def preProcessChunk(chunkID):
print 'Connecting to Mongodb..'
tableName = 'jobs_status_check'
monconn_status_check = MongoConnect(tableName,
host='localhost',
database='jam_status')
monconn_status_check_cur = monconn_status_check.getCursor()
######################################
'''Fetching the Jobs from SQL'''
######################################
#Connect to SQL table and get the jobs data
host = "172.22.65.157"
user = "******"
password = "******"
database = "SumoPlus"
unix_socket = "/tmp/mysql.sock"
port = 3308
print "Loading Jobs From MySql...."
mysql_conn = MySQLConnect(database, host, user, password, unix_socket,
port)
#cmd = '''SELECT rj.jobid as Jobid,rj.jobtitle as JobTitle,rj.description as JD,la1.text_value_MAX as SalaryMax,la2.text_value_MIN as SalaryMin,le1.display as ExpMin,le2.display as ExpMax,li.industry_desc as Industry,c.AttValueCustom as keySkills,l.city_desc as location,fn.field_enu as function,fn.sub_field_enu as subfunction from recruiter_job AS rj left join lookup_annualsalary AS la1 on rj.salarymax = la1.salary_id left join lookup_annualsalary AS la2 on rj.salarymin = la2.salary_id left join lookup_experience AS le1 on rj.minexperience = le1.value left join lookup_experience AS le2 on rj.maxexperience = le2.value left join recruiter_jobattribute as c on rj.jobid = c.jobid_id left join lookup_industry AS li on rj.industry=li.industry_id left join lookup_subfunctionalarea_new163 AS fn on fn.sub_field_id = c.AttValue AND c.AttType = 12 left join lookup_city_new512 AS l on l.city_id = c.AttValue AND c.AttType = 13 WHERE rj.jobstatus in (3,5,6,9) and c.AttType in (3,12,13) and (DATEDIFF( CURDATE(),DATE(rj.publisheddate)) < 4 OR DATEDIFF( CURDATE(),DATE(rj.republisheddate)) < 4) and rj.jobid%''' + str(numChunks) + '=' + str(chunkID)
#cmd = '''SELECT rj.jobid as Jobid,rj.jobtitle as JobTitle,rj.description as JD,la1.text_value_MAX as SalaryMax,la2.text_value_MIN as SalaryMin,le1.display as ExpMin,le2.display as ExpMax,li.industry_desc as Industry,c.AttValueCustom as keySkills,l.city_desc as location,fn.field_enu as function,fn.sub_field_enu as subfunction from recruiter_job AS rj left join lookup_annualsalary AS la1 on rj.salarymax = la1.salary_id left join lookup_annualsalary AS la2 on rj.salarymin = la2.salary_id left join lookup_experience AS le1 on rj.minexperience = le1.value left join lookup_experience AS le2 on rj.maxexperience = le2.value left join recruiter_jobattribute as c on rj.jobid = c.jobid_id left join lookup_industry AS li on rj.industry=li.industry_id left join lookup_subfunctionalarea_new163 AS fn on fn.sub_field_id = c.AttValue AND c.AttType = 12 left join lookup_city_new512 AS l on l.city_id = c.AttValue AND c.AttType = 13 WHERE rj.jobstatus in (3,5,6,9) and c.AttType in (3,12,13) and (DATEDIFF( CURDATE(),DATE(rj.publisheddate)) < 4 OR DATEDIFF( CURDATE(),DATE(rj.republisheddate)) < 4)'''
#print cmd
cmd1 = '''drop table if exists SumoPlus.XY'''
cmd2 = '''create table SumoPlus.XY as
SELECT company_account_id,SUM(final_sale_price)as price,enabled,MAX(expiry_date)as expiry_date
from SumoPlus.backoffice_accountsales a1
where enabled in
(select min(enabled) from SumoPlus.backoffice_accountsales where a1.company_account_id=company_account_id)
group by 1
'''
cmd3 = '''ALTER TABLE SumoPlus.XY add index company_account_id (company_account_id)'''
cmd4 = '''SELECT
rj.jobid as Jobid,
rj.jobtitle as JobTitle,
rj.description as JD,
rj.isbocreated as back_office_job,
rj.publisheddate as publisheddate,
rj.republisheddate as republisheddate,
rj.companyid_id as Company_id,
rj.displayname as Company_name,
la1.text_value_MAX as SalaryMax,
la2.text_value_MIN as SalaryMin,
le1.display as ExpMin,
le2.display as ExpMax,
li.industry_desc as Industry,
group_concat(c.AttValueCustom,'') as keySkills,
group_concat(fn.field_enu,'') as function,
group_concat(l.city_desc,'') as location,
group_concat(fn.sub_field_enu,'') as subfunction,
case account_type
when 0 THEN "Company"
when 1 THEN "Consultant"
when 2 THEN "Others"
when 3 THEN "Enterprise"
ELSE "Not Specified"
END AS account_type,
IF(XY.enabled = 1 AND XY.price != 0 AND XY.expiry_date > CURDATE(),'Paid','Free') AS 'flag'
from
(select * from recruiter_job
where recruiter_job.jobstatus in (3,9)
and (DATEDIFF( CURDATE(),DATE(recruiter_job.publisheddate)) < 8 OR DATEDIFF( CURDATE(),DATE(recruiter_job.republisheddate)) < 8)
