def build_default_item_curs(self,
line,
unit_str,
date_frmt=None,
no_curses=[],
comp_date=True):
"""
Allow to build a default item easily
Take in param the unit and the date format (which allow us to convert in datetime format)
"""
item = IOItem()
row_datetime = self._extract_date(self.rowl_at(line, 'date'),
date_frmt)
if comp_date:
if self.start_date is not None and row_datetime < self.start_date:
return None
item['unit'] = unit_str
for key, value in self.row.items():
take_data = True
for no_curs in no_curses:
if no_curs == key:
take_data = False
if take_data:
item[key] = self.rowl_at(line, key)
item['src_file'] = self.url
item['date'] = create_str_from_time(row_datetime)
return item
def parse_row(self, line):
item = IOItem()
row_datetime = xldate_to_datetime(self.rowl_at(line, 'date'),
self.book.datemode)
if self.start_date is not None and row_datetime < self.start_date:
return None
item['unit'] = self.unit_str
item['date'] = create_str_from_time(row_datetime)
item['level_o'] = str_to_float(self.rowl_at(line, 'level_o'))
item['input_o'] = self.rowl_at(line, 'input_o')
item['output_o'] = self.rowl_at(line, 'output_o')
item['src_file'] = self.url
return item
def process_item(raw_item):
"""Map and normalize raw_item into a usable event.
Args:
raw_item (dict[str, str]):
Returns:
Dict:
"""
item = map_keys(raw_item, portcall_mapping(), skip_missing=True)
# build vessel sub model
item['vessel'] = {
'name': item.pop('vessel_name'),
'length': item.pop('vessel_loa', None),
'beam': item.pop('vessel_beam', None),
}
# build cargo sub model
item['cargoes'] = list(
normalize_cargo(item.pop('movement', None), item.pop('volume', None),
item.pop('product', None)))
if not item['cargoes']:
return
# if there's eta time
if item.get('eta_time'):
eta_time = item.pop('eta_time')
hour, minute = eta_time.split(':')
item['eta'] = create_str_from_time(
may_parse_date_str(item['eta'],
ISO8601_FORMAT).replace(hour=int(hour),
minute=int(minute)),
ISO8601_FORMAT,
)
# berth
berth_from = item.pop('berth_from', None)
berth_to = item.pop('berth_to', None)
item['berth'] = berth_from or berth_to or item.get('berth', None)
return item
def parse_simple_page(self, response):
table = response.xpath('//table[@class="flux"]//tr')
# Skip header
for row in table[1:-1]:
item = IOItem(unit=DEFAULT_UNIT)
extracted_row = row.xpath('td/text()').extract()
io_datetime = may_parse_date_str(extracted_row[0], DATE_FORMAT)
item['date'] = create_str_from_time(io_datetime +
timedelta(hours=18))
lvl = extracted_row[1].replace(' ', '')
if lvl != EMPTY_VALUE:
item['level_o'] = self.level_o_check(lvl, io_datetime)
else:
# If the stock level is not set,
# scrapping output is meaningless as website display 0
continue
# Scraping Nominated quantities as output_forcast and
# Allocated quan. as output (SENDOUT)
out = extracted_row[3].replace(' ', '')
if out != EMPTY_VALUE:
item['output_o'] = out
out_forecast = extracted_row[2].replace(' ', '')
if out_forecast != EMPTY_VALUE:
item['output_forecast_o'] = out_forecast
# if '-', take the output_forecast value into output
if item.get('output_o') is None:
item['output_o'] = item.get('output_forecast_o')
yield item
pagination_bloc = response.xpath(
'//div[@class="pagination-bloc"]/a/text()').extract()
if 'Suivante >' in pagination_bloc:
self.page += 1
yield self._build_request()
def parse_row(self, line):
item = IOItem()
(month, year) = self.get_month_from_itstr(self.sheet.name)
day = self.rowl_at(line, 'date')
date_str = str(year) + '-' + str(month) + '-' + str(
int(day)) + ' 00:00:00'
row_datetime = may_parse_date_str(date_str) + timedelta(hours=6)
if self.start_date is not None and row_datetime < self.start_date:
return None
item['date'] = create_str_from_time(row_datetime)
item['unit'] = 'KWH'
item['level_o'] = str_to_float(self.rowl_at(line, 'level_o'))
item['output_o'] = self.rowl_at(line, 'output')
item['src_file'] = self.url
cargo_val = str_to_float(self.rowl_at(line, 'boat_io'))
if cargo_val is not None:
item['input_cargo'] = cargo_val
else:
item['input_cargo'] = 0
return item
def test_create_date_with_custom_format(self):
whatever_date = datetime(2016, 10, 1)
some_fmt = '%Y-%m-%d'
str_date = create_str_from_time(whatever_date, some_fmt)
self.assertEqual(str_date, '2016-10-01')
def test_create_date_with_default_format(self):
whatever_date = datetime(2016, 10, 1, 16, 23, 4)
str_date = create_str_from_time(whatever_date)
self.assertEqual(str_date, '2016-10-01 16:23:04')
def get_last_spider_exec_strfmt(self, frmt, day_diff=2):
time = self.get_last_spider_exec(day_diff=day_diff)
if time is None:
return None
return create_str_from_time(time, format=frmt)
