols=None,
col_preprocess_func=None):
"""
根据起止时间，将指定目录下的text文件，都存储为tfRecord格式
参数见：transfer_single_text_2_tfRecord
"""
raw_feature_folder_name = os.path.basename(raw_feature_file_path)
tf_record_folder_name = os.path.basename(tf_record_file_path)
raw_feature_file = os.path.join(raw_feature_file_path, raw_feature_folder_name + "_%s")
tf_record_file = os.path.join(tf_record_file_path, tf_record_folder_name + "_%s.tfrecord")
data_info_csv_path = os.path.join(tf_record_file_path, "data_info.csv")
if not os.path.isdir(tf_record_file_path):
os.makedirs(tf_record_file_path)
date_ls = DateUtil.get_every_date(start_date, end_date)
for date in date_ls:
print(date)
self.transfer_single_text_2_tfRecord(raw_feature_file % date, tf_record_file % date, data_info_csv_path,
column_names,
label_name, need_feature_cols, negative_ratio, var_length_cols,
col_preprocess_func)
def load_wechat_2_dict(start_date, end_date):
date_ls = DateUtil.get_every_date(start_date, end_date)
dict_wechat = OrderedDict() #
wechat_segment_data_dir = os.path.join(home_dir, "project_data/xgb",
"raw_data/aggregated_wechat_segment_data")
wechat_segment_data = os.path.join(wechat_segment_data_dir, "aggregated_wechat_segment_data_%s")
log.info(date_ls)
for tmp_date in date_ls:
log.info(tmp_date)
dict_wechat[tmp_date] = OrderedDict()
with codecs.open(wechat_segment_data % tmp_date, "r", "utf-8") as fin:
for line in fin:
arr = line.strip().split("\t")
if len(arr) != 2:
continue
opp_id = arr[0].strip()
chat_ls = arr[1].strip()
try:
chat_ls = json.loads(chat_ls)
except Exception as e:
log.info(e)
continue
dict_wechat[tmp_date][opp_id] = chat_ls
return dict_wechat
def load_multi_day_order(start_date, end_date):
multi_day_dict = {}
date_ls = DateUtil.get_every_date(start_date, end_date)
date_ls = sorted(date_ls, reverse=True) # 降序排列，如果出现多次，用最小的订单时间覆盖
for date in date_ls:
log.info(date)
one_day_dict = load_one_day_order(date)
multi_day_dict.update(one_day_dict)
return multi_day_dict
def get_hist_wechat_segment(start_date, end_date):
date_ls = DateUtil.get_every_date(start_date, end_date)
for tmp_date in date_ls:
start_time = time.time()
get_one_day_wechat_segment(tmp_date)
print("{0} wechat segment cost time: {1}".format(
tmp_date,
time.time() - start_time))
def testNextScheduleSectionOverLongBatch(self):
'''
Test get next schedule Section
Test invalid over long interval
'''
self.engine._fillclock(11, 4) # Get clock list [3, 7, 11, 15, 19, 23]
try:
self.engine._getNextScheduleSection(DateUtil.datetime('2014-02-14 15:06:34'), 23)
except Exception:
raise
def update_employee(self, emp: Employee):
try:
emp_result = self.fetch_employee_by_id(emp.emp_id)
try:
if emp.__eq__(emp_result):
#
print("Successfully Updated Employee- " + emp.emp_id +
"\n")
else:
date_util = DateUtil()
if date_util.check_date_of_birth(emp.dob):
self.db.execute_dynamic_query(
"update employee set fname = ?, lname =?, dob = ? , dept_id = ? where emp_id=?",
emp.fname, emp.lname, emp.dob, emp.dept.dept_id,
emp.emp_id)
self.db.connection.commit()
print("Successfully Updated Employee- " + emp.emp_id +
"\n")
else:
print("Sorry!! Unable to update Employee- " +
emp.emp_id + "\n")
except AttributeError:
if "fname" in str(sys.exc_info()[1]):
