def getInsightViews(start_date, end_date):
logger.info(" extral data from Dashboard.matchedVideoViewCountCompletion start")
date_dict ={"start_date": start_date, "end_date": end_date}
try:
get_data_sql = """
select matchedVideo_id, trackingWebsite_id, trackingMeta_id,
company_id, report_at, view_count, current_timestamp as ETLDate
from matchedVideoViewCountCompletion
where report_at > "%(start_date)s" and report_at <= "%(end_date)s"
""" %date_dict
insight_mysql = MySQLHelper(host=insight_host, user=insight_user,passwd=insight_passwd, port = insight_port, db_name = insight_db)
data = insight_mysql.queryCMD(get_data_sql)
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, \
db_name = target_db, port = target_port, charset = 'utf8')
insert_sql = """insert into matchedVideoViewCountCompletion
(matchedVideo_id, trackingWebsite_id, trackingMeta_id, company_id, report_at, view_count, ETLDate)
values(%s, %s, %s, %s, %s, %s, %s)
"""
target_mysql.executeManyCMD(insert_sql, data)
target_mysql.commit()
except Exception, e:
logger.debug("extral data from Dashboard.matchedVideoViewCountCompletion, %s" %e)
sendToMe(subject = "titleBased_matchedVideoViewCountCompletio ERROR", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
def updateTrackingWebsite():
sendToMe(subject = "update TrackingWebsite start", body = "update TrackingWebsite table: SiteBased TitleBased1 TitleBasedRemoveNum1")
logger.info("update TrackingWebsite start")
update_SiteBased_SQL = """
update SiteBased as a, TitleBasedTrackingWebsite as b
set a.websiteName = b.websiteName, a.websiteDomain = b.websiteDomain, a.websiteType = b.websiteType
where a.trackingWebsite_id = b.trackingWebsite_id
"""
update_TitleBased1_SQL = """
update TitleBased1 as a, TitleBasedTrackingWebsite as b
set a.websiteName = b.websiteName, a.websiteType = b.websiteType
where a.trackingWebsite_id = b.trackingWebsite_id
"""
update_TitleBasedRemoveNum1_SQL = """
update TitleBasedRemoveNum1 as a, TitleBasedTrackingWebsite as b
set a.websiteName = b.websiteName, a.websiteType = b.websiteType
where a.trackingWebsite_id = b.trackingWebsite_id
"""
try:
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, db_name = target_db, port = target_port, charset = 'utf8')
target_mysql.queryCMD(update_SiteBased_SQL)
target_mysql.commit()
target_mysql.queryCMD(update_TitleBased1_SQL)
target_mysql.commit()
target_mysql.queryCMD(update_TitleBasedRemoveNum1_SQL)
target_mysql.commit()
except Exception, e:
sendToMe(subject = "update TrackingWebsite ERROR", body = e)
logger.DEBUG("update TrackingWebsite data %s" %e)
def dataTo_matchedVideoTmp():
logger.info("load data target matchedVideoTmp start")
sendToMe(subject = "matchedVideoTmp start", body = "matchedVideoTmp start")
target_db = "FOX_DASHBOARD"
min_reportDate = getMinDatePara("matchedVideoTmp", "created_at", db = target_db)
min_updateDate = getMinDatePara("matchedVideoTmp", "updated_at", db = target_db)
if not min_reportDate:
min_reportDate = "2015-03-01"
if not min_updateDate:
min_updateDate = "0000-00-00 00:00:00"
data = get_matchedVideo(min_reportDate, min_updateDate)
insert_sql = """
insert into matchedVideoTmp
(id, company_id, trackingMeta_id, trackingWebsite_id, view_count,
count_send_notice, first_send_notice_date, takeoff_time, hide_flag,
clip_duration, matchedFile_id, meta_title, created_at, updated_at)
values(%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s)
ON DUPLICATE KEY UPDATE
trackingMeta_id = values(trackingMeta_id), trackingWebsite_id = values(trackingWebsite_id),
view_count = values(view_count), count_send_notice = values(count_send_notice),
first_send_notice_date = values(first_send_notice_date), takeoff_time = values(takeoff_time),
hide_flag = values(hide_flag), clip_duration = values(clip_duration),
matchedFile_id = values(matchedFile_id), meta_title = values(meta_title),
created_at = values(created_at), updated_at = values(updated_at)
"""
dataToTarget(data, target_db, insert_sql)
sendToMe(subject = "matchedVideoTmp end", body = "matchedVideoTmp end")
logger.info("load data target matchedVideoTmp end")
def getDataFromVT():
logger.info(": extract data from tracker2 start")
# extract dat from vtweb
vt_TitleBasedTrackingWebsite_SQL = """
select
a.trackingWebsite_id,
b.website_type as websiteType,
a.display_name as websiteName,
b.website_domain as websiteDomain,
b.country_id,
CURRENT_TIMESTAMP as ETLDate
from trackingWebsiteExtraInfo as a, mddb.trackingWebsite as b
where a.trackingWebsite_id = b.id
"""
#vtweb_tracker2_section = "vtweb_tracker2"
vtweb_tracker2_section = "vtweb_staging"
try:
vt_host, vt_user, vt_passwd, vt_port, vt_db = getConfMysqlInfo(vtweb_tracker2_section)
vtweb_mysql = MySQLHelper(host=vt_host, user=vt_user,passwd=vt_passwd, port = vt_port, db_name = vt_db)
result = vtweb_mysql.queryCMD(vt_TitleBasedTrackingWebsite_SQL)
except Exception, e:
logger.debug(": extract data from vt for dimension trackingWebsite, %s" %e)
sendToMe(subject = "TitleBasedTrackingWebsite ERROR", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
def dataTo_matchedFileItem():
logger.info("load data target matchedFileItem start")
sendToMe(subject = "matchedFileItem start", body = "matchedFileItem start")
get_data_sql = """
select b.*
from tracker2.matchedVideo as a, tracker2.matchedFileItem as b, mddb.trackingWebsite as c
where a.matchedFile_id = b.matchedFile_id
and (c.website_type = "cyberlocker" or c.website_type = "hybrid")
and a.trackingWebsite_id = c.id
and a.company_id = 10
and a.created_at >= "2015-03-01"
"""
insert_sql = """
insert into matchedFileItem
(id, matchedFile_id, trackingWebsite_id, key_id, file_name, file_size, clip_url, takeoff_time, takeoff_type)
values(%s, %s, %s, %s, %s, %s, %s, %s, %s)
ON DUPLICATE KEY UPDATE
file_name = values(file_name), file_size = values(file_size), clip_url = values(clip_url),
takeoff_time = values(takeoff_time), takeoff_type = values(takeoff_type)
"""
target_db = "FOX_DASHBOARD"
data0 = getData(get_data_sql)
data = [(item[0], item[1], item[2], item[3], item[4], item[5], item[6], item[7] if item[7] else "0000-00-00 00:00:00", item[8]) for item in data0]
dataToTarget(data, target_db, insert_sql)
sendToMe(subject = "matchedFileItem end", body = "matchedFileItem end")
logger.info("load data target matchedFileItem end")
def dataTo_matchedVideo():
logger.info("load data target matchedVideo start")
sendToMe(subject = "matchedVideo start", body = "matchedVideo start")
try:
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, \
db_name = "FOX_DASHBOARD", port = target_port, charset = 'utf8')
sql = """
insert into FOX_DASHBOARD.matchedVideo
(id, company_id, trackingMeta_id, trackingWebsite_id, view_count,
count_send_notice, first_send_notice_date, takeoff_time, hide_flag,
clip_duration, matchedFile_id, meta_title, created_at, updated_at)
select
a.id, a.company_id, trackingMeta_id, trackingWebsite_id, view_count,
count_send_notice, first_send_notice_date, takeoff_time, hide_flag,
clip_duration, matchedFile_id, meta_title, created_at, updated_at
from FOX_DASHBOARD.matchedVideoTmp as a, FOX_DASHBOARD.matchedVideoSequence as b
where a.id = b.id and a.company_id = b.company_id
"""
target_mysql.queryNoData("delete from FOX_DASHBOARD.matchedVideo")
