def compute_precision(invoice, ground_truth):
"""
:param invoice: Invoice object to compare
:param ground_truth: Ground truth Invoice object
:return: Precision metrics
"""
precisions = {
"__pre__total_incl":
Benchmark.scalar_precision_score(invoice.total_incl,
ground_truth.total_incl),
"__pre__total_excl":
Benchmark.scalar_precision_score(invoice.total_excl,
ground_truth.total_excl),
"__pre__invoice_date":
Benchmark.scalar_precision_score(invoice.invoice_date,
ground_truth.invoice_date),
"__pre__invoice_number":
Benchmark.scalar_precision_score(invoice.invoice_number,
ground_truth.invoice_number),
"__pre__due_date":
Benchmark.scalar_precision_score(invoice.due_date,
ground_truth.due_date),
"__pre__total_tax":
Benchmark.scalar_precision_score(invoice.total_tax,
ground_truth.total_tax)
}
if len(invoice.taxes) == 0:
precisions["__pre__taxes"] = None
else:
precisions["__pre__taxes"] = Tax.compare_arrays(
invoice.taxes, ground_truth.taxes)
return precisions
def test_constructor_no_amount():
field_dict = {
'value': "NA",
'rate': "AA",
'code': "N/A",
'probability': 0.1
}
tax = Tax(field_dict)
assert tax.value is None
assert type(str(tax)) == str
def compute_accuracy(receipt, ground_truth):
"""
:param receipt: Receipt object to compare
:param ground_truth: Ground truth Receipt object
:return: Accuracy metrics
"""
return {
"__acc__total_incl": ground_truth.total_incl == receipt.total_incl,
"__acc__total_excl": ground_truth.total_excl == receipt.total_excl,
"__acc__receipt_date": ground_truth.date == receipt.date,
"__acc__total_tax": ground_truth.total_tax == receipt.total_tax,
"__acc__taxes": Tax.compare_arrays(receipt.taxes, ground_truth.taxes),
}
def test_constructor():
field_dict = {
'value': "2",
'rate': 0.2,
'code': "QST",
'probability': 0.1,
'segmentation': {
"bounding_box": [[0.016, 0.707], [0.414, 0.707], [0.414, 0.831],
[0.016, 0.831]]
}
}
tax = Tax(field_dict, value_key="value")
assert tax.value == 2
assert tax.probability == 0.1
assert tax.rate == 0.2
assert len(tax.bbox) > 0
assert type(str(tax)) == str
def compute_accuracy(invoice, ground_truth):
"""
:param invoice: Invoice object to compare
:param ground_truth: Ground truth Invoice object
:return: Accuracy metrics
"""
return {
"__acc__total_incl": ground_truth.total_incl == invoice.total_incl,
"__acc__total_excl": ground_truth.total_excl == invoice.total_excl,
"__acc__invoice_date":
ground_truth.invoice_date == invoice.invoice_date,
"__acc__invoice_number":
ground_truth.invoice_number == invoice.invoice_number,
"__acc__due_date": ground_truth.due_date == invoice.due_date,
"__acc__total_tax": ground_truth.total_tax == invoice.total_tax,
"__acc__taxes": Tax.compare_arrays(invoice.taxes,
ground_truth.taxes),
}
def build_from_api_prediction(self, api_prediction, page_n=0):
"""
:param api_prediction: Raw prediction from HTTP response
:param page_n: Page number for multi pages pdf input
:return: (void) set the object attributes with api prediction values
"""
self.locale = Locale(api_prediction["locale"], page_n=page_n)
self.total_incl = Amount(api_prediction["total_incl"], value_key="value", page_n=page_n)
self.date = Date(api_prediction["date"], value_key="value", page_n=page_n)
self.category = Field(api_prediction["category"], page_n=page_n)
self.merchant_name = Field(api_prediction["supplier"], value_key="value", page_n=page_n)
self.time = Field(api_prediction["time"], value_key="value", page_n=page_n)
self.taxes = [
Tax(tax_prediction, page_n=page_n, value_key="value", rate_key="rate", code_key="code")
for tax_prediction in api_prediction["taxes"]]
self.orientation = Orientation(api_prediction["orientation"], page_n=page_n)
self.total_tax = Amount({"value": None, "probability": 0.}, value_key="value", page_n=page_n)
self.total_excl = Amount({"value": None, "probability": 0.}, value_key="value", page_n=page_n)
def build_from_api_prediction(self, api_prediction, page_n=0):
"""
:param api_prediction: Raw prediction from HTTP response