print("First Name cannot be null")
elif "lname" in str(sys.exc_info()[1]):
print("Last Name cannot be null")
elif "dob" in str(sys.exc_info()[1]):
print("Date of birth is not in correct format")
except ValueError:
print(sys.exc_info()[1])
except:
print("Unable to update employee Id - " + str(emp.emp_id) +
". Check employee Id.\n")
print(sys.exc_info())
print("\n")
def load_hist_wechat_record_dict(date):
wechat_dict = defaultdict(list)
start_date = DateUtil.get_relative_delta_time_str(
date, day=-HISTORY_WECHAT_RECORD_DELTA_DAY)
end_date = DateUtil.get_relative_delta_time_str(date, -1)
date_ls = sorted(DateUtil.get_every_date(start_date,
end_date)) # 时间在后的聊天追加在后面
wechat_full_sentence_data_dir = os.path.join(PROJECT_DATA_DIR, "raw_data",
"wechat_full_sentence_data")
wechat_full_sentence_data_file = os.path.join(
wechat_full_sentence_data_dir, "wechat_full_sentence_data_%s")
for date in date_ls:
log.info(date)
wechat_full_sentence_data = wechat_full_sentence_data_file % date
with codecs.open(wechat_full_sentence_data, 'r', 'utf-8') as fin:
for line in fin:
arr = line.strip().split("\t")
if len(arr) != 5:
continue
opp_id, student_chat_num, teacher_chat_num, all_chat_num, chat_content = arr
student_chat_num, teacher_chat_num, all_chat_num = int(
student_chat_num), int(teacher_chat_num), int(all_chat_num)
if opp_id not in wechat_dict:
wechat_dict[opp_id] = {
"stat_info": [0, 0, 0],
"chat_content": ""
}
wechat_dict[opp_id]["chat_content"] = wechat_dict[opp_id][
"chat_content"] + chat_content
wechat_dict[opp_id]["stat_info"] = [
x + y for x, y in
zip([student_chat_num, teacher_chat_num, all_chat_num],
wechat_dict[opp_id]["stat_info"])
]
return wechat_dict
def testMatchTimePattern(self):
self.assertEqual(True, TimeMatcher.matchTimePattern(TimePattern("* * * * *"),
DateUtil.datetime("2014-02-17 20:28:35")))
self.assertEqual(True, TimeMatcher.matchTimePattern(TimePattern("28 * * * *"),
DateUtil.datetime("2014-02-17 20:28:35")))
self.assertEqual(True, TimeMatcher.matchTimePattern(TimePattern("* 20 * * *"),
DateUtil.datetime("2014-02-17 20:28:35")))
self.assertEqual(True, TimeMatcher.matchTimePattern(TimePattern("* * 17 * *"),
DateUtil.datetime("2014-02-17 20:28:35")))
self.assertEqual(True, TimeMatcher.matchTimePattern(TimePattern("* * * 2 *"),
DateUtil.datetime("2014-02-17 20:28:35")))
self.assertEqual(True, TimeMatcher.matchTimePattern(TimePattern("* * * * 1"),
DateUtil.datetime("2014-02-17 20:28:35")))
self.assertEqual(True, TimeMatcher.matchTimePattern(TimePattern("28 20 17 2 1"),
DateUtil.datetime("2014-02-17 20:28:35")))
self.assertEqual(True, TimeMatcher.matchTimePattern(TimePattern("*/2 * * * *"),
DateUtil.datetime("2014-02-17 20:28:35")))
self.assertEqual(False, TimeMatcher.matchTimePattern(TimePattern("*/3 * * * *"),
DateUtil.datetime("2014-02-17 20:28:35")))
def get_hist_wechat_tf_feature(start_date, end_date):
date_ls = DateUtil.get_every_date(start_date, end_date)
log.info("initial past n day wechat segment data...")
s_date = (datetime.strptime(start_date, "%Y%m%d") - timedelta(days=PAST_DAYS_LIMIT)).strftime("%Y%m%d")
t_date = start_date
log.info("load past days wechat segment data...")
dict_wechat = load_wechat_2_dict(s_date, t_date)
log.info("extract tf feature...")
