def file2db(file, zhinengleibie, province):
conn = get_conn()
filename = path.split(file)[-1]
job_id = filename.replace(".html", "")
exists = conn.execute(
f"select count(1) from jobs where job_id='{job_id}' and year_month={year_month}"
).fetchall()[0][0]
if exists:
try_rename(file)
return
#print(file)
job = file2job(file, zhinengleibie, province)
if not job:
try_rename(file)
return
data = pd.DataFrame(columns=get_featurenames(job))
l = object2list(job)
data.loc[job.job_id] = l
data.to_sql("jobs", conn, if_exists="append", index=False)
conn.close()
try_rename(file)
def file2db(file, city):
conn = get_conn()
try:
filename = path.split(file)[-1]
job_id = filename.replace(".html", "")
exists = conn.execute(
"select count(1) from _51jobs where job_id='{0}'".format(
job_id)).fetchall()[0][0]
if exists:
os.rename(file, file.replace("51jobs", "51jobs_back"))
return
print(file)
job = file2job(file, city)
if not job:
return
# if not job.check_all():
# return
data = pd.DataFrame(columns=get_featurenames(job))
l = object2list(job)
data.loc[job.job_id] = l
data.to_sql("_51jobs", conn, if_exists="append", index=False)
except Exception:
pass
finally:
conn.close()