target_mysql.queryNoData(sql)
target_mysql.commit()
except Exception, e:
logger.debug("extract data from 192.168.111.235 tracker2 error, %s" %e)
sendToMe(subject = "extract data from 192.168.111.235 tracker2", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
def updateSiteBasedAlexa():
sendToMe(subject = "update alexa data start", body = "update alexa data start")
logger.info(" updateSiteBasedAlexa start")
update_SQL1 = """
update SiteBased as a, SiteBasedAlexa as b
set a.alexaGlobalRank = b.alexaGlobalRank, a.alexaTopCountry = b.alexaTopCountry
where a.trackingWebsite_id = b.trackingWebsite_id
and b.reportDate = (select max(reportDate) from SiteBasedAlexa)
and a.reportDate > (select max(reportDate) from SiteBasedAlexa);
"""
update_SQL2 = """
update SiteBased as a, SiteBasedAlexa as b
set a.alexaGlobalRank = b.alexaGlobalRank, a.alexaTopCountry = b.alexaTopCountry
where a.trackingWebsite_id = b.trackingWebsite_id
and b.reportDate = (select max(reportDate) from SiteBasedAlexa)
and a.alexaTopCountry = "unknown"
and a.alexaGlobalRank = 0;
"""
try:
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, db_name = target_db, port = target_port, charset = 'utf8')
target_mysql.queryCMD(update_SQL1)
target_mysql.commit()
target_mysql.queryCMD(update_SQL2)
target_mysql.commit()
except Exception, e:
sendToMe(subject = "update alexa data ERROR", body = e)
logger.DEBUG("update alexa data %s" %e)
def dataTo_country():
logger.info("load data target country start")
sendToMe(subject = "country start", body = "country start")
get_data_sql = """
select
id, region, country_code, country_name, language_id,
national_flag, longitude, latitude, created_at, updated_at
from mddb.country;
"""
insert_sql = """
insert into country
(id, region, country_code, country_name, language_id,
national_flag, longitude, latitude, created_at, updated_at)
values(%s, %s, %s, %s, %s, %s, %s, %s, %s, %s)
ON DUPLICATE KEY UPDATE
region = values(region), country_code = values(country_code), country_name = values(country_name),
language_id = values(language_id), national_flag = values(national_flag), longitude = values(longitude),
latitude = values(latitude), created_at = values(created_at), updated_at = values(updated_at)
"""
target_db = "FOX_DASHBOARD"
data = getData(get_data_sql)
dataToTarget(data, target_db, insert_sql)
sendToMe(subject = "country end", body = "country end")
logger.info("load data target country end")
def main():
today = time.strftime("%Y-%m-%d")
subject_start = "update alexa data start " + today
sendToMe(subject = subject_start, body = "update alexa data (SiteBased table) start")
updateSiteBasedAlexa()
subject_end = "update alexa data end " + today
sendToMe(subject_end, body = "update alexa data (SiteBased table) end")
def dataToTarget(data, db):
logger.info("load data target matchedVideo start")
try:
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, \
db_name = db, port = target_port, charset = 'utf8')
insert_sql = """
insert into Test
(id, company_id, trackingMeta_id, trackingWebsite_id, view_count,
count_send_notice, first_send_notice_date, takeoff_time, hide_flag,
clip_duration, matchedFile_id, meta_title, created_at, updated_at)
values(%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s)
ON DUPLICATE KEY UPDATE
trackingMeta_id = values(trackingMeta_id), trackingWebsite_id = values(trackingWebsite_id),
view_count = values(view_count), count_send_notice = values(count_send_notice),
first_send_notice_date = values(first_send_notice_date), takeoff_time = values(takeoff_time),
hide_flag = values(hide_flag), clip_duration = values(clip_duration),
matchedFile_id = values(matchedFile_id), meta_title = values(meta_title),
created_at = values(created_at), updated_at = values(updated_at)
"""
commitInTurn(commit_num = 50000, data = data, executeFun = target_mysql.executeManyCMD, \
commitFun = target_mysql.commit, executeSQL =insert_sql)
#target_mysql.executeManyCMD(insert_sql, data)
#target_mysql.commit()
except Exception, e:
logger.debug("extral data from tracker2.matchedVideo, %s" %e)
sendToMe(subject = "matchedVideo ERROR", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
def main():
sendToMe(subject = "titleBased_infringAllViews start", body = "titleBased_infringAllViews start")
matchedVideo_start_date = getMinDatePara("matchedVideo", "firstSendNoticeDate")
if not matchedVideo_start_date:
matchedVideo_start_date = "2015-02-28"
while True:
if str(matchedVideo_start_date) >= str(time.strftime('%Y-%m-%d',time.localtime(time.time() - 24 * 60 * 60 * 365))):
break
matchedVideo_end_date = time.strftime("%Y-%m-%d", time.localtime(time.mktime(time.strptime(str(matchedVideo_start_date), "%Y-%m-%d")) + 1 * 24 * 60 * 60))
getMatchedVideo(start_date = matchedVideo_start_date, end_date = matchedVideo_end_date)
matchedVideo_start_date = getMinDatePara("matchedVideo", "firstSendNoticeDate")
# ---------------------------------------------------------------------------------------------------------------------
insight_start_date = getMinDatePara("matchedVideoViewCountCompletion", "report_at")
if not insight_start_date:
insight_start_date = "2015-02-28"
while True:
if str(insight_start_date) >= str(time.strftime('%Y-%m-%d',time.localtime(time.time() - 24 * 60 * 60 * 365))):
break
insight_end_date = time.strftime("%Y-%m-%d", time.localtime(time.mktime(time.strptime(str(insight_start_date), "%Y-%m-%d")) + 1 * 24 * 60 * 60))
getInsightViews(start_date = insight_start_date, end_date = insight_end_date)
insight_start_date = getMinDatePara("matchedVideoViewCountCompletion", "report_at")
sendToMe(subject = "titleBased_infringAllViews end", body = "titleBased_infringAllViews end")
def getInsightViews(start_date, end_date):
logger.info(" extral data from Dashboard.matchedVideoViewCountCompletionAll start")
date_dict ={"start_date": start_date, "end_date": end_date}
target_server_section = "target_server_staging"
insight_server_section = "insight"
target_host, target_user, target_passwd, target_port, target_db = getConfMysqlInfo(target_server_section)
insight_host, insight_user, insight_passwd, insight_port, insight_db = getConfMysqlInfo(insight_server_section)
try:
get_data_sql = """
select matchedVideo_id, trackingWebsite_id, trackingMeta_id,
company_id, report_at, view_count, hide_flag, current_timestamp as ETLDate
from matchedVideoViewCountCompletionAll
where report_at > "%(start_date)s" and report_at <= "%(end_date)s"
""" %date_dict
insight_mysql = MySQLHelper(host=insight_host, user=insight_user,passwd=insight_passwd, port = insight_port, db_name = insight_db)
data = insight_mysql.queryCMD(get_data_sql)
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, \
db_name = target_db, port = target_port, charset = 'utf8')
insert_sql = """
insert into matchedVideoViewCountCompletionAll
(matchedVideo_id, trackingWebsite_id, trackingMeta_id, company_id, report_at, view_count, hide_flag, ETLDate)
values(%s, %s, %s, %s, %s, %s, %s, %s)
"""
commitInTurn(commit_num=100000, data=data, executeFun=target_mysql.executeManyCMD, \
commitFun=target_mysql.commit, executeSQL=insert_sql)
#target_mysql.executeManyCMD(insert_sql, data)
#target_mysql.commit()
except Exception, e:
logger.debug("extral data from Dashboard.matchedVideoViewCountCompletionAll, %s" %e)
