def parse_sheet_rows(self, sheet, post_process=lambda x: x):
"""Parse and process rows of a spreadsheet.
Args:
sheet (xlrd.Sheet):
post_process (Callable[Any, List[str]]):
Yields:
Dict[str, str]:
"""
for idx, row in enumerate(sheet.get_rows()):
row = [
xldate_to_datetime(cell.value, sheet.book.datemode).isoformat()
if is_xldate(cell)
else may_strip(cell.value)
for cell in row
]
# headers will always be the first row of any sheet from this source
if idx == 0:
header = row
continue
# post-process row according to caller specifications
row = post_process(row)
if row:
yield {head: row[head_idx] for head_idx, head in enumerate(header)}
def parse_attachment(
self, attachment_doc, reported_date, sheet_list, get_header, pn_row, pn, doc_name
):
for sheet in xlrd.open_workbook(file_contents=attachment_doc, on_demand=True).sheets():
# only parse relevant sheets
if any(sub in sheet.name.lower() for sub in ['sheet1', 'new']) or not any(
sub in sheet.name.lower() for sub in sheet_list
):
continue
first_row, second_row = None, None
port_name = pn if pn else pn_row
for idx, raw_row in enumerate(sheet.get_rows()):
row = []
# to handle is xldate exception
for cell in raw_row:
if is_xldate(cell):
try:
cell = xldate_to_datetime(cell.value, sheet.book.datemode).isoformat()
except Exception:
cell = str(cell.value)
else:
cell = str(cell.value)
row.append(cell)
# get port name
if isinstance(port_name, int) and idx == pn_row:
port_name = row[1]
# format headers because it is split into 2, make it adaptable
# across sheets instead of hardcoding
if idx in get_header:
if idx == get_header[0]:
first_row = row
continue
if idx == get_header[1]:
second_row = row
continue
if first_row and second_row:
headers = self.combine_rows(first_row, second_row)
# only process relevant rows after header row
if idx > get_header[1]:
raw_item = {may_strip(head): row[idx] for idx, head in enumerate(headers)}
# contextualise raw item with meta info
raw_item.update(
reported_date=reported_date,
provider_name=self.provider,
port_name=port_name,
file_name=doc_name,
)
yield normalize.process_item(raw_item)
def parse_mail(self, mail):
"""The method will be called for every mail the search_term matched.
Args:
mail (Mail):
Yields:
Dict[str, str]:
"""
for attachment in mail.attachments():
for sheet in xlrd.open_workbook(file_contents=attachment.body, on_demand=True).sheets():
# only process the sheet we want
if not any(sub in sheet.name.lower() for sub in ('export', 'import')):
continue
sheet_name = sheet.name.lower()
for idx, raw_row in enumerate(sheet.get_rows()):
# Include handling of xlrd.xldate.XLDateAmbiguous cases
row = []
for cell in raw_row:
if is_xldate(cell):
try:
cell = xldate_to_datetime(
cell.value, sheet.book.datemode
).isoformat()
except Exception:
cell = str(cell.value)
else:
cell = str(cell.value)
row.append(cell)
if idx == 0:
reported_date = to_isoformat(row[1], dayfirst=True)
continue
# second row will always contain header; extract it
if idx == 1:
header = row
continue
# extract data row
raw_item = {head: row[head_idx] for head_idx, head in enumerate(header)}
raw_item.update(
reported_date=reported_date,
provider_name=self.provider,
sheet_name=sheet_name,
)
if DataTypes.SpotCharter in self.produces:
print(raw_item)
yield normalize_charters.process_item(raw_item)
else:
yield normalize_grades.process_item(raw_item)
def parse_mail(self, mail):
"""Extract mail found with specified email filters in spider arguments.
Args:
mail (Mail):
Yields:
Dict[str, str]:
"""
for attachment in mail.attachments():
# memoise reported date so it won't need to be called repeatedly later
_match = re.match(r'.*DBE\s(\d+)', attachment.name)
if not _match:
raise ValueError(f'Unknown reported date format: {attachment.name}')
reported_date = _match.group(1)
# each xlsx file by this provider will only have one sheet
sheet = xlrd.open_workbook(
file_contents=attachment.body, on_demand=True
).sheet_by_index(0)
# extract raw data from sheet
for idx, row in enumerate(sheet.get_rows()):
row = [
xldate_to_datetime(cell.value, sheet.book.datemode).isoformat()
if is_xldate(cell)
else str(cell.value)
for cell in row
]
# initialise headers
if idx == HEADER_ROW_IDX:
header = row
continue
# skip vessels with () inside, i.e., vessel has not been identified by provider
if '(' in row[VESSEL_COLUMN_INDEX]:
continue
# FIXME timecharter type
if 'timecharter' in row[COMMODITY_COLUMN_INDEX].lower():
continue
raw_item = {head: row[idx] for idx, head in enumerate(header)}
# contextualise raw item with metadata
raw_item.update(
provider_name='MRI', reported_date=reported_date, spider_name=self.name
)
if DataTypes.SpotCharter in self.produces:
yield normalize_charters.process_item(raw_item)
# FIXME supposed to be `DataTypes.PortCall` here, but we don't want
# data-dispatcher to consume data from these spiders and the ETL to create PCs
else:
yield normalize_grades.process_item(raw_item)
def parse_mail(self, mail):
"""This method will be called for every mail the search_term matched.
