def bidding_info_service(announce_id, url):
participator = None
if announce_id:
participator = sc_crawler_query_all(company_bidding_info_by_id_query,
(announce_id, ))
elif url:
participator = sc_crawler_query_all(company_bidding_info_by_url_query,
(url, ))
if participator:
participator_object = collect_participator(participator)
result = {
"announceID": participator[0]["announceID"],
"title": participator[0]["title"],
"url": participator[0]["url"],
"sourceKey":
get_content_source_key(participator[0]["contentSource"]),
"website": participator[0]["website"],
"publishedDateTime": participator[0]["publishedDateTime"],
"winningDate": participator[0]["winningDate"],
"announceType": participator[0]["announceType"],
"participatorObject": participator_object,
"participator": participator_to_string(participator_object)
}
return result
return None
def bidding_statistic_by_company_service(company_name, begin, end):
parameters = {
"company_name": company_name,
"start_date": begin,
"end_date": end
}
distribute = sc_crawler_query_all(
bidding_result_statistic_by_company_query, parameters)
statistic = {
"bidCount": 0,
"bidMoneyAmount": 0,
"winCount": 0,
"winMoneyAmount": 0
}
for item in distribute:
for key in statistic.keys():
statistic[key] += item[key] if item[key] else 0
distribute.append(statistic)
distribute = _format_statistic_by_company_result_data(distribute)
distribute, statistic = distribute[:-1], distribute[-1]
# 如果 statistic 中总投标次数为0时，返回None
if statistic.get("bidCount") == 0:
result = None
else:
result = {"provinceDistribution": distribute, "statistic": statistic}
return result
def bidding_statistic_by_company_trend_service(company_name, base):
parameters = {"company_name": company_name}
trend_data = sc_crawler_query_all(
bidding_result_statistic_by_company_trend_query, parameters)
trend = {}
for item in trend_data:
year_month = item.pop("yearMonth")
month = int(year_month[-2:])
month = int((month - 1) / base) * base + 1
year_month = year_month[:-2] + "%02d" % month
if year_month not in trend:
trend[year_month] = []
trend[year_month].append(item)
result = {}
for year_month, distribute in trend.items():
statistic = {
"bidCount": 0,
"bidMoneyAmount": 0,
"winCount": 0,
"winMoneyAmount": 0
}
distribute_collect = {}
for item in distribute:
province = item.get("province")
if province not in distribute_collect:
distribute_collect[province] = item
for key in statistic.keys():
distribute_collect[province][
key] = item[key] if item[key] else 0
else:
for key in statistic.keys():
distribute_collect[province][
key] += item[key] if item[key] else 0
distribute = distribute_collect.values()
for item in distribute:
for key in statistic.keys():
statistic[key] += item[key] if item[key] else 0
distribute.append(statistic)
distribute = _format_statistic_by_company_result_data(distribute)
distribute, statistic = distribute[:-1], distribute[-1]
result[year_month] = {
"provinceDistribution": distribute,
"statistic": statistic
}
return result
def bidding_company_info_service(company_name, st_datetime, ed_datetime):
items = sc_crawler_query_all(company_bidding_info_query,
(company_name, st_datetime, ed_datetime))
coding_filter(items)
announce_id, participator, result, results = None, [], {}, []
for item in items:
if announce_id != item["announceID"]:
announce_id = item["announceID"]
if participator and result:
result["participatorObject"] = collect_participator(
participator)
result["participator"] = participator_to_string(
result["participatorObject"])
results.append(result)
participator, result = [], {}
if not result:
result["announceID"] = item["announceID"]
result["title"] = item["title"]
result["url"] = item["url"]
result["sourceKey"] = get_content_source_key(item["contentSource"])
result["website"] = item["website"]
result["publishedDateTime"] = item["publishedDateTime"]
result["winningDate"] = item["winningDate"]
result["announceType"] = item["announceType"]
part = {
"resultID": item["resultID"],
"winningCompany": item["winningCompany"],
"winningAmount": item["winningAmount"],
"unit": item["unit"],
"currency": item["currency"],
"roleName": item["roleName"],
"candidateRank": item["candidateRank"],
"segmentName": item["segmentName"]
}
participator.append(part)
if participator and result:
result["participatorObject"] = collect_participator(participator)
result["participator"] = participator_to_string(
result["participatorObject"])
# result["participator"] = participator_to_string(participator)
results.append(result)
return results
def bidding_statistic_service(op,
begin,
end,
project_type=None,
province=None,
city=None,
county=None,
top=10,
page=0,
page_size=10):
parameters = {
"start_date": begin,
"end_date": end,
"project_type": project_type,
"skip_n": page * page_size,
"province": province,
"city": city,
"county": county,
"page_size": page_size,
"top_n": top
}
sc_crawler_non_query(set_work_mem_32m_non_query)
if op == "amountCountRegion":
if not province:
query = get_statistic_query(
nation_amount_count_base_on_province_query,
project_type,
province,
city=None,
county=None)
result = sc_crawler_query_all(query, parameters)
elif province in (u"北京市", u"上海市", u"天津市", u"重庆市"):
query = get_statistic_query(
province_amount_count_base_on_county_query,
project_type,
province,
city=None,
county=None)
result = sc_crawler_query_all(query, parameters)
region, default_region_name = u"county", u"市级"
t = [i for i in range(len(result)) if result[i][region] is None]
for i in t:
result[i][region] = default_region_name
for i, it in enumerate(result):
result[i]["city"] = it["county"]
del result[i]["county"]
else:
query = get_statistic_query(
province_amount_count_base_on_city_query,
project_type,
province,
city=None,
county=None)
result = sc_crawler_query_all(query, parameters)
region, default_region_name = u"city", u"省级"
t = [i for i in range(len(result)) if result[i][region] is None]
for i in t:
result[i][region] = default_region_name
for i, it in enumerate(result):
for reg in ["province", "city", "county"]:
if reg in it:
result[i]["shortName"] = get_keyword(it[reg])
break
return transform_amount(result)
elif op == "amountCountDate":
query = get_statistic_query(amount_count_base_on_date_query,
project_type, province, city, county)
result = sc_crawler_query_all(query, parameters)
return transform_amount(result)
elif op == "topRole":
query = get_statistic_query(top_n_amount_base_on_purchaser_query,
project_type, province, city, county)
top_pur = transform_amount(sc_crawler_query_all(query, parameters))
query = get_statistic_query(top_n_amount_base_on_agent_query,
project_type, province, city, county)
top_agt = transform_amount(sc_crawler_query_all(query, parameters))
query = get_statistic_query(top_n_amount_base_on_winner_query,
project_type, province, city, county)
top_win = transform_amount(sc_crawler_query_all(query, parameters))
if len(top_pur) > 0 or len(top_agt) or len(top_win) > 0:
result = {
"topPurchasers": top_pur,
"topAgents": top_agt,
"topWinners": top_win
}
else:
result = None
return result
elif op == "winnerInfo":
query = get_statistic_query(winner_info_query, project_type, province,
city, county)
result = sc_crawler_query_all(query, parameters)
for i, item in enumerate(result):
if province in (u"北京市", u"上海市", u"天津市", u"重庆市"):
result[i]["city"] = item["county"]
del result[i]["county"]
return transform_amount(result)
elif op == "winnerInfoCount":
query = get_statistic_query(winner_info_count_query, project_type,
province, city, county)
result = sc_crawler_query_one(query, parameters)
return result
else:
return []