for tmp_date in date_ls:
log.info("extract %s tf feature..." % tmp_date)
start_time = time.time()
get_one_day_wechat_tf_feature(tmp_date, dict_wechat)
log.info("extract {0} wechat tf feature cost time:{1}".format(tmp_date, time.time()-start_time))
del_date = (datetime.strptime(tmp_date, "%Y%m%d") - timedelta(days=PAST_DAYS_LIMIT)).strftime("%Y%m%d")
add_date = (datetime.strptime(tmp_date, "%Y%m%d") + timedelta(days=1)).strftime("%Y%m%d")
if add_date <= end_date:
log.info("update past days wechat segment data [del %s, add %s]..." % (del_date, add_date))
dict_wechat = update_wechat_dict(dict_wechat, del_date, add_date)
log.info("=======" * 3)
def gen_bench_mark_multi_day(start_date, end_date):
bench_mark_text_file_tmp = os.path.join(bench_mark_text_file_path, "total_chat_num_%s", "total_chat_num_%s_%s")
for date in DateUtil.get_every_date(start_date, end_date):
print(date)
# 按chat_num分组
source_file_path = os.path.join(raw_feature_path, "wechat_basic_feature_%s" % date)
chat_group_dict = {x: [] for x in chat_num_ls}
with codecs.open(source_file_path, 'r', 'utf-8') as fin:
for line in fin:
arr = line.strip().split("\t")
column_names = ["label", "opp_id", "acc_id", "create_time", "today_student_chat_num",
"today_teacher_chat_num", "today_total_chat_num", "hist_student_chat_num",
"hist_teacher_chat_num", "hist_total_chat_num", "chat_content"]
tmp_dict = {key: value for key, value in zip(column_names, arr)}
total_student_chat_num = int(tmp_dict["today_student_chat_num"]) + int(
tmp_dict["hist_student_chat_num"])
if total_student_chat_num in chat_group_dict:
chat_group_dict[total_student_chat_num].append(line.strip())
for chat_num, line_ls in chat_group_dict.items():
if not line_ls:
continue
bench_mark_text_file = bench_mark_text_file_tmp % (chat_num, chat_num, date)
tmp_text_folder_path = os.path.dirname(bench_mark_text_file)
tmp_tf_record_folder_path = tmp_text_folder_path.replace("feature_file", "tf_record")
if not os.path.isdir(tmp_text_folder_path):
os.makedirs(tmp_text_folder_path)
with codecs.open(bench_mark_text_file, "w", "utf-8") as fout:
for line in line_ls:
fout.write(line + "\n")
# 将转换好的text bench mark 文本文件 ， 转换为tfRecord文件
tfRecorder.transfer_texts_2_tfRecord_default(date, date, tmp_text_folder_path, tmp_tf_record_folder_path)
LOG_PATH = os.path.join(PROJECT_DATA_DIR, "log")
def get_logger(path):
log_file = os.path.join(LOG_PATH, path)
logger = logging.getLogger(PROJECT_NAME) # 程序顶级目录的名字
fmt = '[%(asctime)s] - %(filename)s:%(lineno)s - %(name)s - %(message)s'
formatter = logging.Formatter(fmt) # 实例化formatter
handler = logging.handlers.RotatingFileHandler(log_file,
maxBytes=1024 * 1024,
backupCount=5) # 实例化handler
handler.setFormatter(formatter) # 为handler添加formatter
logger.addHandler(handler) # 为logger添加handler
logger.setLevel(logging.DEBUG)
consoleHandle = logging.StreamHandler()
consoleHandle.setFormatter(formatter)
logger.addHandler(consoleHandle)
return logger
G_LOG = get_logger("log_%s" % DateUtil.get_relative_delta_time_str())
if __name__ == "__main__":
logger = get_logger("log_20170523")
logger.info('first info message')
logger.debug('first debug message')
logger.debug('-----------')
def get_hist_wechat_full_sentence(start_date, end_date):
date_ls = DateUtil.get_every_date(start_date, end_date)
for tmp_date in date_ls:
log.info("extract %s basic feature..." % tmp_date)
get_one_day_wechat_full_sentence(tmp_date)
def get_one_day_wechat_basic_feature(date):
log.info("get %s wechat basic feature..." % date)
wechat_basic_feature_dir = os.path.join(PROJECT_DATA_DIR, "feature_file",
"wechat_basic_feature")
aggregated_wechat_data_dir = os.path.join(PROJECT_DATA_DIR, "raw_data",
"aggregated_wechat_data")
wechat_basic_feature_data = os.path.join(wechat_basic_feature_dir,
"wechat_basic_feature_%s" % date)
aggregated_wechat_data = os.path.join(aggregated_wechat_data_dir,
"aggregated_wechat_data_%s" % date)
log.info("prepare hist wechat chat dict...")