Each vessel movement has an associated uuid that is linked to a cargo's uuid,
allowing for easy retrieval of a vessel's cargo movement.
However, each vessel may contain multiple cargo movements with the same uuid,
therefore we store the products as a list value against the uuid key.
Args:
mail (Mail):
Yields:
Dict[str, str]:
"""
# memoise reported date so it does not need to be called repeatedly later
self.reported_date = to_isoformat(mail.envelope['date'])
for attachment in mail.attachments():
# sanity check in case file is not a spreadsheet
if not attachment.is_spreadsheet:
continue
sheet = xlrd.open_workbook(file_contents=attachment.body,
on_demand=True).sheet_by_index(0)
header = None
for idx, row in enumerate(sheet.get_rows()):
row = [
xldate_to_datetime(cell.value,
sheet.book.datemode).isoformat()
if is_xldate(cell) else cell.value
for cell_idx, cell in enumerate(row)
]
if 'approach' in row[3]:
# sheet has 2 tables with different headers,
# first table has header split into 2
header = row
header[2] = 'Vessel'
header[5] = 'Cargo'
continue
if 'Status' in row[1]:
header = row
continue
if header and len(header) == len(row):
raw_item = {h: row[idx] for idx, h in enumerate(header)}
raw_item.update(provider_name=self.provider,
reported_date=self.reported_date)
yield normalize.process_item(raw_item)
def parse_file(self, response):
sheet = xlrd.open_workbook(file_contents=response.body,
on_demand=True).sheet_by_index(0)
headers = None
do_process = False
for idx, raw_row in enumerate(sheet.get_rows()):
row = [
xldate_to_datetime(cell.value,
sheet.book.datemode).isoformat()
if is_xldate(cell) else may_strip(str(cell.value))
for cell in raw_row
][:MAX_ROW_LENGTH]
# filtering unwanted portions for easier processing of remarks field (to retrieve
# cargo information)
# this hold true when the field names don't change
# if any(row[0] == alias for alias in NO_PROCESSING):
# processing = False
if row and any(sub in row[0] for sub in RESUME_PROCESSING):
do_process = True
if row and any(sub in row[0] for sub in PAUSE_PROCESSING):
do_process = False
if not do_process:
continue
# vessels expected table
if HEADER_SIGN in row:
headers = row
continue
if headers and row[ETA_COL_IDX]:
raw_item = {
headers[cell_idx]: cell
for cell_idx, cell in enumerate(row)
}
raw_item.update(self.meta_field)
yield normalize.process_item(raw_item)
# waiters info in text
waiters = self.parse_waiters(row[0])
if waiters:
raw_item = {
str(cell_idx): cell
for cell_idx, cell in enumerate(waiters)
}
raw_item.update(self.meta_field)
yield normalize.process_item(raw_item)
def parse_mail(self, mail):
"""This method will be called for every mail the search_term matched.
Each vessel movement has an associated uuid that is linked to a cargo's uuid,
allowing for easy retrieval of a vessel's cargo movement.
However, each vessel may contain multiple cargo movements with the same uuid,
therefore we store the products as a list value against the uuid key.
Args:
mail (Mail):
Yields:
Dict[str, str]:
"""
# memoise reported date so it does not need to be called repeatedly later
self.reported_date = to_isoformat(mail.envelope['date'])
for attachment in mail.attachments():
# sanity check in case file is not a spreadsheet
if not attachment.is_spreadsheet:
continue
sheet = xlrd.open_workbook(file_contents=attachment.body,
on_demand=True).sheet_by_index(0)
header = None
for idx, row in enumerate(sheet.get_rows()):
# report seperates date and timestamp into 2 cells,
# this would cause the cell to try to convert the timestamps into a date
# hence throwing an error (index affected 2,5,9)
row = [
xldate_to_datetime(cell.value,
sheet.book.datemode).isoformat()
if is_xldate(cell) and cell_idx not in (2, 5, 9) else
cell.value for cell_idx, cell in enumerate(row)
]
if idx == 0:
header = row
continue
if header:
raw_item = {h: row[idx] for idx, h in enumerate(header)}
raw_item.update(provider_name='Interadria',
reported_date=self.reported_date)
yield normalize.process_item(raw_item)
else:
self.logger.warning('No headers are found.')
def parse_mail(self, mail):
"""Extract data from each mail matched by the query spider argument.