) AS rj
left join lookup_annualsalary AS la1 on rj.salarymax = la1.salary_id
left join lookup_annualsalary AS la2 on rj.salarymin = la2.salary_id
left join lookup_experience AS le1 on rj.minexperience = le1.value
left join lookup_experience AS le2 on rj.maxexperience = le2.value
left join recruiter_jobattribute as c on rj.jobid = c.jobid_id
left join lookup_industry AS li on rj.industry=li.industry_id
left join lookup_subfunctionalarea_new163 AS fn on fn.sub_field_id = c.AttValue AND c.AttType = 12
left join lookup_city_new512 AS l on l.city_id = c.AttValue AND c.AttType = 13
left join SumoPlus.XY AS XY on XY.company_account_id = rj.companyid_id
left join SumoPlus.backoffice_companyaccount AS F on F.id= rj.companyid_id
WHERE
c.AttType in (3,12,13)
group by rj.jobid
'''
cmd5 = '''drop table if exists SumoPlus.XY
'''
print 'chnukID:', chunkID, ': Loading jobs from SQL....', time.ctime()
mysql_conn.query(cmd1)
mysql_conn.query(cmd2)
mysql_conn.query(cmd3)
jobs = mysql_conn.query(cmd4)
mysql_conn.query(cmd5)
print 'chunkID:', chunkID, ': Loading jobs from SQL....completed..', time.ctime(
)
print 'chunkid:', chunkID, ' : Number of jobs loaded: ', len(jobs)
######################################
'''Connecting to Mongo 233 Server'''
######################################
print 'Connecting to Mongodb..'
tableName = 'jobs_processed'
monconn_jobs_local = MongoConnect(tableName,
host='localhost',
database='JobAlerts')
monconn_jobs_local_cur = monconn_jobs_local.getCursor()
print 'Connecting to Mongodb...finished'
######################################
'''Processing the Jobs'''
######################################
i = 0
for job in jobs:
#pprint(job)
#print i
if i % 1000 == 0:
print '\tchunkID:', chunkID, ' numRecords:', i, ' completed in ', time.time(
) - start_time, ' seconds'
job_id = job['Jobid']
job_title = cleanToken(job['JobTitle'])
job_maxexp = cleanToken(job['ExpMax'])
job_minexp = cleanToken(job['ExpMin'])
job_maxsal = cleanToken(job['SalaryMax'])
job_minsal = cleanToken(job['SalaryMin'])
job_jd = cleanHTML(cleanToken(job['JD']))
job_industry = cleanToken(job['Industry'])
job_location = removeDup(job['location'])
job_subfunction = removeDup(cleanToken(job['subfunction']))
job_function = removeDup(cleanToken(job['function']))
job_skills = removeDup(cleanToken(job['keySkills']))
job_flag = job['flag']
job_accounttype = job['account_type']
job_company_id = job['Company_id']
job_company_name = cleanToken(job['Company_name'])
job_published_date = job['publisheddate']
job_republished_date = job['republisheddate']
job_back_office = int(job['back_office_job'])
if job_company_id == 421880: ################## Altimetrik Jobs removed ##########################
continue
job_location = job_location.replace(', ', ',').lower().split(',')
##Extract additional fields like bow
text = 5 * (" " + job_title) + ' ' + 5 * (
" " + job_skills) + ' ' + 1 * (" " + job_jd) + ' ' + 2 * (
" " + job_industry) + ' ' + 2 * (
" " + job_function) + ' ' + 2 * (" " + job_subfunction)
text = text.replace('candidates', ' ')
'''
try:
text = 5*(" "+job_title) + ' ' + 3*(" "+job_skills) + ' ' + 1*(" "+job_jd) +' '+2*(" "+job_industry)+' '+2*(" "+job_function)+' '+2*(" "+job_subfunction)
text = text.replace('candidates', ' ')
except:
text = 5*(" "+job_title) + ' ' + 3*(" "+job_skills) + ' ' + 1*(" "+job_jd)
text = text.replace('candidates', ' ')
'''
job_bow = mb.getBow(text, getbowdict=0)
#job_keySkills = ','.join([x for x in jobKeySkills.split(',') if x.strip() != ''])
#pprint(job_bow)
document = {'job_id': job_id, 'job_title': job_title,'job_function':job_function, \
'job_maxexp': job_maxexp, 'job_minexp': job_minexp,\
'job_location':job_location, 'job_subfunction':job_subfunction,\
'job_maxsal':job_maxsal,'job_minsal':job_minsal, 'job_skills': job_skills, \
'job_bow': job_bow, 'job_industry': job_industry, 'job_jd': job_jd, \
'job_flag':job_flag,'job_accounttype':job_accounttype, \
'job_company_id':job_company_id,'job_company_name':job_company_name,
'job_published':job_published_date,'job_republished':job_republished_date,'job_back_office':job_back_office
}
monconn_jobs_local.saveToTable(document)
i += 1
print "Processing finished....."
print 'chunkID:', chunkID, ' Total time taken is: ', time.time(
) - start_time, ' seconds.'