sendToMe(subject = "titleBased_matchedVideoViewCountCompletionAll ERROR", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
def loadDataToTitleBasedRemoveNum1():
logger.info(":extract data from TitleBasedRemoveNum start")
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
try:
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, db_name = target_db, port = target_port, charset = 'utf8')
aggregate_SQL = """
select
a.reportDate,
a.takeoffDate,
a.trackingWebsite_id,
ifnull(c.displayName, a.websiteName) as websiteName,
a.websiteType,
ifnull(b.mapTitle, a.title) as title,
sum(removedNum) as removedNum,
sum(complianceTime) as complianceTime,
current_timestamp as ETLDate
from
(select
a.reportDate,
a.takeoffDate,
a.trackingWebsite_id,
c.websiteName,
c.websiteType,
b.title,
sum(removedNum) as removedNum,
sum(complianceTime) as complianceTime
from TitleBasedRemoveNumTmp as a, TitleBasedMeta as b, TitleBasedTrackingWebsite as c
where a.trackingWebsite_id = c.trackingWebsite_id
and a.trackingMeta_id = b.trackingMeta_id
group by 1, 2, 3, 4, 5, 6) as a
left join MetaTitleMapTitle as b
on a.title = b.metaTitle
left join SiteMap as c
on a.trackingWebsite_id = c.trackingWebsite_id
group by 1, 2, 3, 4, 5, 6
"""
target_mysql.queryNoData("delete from TitleBasedRemoveNum")
aggregate_result = target_mysql.queryCMD(aggregate_SQL)
insertUpdate_SQL = """
INSERT INTO TitleBasedRemoveNum
(reportDate, takeoffDate, trackingWebsite_id, websiteName,
websiteType, title, removedNum, complianceTime, ETLDate)
VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s)
ON DUPLICATE KEY UPDATE
removedNum = VALUES(removedNum), complianceTime = VALUES(complianceTime), ETLDate = VALUES(ETLDate)
"""
target_mysql.queryNoData("delete from TitleBasedRemoveNum")
commitInTurn(commit_num = 50000, data = aggregate_result, executeFun = target_mysql.insertUpdateCMD, \
commitFun = target_mysql.commit, executeSQL = insertUpdate_SQL)
#target_mysql.insertUpdateCMD(insertUpdate_SQL, aggregate_result)
#target_mysql.commit()
except Exception, e:
logger.debug(" load data to TitleBasedRemoveNum, %s" %e)
sendToMe(subject = "TitleBasedRemove ERROR", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
def main():
sendToMe(subject = "metaExtraInfo start", body = "metaExtraInfo start")
cfg_file = "/Job/VIACOM/Dashboard/TitleBased/conf/viacom_dashboard.cfg"
job_name = "metaExtraInfo"
judgeFileExist(cfg_file, job_name)
target_db = "tracker2"
data = getData()
dataToTarget(data, target_db)
sendToMe(subject = "metaExtraInfo end", body = "metaExtraInfo end")
def updateViews(start_date, end_date):
logger.info(" aggregate data from matchedVideoViewCountCompletion start")
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db = getConfMysqlInfo(target_server_section)
date_dict = {"start_date": start_date, "end_date": end_date}
try:
get_data_sql = """
select
a.reportDate,
a.trackingWebsite_id,
ifnull(c.displayName, a.websiteName) as websiteName,
a.websiteType,
ifnull(b.mapTitle, a.title) title,
sum(a.infringingViews) as infringingViews,
sum(a.reportedViews) as reportedViews,
current_timestamp as ETLDate
from
(select
a.report_at as reportDate,
a.trackingWebsite_id,
c.websiteName,
c.websiteType,
b.title,
sum(if(a.hide_flag = 2, a.view_count, 0)) as infringingViews,
sum(a.view_count) as reportedViews
from matchedVideoViewCountCompletionAll as a, TitleBasedMeta as b, TitleBasedTrackingWebsite as c
where a.trackingWebsite_id = c.trackingWebsite_id
and a.trackingMeta_id = b.trackingMeta_id
and c.websiteType = 'ugc'
and a.report_at > "%(start_date)s"
and a.report_at <= "%(end_date)s"
group by 1, 2, 3, 4, 5) as a
left join MetaTitleMapTitle as b on a.title = b.metaTitle
left join SiteMap as c on a.trackingWebsite_id = c.trackingWebsite_id
group by 1, 2, 3, 4, 5
""" % date_dict
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, \
db_name=target_db, port=target_port, charset='utf8')
data = target_mysql.queryCMD(get_data_sql)
insert_sql = """insert into TitleBased1
(reportDate, trackingWebsite_id, websiteName, websiteType, title, infringingViews, reportedViews, ETLDate)
values(%s, %s, %s, %s, %s, %s, %s, %s)
ON DUPLICATE KEY UPDATE
infringingViews = values(infringingViews), reportedViews = values(reportedViews),
ETLDate = values(ETLDate), websiteName = VALUES(websiteName)
"""
commitInTurn(commit_num=100000, data=data, executeFun=target_mysql.executeManyCMD, \
commitFun=target_mysql.commit, executeSQL=insert_sql)
except Exception, e:
logger.debug("aggregate data to TitleBased1 ERROR , %s" % e)
sendToMe(subject="titleBased_infringAllViews ERROR", body=re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
def getData(sql):
logger.info(" extract data from tracker2.metaExtraInfo start")
try:
vtweb_tracker2_section = "vtweb"
vt_host, vt_user, vt_passwd, vt_port, vt_db = getConfMysqlInfo(vtweb_tracker2_section)
vtweb_mysql = MySQLHelper(host=vt_host, user=vt_user,passwd=vt_passwd, port = vt_port, db_name = vt_db)
data = vtweb_mysql.queryCMD(sql)
except Exception, e:
logger.debug("extract data from tracker2.metaExtraInfo, %s" %e)
sendToMe(subject = "metaExtraInfo ERROR", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
def getMinDatePara(table_name, date_para, db = "FOX_DASHBOARD"):
#get max report date from table TitleBased
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db = getConfMysqlInfo(target_server_section)
try:
target_mysql = MySQLHelper(host=target_host, user=target_user,passwd=target_passwd, port = target_port, db_name = db)
sql = "select max(%s) from %s" %(date_para, table_name)
min_date = target_mysql.queryCMD(sql)[0][0]
#if min_date == None:
# min_date = time.strftime('%Y-%m-%d',time.localtime(time.time() - 24 * 60 * 60 * 365))
except Exception , e:
logger.debug(": get last max report date from %s, %s" %(table_name, e))
sendToMe(subject = table_name, body = re.sub(r'\'|"|!', "", str(e)))
def main():
cfg_file = "/Job/FOX/Dashboard/TitleBased/conf/viacom_dashboard.cfg"
if not os.path.exists(cfg_file):
logging.debug(": config file not exists; file_name %s" %cfg_file)
sendToMe(subject = "SiteBased_alexa ERROR", body = "config file not exists")
sys.exit(0)
logger.info(": extract data from siteBased start")
socket.setdefaulttimeout(10.0)
sendToMe(subject = "SiteBased_alexa start", body = "extract data from siteBased start")
target_server_section = "staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
try:
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, db_name = target_db, port = target_port, charset = 'utf8')
if True:
f = open("id_dis_domain", "r")
for line in f.readlines():
line = line.strip("\n")
base_url = "http://www.alexa.com/siteinfo/"
id, display_name, domain = line.split(",")[1], line.split(",")[2], line.split(",")[3]
url = base_url + domain
run_time = 0
alexaGlobalRank, alexaTopCountry = getAlexaInfo(url)
while True:
run_time += 1
if alexaGlobalRank == 0 or alexaTopCountry == "unknown":
alexaGlobalRank, alexaTopCountry = getAlexaInfo(url)
else:
break
if run_time == 3:
break
time.sleep(random.randint(5, 8))
time.sleep(random.randint(5, 8))
alexa_info_tuple = [("2016-05-31", id, display_name, alexaGlobalRank,alexaTopCountry,1)]
if not (alexaGlobalRank == 0 and alexaTopCountry == "unknown"):