Args:
mail (Mail):
Yields:
Dict[str, str]:
"""
# memoise reported_date so it won't need to be called repeatedly later
reported_date = to_isoformat(mail.envelope['date'])
for attachment in mail.attachments():
sheet = xlrd.open_workbook(
file_contents=attachment.body, on_demand=True
).sheet_by_index(0)
# store state of the table, in order to get relevant rows to extract
for idx, row in enumerate(sheet.get_rows()):
row = [
xldate_to_datetime(cell.value, sheet.book.datemode).isoformat()
if is_xldate(cell)
else cell.value
for cell in row
]
# ignore unnecessary rows
if idx < 3:
continue
# locate port row
if idx == 3:
port = row[0].lower().replace('port: ', '')
continue
# locate header row
if idx == 4:
header = row
continue
raw_item = {head: row[idx] for idx, head in enumerate(header)}
# contextualise raw item with meta info
raw_item.update(
reported_date=reported_date,
provider_name=self.provider,
port_name=port,
attachment_name=attachment.name,
)
yield from normalize.process_item(raw_item)
def parse_attachment(self, attachment_doc, reported_date):
# each xlsx file by this provider will only have one sheet
sheet = xlrd.open_workbook(file_contents=attachment_doc, on_demand=True).sheet_by_name(
'Market Data Report'
)
# for this spot charter we don't look at cargo information (already saved
# in corresponding cargo movement)
# when several rows for a vessel with same charter values,port values and
# arrival values we only build one item
spot_charters_seen = {}
for idx, row in enumerate(sheet.get_rows()):
row = [
xldate_to_datetime(cell.value, sheet.book.datemode).isoformat()
if is_xldate(cell)
else str(cell.value)
for cell in row
]
# remove empty rows before and after the main data table
# denoted by a falsy value
if not row[EMPTY_ROW_INDICATOR_INDEX]:
continue
# initialise headers
if HEADER_PATTERN in row:
header = row
continue
# extract data row
raw_item = {head: row[head_idx] for head_idx, head in enumerate(header)}
raw_item.update(reported_date=reported_date, provider_name=self.provider)
if DataTypes.SpotCharter in self.produces:
if (
raw_item['Charterer Name']
+ raw_item['Vessel']
+ raw_item['Arrival'].split(' ')[0]
+ raw_item['Port Name']
in spot_charters_seen.keys()
):
continue
spot_charters_seen[
raw_item['Charterer Name']
+ raw_item['Vessel']
+ raw_item['Arrival'].split(' ')[0]
+ raw_item['Port Name']
] = True
yield normalize_charters.process_item(raw_item)
elif DataTypes.Cargo in self.produces:
yield normalize_grades.process_item(raw_item)
def parse_mail(self, mail):
"""Extract data from each mail matched by the query spider argument.
Args:
mail (Mail):
Yields:
Dict[str, str]:
"""
# memoise reported_date so it won't need to be called repeatedly later
reported_date = to_isoformat(mail.envelope['date'])
for attachment in mail.attachments():
# only one sheet in the excel file is relevant
sheet = xlrd.open_workbook(
file_contents=attachment.body,
on_demand=True).sheet_by_index(RELEVANT_SHEET_INDEX)
for idx, row in enumerate(sheet.get_rows()):
row = [
xldate_to_datetime(cell.value,
sheet.book.datemode).isoformat()
if is_xldate(cell) else cell.value for cell in row
]
# initialise headers and standardise as headers may have variations
if HEADER_PATTERN in str(row):
header = [cell for cell in row if cell]
continue
# remove empty, useless rows
if not row[RELEVANT_ROW_INDICATOR]:
continue
raw_item = {
head: row[col_idx]
for col_idx, head in enumerate(header)
}
# contextualise raw item with metadata
raw_item.update(provider_name=self.provider,
reported_date=reported_date)
yield normalize.process_item(raw_item)
def parse_mail(self, mail):
"""Extract data from each mail matched by the query spider argument.