end_time = time.time()
time_taken = end_time - start_time
os.system(
' echo "Jobs Processed ' + str(i) + ' in :' +
str(end_time - start_time) + ' seconds' +
' " | mutt -s "Job Alert Mailer " [email protected] ,[email protected]'
)
del (monconn_jobs_local)
del (mysql_conn)
monconn_status_check.saveToTable({'_id': 1, 'status': 1})
del (monconn_status_check)
try:
datadict["user_edom"] = (doc['sEm']).split("@")[1]
except:
datadict["user_edom"] = None
try:
datadict["user_skills"] = ', '.join(doc['sS'])
except:
datadict["user_skills"] = None
#########################################################################################################
############----------------- Creating the Bag of Words of Candidate's Details
#########################################################################################################
text = 5 * (" " + cleanToken(
datadict["user_profiletitle"])) + ' ' + 5 * (
" " + cleanToken(datadict["user_skills"])
) + ' ' + 2 * (" " + cleanToken(
datadict["user_jobtitle"])) + ' ' + 1 * (
" " + cleanToken(datadict["user_industry"])
) + ' ' + 1 * (" " + cleanToken(
datadict["user_functionalarea"]))
datadict["user_bow"] = mb.getBow(text, getbowdict=0)
except Exception as e:
print e
continue
datadict["p"] = pidcounter
datadict["s"] = False
#user_id_list.append(doc['id'])
insert.append(datadict)
def preProcessChunk(chunkID):
#########################################################################################################
############----------------- SQL Credentials
#########################################################################################################
'''
host="172.22.65.157"
user="******"
password="******"
database="SumoPlus"
unix_socket="/tmp/mysql.sock"
port = 3308
'''
host = "172.22.66.204"
user = "******"
password = "******"
database = "SumoPlus"
unix_socket = "/tmp/mysql.sock"
port = 3306
#########################################################################################################
############----------------- Creating the SQL Query
#########################################################################################################
print "Loading Jobs From MySql...."
mysql_conn = MySQLConnect(database, host, user, password, unix_socket,
port)
cmd1 = '''drop table if exists SumoPlus.XY'''
cmd2 = '''create table SumoPlus.XY as
SELECT company_account_id,SUM(final_sale_price)as price,enabled,MAX(expiry_date)as expiry_date
from SumoPlus.backoffice_accountsales a1
where enabled in
(select min(enabled) from SumoPlus.backoffice_accountsales where a1.company_account_id=company_account_id)
group by 1
'''
cmd3 = '''ALTER TABLE SumoPlus.XY add index company_account_id (company_account_id)'''
cmd4 = '''SELECT
rj.jobid as Jobid,
rj.jobtitle as JobTitle,
rj.description as JD,
rj.isbocreated as back_office_job,
rj.publisheddate as publisheddate,
rj.republisheddate as republisheddate,
rj.companyid_id as Company_id,
rj.displayname as Company_name,
la1.text_value_MAX as SalaryMax,
la2.text_value_MIN as SalaryMin,
le1.display as ExpMin,
le2.display as ExpMax,
li.industry_desc as Industry,
group_concat(c.AttValueCustom,'') as keySkills,
group_concat(fn.field_enu,'') as function,
group_concat(l.city_desc,'') as location,
group_concat(fn.sub_field_enu,'') as subfunction,
case account_type
when 0 THEN "Company"
when 1 THEN "Consultant"
when 2 THEN "Others"
when 3 THEN "Enterprise"
ELSE "Not Specified"
END AS account_type,
IF(XY.enabled = 1 AND XY.price != 0 AND XY.expiry_date > CURDATE(),'Paid','Free') AS 'flag'
from
(select * from recruiter_job
where recruiter_job.jobstatus in (3,9)
and (DATEDIFF( CURDATE(),DATE(recruiter_job.publisheddate)) < 16 OR DATEDIFF( CURDATE(),DATE(recruiter_job.republisheddate)) < 16)
) AS rj
left join lookup_annualsalary AS la1 on rj.salarymax = la1.salary_id
left join lookup_annualsalary AS la2 on rj.salarymin = la2.salary_id
left join lookup_experience AS le1 on rj.minexperience = le1.value
left join lookup_experience AS le2 on rj.maxexperience = le2.value
left join recruiter_jobattribute as c on rj.jobid = c.jobid_id
left join lookup_industry AS li on rj.industry=li.industry_id
left join lookup_subfunctionalarea_new163 AS fn on fn.sub_field_id = c.AttValue AND c.AttType = 12
left join lookup_city_new512 AS l on l.city_id = c.AttValue AND c.AttType = 13
left join SumoPlus.XY AS XY on XY.company_account_id = rj.companyid_id
left join SumoPlus.backoffice_companyaccount AS F on F.id= rj.companyid_id
WHERE
c.AttType in (3,12,13)
group by rj.jobid
'''
cmd5 = '''drop table if exists SumoPlus.XY
'''
#########################################################################################################
############----------------- Executing the SQL Query
#########################################################################################################
print 'chnukID:', chunkID, ': Loading jobs from SQL....', time.ctime()
mysql_conn.query(cmd1)
mysql_conn.query(cmd2)
mysql_conn.query(cmd3)
jobs = mysql_conn.query(cmd4)
mysql_conn.query(cmd5)
print 'chunkID:', chunkID, ': Loading jobs from SQL....completed..', time.ctime(
)
print 'chunkid:', chunkID, ' : Number of jobs loaded: ', len(jobs)
#########################################################################################################
############-----------------Connecting to Jobs Collections Mongo (172.22.66.233)
#########################################################################################################
print 'Connecting to Mongodb..'
