def test_crawler(self):
"""
Test baidu crawler
:return:
"""
from core.crawler.crawl import kwquery
from core.crawler.crawl import jieba_initialize
jieba_initialize()
query = "回锅肉属于什么菜系"
query = "北京奥运会是什么时候"
ans = kwquery(query)
print("~~~~~~~")
for a in ans:
print(a)
print("~~~~~~~")
def test_crawler(self):
"""
Test baidu crawler
:return:
"""
from core.crawler.crawl import kwquery
from core.crawler.crawl import jieba_initialize
jieba_initialize()
query = "回锅肉属于什么菜系"
query = "北京奥运会是什么时候"
ans = kwquery(query)
print("~~~~~~~")
for a in ans:
print(a)
print("~~~~~~~")
from core.crawler.crawl import jieba_initialize, crawler_daemon
from core.crawler.pmi import baidu_count_daemon
from core.ocr.baiduocr import get_text_from_image as bai_get_text
from core.ocr.spaceocr import get_text_from_image as ocrspace_get_text
from utils import stdout_template
from utils.backup import save_question_answers_to_file, get_qa_list, upload_to_cloud
from utils.process_stdout import ProcessStdout
logger = logging.getLogger("assistant")
handler = logging.handlers.WatchedFileHandler("assistant.log")
formatter = logging.Formatter("%(asctime)s - %(levelname)s - %(message)s")
handler.setFormatter(formatter)
logger.addHandler(handler)
## jieba init
jieba_initialize()
if prefer[0] == "baidu":
get_text_from_image = partial(bai_get_text,
app_id=app_id,
app_key=app_key,
app_secret=app_secret,
api_version=api_version,
timeout=5)
elif prefer[0] == "ocrspace":
get_text_from_image = partial(ocrspace_get_text, api_key=api_key)
def parse_args():
parser = ArgumentParser(description="Million Hero Assistant")
from core.crawler.crawl import jieba_initialize, crawler_daemon
from core.crawler.pmi import baidu_count_daemon
from core.ocr.baiduocr import get_text_from_image as bai_get_text
from core.ocr.spaceocr import get_text_from_image as ocrspace_get_text
from utils import stdout_template
from utils.backup import save_question_answers_to_file, get_qa_list, upload_to_cloud
from utils.process_stdout import ProcessStdout
logger = logging.getLogger("assistant")
handler = logging.handlers.WatchedFileHandler("assistant.log")
formatter = logging.Formatter("%(asctime)s - %(levelname)s - %(message)s")
handler.setFormatter(formatter)
logger.addHandler(handler)
## jieba init
jieba_initialize()
if prefer[0] == "baidu":
get_text_from_image = partial(bai_get_text,
app_id=app_id,
app_key=app_key,
app_secret=app_secret,
api_version=api_version,
timeout=5)
elif prefer[0] == "ocrspace":
get_text_from_image = partial(ocrspace_get_text, api_key=api_key)
def parse_args():
parser = ArgumentParser(description="Million Hero Assistant")