Args:
mail (Mail):
Yields:
Dict[str, str]:
"""
for attachment in mail.attachments():
# each xls file only has one sheet
sheet = xlrd.open_workbook(file_contents=attachment.body,
on_demand=True).sheet_by_index(0)
# store state of the table, in order to get relevant rows to extract
is_relevant = False
for row in sheet.get_rows():
row = [
xldate_to_datetime(cell.value,
sheet.book.datemode).isoformat()
if is_xldate(cell) else cell.value for cell in row
]
# discard irrelavant rows until we see the start pattern
if any(sub in row
for sub in ('Voyage Reference', 'Vessel Name')):
is_relevant = True
header = row
continue
# sanity check using first element in case of empty row
if is_relevant and row[0]:
raw_item = {
head: row[idx]
for idx, head in enumerate(header)
}
# contextualise raw item with some meta info
raw_item.update(
reported_date=to_isoformat(mail.envelope['date']),
provider_name=self.provider,
)
yield from normalize.process_item(raw_item)
def parse_mail(self, mail):
"""Extract mail found with specified email filters in spider arguments.
Args:
mail (Mail):
Yields:
Dict[str, str]:
"""
for attachment in mail.attachments():
# each xlsx file by this provider will only have one sheet
sheet = xlrd.open_workbook(file_contents=attachment.body,
on_demand=True).sheet_by_index(0)
for idx, row in enumerate(sheet.get_rows()):
row = [
xldate_to_datetime(cell.value,
sheet.book.datemode).isoformat()
if is_xldate(cell) else cell.value for cell in row
]
# remove empty filler rows before and after the main data table
if not row[EMPTY_ROW_INDICATOR_INDEX]:
continue
# initialise headers
if HEADER_PATTERN in row:
header = row
continue
# sanity check, just in case we somehow miss the headers
# due to changes in table structures
if 'header' in locals():
raw_item = {
head: row[idx]
for idx, head in enumerate(header) if head
}
# contextualise raw item with metadata
raw_item.update(provider_name='Affinity',
spider_name=self.name)
yield normalize.process_item(raw_item)
def parse_sheet(self, sheet):
"""Extract raw table data from specified sheet.
Args:
sheet (xlrd.Sheet):
Yields:
Dict[str, str]:
"""
header = None
for idx, row in enumerate(sheet.get_rows()):
# first row will always be the header
if idx == 0:
header = [cell.value for cell in row]
continue
row = [
xldate_to_datetime(cell.value,
sheet.book.datemode).isoformat()
if is_xldate(cell) else cell.value for cell in row
]
yield {head: row[head_idx] for head_idx, head in enumerate(header)}
def parse_mail(self, mail):
"""Extract data from each mail matched by the query spider argument.
Args:
mail (Mail):
Yields:
Dict[str, str]:
"""
# memoise reported_date so it won't need to be called repeatedly later
reported_date = to_isoformat(mail.envelope['date'])
for attachment in mail.attachments():
# sheet 0 contains historical info, but we want monthly info in sheet 1
sheet = xlrd.open_workbook(file_contents=attachment.body,
on_demand=True).sheet_by_index(1)
# store state of the table, in order to get relevant rows to extract
for idx, row in enumerate(sheet.get_rows()):
row = [
xldate_to_datetime(cell.value,
sheet.book.datemode).isoformat()
if is_xldate(cell) else cell.value for cell in row
]
# initialise headers
if idx == 0:
header = row
continue
raw_item = {head: row[idx] for idx, head in enumerate(header)}
# contextualise raw item with meta info
raw_item.update(reported_date=reported_date,
provider_name=self.provider)
yield normalize.process_item(raw_item)
def parse_attachment(self, attachment_doc, reported_date):
workbook = xlrd.open_workbook(file_contents=attachment_doc,
on_demand=True)
for sheet in workbook.sheets():
if sheet.name.lower() not in BLACKLIST_SHEETS:
# assign variable to detect which row to start processing
start_processing = False
raw_port_name = None
# store state of the table, in order to get relevant rows to extract
for raw_row in sheet.get_rows():
# Include handling of xlrd.xldate.XLDateAmbiguous cases
row = []
for cell in raw_row:
if is_xldate(cell):
try:
cell = xldate_to_datetime(
cell.value,
sheet.book.datemode).isoformat()
except Exception:
cell = str(cell.value)
else:
cell = str(cell.value)
row.append(cell)
# detect portname as some tabs/attachments have the port name above
# the data row
if 'Port' in row[0]:
raw_port_name = row[0]
continue
# detect relevant row
if row[0] in START_PROCESSING_WORD:
start_processing = True
header = row
continue
# detect irrelevant row
if row[0] == STOP_PROCESSING_WORD:
start_processing = False
continue
if start_processing:
# remove unnecessary rows
if row[0] == '' or 'Without' in row[0] or ':' in row[0]:
continue
raw_item = {
head: row[idx]
for idx, head in enumerate(header)
}
# contextualise raw item with meta info
raw_item.update(
reported_date=reported_date,
provider_name=self.provider,
raw_port_name=raw_port_name,
)
if DataTypes.SpotCharter in self.produces:
yield normalize_charters.process_item(raw_item)
if DataTypes.Cargo in self.produces:
yield from normalize_grades.process_item(raw_item)
def parse_workbook_content(self, response):
"""Parse workbook content.