hist_wechat_chat_dict = load_hist_wechat_record_dict(date)
log.info("prepare hist and future order dict...")
hist_order_dict = load_multi_day_order(
DateUtil.get_relative_delta_time_str(date,
day=-HISTORY_ORDER_DELTA_DAY),
DateUtil.get_relative_delta_time_str(date, day=-1))
future_order_dict = load_multi_day_order(
date,
DateUtil.get_relative_delta_time_str(date, day=FUTURE_ORDER_DELTA_DAY))
log.info("start 2 gen wechat basic feature...")
with codecs.open(aggregated_wechat_data, "r", "utf-8") as fin, \
codecs.open(wechat_basic_feature_data, "w", "utf-8") as fout:
for line in fin:
arr = line.strip().split("\t")
if len(arr) != 2:
continue
opp_id = arr[0].strip()
chat_ls = arr[1].strip()
try:
chat_ls = json.loads(chat_ls, encoding="utf-8")
except Exception as e:
log.info(e)
continue
if opp_id in hist_order_dict:
continue
student_chat_idx = []
for idx, chat_dict in enumerate(chat_ls):
send_type = chat_dict["send_type"]
if send_type == "1":
student_chat_idx.append(idx)
if not student_chat_idx:
continue
# 随机选取一句学生对话，作为样本生成点
sample_idx = np.random.choice(student_chat_idx, 1)[0]
sample_chat = chat_ls[sample_idx]
create_time = sample_chat["create_time"]
account = sample_chat["account"]
order_time = future_order_dict.get(opp_id, None) # 是否最近成单
label = judge_label(order_time, create_time)
if label == "-1":
continue
sample_chat_ls = chat_ls[:sample_idx + 1]
cleared_chat_sentence = clear_sentence(sample_chat_ls)
chat_stat_ls = stat_sentence(sample_chat_ls)
hist_chat_stat_ls = [0, 0, 0]
hist_wechat_chat = hist_wechat_chat_dict.get(opp_id,
None) # 是否有历史聊天记录
if hist_wechat_chat: # 拼接历史聊天信息
hist_chat_stat_ls = hist_wechat_chat["stat_info"]
cleared_chat_sentence = hist_wechat_chat[
"chat_content"] + cleared_chat_sentence
today_stat_str = "\t".join(map(str, chat_stat_ls))
hist_stat_str = "\t".join(map(str, hist_chat_stat_ls))
result = "\t".join([
label, opp_id, account, create_time, today_stat_str,
hist_stat_str, cleared_chat_sentence
])
fout.write(result + "\n")
# 不抽样，剖析每个对话，学生说话则触发一次样本生成
# for idx, chat_dict in enumerate(chat_ls):
# send_type = chat_dict["send_type"]
# create_time = chat_dict["create_time"]
# accout = chat_dict["account"]
#
# if send_type == "0": # 老师会话不做样本选取点
# continue
#
# label = judge_label(order_time, create_time)
# if label == "-1":
# continue
#
# cleared_chat_sentence = clear_sentence(chat_ls[:idx + 1])
# chat_stat_ls = stat_sentence(chat_ls[:idx + 1])
# hist_chat_stat_ls = [0, 0, 0]
#
# if hist_wechat_chat: # 拼接历史聊天信息
# hist_chat_stat_ls = hist_wechat_chat["stat_info"]
# cleared_chat_sentence = hist_wechat_chat["chat_content"] + cleared_chat_sentence
#
# today_stat_str = "\t".join(map(str, chat_stat_ls))
# hist_stat_str = "\t".join(map(str, hist_chat_stat_ls))
# result = "\t".join(
# [label, opp_id, accout, create_time, today_stat_str, hist_stat_str, cleared_chat_sentence])
# fout.write(result + "\n")
log.info("finished, write feature to file : %s" %
wechat_basic_feature_data)