insert_SiteBasedAlexa_SQL = """
insert into Website_Alexa_Info
(CreateDate, WebsiteId, DisplayName, Rank, TopOneCountry, IsEndOfMonth)
values (%s, %s, %s, %s, %s, %s)
on duplicate key update
Rank = values(Rank),
TopOneCountry = values(TopOneCountry)
"""
try:
target_mysql.insertUpdateCMD(insert_SiteBasedAlexa_SQL, alexa_info_tuple)
print alexa_info_tuple
target_mysql.commit()
except MySQLdb.Error, e:
logger.debug(e)
sendToMe(subject = "update SiteBasedAlexa Error", body = re.sub(r'\'|"|!', "", str(e)))
continue
else:
logger.info("has no data %s" %alexa_date_max)
except Exception, e:
logger.debug(": load data to SiteBasedAlexa %s" %e)
sendToMe(subject = "SiteBasedAlexa ERROR", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
def addAlexa(d):
sendToMe(subject = "add alexa data start", body = "add alexa data start")
logger.info(" addSiteBasedAlexa start")
get_alexa_sql = "select '%s', trackingWebsite_id, websiteDomain, alexaGlobalRank, alexaTopCountry, current_timestamp() from SiteBasedAlexa where reportDate = (select max(reportDate) from SiteBasedAlexa);" %d
insert_alexa_sql = "insert into SiteBasedAlexa(reportDate,trackingWebsite_id, websiteDomain, alexaGlobalRank, alexaTopCountry,ETLDate) VALUES(%s,%s,%s,%s,%s,%s);"
try:
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, db_name = target_db, port = target_port, charset = 'utf8')
alexa_result = target_mysql.queryCMD(get_alexa_sql)
target_mysql.insertUpdateCMD(insert_alexa_sql, alexa_result)
target_mysql.commit()
except Exception, e:
sendToMe(subject = "add alexa data ERROR", body = e)
logger.DEBUG("add alexa data %s" %e)
def main():
sendToMe(subject = "matchedVideoViewCountCompletion start", body = "matchedVideoViewCountCompletion start")
insight_start_date = getMinDatePara("matchedVideoViewCountCompletion", "report_at")
if not insight_start_date:
insight_start_date = "2015-02-28"
while True:
if str(insight_start_date) >= str(time.strftime('%Y-%m-%d',time.localtime(time.time() - 2 * 24 * 60 * 60))):
break
insight_end_date = time.strftime("%Y-%m-%d", time.localtime(time.mktime(time.strptime(str(insight_start_date), "%Y-%m-%d")) + 1 * 24 * 60 * 60))
getInsightViews(start_date = insight_start_date, end_date = insight_end_date)
insight_start_date = getMinDatePara("matchedVideoViewCountCompletion", "report_at")
sendToMe(subject = "matchedVideoViewCountCompletion end", body = "matchedVideoViewCountCompletion end")
def dataToTarget(data, db, insert_sql):
logger.info("load data target metaExtraInfo start")
try:
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, \
db_name = db, port = target_port, charset = 'utf8')
commitInTurn(commit_num = 50000, data = data, executeFun = target_mysql.executeManyCMD, \
commitFun = target_mysql.commit, executeSQL =insert_sql)
#target_mysql.executeManyCMD(insert_sql, data)
#target_mysql.commit()
except Exception, e:
logger.debug("extral data from tracker2.metaExtraInfo, %s" %e)
sendToMe(subject = "metaExtraInfo ERROR", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
def getData():
logger.info(" extract data from tracker2.metaExtraInfo start")
try:
get_data_sql = """
select
meta_id, company_id, display_name, priority_type, created_at, updated_at
from tracker2.metaExtraInfo
where company_id = 14
"""
vtweb_tracker2_section = "vtweb"
vt_host, vt_user, vt_passwd, vt_port, vt_db = getConfMysqlInfo(vtweb_tracker2_section)
vtweb_mysql = MySQLHelper(host=vt_host, user=vt_user,passwd=vt_passwd, port = vt_port, db_name = vt_db)
data = vtweb_mysql.queryCMD(get_data_sql)
except Exception, e:
logger.debug("extract data from tracker2.metaExtraInfo, %s" %e)
sendToMe(subject = "metaExtraInfo ERROR", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
def updateCountry():
sendToMe(subject = "update Country start", body = "update Country table: SiteBased")
logger.info("update Country for table SiteBased start")
update_SiteBased_SQL = """
update SiteBased as a, TitleBasedCountry as b
set a.hostCountry = b.countryName
where a.country_id = b.country_id
"""
try:
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, db_name = target_db, port = target_port, charset = 'utf8')
target_mysql.queryCMD(update_SiteBased_SQL)
target_mysql.commit()
except Exception, e:
sendToMe(subject = "update TrackingWebsite ERROR", body = e)
logger.DEBUG("update TrackingWebsite data %s" %e)
def main():
sendToMe(subject = "matchedVideo start", body = "matchedVideo start")
cfg_file = "/Job/VIACOM/Dashboard/TitleBased/conf/viacom_dashboard.cfg"
job_name = "matchedVideo"
judgeFileExist(cfg_file, job_name)
target_db = "tracker2"
min_reportDate = getMinDatePara("Test", "created_at", db = target_db)
min_updateDate = getMinDatePara("Test", "updated_at", db = target_db)
if not min_reportDate:
min_reportDate = "2015-03-01"
if not min_updateDate:
min_updateDate = "0000-00-00 00:00:00"
min_reportDate = '2016-05-08'
data = getMatchedVideo(min_reportDate, min_updateDate)
dataToTarget(data, target_db)
sendToMe(subject = "matchedVideo end", body = "matchedVideo end")
def loadDataTo123(vtweb_data):
logger.info(":load data to TitleBasedMeta start")
target_server_section = "target_server_staging"
try:
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, db_name = target_db, port = target_port, charset = 'utf8')
insertUpdate_SQL = """
INSERT INTO TitleBasedMeta(trackingMeta_id, metaType, title, ETLDate) VALUES(%s, %s, %s, %s) on duplicate key update title = values(title),
ETLDate = values(ETLDate), metaType = values(metaType)
"""
commitInTurn(commit_num = 50000, data = vtweb_data, executeFun = target_mysql.insertUpdateCMD, \
commitFun = target_mysql.commit, executeSQL = insertUpdate_SQL)
#target_mysql.insertUpdateCMD(insertUpdate_SQL, result)
#target_mysql.commit()
except Exception, e:
logger.debug(": load data to TitleBasedMeta, %s" %e)
sendToMe(subject = "titleBased_meta ERROR", body = e)
sys.exit(0)
def dataToMetaExtraInfo():
sendToMe(subject = "metaExtraInfo start", body = "metaExtraInfo start")
get_data_sql = """
select meta_id, company_id, display_name, priority_type, created_at, updated_at
from tracker2.metaExtraInfo
where company_id = 14
"""
insert_sql = """
insert into metaExtraInfo
(meta_id, company_id, display_name, priority_type, created_at, updated_at)
values(%s, %s, %s, %s, %s, %s)
ON DUPLICATE KEY UPDATE
display_name = values(display_name), priority_type = values(priority_type),
created_at = values(created_at), updated_at = values(updated_at)
"""
target_db = "tracker2"
data = getData(get_data_sql)
dataToTarget(data, target_db, insert_sql)
sendToMe(subject = "metaExtraInfo end", body = "metaExtraInfo end")
def dataTo_trackingWebsiteExtraInfo():
logger.info("load data target trackingWebsiteExtraInfo start")
sendToMe(subject = "trackingWebsiteExtraInfo start", body = "trackingWebsiteExtraInfo start")
get_data_sql = """
select trackingWebsite_id, display_name, created_at, updated_at
from tracker2.trackingWebsiteExtraInfo
"""
insert_sql = """
insert into trackingWebsiteExtraInfo
(trackingWebsite_id, display_name, created_at, updated_at)
values(%s, %s, %s, %s)
ON DUPLICATE KEY UPDATE
display_name = values(display_name), created_at = values(created_at), updated_at = values(updated_at)
"""
target_db = "FOX_DASHBOARD"