tableName = 'jobs_processed'
monconn_jobs_local = MongoConnect(tableName,
host='localhost',
database='mailer_weekly')
monconn_jobs_local_cur = monconn_jobs_local.getCursor()
print 'Connecting to Mongodb...finished'
#########################################################################################################
############-----------------Processing the Jobs data extracted from SQL
#########################################################################################################
i = 0
for job in jobs:
if i % 1000 == 0:
print '\tchunkID:', chunkID, ' numRecords:', i, ' completed in ', time.time(
) - start_time, ' seconds'
job_id = job['Jobid']
job_title = cleanToken(job['JobTitle'])
job_maxexp = cleanToken(job['ExpMax'])
job_minexp = cleanToken(job['ExpMin'])
job_maxsal = cleanToken(job['SalaryMax'])
job_minsal = cleanToken(job['SalaryMin'])
job_jd = cleanHTML(cleanToken(job['JD']))
job_industry = cleanToken(job['Industry'])
job_location = removeDup(job['location'])
job_subfunction = removeDup(cleanToken(job['subfunction']))
job_function = removeDup(cleanToken(job['function']))
job_skills = removeDup(cleanToken(job['keySkills']))
job_flag = job['flag']
job_accounttype = job['account_type']
job_company_id = job['Company_id']
job_company_name = cleanToken(job['Company_name'])
job_published_date = job['publisheddate']
job_republished_date = job['republisheddate']
job_back_office = int(job['back_office_job'])
job_location = job_location.replace(', ', ',').lower().split(',')
if job_company_id == 421880: #######---------- Altimetrik Jobs removed
continue
#########################################################################################################
############-----------------Creating Bag of Words for Text
#########################################################################################################
text = 5 * (" " + job_title) + ' ' + 5 * (
" " + job_skills) + ' ' + 1 * (" " + job_jd) + ' ' + 2 * (
" " + job_industry) + ' ' + 2 * (
" " + job_function) + ' ' + 2 * (" " + job_subfunction)
text = text.replace('candidates', ' ')
job_bow = mb.getBow(text, getbowdict=0)
#########################################################################################################
############-----------------Creating Job document to be saved in Mongo
#########################################################################################################
document = {'job_id': job_id, 'job_title': job_title,'job_function':job_function, \
'job_maxexp': job_maxexp, 'job_minexp': job_minexp,\
'job_location':job_location, 'job_subfunction':job_subfunction,\
'job_maxsal':job_maxsal,'job_minsal':job_minsal, 'job_skills': job_skills, \
'job_bow': job_bow, 'job_industry': job_industry, 'job_jd': job_jd, \
'job_flag':job_flag,'job_accounttype':job_accounttype, \
'job_company_id':job_company_id,'job_company_name':job_company_name,
'job_published':job_published_date,'job_republished':job_republished_date,'job_back_office':job_back_office
}
#########################################################################################################
############-----------------Saving the document in Job collection Mongo (172.22.66.233)
#########################################################################################################
monconn_jobs_local.saveToTable(document)
i += 1
print "Processing finished....."
print 'chunkID:', chunkID, ' Total time taken is: ', time.time(
) - start_time, ' seconds.'
end_time = time.time()
time_taken = end_time - start_time
send_email([
'*****@*****.**',
'*****@*****.**'
], "Revival Mailer Weekly", 'Jobs Processed ' + str(i) + ' in :' +
str(end_time - start_time) + ' seconds')
#########################################################################################################
############-----------------Changing the status of completion and deleting the mongo connections
#########################################################################################################
del (monconn_jobs_local)
del (mysql_conn)
try:
#count=count+1
querycount = querycount + 1
try:
datadict["_id"] = (doc['id'])
datadict["user_ctc"] = (doc['sAS'])
datadict["user_experience"] = (doc['sEx'])
datadict["user_id"] = unicode(doc['id'])
datadict["user_email"] = unicode(doc['sEm'])
datadict["user_location"] = (doc['sCL'])
datadict["user_lastlogin"] = unicode(doc['sLH'])
try:
datadict["user_functionalarea"] = cleanToken(doc['sCFA'])
except:
datadict["user_functionalarea"] = ''
try:
datadict["user_industry"] = cleanToken(doc['sCIT'])
except:
datadict["user_industry"] = ''
try:
datadict["user_skills"] = ', '.join(doc['sS'])
except:
datadict["user_skills"] = ''
try:
datadict["user_jobtitle"] = cleanToken(doc['sCTT'])
except:
user_ctc_list.append(user_ctc_dict[row.get(
'v', '')])
except:
user_ctc = ""
if row.get('t') == 10:
try:
user_functionalarea_list.append(
user_functionalarea_dict[row.get('v', '')])
except:
user_functionalarea = " "
user_skills = ""
try:
user_functionalarea = cleanToken(
','.join(user_functionalarea_list))
except:
user_functionalarea = ""
try:
user_ctc = ','.join(user_ctc_list)
except:
user_ctc = ""
try:
user_industry = cleanToken(','.join(user_industry_list))
except:
user_industry = ""
try:
user_location = ','.join(user_location_list)
except:
user_location = ""
skills_list = user.get('_skills', [])
skills_extracted = []
for skill_dict in skills_list:
skills_extracted.append(skill_dict['nm'])
skills = ','.join(skills_extracted)
jobs_list = user.get('_jobs', [])
if len(jobs_list) != 0:
user_job = jobs_list[0].get('jt', '')
else:
user_job = ''
document = {
'_id': user['cid'],
'user_creation_date': user['cd'],
'user_jobtitle_resume': cleanToken(user_job),
'user_skills_resume': cleanToken(skills)
}
monconn_resume.saveToTable(document)
send_email(
[
'*****@*****.**',
'*****@*****.**'
], "Resume Dump Incremental - JAM",
'Resume Details Dump Incremental Completed!! \n Collection Name : candidate_data \n DB Name : ResumeDump'
)
del (monconn_users)
del (monconn_resume)
except:
send_email(
def preProcessChunk(chunkId1, chunkId2):
######################################
'''Fetching the Jobs from SQL'''
######################################
#host="172.22.65.157"
host = "172.22.66.204"
user = "******"
password = "******"
database = "SumoPlus"
unix_socket = "/tmp/mysql.sock"
port = 3306
print "Loading Jobs From MySql...."
mysql_conn = MySQLConnect(database, host, user, password, unix_socket,
port)
#cmd = '''SELECT rj.jobid as Jobid,rj.jobtitle as JobTitle,rj.description as JD,la1.text_value_MAX as SalaryMax,la2.text_value_MIN as SalaryMin,le1.display as ExpMin,le2.display as ExpMax,li.industry_desc as Industry,c.AttValueCustom as keySkills,l.city_desc as location,fn.field_enu as function,fn.sub_field_enu as subfunction from recruiter_job AS rj left join lookup_annualsalary AS la1 on rj.salarymax = la1.salary_id left join lookup_annualsalary AS la2 on rj.salarymin = la2.salary_id left join lookup_experience AS le1 on rj.minexperience = le1.value left join lookup_experience AS le2 on rj.maxexperience = le2.value left join recruiter_jobattribute as c on rj.jobid = c.jobid_id left join lookup_industry AS li on rj.industry=li.industry_id left join lookup_subfunctionalarea_new163 AS fn on fn.sub_field_id = c.AttValue AND c.AttType = 12 left join lookup_city_new512 AS l on l.city_id = c.AttValue AND c.AttType = 13 WHERE rj.jobstatus in (3,5,6,9) and c.AttType in (3,12,13) and (DATEDIFF( CURDATE(),DATE(rj.publisheddate)) < 4 OR DATEDIFF( CURDATE(),DATE(rj.republisheddate)) < 4) and rj.jobid%''' + str(numChunks) + '=' + str(chunkID)
#cmd = '''SELECT rj.jobid as Jobid,rj.jobtitle as JobTitle,rj.description as JD,la1.text_value_MAX as SalaryMax,la2.text_value_MIN as SalaryMin,le1.display as ExpMin,le2.display as ExpMax,li.industry_desc as Industry,c.AttValueCustom as keySkills,l.city_desc as location,fn.field_enu as function,fn.sub_field_enu as subfunction from recruiter_job AS rj left join lookup_annualsalary AS la1 on rj.salarymax = la1.salary_id left join lookup_annualsalary AS la2 on rj.salarymin = la2.salary_id left join lookup_experience AS le1 on rj.minexperience = le1.value left join lookup_experience AS le2 on rj.maxexperience = le2.value left join recruiter_jobattribute as c on rj.jobid = c.jobid_id left join lookup_industry AS li on rj.industry=li.industry_id left join lookup_subfunctionalarea_new163 AS fn on fn.sub_field_id = c.AttValue AND c.AttType = 12 left join lookup_city_new512 AS l on l.city_id = c.AttValue AND c.AttType = 13 WHERE rj.jobstatus in (3,5,6,9) and c.AttType in (3,12,13) and (DATEDIFF( CURDATE(),DATE(rj.publisheddate)) < 4 OR DATEDIFF( CURDATE(),DATE(rj.republisheddate)) < 4)'''
#print cmd
cmd1 = '''drop table if exists SumoPlus.XY'''
cmd2 = '''create table SumoPlus.XY as
SELECT company_account_id,SUM(final_sale_price)as price,enabled,MAX(expiry_date)as expiry_date
from SumoPlus.backoffice_accountsales a1
where enabled in
(select min(enabled) from SumoPlus.backoffice_accountsales where a1.company_account_id=company_account_id)
group by 1
'''
cmd3 = '''ALTER TABLE SumoPlus.XY add index company_account_id (company_account_id)'''
cmd4 = '''SELECT
rj.jobid as Jobid,
rj.jobtitle as JobTitle,
rj.description as JD,
rj.companyid_id as Company_id,
rj.displayname as Company_name,
rj.publisheddate as Published_Date,
rj.republisheddate as RePublished_Date,
rj.expirydate as Expiry_Date,
la1.text_value_MAX as SalaryMax,
la2.text_value_MIN as SalaryMin,
le1.display as ExpMin,
le2.display as ExpMax,
li.industry_desc as Industry,
group_concat(c.AttValueCustom,'') as keySkills,
group_concat(fn.field_enu,'') as function,
group_concat(l.city_desc,'') as location,
group_concat(fn.sub_field_enu,'') as subfunction,
