def save(params, chunk, chunks_info):
global latest_collection_name, history_collection_name
data = File.readChunkData(chunk)
dbManager = SharedMemoryManager.getInstance()
db = dbManager.query()
current_index = fn.getNestedElement(chunks_info, 'current', 0)
total_index = fn.getNestedElement(chunks_info, 'total', len(data))
date = fn.getNestedElement(params, 'date')
datetime = DateTime.convertDateTimeFromString(date)
total_length = len(data)
queue_info = chunks_info['queue']
# Logger.v('Running Index:', chunks_info['queue']['running']);
chunks_info['queue']['current'] += 1
# Logger.v('Saving from... {0}/{1}, current package: {2}'.format(current_index, total_index, total_length) );
fn.printProgressBar(queue_info['current'], queue_info['total'],
'Processing Chunk Insertion')
for idx in range(0, total_length):
# insert stock_latest
row = data[idx]
obj_ = transformToLowercase(data=row, datetime=datetime)
ModelStockIntegrity.update(data=obj_)
dbManager.addBulkInsert(latest_collection_name, obj_, batch=True)
# dbManager.addBulkInsert(history_collection_name, obj_, batch=True); # temporary off (need 7 day data only)
# insert items
# d = data[idx];
ModelItem.saveItem(row)
# fn.printProgressBar(current_index+idx, total_index, 'Processing Item Insertion');
#ensure all data is save properly
# dbManager.executeBulkOperations(history_collection_name); # temporary off (need 7 day data only)
dbManager.executeBulkOperations(latest_collection_name)
return chunks_info
def getMonthRange(params):
start_month = fn.getNestedElement(params, 'start_month');
number_of_month = fn.getNestedElement(params, 'number_of_month', 1);
month_range = [start_month];
new_month = '{0}-01'.format(start_month);
for month_count in range(0, number_of_month - 1): # included start_month, so total month less 1
new_month = DateTime.getNextMonth(DateTime.convertDateTimeFromString(new_month));
year_month = DateTime.getDateCategoryName(new_month, element='year_month_digit');
month_range.append(year_month);
return month_range;
def getCollectionName(params):
global latest_collection_name
latest_collection_name = 'stock_latest'
# set default;
dbManager = SharedMemoryManager.getInstance()
db = dbManager.query()
data = list(db[latest_collection_name].find({}, {
'_id': 0,
'date': 1
}).limit(1))
if data:
latest_date_string = DateTime.toString(data[0]['date'])
latest_date = DateTime.convertDateTimeFromString(latest_date_string)
date_string = fn.getNestedElement(params, 'date', None)
if date_string:
date = DateTime.convertDateTimeFromString(date_string)
different = latest_date - date
day_diff = math.floor(different.total_seconds() / float(86400))
if day_diff > 0:
latest_collection_name = 'stock_{0}'.format(day_diff)
def getMissingDates(data):
dbManager = SharedMemoryManager.getInstance();
db = dbManager.query();
missing_dates = {};
today = DateTime.now(tzinfo=msia_tz); # date only
state_by = 'state_code';
states = list(db['state'].find({},{'_id': 0, state_by: 1}));
current_year = DateTime.getDateCategoryName(date=DateTime.now(tzinfo=msia_tz), element='year');
for rk in data:
row = data[rk];
if rk not in missing_dates:
missing_dates[rk] = [];
dates = groupDates(params={'states': states, 'state_by': state_by}, data=row);
for date in dates['missing']:
end_date_of_month = DateTime.getDaysAgo(days_to_crawl=1, datefrom=DateTime.getNextMonth(DateTime.convertDateTimeFromString(date)));
day_diff = DateTime.getDifferenceBetweenDuration([today, end_date_of_month]);
if day_diff >= 0:
date_str = DateTime.toString(today);
else:
date_str = DateTime.toString(end_date_of_month);
if date_str not in dates['crawled']:
missing_dates[rk].append(date_str);
# Logger.v('day_diff', day_diff);
# Logger.v('date', DateTime.getDaysAgo(days_to_crawl=1, datefrom=DateTime.getNextMonth(DateTime.convertDateTimeFromString(ed))));
missing_dates[rk] = sorted(list(set(missing_dates[rk])), reverse=True);
return missing_dates;
def generateTemplate(params):
result = {};
report_keys = fn.getNestedElement(params, 'keys.report', ['procurement', 'budget']);
first_date = fn.getNestedElement(params, 'first_date');
last_date = fn.getNestedElement(params, 'last_date');
state_by = fn.getNestedElement(params, 'state_by');
states = fn.getNestedElement(params, 'states');
today = DateTime.now(tzinfo=msia_tz); # date only
for rk in report_keys:
if rk not in result:
result[rk] = {};
for date in DateTime.getBetween([first_date, last_date], element='date')['order']:
end_date_of_month = DateTime.getDaysAgo(days_to_crawl=1, datefrom=DateTime.getNextMonth(DateTime.convertDateTimeFromString(date)));
year_month = date[:7];
day_diff = DateTime.getDifferenceBetweenDuration([today, end_date_of_month]);
if day_diff >= 0:
date_str = DateTime.toString(today);
else:
date_str = DateTime.toString(end_date_of_month);
if date_str not in result[rk]:
result[rk][date_str] = {};
result[rk][date_str].update({
'date': date_str,
})
for idx in range(0, len(states)):
state = states[idx][state_by];
result[rk][date_str].update({
state: 0,
});
return result;
def generateCrawlParam(params):
Debug = DebugManager.DebugManager()
Debug.start()
global pass_month_quantity
dbManager = SharedMemoryManager.getInstance()
db = dbManager.query()
crawl_params = {}
limit_for_test = 10
report_keys = fn.getNestedElement(params, 'keys.report',
['budget', 'procurement'])
interval = fn.getNestedElement(params, 'interval', 1)
filter_facility_code = fn.getNestedElement(params, 'filter.facility_code',
True)
check_empty = fn.getNestedElement(params, 'schedule_params.check_empty',
False)
today = fn.getNestedElement(
params, 'schedule_params.today',
DateTime.toString(DateTime.now(tzinfo=msia_tz)))
# Logger.v('filter_facility_code', filter_facility_code);
if check_empty:
# past_dates = DateTime.getPastDate(count=pass_month_quantity, duration=interval);
past_dates = DateTime.getPastDate(
count=pass_month_quantity,
duration=interval,
end=DateTime.convertDateTimeFromString(today))
# Logger.v('past_dates', past_dates);
# exit();
else:
past_dates = DateTime.getPastDate(count=pass_month_quantity,
duration=interval)
# Logger.v('past_dates', past_dates);
state_codes = retrieveOption(collection_name='state',
show_keys=['state_code'],
hide_keys=['_id'])
state_code = extractListByKey(data=state_codes, key='state_code')
facility_codes = retrieveOption(collection_name='facility',
show_keys=['facility_code'],
hide_keys=['_id'])
facility_code = extractListByKey(data=facility_codes, key='facility_code')
for key in report_keys:
# Logger.v('collection', key, past_dates[0]);
Debug.trace()
if key not in crawl_params:
crawl_params[key] = []
mongo_data = list(db[key].find({}, {}))
if len(mongo_data) == 0:
dates = past_dates[0][:]
else:
dates = past_dates[0][:1]
year = extractYear(data=dates)
# Logger.v('year', year);
# Logger.v('filter_facility_code', filter_facility_code);
if key == 'budget':
if not filter_facility_code:
iteration = 0
total = len(year) * len(state_code)
# fn.printProgressBar(iteration=iteration, total=total);
for y in year:
for sc in state_code:
obj_ = {
'financial_year': y,
'state_code': sc,
'page_type': key,
'upid': '_'.join([sc, y]),
'url': api_links[key].format(sc, y, ''),
'start_date': today,
'end_date': today,
}
if obj_ not in crawl_params[key]:
crawl_params[key].append(obj_)
# Logger.v('len(crawl_param])', len(crawl_params[key]));
iteration += 1
# fn.printProgressBar(iteration=iteration, total=total);
else:
iteration = 0
total = len(year) * len(state_code) * len(
facility_code[:limit_for_test])
# fn.printProgressBar(iteration=iteration, total=total);
for y in year:
for sc in state_code:
for fc in facility_code[:limit_for_test]:
obj_ = {
'financial_year': y,
'state_code': sc,
'page_type': key,
'upid': '_'.join([sc, y, fc]),
'facility_code': fc,
'url': api_links[key].format(sc, y, fc),
'start_date': today,
'end_date': today,
}
if obj_ not in crawl_params[key]:
crawl_params[key].append(obj_)
# Logger.v('len(crawl_param])', len(crawl_params[key]));
iteration += 1
# fn.printProgressBar(iteration=iteration, total=total);
elif key == 'procurement':
if not filter_facility_code:
for past_duration in dates:
start_date = DateTime.toString(
DateTime.getDaysAgo(days_to_crawl=-1,
datefrom=past_duration[0]))
end_date = DateTime.toString(
DateTime.getDaysAgo(days_to_crawl=1,
datefrom=past_duration[1]))
for sc in state_code:
obj_ = {
'state_code':
sc,
'start_date':
start_date,
'end_date':
end_date,
'page_type':
key,
'upid':
'_'.join([sc, start_date, end_date]),
'url':
api_links[key].format(sc,
start_date.replace('-', ''),
end_date.replace('-', ''),
''),
}
if obj_ not in crawl_params[key]:
crawl_params[key].append(obj_)
# Logger.v('len(crawl_param])', len(crawl_params[key]));
else:
for past_duration in dates:
start_date = DateTime.toString(
DateTime.getDaysAgo(days_to_crawl=-1,
datefrom=past_duration[0]))
end_date = DateTime.toString(
DateTime.getDaysAgo(days_to_crawl=1,
datefrom=past_duration[1]))
for sc in state_code:
for fc in facility_code[:limit_for_test]:
obj_ = {
'state_code':
sc,
'start_date':
start_date,
'end_date':
end_date,
'page_type':
key,
'facility_code':
fc,
'upid':
'_'.join([sc, start_date, end_date, fc]),
'url':
api_links[key].format(
sc, start_date.replace('-', ''),
end_date.replace('-', ''), fc)
}
if obj_ not in crawl_params[key]:
crawl_params[key].append(obj_)
# Logger.v('len(crawl_param])', len(crawl_params[key]));
for c in crawl_params:
# Logger.v('crawl_params', c, len(crawl_params[c]));
fn.writeExcelFile(filename='{0}/{1}'.format(test_folder, c),
data=crawl_params[c])
Logger.v('crawl_params', len(crawl_params))
Debug.show('Generate Crawl Params')
return crawl_params