data = getData(get_data_sql)
dataToTarget(data, target_db, insert_sql)
sendToMe(subject = "trackingWebsiteExtraInfo end", body = "trackingWebsiteExtraInfo end")
logger.info("load data target trackingWebsiteExtraInfo end")
def main():
sendToMe(subject = "matchedVideo start", body = "matchedVideo start")
while True:
min_reportDate = getMinDatePara("matchedVideo", "reportDate")
if not min_reportDate:
min_reportDate = "2015-02-28"
max_reportDate = time.strftime("%Y-%m-%d", \
time.localtime(time.mktime(time.strptime(str(min_reportDate), "%Y-%m-%d")) + 1 * 24 * 60 * 60))
min_updateDate = getMinDatePara("matchedVideo", "updateDate")
if not min_updateDate:
min_updateDate = "2015-02-28"
if str(max_reportDate) >= str(time.strftime('%Y-%m-%d',time.localtime(time.time() - 1 * 24 * 60 * 60))):
break
getMatchedVideo(min_reportDate, max_reportDate, min_updateDate)
sendToMe(subject = "matchedVideo end", body = "matchedVideo end")
def loadDataTo123(vtweb_data):
logger.info(":load data to TitleBasedTrackingWebsite start")
target_server_section = "target_server_staging"
try:
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, db_name = target_db, port = target_port, charset = 'utf8')
insertUpdate_SQL = """
INSERT INTO TitleBasedTrackingWebsite
(trackingWebsite_id, websiteType, websiteName, websiteDomain, country_id, ETLDate)
VALUES(%s, %s, %s, %s, %s, %s)
on duplicate key update websiteDomain = values(websiteDomain),
websiteType = values(websiteType), ETLDate = values(ETLDate), websiteName = values(websiteName), country_id = values(country_id)
"""
result1 = [(item[0], item[1], item[2], re.sub(r"_ugc|_hybrid|_Hybrid|_cyberlocker|_UGC|_Cy|_cy", '', item[3]), item[4], item[5]) for item in vtweb_data]
commitInTurn(commit_num = 50000, data = result1, executeFun = target_mysql.insertUpdateCMD, \
commitFun = target_mysql.commit, executeSQL = insertUpdate_SQL)
except Exception, e:
logger.debug(": load data to TitleBasedTrackingWebsite, %s" %e)
sendToMe(subject = "TitleBasedTrackingWebsite ERROR", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
def getDataFromVT():
logger.info(": extract data from tracker2 start")
# extract dat from vtweb
vt_TitleBasedMeta_SQL = """
select a.id as trackingMeta_id, a.meta_type, a.meta_title as title, CURRENT_TIMESTAMP as ETLDate
from mddb.meta as a, tracker2.metaExtraInfo as b
where a.company_id = 14
and b.company_id = 14
and a.id = b.meta_id
"""
#vtweb_tracker2_section = "vtweb_tracker2"
vtweb_tracker2_section = "vtweb_staging"
try:
vt_host, vt_user, vt_passwd, vt_port, vt_db = getConfMysqlInfo(vtweb_tracker2_section)
vtweb_mysql = MySQLHelper(host=vt_host, user=vt_user,passwd=vt_passwd, port = vt_port, db_name = vt_db)
result = vtweb_mysql.queryCMD(vt_TitleBasedMeta_SQL)
except Exception, e:
logger.debug(": extract data from vt for dimension meta, %s" %e)
sendToMe(subject = "titleBased_meta ERROR", body = e)
sys.exit(0)
logger = logging.getLogger("titleBased_CMS")
logger.setLevel(logging.DEBUG)
log_file = '/Job/VIACOM/Dashboard/TitleBased/log/titleBased_CMS.log'
filehandler = logging.handlers.RotatingFileHandler(filename=log_file,
maxBytes=5 * 1024 * 1024,
backupCount=10,
mode='a')
formatter = logging.Formatter(
'%(asctime)s %(filename)s[line:%(lineno)d] %(levelname)s %(message)s')
filehandler.setFormatter(formatter)
logger.addHandler(filehandler)
cfg_file = "/Job/VIACOM/Dashboard/TitleBased/conf/viacom_dashboard.cfg"
if not os.path.exists(cfg_file):
logging.debug(": config file not exists")
sendToMe(subject="titleBased_CMS ERROR", body="config file not exists")
sys.exit(0)
#################################################################################################################################
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db = getConfMysqlInfo(
target_server_section)
#################################################################################################################################
sendToMe(
subject="titleBased_CMS start",
body=
"aggregate data from DM_VIACOM.SelfService_Aggregate_ByNoticedDate (CMS data) start"
)
##CMS data and reportedViews (UGC) from SelfService_Aggregate_ByNoticedDate aggregate to TitleBased1
#--------------------------------------------------------------------------------------------------------------------------------
logger.info(
" aggregate data from DM_VIACOM.SelfService_Aggregate_ByNoticedDate (CMS data) start"
from titleBased import getConfMysqlInfo, getMinDatePara
from sendMail import sendToMe
import re
logger = logging.getLogger("titleBased_matchedVideo")
logger.setLevel(logging.DEBUG)
log_file = '/Job/VIACOM/Dashboard/TitleBased/log/titleBased_matchedVideo.log'
filehandler = logging.handlers.RotatingFileHandler(filename=log_file, maxBytes=5*1024*1024, backupCount=10, mode='a')
formatter = logging.Formatter('%(asctime)s %(filename)s[line:%(lineno)d] %(levelname)s %(message)s')
filehandler.setFormatter(formatter)
logger.addHandler(filehandler)
cfg_file = "/Job/VIACOM/Dashboard/TitleBased/conf/viacom_dashboard.cfg"
if not os.path.exists(cfg_file):
logging.debug(": config file not exists")
sendToMe(subject = "titleBased_matchedVideo ERROR", body = "config file not exists")
sys.exit(0)
#################################################################################################################################
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
insight_server_section = "insight"
insight_host, insight_user, insight_passwd, insight_port, insight_db= getConfMysqlInfo(insight_server_section)
vtweb_tracker2_section = "vtweb_staging"
vt_host, vt_user, vt_passwd, vt_port, vt_db = getConfMysqlInfo(vtweb_tracker2_section)
vtweb_mysql = MySQLHelper(host=vt_host, user=vt_user,passwd=vt_passwd, port = vt_port, db_name = vt_db)
#################################################################################################################################
def getMatchedVideo(min_reportDate, max_reportDate, min_updateDate):
logger.info(" extral data from tracker2.matchedVideo start")
def main():
cfg_file = "/Job/HBO/Dashboard/TitleBased/conf/viacom_dashboard.cfg"
if not os.path.exists(cfg_file):
logging.debug(": config file not exists; file_name %s" %cfg_file)
sendToMe(subject = "SiteBased_alexa ERROR", body = "config file not exists")
sys.exit(0)
logger.info(": extract data from siteBased start")
socket.setdefaulttimeout(10.0)
sendToMe(subject = "SiteBased_alexa start", body = "extract data from siteBased start")
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
try:
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, db_name = target_db, port = target_port, charset = 'utf8')
alexa_date_min = getMinDatePara("SiteBasedAlexa", "reportDate")
alexa_date_min = '2016-04-12'
if alexa_date_min == None:
alexa_date_min = time.strftime("%Y-%m-%d", time.localtime(time.time() - 10 * 24 * 60 * 60))
alexa_date_max = time.strftime("%Y-%m-%d", time.localtime(time.time() - 1 * 24 * 60 * 60))
delt = time.mktime(time.strptime(str(alexa_date_max), "%Y-%m-%d")) - time.mktime(time.strptime(str(alexa_date_min), "%Y-%m-%d"))
if delt >= 10*24*60*60:
alexa_date_dict = {"alexa_date_min": alexa_date_min, "alexa_date_max": alexa_date_max}
site_SQL = """