lj.Applications as Application_Number,
case account_type
when 0 THEN "Company"
when 1 THEN "Consultant"
when 2 THEN "Others"
when 3 THEN "Enterprise"
ELSE "Not Specified"
END AS account_type,
IF(XY.enabled = 1 AND XY.price != 0 AND XY.expiry_date > CURDATE(),'Paid','Free') AS 'flag'
from
(select * from recruiter_job
where ( (DATEDIFF( CURDATE(),DATE(recruiter_job.publisheddate)) > %s AND DATEDIFF( CURDATE(),DATE(recruiter_job.publisheddate)) <= %s) OR (DATEDIFF( CURDATE(),DATE(recruiter_job.republisheddate)) > %s AND DATEDIFF( CURDATE(),DATE(recruiter_job.republisheddate)) <= %s))) AS rj
left join lookup_annualsalary AS la1 on rj.salarymax = la1.salary_id
left join lookup_annualsalary AS la2 on rj.salarymin = la2.salary_id
left join lookup_experience AS le1 on rj.minexperience = le1.value
left join lookup_experience AS le2 on rj.maxexperience = le2.value
left join recruiter_jobattribute as c on rj.jobid = c.jobid_id
left join lookup_industry AS li on rj.industry=li.industry_id
left join lookup_subfunctionalarea_new163 AS fn on fn.sub_field_id = c.AttValue AND c.AttType = 12
left join lookup_city_new512 AS l on l.city_id = c.AttValue AND c.AttType = 13
left join SumoPlus.XY AS XY on XY.company_account_id = rj.companyid_id
left join SumoPlus.backoffice_companyaccount AS F on F.id= rj.companyid_id
left join ShineReport.LiveJobsApplications AS lj on rj.jobid = lj.JobId
WHERE
c.AttType in (3,12,13)
group by rj.jobid
''' % (chunkId1, chunkId2, chunkId1, chunkId2)
cmd5 = '''drop table if exists SumoPlus.XY
'''
print 'chnukID:', chunkId1, ': Loading jobs from SQL....', time.ctime()
mysql_conn.query(cmd1)
print 'cmd1'
mysql_conn.query(cmd2)
print 'cmd2'
mysql_conn.query(cmd3)
print 'cmd3'
jobs = mysql_conn.query(cmd4)
print 'jobs'
mysql_conn.query(cmd5)
print 'chunkID:', chunkId1, ': Loading jobs from SQL....completed..', time.ctime(
)
print 'chunkid:', chunkId1, ' : Number of jobs loaded: ', len(jobs)
######################################
'''Connecting to Mongo 233 Server'''
######################################
print 'Connecting to Mongodb..'
tableName = 'jobs_processed_9months'
monconn_jobs_local = MongoConnect(tableName,
host='172.22.66.198',
database='SimilarJobs')
monconn_jobs_local_cur = monconn_jobs_local.getCursor()
print 'Connecting to Mongodb...finished'
######################################
'''Processing the Jobs'''
######################################
global i
#i = 0
for job in jobs:
#pprint(job)
#print i
if i % 1000 == 0:
print '\tchunkID:', chunkId1, ' numRecords:', i, ' completed in ', time.time(
) - start_time, ' seconds'
job_id = job['Jobid']
job_title = cleanToken(job['JobTitle'])
job_maxexp = cleanToken(job['ExpMax'])
job_minexp = cleanToken(job['ExpMin'])
job_maxsal = cleanToken(job['SalaryMax'])
job_minsal = cleanToken(job['SalaryMin'])
job_jd = cleanHTML(cleanToken(job['JD']))
job_industry = cleanToken(job['Industry'])
job_location = removeDup(job['location'])
job_subfunction = removeDup(cleanToken(job['subfunction']))
job_function = removeDup(cleanToken(job['function']))
job_skills = removeDup(cleanToken(job['keySkills']))
job_flag = job['flag']
job_accounttype = job['account_type']
job_company_id = job['Company_id']
job_company_name = cleanToken(job['Company_name'])
job_index = i
job_publishedate = job['Published_Date']
job_repubslisheddate = job['RePublished_Date']
job_expirydate = job['Expiry_Date']
pid = i % 5000
job_applications = job['Application_Number']
job_location = job_location.replace(', ', ',').lower().split(',')
#################################################
'''Creating Bag of Words from the text fields'''
#################################################
text = 5 * (" " + job_title) + ' ' + 3 * (
" " + job_skills) + ' ' + 1 * (" " + job_jd) + ' ' + 2 * (
" " + job_industry) + ' ' + 2 * (
" " + job_function) + ' ' + 2 * (" " + job_subfunction)
text = text.replace('candidates', ' ')
job_bow = mb.getBow(text, getbowdict=0)
##################################################
'''Dumping Job Details in Mongo (172.22.66.253)'''
##################################################
document = {'job_id': job_id, 'job_title': job_title,'job_function':job_function, \
'job_maxexp': job_maxexp, 'job_minexp': job_minexp,\
'job_location':job_location, 'job_subfunction':job_subfunction,\
'job_maxsal':job_maxsal,'job_minsal':job_minsal, 'job_skills': job_skills, \
'job_bow': job_bow, 'job_industry': job_industry, 'job_jd': job_jd, \
'job_flag':job_flag,'job_accounttype':job_accounttype, \
'job_company_id':job_company_id,'job_company_name':job_company_name,'job_index':job_index, \
'application_number': job_applications,'pid':pid,'job_publishedate':job_publishedate , \
'job_repubslisheddate':job_repubslisheddate,'job_expirydate':job_expirydate
}
monconn_jobs_local.saveToTable(document)
i += 1
print "Processing finished....."