Excel file contains 3 tables: berthed table, arrival table and eta table in that order.
Both arrival and berthed tables use the same headers.
Args:
response (scrapy.HtmlResponse):
Returns:
dict[str, str]:
"""
sheet = parser.get_xlsx_sheet(response)
for idx, row in enumerate(sheet.get_rows()):
row = [
xldate_to_datetime(cell.value,
sheet.book.datemode).isoformat()
if is_xldate(cell) else str(cell.value) for cell in row
]
# flag to indicate extraction of arrival table
if not self.arrival_table and 'No' in row[0]:
self.arrival_table = True
self.berthed_table = False
self.eta_table = False
self.arrival_header = row
continue
# flag to indicate extraction of berthed table
if not self.berthed_table and 'NAME OF SHIP' in row[2]:
self.arrival_table = False
self.berthed_table = True
self.eta_table = False
self.berthed_header = row
continue
# flag to indicate extraction of eta table
if not self.eta_table and 'EXPECTED VESSEL' in row[2]:
self.arrival_table = False
self.berthed_table = False
self.eta_table = True
# eta table does not have headers; they are derived from arrival table
self.eta_header = self.arrival_header
continue
# third table row contains `reported_date`
if idx == 2:
reported_date = parser._extract_reported_date(row[0])
continue
# extract berthed rows, berthed rows have a value for the 8th cell
if self.berthed_table and row[1] and 'THIS MAY ALTER' not in row[1]:
# vessels that are 'ships to follow' are processed in the other tables
if 'AT BERTH' not in row[1]:
continue
raw_item = parser._map_row_to_dict(
row,
self.berthed_header,
event='berthed',
port_name=self.provider,
provider_name=self.provider,
reported_date=reported_date,
)
yield normalize.process_item(raw_item)
# extract eta rows, eta rows have a value for the 4th cell
if self.eta_table and row[3]:
raw_item = parser._map_row_to_dict(
row,
self.eta_header,
event='eta',
port_name=self.provider,
provider_name=self.provider,
reported_date=reported_date,
)
yield normalize.process_item(raw_item)
# extract arrival rows, arrival rows have a value for the 4th cell
if self.arrival_table and row[3]:
raw_item = parser._map_row_to_dict(
row,
self.arrival_header,
event='arrival',
port_name=self.provider,
provider_name=self.provider,
reported_date=reported_date,
)
yield normalize.process_item(raw_item)
def parse_mail(self, mail):
"""Extract data from each mail matched by the query spider argument.
Args:
mail (Mail):
Yields:
Dict[str, str]:
"""
# memoise reported_date so it won't need to be called repeatedly later
reported_date = self.extract_reported_date(mail.envelope['subject'])
for attachment in mail.attachments():
if not attachment.is_spreadsheet:
continue
for sheet in xlrd.open_workbook(file_contents=attachment.body,
on_demand=True).sheets():
if sheet.name.lower() not in RELEVANT_NAMES:
continue
# store state of the table, in order to get relevant rows to extract
for idx, raw_row in enumerate(sheet.get_rows()):
row = []
# to handle is xldate exception
for cell in raw_row:
if is_xldate(cell):
try:
cell = xldate_to_datetime(
cell.value,
sheet.book.datemode).isoformat()
except Exception:
cell = str(cell.value)
else:
cell = str(cell.value)
row.append(cell)
# retrieve static information
_static_info = parser.HEADER_SHEET_MAPPING.get(
sheet.name.lower())
# ignore unnecessary rows before header row
if idx < _static_info[0]:
continue
header = _static_info[2]
# ignore rows where vessel column is empty
# can be put in normalize but this file has a lot of empty rows
# at the end of the report. Putting it here would cut the noise
if not row[_static_info[1]]:
continue
raw_item = {
head: row[idx]
for idx, head in enumerate(header)
}
# contextualise raw item with meta info
raw_item.update(reported_date=reported_date,
provider_name=self.provider)
yield from normalize.process_item(raw_item)