select distinct trackingWebsite_id, websiteDomain from SiteBased
where reportDate <= "%(alexa_date_max)s"
and reportDate > "%(alexa_date_min)s"
and alexaGlobalRank = 0
and alexaTopCountry = "unknown"
""" %alexa_date_dict
site_info = target_mysql.queryCMD(site_SQL)
if site_info:
base_url = "http://www.alexa.com/siteinfo/"
for site in site_info:
domain = site[1]
url = base_url + domain
run_time = 0
alexaGlobalRank, alexaTopCountry = getAlexaInfo(url)
while True:
run_time += 1
if alexaGlobalRank == 0 or alexaTopCountry == "unknown":
alexaGlobalRank, alexaTopCountry = getAlexaInfo(url)
else:
break
if run_time == 3:
break
time.sleep(random.randint(5, 8))
time.sleep(random.randint(5, 8))
alexa_info_tuple = [(alexa_date_max, site[0], site[1], alexaGlobalRank, alexaTopCountry, time.strftime("%Y-%m-%d %H:%M:%S"))]
if not (alexaGlobalRank == 0 and alexaTopCountry == "unknown"):
insert_SiteBasedAlexa_SQL = """
insert into SiteBasedAlexa
(reportDate, trackingWebsite_id, websiteDomain, alexaGlobalRank, alexaTopCountry, ETLDate)
values (%s, %s, %s, %s, %s, %s)
on duplicate key update
alexaGlobalRank = values(alexaGlobalRank),
alexaTopCountry = values(alexaTopCountry), ETLDate = values(ETLDate)
"""
try:
target_mysql.insertUpdateCMD(insert_SiteBasedAlexa_SQL, alexa_info_tuple)
print alexa_info_tuple
target_mysql.commit()
except MySQLdb.Error, e:
logger.debug(e)
sendToMe(subject = "update SiteBasedAlexa Error", body = re.sub(r'\'|"|!', "", str(e)))
continue
else:
logger.info("has no data %s" %alexa_date_max)
except Exception, e:
logger.debug(": load data to SiteBasedAlexa %s" %e)
sendToMe(subject = "SiteBasedAlexa ERROR", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
import time
from titleBased import getConfMysqlInfo, commitInTurn
from sendMail import sendToMe
logger = logging.getLogger("titleBased_meta")
logger.setLevel(logging.DEBUG)
log_file = '/Job/FOX/Dashboard/TitleBased/log/titleBased_meta.log'
filehandler = logging.handlers.RotatingFileHandler(filename=log_file, maxBytes=5*1024*1024, backupCount=10, mode='a')
formatter = logging.Formatter('%(asctime)s %(filename)s[line:%(lineno)d] %(levelname)s %(message)s')
filehandler.setFormatter(formatter)
logger.addHandler(filehandler)
cfg_file = "/Job/FOX/Dashboard/TitleBased/conf/viacom_dashboard.cfg"
if not os.path.exists(cfg_file):
logging.debug(": config file not exists")
sendToMe(subject = "titleBased_meta Error", body = "config file not exists")
sys.exit(0)
#################################################################################################################################
def getDataFromVT():
logger.info(": extract data from tracker2 start")
# extract dat from vtweb
vt_TitleBasedMeta_SQL = """
select a.id as trackingMeta_id, a.meta_type, a.meta_title as title, CURRENT_TIMESTAMP as ETLDate
from meta as a, metaExtraInfo as b
where a.company_id = 10
and b.company_id = 10
and a.id = b.meta_id
"""
#vtweb_tracker2_section = "vtweb_tracker2"
vtweb_tracker2_section = "vtweb_staging"
try:
import time
from titleBased import getConfMysqlInfo, getMinDatePara
from sendMail import sendToMe
logger = logging.getLogger("titleBased_views")
logger.setLevel(logging.DEBUG)
log_file = '/Job/HBO/Dashboard/TitleBased/log/titleBased_views.log'
filehandler = logging.handlers.RotatingFileHandler(filename=log_file, maxBytes=5*1024*1024, backupCount=10, mode='a')
formatter = logging.Formatter('%(asctime)s %(filename)s[line:%(lineno)d] %(levelname)s %(message)s')
filehandler.setFormatter(formatter)
logger.addHandler(filehandler)
cfg_file = "/Job/HBO/Dashboard/TitleBased/conf/viacom_dashboard.cfg"
if not os.path.exists(cfg_file):
logging.debug(": config file not exists")
sendToMe(subject = "titleBased_views ERROR", body = "config file not exists")
sys.exit(0)
#################################################################################################################################
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
#################################################################################################################################
##CMS data and reportedViews (UGC) from SelfService_Aggregate_ByNoticedDate aggregate to TitleBased1
#--------------------------------------------------------------------------------------------------------------------------------
logger.info(" aggregate data from DM_HBO.SelfService_Aggregate_ByNoticedDate, extract reportedViews start")
sendToMe(subject = "titleBased_views start", body = "aggregate data from DM_HBO.SelfService_Aggregate_ByNoticedDate, extract reportedViews start")
date_para_reportedViews_min = time.strftime("%Y-%m-%d", time.localtime(time.time()- (65) * 24 * 60 * 60))
date_para_reportedViews_min = "2015-03-01"
date_para_reportedViews_max = time.strftime("%Y-%m-%d", time.localtime(time.time()- 1 * 24 * 60 * 60))
logger.setLevel(logging.DEBUG)
log_file = '/Job/FOX/Dashboard/TitleBased/log/titleBased.log'
filehandler = logging.handlers.RotatingFileHandler(filename=log_file,
maxBytes=5 * 1024 * 1024,
backupCount=10,
mode='a')
formatter = logging.Formatter(
'%(asctime)s %(filename)s[line:%(lineno)d] %(levelname)s %(message)s')
filehandler.setFormatter(formatter)
logger.addHandler(filehandler)
cfg_file = "/Job/FOX/Dashboard/TitleBased/conf/viacom_dashboard.cfg"
if not os.path.exists(cfg_file):
logger.debug(": config file not exists")
sendToMe(subject="titleBasedTmp ERROR", body="config file not exists")
sys.exit(0)
def getConfMysqlInfo(section_name):
conf = CfgParser(cfg_file).parse()
try:
host = conf[section_name]["host"]
user = conf[section_name]["user"]
passwd = conf[section_name]["passwd"]
port = conf[section_name]["port"]
db = conf[section_name]["db"]
return host, user, passwd, port, db
except KeyError, e:
logger.debug(": mysql config file section or option not exists, %s" %
logger = logging.getLogger("titleBased_matchedVideoViewCountCompletio")
logger.setLevel(logging.DEBUG)
log_file = '/Job/FOX/Dashboard/TitleBased/log/titleBased_matchedVideoViewCountCompletion.log'
filehandler = logging.handlers.RotatingFileHandler(filename=log_file,
maxBytes=5 * 1024 * 1024,
backupCount=10,
mode='a')
formatter = logging.Formatter(
'%(asctime)s %(filename)s[line:%(lineno)d] %(levelname)s %(message)s')
filehandler.setFormatter(formatter)
logger.addHandler(filehandler)
cfg_file = "/Job/FOX/Dashboard/TitleBased/conf/viacom_dashboard.cfg"
if not os.path.exists(cfg_file):
logging.debug(": config file not exists")
sendToMe(subject="titleBased_matchedVideoViewCountCompletio ERROR",
body="config file not exists")
sys.exit(0)
#################################################################################################################################
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db = getConfMysqlInfo(
target_server_section)
insight_server_section = "insight"
insight_host, insight_user, insight_passwd, insight_port, insight_db = getConfMysqlInfo(
insight_server_section)
vtweb_tracker2_section = "vtweb_staging"
vt_host, vt_user, vt_passwd, vt_port, vt_db = getConfMysqlInfo(
vtweb_tracker2_section)
vtweb_mysql = MySQLHelper(host=vt_host,
logger = logging.getLogger("titleBased_trackingWebsite")
logger.setLevel(logging.DEBUG)
log_file = '/Job/VIACOM/Dashboard/TitleBased/log/titleBased_trackingWebsite.log'