:param page_n: Page number for multi pages pdf input
:return: (void) set the object attributes with api prediction values
"""
self.company_number = [
Field(company_reg, extra_fields={"type"}, page_n=page_n)
for company_reg in api_prediction["company_registration"]
]
self.invoice_date = Date(api_prediction["date"],
value_key="value",
page_n=page_n)
self.due_date = Date(api_prediction["due_date"],
value_key="value",
page_n=page_n)
self.invoice_number = Field(api_prediction["invoice_number"],
page_n=page_n)
self.locale = Locale(api_prediction["locale"],
value_key="language",
page_n=page_n)
self.orientation = Orientation(api_prediction["orientation"],
page_n=page_n)
self.supplier = Field(api_prediction["supplier"], page_n=page_n)
self.taxes = [
Tax(tax_prediction, page_n=page_n, value_key="value")
for tax_prediction in api_prediction["taxes"]
]
self.payment_details = [
PaymentDetails(payment_detail, page_n=page_n)
for payment_detail in api_prediction["payment_details"]
]
self.total_incl = Amount(api_prediction["total_incl"],
value_key="value",
page_n=page_n)
self.total_excl = Amount(api_prediction["total_excl"],
value_key="value",
page_n=page_n)
self.total_tax = Amount({
"value": None,
"probability": 0.
},
value_key="value",
page_n=page_n)
def compute_precision(receipt, ground_truth):
"""
:param receipt: Receipt object to compare
:param ground_truth: Ground truth Receipt object
:return: Precision metrics
"""
precisions = {
"__pre__total_incl": Benchmark.scalar_precision_score(
receipt.total_incl, ground_truth.total_incl),
"__pre__total_excl": Benchmark.scalar_precision_score(
receipt.total_excl, ground_truth.total_excl),
"__pre__receipt_date": Benchmark.scalar_precision_score(
receipt.date, ground_truth.date),
"__pre__total_tax": Benchmark.scalar_precision_score(
receipt.total_tax, ground_truth.total_tax)}
if len(receipt.taxes) == 0:
precisions["__pre__taxes"] = None
else:
precisions["__pre__taxes"] = Tax.compare_arrays(receipt.taxes, ground_truth.taxes)
return precisions
def compute_accuracy(financial_document, ground_truth):
"""
:param financial_document: FinancialDocument object to compare
:param ground_truth: Ground truth FinancialDocument object
:return: Accuracy metrics
"""
return {
"__acc__total_incl":
ground_truth.total_incl == financial_document.total_incl,
"__acc__total_excl":
ground_truth.total_excl == financial_document.total_excl,
"__acc__invoice_date":
ground_truth.date == financial_document.date,
"__acc__invoice_number":
ground_truth.invoice_number == financial_document.invoice_number,
"__acc__due_date":
ground_truth.due_date == financial_document.due_date,
"__acc__total_tax":
ground_truth.total_tax == financial_document.total_tax,
"__acc__taxes":
Tax.compare_arrays(financial_document.taxes, ground_truth.taxes),
}
def test_constructor_no_rate():
field_dict = {'value': "2", 'rate': "AA", 'probability': 0.1}
tax = Tax(field_dict)
assert tax.rate is None
assert len(tax.bbox) == 0
def __init__(self,
api_prediction=None,
input_file=None,
locale=None,
total_incl=None,
total_excl=None,
invoice_date=None,
invoice_number=None,
due_date=None,
taxes=None,
supplier=None,
payment_details=None,
company_number=None,
vat_number=None,
orientation=None,
total_tax=None,
page_n=0):
"""
:param api_prediction: Raw prediction from HTTP response
:param input_file: Input object
:param locale: locale value for creating Invoice object from scratch
:param total_incl: total_incl value for creating Invoice object from scratch
:param total_excl: total_excl value for creating Invoice object from scratch
:param invoice_date: invoice_date value for creating Invoice object from scratch
:param invoice_number: invoice_number value for creating Invoice object from scratch
:param due_date: due_date value for creating Invoice object from scratch
:param taxes: taxes value for creating Invoice object from scratch
:param supplier: supplier value for creating Invoice object from scratch
:param payment_details: payment_details value for creating Invoice object from scratch