print 'chunkID:', chunkId1, ' Total time taken is: ', time.time(
) - start_time, ' seconds.'
end_time = time.time()
time_taken = end_time - start_time
monconn_jobs_local.doIndexing('pid')
#send_email(['*****@*****.**', '*****@*****.**','*****@*****.**'],"Similar Jobs Mailer 9 Month Jobs",'Jobs Processing 9 Months Completed !!\nJobs Processed '+str(i)+' in :' + str(end_time - start_time) + ' seconds')
#os.system(' echo "Jobs Processing 9 Months Completed !!\nJobs Processed '+str(i)+' in :' + str(end_time - start_time) + ' seconds' +' " | mutt -s "Similar Jobs Mailer" [email protected], [email protected], [email protected]')
del (monconn_jobs_local)
del (mysql_conn)
def preProcessChunk(chunkID):
#########################################################################################################
############----------------- SQL Credentials
#########################################################################################################
#Connect to SQL table and get the jobs data
#host="172.16.66.64"
#user="******"
#password="******"
'''
host="172.22.65.157"
user="******"
password="******"
database="SumoPlus"
unix_socket="/tmp/mysql.sock"
port = 3308
'''
host="172.22.66.204"
user="******"
password="******"
database="SumoPlus"
unix_socket="/tmp/mysql.sock"
port = 3306
#########################################################################################################
############----------------- Creating the SQL Query
#########################################################################################################
print "Loading Jobs From MySql...."
mysql_conn = MySQLConnect(database, host, user, password, unix_socket, port)
#cmd = '''SELECT rj.jobid as Jobid,rj.jobtitle as JobTitle,rj.description as JD,la1.text_value_MAX as SalaryMax,la2.text_value_MIN as SalaryMin,le1.display as ExpMin,le2.display as ExpMax,li.industry_desc as Industry,c.AttValueCustom as keySkills,l.city_desc as location,fn.field_enu as function,fn.sub_field_enu as subfunction from recruiter_job AS rj left join lookup_annualsalary AS la1 on rj.salarymax = la1.salary_id left join lookup_annualsalary AS la2 on rj.salarymin = la2.salary_id left join lookup_experience AS le1 on rj.minexperience = le1.value left join lookup_experience AS le2 on rj.maxexperience = le2.value left join recruiter_jobattribute as c on rj.jobid = c.jobid_id left join lookup_industry AS li on rj.industry=li.industry_id left join lookup_subfunctionalarea_new163 AS fn on fn.sub_field_id = c.AttValue AND c.AttType = 12 left join lookup_city_new512 AS l on l.city_id = c.AttValue AND c.AttType = 13 WHERE rj.jobstatus in (3,5,6,9) and c.AttType in (3,12,13) and (DATEDIFF( CURDATE(),DATE(rj.publisheddate)) < 4 OR DATEDIFF( CURDATE(),DATE(rj.republisheddate)) < 4) and rj.jobid%''' + str(numChunks) + '=' + str(chunkID)
#cmd = '''SELECT rj.jobid as Jobid,rj.jobtitle as JobTitle,rj.description as JD,la1.text_value_MAX as SalaryMax,la2.text_value_MIN as SalaryMin,le1.display as ExpMin,le2.display as ExpMax,li.industry_desc as Industry,c.AttValueCustom as keySkills,l.city_desc as location,fn.field_enu as function,fn.sub_field_enu as subfunction from recruiter_job AS rj left join lookup_annualsalary AS la1 on rj.salarymax = la1.salary_id left join lookup_annualsalary AS la2 on rj.salarymin = la2.salary_id left join lookup_experience AS le1 on rj.minexperience = le1.value left join lookup_experience AS le2 on rj.maxexperience = le2.value left join recruiter_jobattribute as c on rj.jobid = c.jobid_id left join lookup_industry AS li on rj.industry=li.industry_id left join lookup_subfunctionalarea_new163 AS fn on fn.sub_field_id = c.AttValue AND c.AttType = 12 left join lookup_city_new512 AS l on l.city_id = c.AttValue AND c.AttType = 13 WHERE rj.jobstatus in (3,5,6,9) and c.AttType in (3,12,13) and (DATEDIFF( CURDATE(),DATE(rj.publisheddate)) < 4 OR DATEDIFF( CURDATE(),DATE(rj.republisheddate)) < 4)'''
#print cmd
cmd='''SELECT
rj.jobid as Jobid,
rj.jobtitle as JobTitle,
rj.description as JD,
la1.text_value_MAX as SalaryMax,
la2.text_value_MIN as SalaryMin,
le1.display as ExpMin,
le2.display as ExpMax,
li.industry_desc as Industry,
group_concat(c.AttValueCustom,'') as keySkills,
group_concat(fn.field_enu,'') as function,
group_concat(l.city_desc,'') as location,
group_concat(fn.sub_field_enu,'') as subfunction
from
(select * from recruiter_job
where recruiter_job.jobstatus in (3,9)
and (DATEDIFF( CURDATE(),DATE(recruiter_job.publisheddate)) < 8 OR DATEDIFF( CURDATE(),DATE(recruiter_job.republisheddate)) < 8)
) AS rj
left join lookup_annualsalary AS la1 on rj.salarymax = la1.salary_id