filehandler = logging.handlers.RotatingFileHandler(filename=log_file,
maxBytes=5 * 1024 * 1024,
backupCount=10,
mode='a')
formatter = logging.Formatter(
'%(asctime)s %(filename)s[line:%(lineno)d] %(levelname)s %(message)s')
filehandler.setFormatter(formatter)
logger.addHandler(filehandler)
cfg_file = "/Job/VIACOM/Dashboard/TitleBased/conf/viacom_dashboard.cfg"
if not os.path.exists(cfg_file):
logging.debug(": config file not exists; file_name %s" % cfg_file)
sendToMe(subject="titleBased_trackingWebsite ERROR",
body="config file not exists")
sys.exit(0)
def getDataFromVT():
logger.info(": extract data from tracker2 start")
# extract dat from vtweb
vt_TitleBasedTrackingWebsite_SQL = """
select
a.trackingWebsite_id,
b.website_type as websiteType,
a.display_name as websiteName,
b.website_domain as websiteDomain,
b.country_id,
CURRENT_TIMESTAMP as ETLDate
(reportDate, trackingWebsite_id, websiteDomain, alexaGlobalRank, alexaTopCountry, ETLDate)
values (%s, %s, %s, %s, %s, %s)
on duplicate key update
alexaGlobalRank = values(alexaGlobalRank),
alexaTopCountry = values(alexaTopCountry), ETLDate = values(ETLDate)
"""
try:
target_mysql.insertUpdateCMD(insert_SiteBasedAlexa_SQL, alexa_info_tuple)
print alexa_info_tuple
target_mysql.commit()
except MySQLdb.Error, e:
logger.debug(e)
sendToMe(subject = "update SiteBasedAlexa Error", body = re.sub(r'\'|"|!', "", str(e)))
continue
else:
logger.info("has no data %s" %alexa_date_max)
except Exception, e:
logger.debug(": load data to SiteBasedAlexa %s" %e)
sendToMe(subject = "SiteBasedAlexa ERROR", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
finally:
target_mysql.closeCur()
target_mysql.closeConn()
logger.info(":load data to SiteBasedAlexa end")
sendToMe(subject = "SiteBasedAlexa End", body = "load data to SiteBasedAlexa end")
if __name__ == "__main__":
main()
#################################################################################################################################
#sed "/\/home\/vobile\/cwj\/ViacomProject\/dashboard\/job/\/Job\/HBO\/Dashboard\/TitleBased/g"
def judgeFileExist(cfg_file, job_name):
if not os.path.exists(cfg_file):
logging.debug(": config file not exists")
sendToMe(subject = job_name +" ERROR", body = "config file not exists")
sys.exit(0)
logger = logging.getLogger("monitorComplianceRate")
logger.setLevel(logging.DEBUG)
log_file = '/Job/VIACOM/Dashboard/TitleBased/log/monitorComplianceRate.log'
filehandler = logging.handlers.RotatingFileHandler(filename=log_file,
maxBytes=5 * 1024 * 1024,
backupCount=10,
mode='a')
formatter = logging.Formatter(
'%(asctime)s %(filename)s[line:%(lineno)d] %(levelname)s %(message)s')
filehandler.setFormatter(formatter)
logger.addHandler(filehandler)
cfg_file = "/Job/VIACOM/Dashboard/TitleBased/conf/viacom_dashboard.cfg"
if not os.path.exists(cfg_file):
logging.debug(": config file not exists")
sendToMe(subject="monitorComplianceRate ERROR",
body="config file not exists")
sys.exit(0)
#################################################################################################################################
def monitor():
logger.info("monitor complaince rate start")
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db = getConfMysqlInfo(
target_server_section)
try:
target_mysql = MySQLHelper(host=target_host,
user=target_user,
passwd=target_passwd,
db_name=target_db,
port=target_port,
clip_duration, matchedFile_id, meta_title, created_at, updated_at
from HBO_DASHBOARD.matchedVideoTmp as a, HBO_DASHBOARD.matchedVideoSequence as b
where a.id = b.id and a.company_id = b.company_id
"""
target_mysql.queryNoData("delete from HBO_DASHBOARD.matchedVideo")
target_mysql.queryNoData(sql)
target_mysql.commit()
except Exception, e:
logger.debug("extract data from 192.168.111.235 tracker2 error, %s" %e)
sendToMe(subject = "extract data from 192.168.111.235 tracker2", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
finally:
target_mysql.closeConn()
target_mysql.closeCur()
sendToMe(subject = "matchedVideo end", body = "matchedVideo end")
logger.info("load data target matchedVideo end")
def dataTo_metaExtraInfo():
logger.info("load data target metaExtraInfo start")
sendToMe(subject = "metaExtraInfo start", body = "metaExtraInfo start")
get_data_sql = """
select meta_id, company_id, display_name, priority_type, created_at, updated_at
from tracker2.metaExtraInfo
where company_id = 34
"""
insert_sql = """
insert into metaExtraInfo
(meta_id, company_id, display_name, priority_type, created_at, updated_at)
values(%s, %s, %s, %s, %s, %s)
target_mysql.commit()
target_mysql.queryCMD(update_TitleBased1_SQL)
target_mysql.commit()
target_mysql.queryCMD(update_TitleBasedRemoveNum1_SQL)
target_mysql.commit()
except Exception, e:
sendToMe(subject="update TrackingWebsite ERROR", body=e)
logger.DEBUG("update TrackingWebsite data %s" % e)
finally:
target_mysql.closeCur()
target_mysql.closeConn()
sendToMe(
subject="update TrackingWebsite end",
body=
"update TrackingWebsite table: SiteBased TitleBased1 TitleBasedRemoveNum1"
)
logger.info(" update TrackingWebsite end")
def updateCountry():
sendToMe(subject="update Country start",
body="update Country table: SiteBased")
logger.info("update Country for table SiteBased start")
update_SiteBased_SQL = """
update SiteBased as a, TitleBasedCountry as b
set a.hostCountry = b.countryName
where a.country_id = b.country_id
"""
def main(): sendToMe(subject = "titleBased_meta start", body = "extract data from tracker2 start") vtweb_data = getDataFromVT() loadDataTo123(vtweb_data) sendToMe(subject = "titleBased_meta End", body = "load data to TitleBasedMeta end")
def getMatchedVideo(min_reportDate, max_reportDate, min_updateDate):
logger.info(" extral data from tracker2.matchedVideo start")
date_dict ={"min_reportDate": min_reportDate, "max_reportDate": max_reportDate, "min_updateDate": min_updateDate}
try:
get_data_sql = """
select
a.id as matchedVideo_id,
a.trackingMeta_id,
a.trackingWebsite_id,
date_format(a.first_send_notice_date, "%%Y-%%m-%%d") as firstSendNoticeDate,
a.hide_flag as hideFlag,
count_send_notice as countSendNotice,
date_format(a.created_at, "%%Y-%%m-%%d") as reportDate,
date_format(a.updated_at, "%%Y-%%m-%%d") as updateDate,
CURRENT_TIMESTAMP as ETLDate
from matchedVideo as a, mddb.trackingWebsite as b
where date_format(a.created_at, "%%Y-%%m-%%d") > "%(min_reportDate)s"
and date_format(a.created_at, "%%Y-%%m-%%d") <= "%(max_reportDate)s"
and date_format(a.updated_at, "%%Y-%%m-%%d") <= "%(max_reportDate)s"
and a.company_id =14
and b.website_type = "ugc"
and a.created_at >= "2015-03-01"
union all
select
a.id as matchedVideo_id,
a.trackingMeta_id,
a.trackingWebsite_id,
date_format(a.first_send_notice_date, "%%Y-%%m-%%d") as firstSendNoticeDate,
a.hide_flag as hideFlag,
count_send_notice as countSendNotice,
date_format(a.created_at, "%%Y-%%m-%%d") as reportDate,
date_format(a.updated_at, "%%Y-%%m-%%d") as updateDate,
CURRENT_TIMESTAMP as ETLDate
from matchedVideo as a, mddb.trackingWebsite as b
where date_format(a.created_at, "%%Y-%%m-%%d") <= "%(min_reportDate)s"
and date_format(a.updated_at, "%%Y-%%m-%%d") > "%(min_updateDate)s"
and date_format(a.updated_at, "%%Y-%%m-%%d") <= "%(max_reportDate)s"
and a.company_id =14
and b.website_type = "ugc"
and a.created_at >= "2015-03-01"
""" %date_dict
vtweb_mysql = MySQLHelper(host=vt_host, user=vt_user,passwd=vt_passwd, port = vt_port, db_name = vt_db)
vtweb_mysql.queryCMD("set time_zone = '-8:00'")
data = vtweb_mysql.queryCMD(get_data_sql)
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd, \
db_name = target_db, port = target_port, charset = 'utf8')
insert_sql = """
insert into matchedVideo
(matchedVideo_id, trackingMeta_id, trackingWebsite_id, firstSendNoticeDate, hideFlag,
countSendNotice, reportDate, updateDate, ETLDate)
values(%s, %s, %s, %s, %s, %s, %s, %s, %s)
ON DUPLICATE KEY UPDATE firstSendNoticeDate = values(firstSendNoticeDate), hideFlag = values(hideFlag),
countSendNotice = values(countSendNotice), reportDate = values(reportDate),
updateDate = values(updateDate), ETLDate =values(ETLDate)
"""
target_mysql.executeManyCMD(insert_sql, data)
target_mysql.commit()
except Exception, e:
logger.debug("extral data from tracker2.matchedVideo, %s" %e)
sendToMe(subject = "titleBased_infringAllViews ERROR", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
def main(): sendToMe(subject = "titleBased_titleBased start", body = "aggregate data from TitleBased to TitleBased1 start") aggregateDataToTitleBased1() sendToMe(subject = "titleBased_titleBased end", body = "aggregate data from TitleBased to TitleBased1 end")
from titleBased import getConfMysqlInfo, getMinDatePara, commitInTurn
from sendMail import sendToMe
import re
logger = logging.getLogger("titleBased_titleBased")
logger.setLevel(logging.DEBUG)
log_file = '/Job/VIACOM/Dashboard/TitleBased/log/titleBased_titleBased.log'
filehandler = logging.handlers.RotatingFileHandler(filename=log_file, maxBytes=5*1024*1024, backupCount=10, mode='a')
formatter = logging.Formatter('%(asctime)s %(filename)s[line:%(lineno)d] %(levelname)s %(message)s')
filehandler.setFormatter(formatter)
logger.addHandler(filehandler)
cfg_file = "/Job/VIACOM/Dashboard/TitleBased/conf/viacom_dashboard.cfg"
if not os.path.exists(cfg_file):
logging.debug(": config file not exists")
sendToMe(subject = "titleBased_titleBased ERROR", body = "config file not exists")
sys.exit(0)
#################################################################################################################################
#target_server_section = "target_server_staging"
#target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
#################################################################################################################################
def aggregateDataToTitleBased1():
logger.info(" aggregate data from TitleBased to TitleBased1 start")
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
try:
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd,
db_name = target_db, port = target_port, charset = 'utf8')
aggregate_TitleBased1_SQL = """
select
a.reportDate,
def aggregateDataToTitleBased1():
logger.info(" aggregate data from TitleBased to TitleBased1 start")
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
try:
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd,
db_name = target_db, port = target_port, charset = 'utf8')
aggregate_TitleBased1_SQL = """
select
a.reportDate,
a.trackingWebsite_id,
a.websiteName,
a.websiteType,
ifnull(b.mapTitle, a.title) title,
min(a.tier) as tier,
sum(matchedNum) as matchedNum,
sum(matchedNumDurationNoZero) as matchedNumDurationNoZero,
sum(infringingNum) as infringingNum,
sum(infringingNumDurationNoZero) as infringingNumDurationNoZero,
sum(clipDurationSum) as clipDurationSum,
sum(clipDurationInfringingSum) as clipDurationInfringingSum,
current_timestamp as ETLDate
from
(select
a.reportDate,
a.trackingWebsite_id,
c.websiteName,
c.websiteType,
b.title,
min(a.tier) tier,
sum(matchedNum) as matchedNum,
sum(matchedNumDurationNoZero) as matchedNumDurationNoZero,
sum(infringingNum) as infringingNum,
sum(infringingNumDurationNoZero) as infringingNumDurationNoZero,
sum(clipDurationSum) as clipDurationSum,
sum(clipDurationInfringingSum) as clipDurationInfringingSum,
current_timestamp as ETLDate
from TitleBasedTmp as a, TitleBasedMeta as b, TitleBasedTrackingWebsite as c
where a.trackingWebsite_id = c.trackingWebsite_id
and a.trackingMeta_id = b.trackingMeta_id
group by 1, 2, 3, 4, 5) as a
left join MetaTitleMapTitle as b
on a. title = b.metaTitle
group by 1, 2, 3, 4, 5
""" #%date_para_TitleBased1_dict
TitleBased1_result = target_mysql.queryCMD(aggregate_TitleBased1_SQL)
insert_TitleBased1_SQL = """
INSERT INTO TitleBased1
(reportDate, trackingWebsite_id, websiteName, websiteType, title, tier,
matchedNum, matchedNumDurationNoZero, infringingNum, infringingNumDurationNoZero,
clipDurationSum, clipDurationInfringingSum, ETLDate)
VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s)
ON DUPLICATE KEY UPDATE matchedNum = VALUES(matchedNum), clipDurationSum = VALUES(clipDurationSum),
infringingNum = VALUES(infringingNum), ETLDate = VALUES(ETLDate), matchedNumDurationNoZero = VALUES(matchedNumDurationNoZero),
matchedNumDurationNoZero = VALUES(matchedNumDurationNoZero), infringingNumDurationNoZero = VALUES(infringingNumDurationNoZero),
clipDurationInfringingSum = VALUES(clipDurationInfringingSum)
"""
commitInTurn(commit_num = 50000, data = TitleBased1_result, executeFun = target_mysql.insertUpdateCMD, \
commitFun = target_mysql.commit, executeSQL = insert_TitleBased1_SQL)
#target_mysql.insertUpdateCMD(insert_TitleBased1_SQL, TitleBased1_result)
#target_mysql.commit()
except Exception, e:
logger.debug(": load data to TitleBased1, %s" %e)
sendToMe(subject = "titleBased_titleBased ERROR", body = re.sub(r'\'|"|!', "", str(e)))
sys.exit(0)
filehandler = logging.handlers.RotatingFileHandler(filename=log_file, maxBytes=5*1024*1024, backupCount=10, mode='a')
formatter = logging.Formatter('%(asctime)s %(filename)s[line:%(lineno)d] %(levelname)s %(message)s')
filehandler.setFormatter(formatter)
logger.addHandler(filehandler)
cfg_file = "/Job/VIACOM/Dashboard/TitleBased/conf/viacom_dashboard.cfg"
if not os.path.exists(cfg_file):
logging.debug(": config file not exists")
sys.exit(0)
#################################################################################################################################
target_server_section = "target_server_staging"
target_host, target_user, target_passwd, target_port, target_db= getConfMysqlInfo(target_server_section)
#################################################################################################################################
#--------------------------------------------------------------------------------------------------------------------------------
logger.info(" aggregate data from TitleBased1 to SiteBased start")
sendToMe(subject = "SiteBased start", body = "aggregate data from TitleBased1 to SiteBased start")
try:
target_mysql = MySQLHelper(host=target_host, user=target_user, passwd=target_passwd,
db_name = target_db, port = target_port, charset = 'utf8')
aggregate_TitleBased1_SQL = """
select
a.reportDate,
a.trackingWebsite_id,
b.websiteName,
b.websiteType,
b.websiteDomain,
b.country_id,
c.countryName as hostCountry,
a.title,
sum(matchedNum) as matchedNum,
sum(matchedNumDurationNoZero) as matchedNumDurationNoZero,