left join lookup_annualsalary AS la2 on rj.salarymin = la2.salary_id
left join lookup_experience AS le1 on rj.minexperience = le1.value
left join lookup_experience AS le2 on rj.maxexperience = le2.value
left join recruiter_jobattribute as c on rj.jobid = c.jobid_id
left join lookup_industry AS li on rj.industry=li.industry_id
left join lookup_subfunctionalarea_new163 AS fn on fn.sub_field_id = c.AttValue AND c.AttType = 12
left join lookup_city_new512 AS l on l.city_id = c.AttValue AND c.AttType = 13
WHERE
c.AttType in (3,12,13)
group by rj.jobid
'''
#########################################################################################################
############----------------- Executing the SQL Query
#########################################################################################################
print 'chnukID:', chunkID, ': Loading jobs from SQL....', time.ctime()
jobs = mysql_conn.query(cmd)
print 'chunkID:', chunkID,': Loading jobs from SQL....completed..', time.ctime()
print 'chunkid:', chunkID, ' : Number of jobs loaded: ', len(jobs)
#########################################################################################################
############-----------------Connecting to Jobs Collections Mongo (172.22.66.233)
#########################################################################################################
print 'Connecting to Mongodb..'
tableName = 'jobs_processed_midout'
monconn_jobs_local = MongoConnect(tableName , host = 'localhost', database = 'Midout_Mailers')
monconn_jobs_local_cur = monconn_jobs_local.getCursor()
print 'Connecting to Mongodb...finished'
#########################################################################################################
############-----------------Processing the Jobs data extracted from SQL
#########################################################################################################
i = 0
for job in jobs:
#pprint(job)
#print i
if i%1000 == 0:
print '\tchunkID:', chunkID, ' numRecords:' , i, ' completed in ', time.time() - start_time, ' seconds'
job_id = job['Jobid']
job_title = cleanToken(job['JobTitle'])
job_maxexp = cleanToken(job['ExpMax'])
job_minexp = cleanToken(job['ExpMin'])
job_maxsal = cleanToken(job['SalaryMax'])
job_minsal = cleanToken(job['SalaryMin'])
job_jd = cleanHTML(cleanToken(job['JD']) )
job_industry = cleanToken(job['Industry'])
job_location=removeDup(job['location'])
job_subfunction=removeDup(job['subfunction'])
job_function=removeDup(job['function'])
job_skills=removeDup(cleanToken(job['keySkills']))
#########################################################################################################
############-----------------Creating Bag of Words for Text
#########################################################################################################
text = 5*(" "+job_title) + ' ' + 5*(" "+job_skills) + ' ' + 1*(" "+job_jd) +' '+2*(" "+job_industry)+' '+2*(" "+job_function)+' '+2*(" "+job_subfunction)
text = text.replace('candidates', ' ')
job_bow = mb.getBow(text, getbowdict = 0)
#########################################################################################################
############-----------------Creating Job document to be saved in Mongo
#########################################################################################################
document = {'job_id': job_id, 'job_title': job_title,'job_function':job_function, \
'job_maxexp': job_maxexp, 'job_minexp': job_minexp,\
'job_location':job_location, 'job_subfunction':job_subfunction,\
'job_maxsal':job_maxsal,'job_minsal':job_minsal, 'job_skills': job_skills, \
'job_bow': job_bow, 'job_industry': job_industry, 'job_jd': job_jd
}
#########################################################################################################
############-----------------Saving the document in Job collection Mongo (172.22.66.233)
#########################################################################################################
monconn_jobs_local.saveToTable(document)
i += 1
print "Processing finished....."
print 'chunkID:', chunkID, ' Total time taken is: ', time.time() - start_time, ' seconds.'
end_time = time.time()
time_taken = end_time - start_time
send_email(['*****@*****.**', '*****@*****.**'],"Midout Mailers",'Jobs Processed '+str(i)+' in :' + str(end_time - start_time) + ' seconds')
#os.system(' echo "Jobs Processed '+str(i)+' in :' + str(end_time - start_time) + ' seconds' +' " | mutt -s "Midout Mailers" [email protected] ,[email protected]')
del(monconn_jobs_local)
del(mysql_conn)