:param company_number: company_number value for creating Invoice object from scratch
:param vat_number: vat_number value for creating Invoice object from scratch
:param orientation: orientation value for creating Invoice object from scratch
:param total_tax: total_tax value for creating Invoice object from scratch
:param page_n: Page number for multi pages pdf input
"""
self.type = "Invoice"
self.locale = None
self.total_incl = None
self.total_excl = None
self.invoice_date = None
self.invoice_number = None
self.due_date = None
self.taxes = []
self.supplier = None
self.payment_details = None
self.company_number = None
self.orientation = None
self.total_tax = None
if api_prediction is not None:
self.build_from_api_prediction(api_prediction, page_n=page_n)
else:
self.locale = Locale({"value": locale},
value_key="value",
page_n=page_n)
self.total_incl = Amount({"value": total_incl},
value_key="value",
page_n=page_n)
self.date = Date({"value": invoice_date},
value_key="value",
page_n=page_n)
self.invoice_date = Date({"value": invoice_date},
value_key="value",
page_n=page_n)
self.due_date = Date({"value": due_date},
value_key="value",
page_n=page_n)
self.supplier = Field({"value": supplier},
value_key="value",
page_n=page_n)
if taxes is not None:
self.taxes = [
Tax({
"value": t[0],
"rate": t[1]
},
page_n=page_n,
value_key="value",
rate_key="rate") for t in taxes
]
self.orientation = Orientation({"value": orientation},
value_key="value",
page_n=page_n)
self.total_tax = Amount({"value": total_tax},
value_key="value",
page_n=page_n)
self.total_excl = Amount({"value": total_excl},
value_key="value",
page_n=page_n)
self.invoice_number = Field({"value": invoice_number},
value_key="value",
page_n=page_n)
self.payment_details = Field({"value": payment_details},
value_key="value",
page_n=page_n)
self.company_number = Field({"value": company_number},
value_key="value",
page_n=page_n)
# Invoke Document constructor
super(Invoice, self).__init__(input_file)
# Run checks
self._checklist()
# Reconstruct extra fields
self._reconstruct()
def __init__(
self,
api_prediction=None,
input_file=None,
locale=None,
total_incl=None,
date=None,
category=None,
merchant_name=None,
time=None,
taxes=None,
orientation=None,
total_tax=None,
total_excl=None,
page_n=0
):
"""
:param api_prediction: Raw prediction from HTTP response
:param input_file: Input object
:param locale: locale value for creating Receipt object from scratch
:param total_incl: total_incl value for creating Receipt object from scratch
:param date: date value for creating Receipt object from scratch
:param category: category value for creating Receipt object from scratch
:param merchant_name: merchant_name value for creating Receipt object from scratch
:param time: time value for creating Receipt object from scratch
:param taxes: taxes value for creating Receipt object from scratch
:param orientation: orientation value for creating Receipt object from scratch
:param total_tax: total_tax value for creating Receipt object from scratch
:param total_excl: total_excl value for creating Receipt object from scratch
:param page_n: Page number for multi pages pdf input
"""
self.type = "Receipt"
self.locale = None
self.total_incl = None
self.date = None
self.category = None
self.merchant_name = None
self.time = None
self.taxes = []
self.orientation = None
self.total_tax = None
self.total_excl = None
if api_prediction is not None:
self.build_from_api_prediction(api_prediction, page_n=page_n)
else:
self.locale = Locale({"value": locale}, value_key="value", page_n=page_n)
self.total_incl = Amount({"value": total_incl}, value_key="value", page_n=page_n)
self.date = Date({"value": date}, value_key="value", page_n=page_n)
self.category = Field({"value": category}, value_key="value", page_n=page_n)
self.merchant_name = Field({"value": merchant_name}, value_key="value", page_n=page_n)
self.time = Field({"value": time}, value_key="value", page_n=page_n)
if taxes is not None:
self.taxes = [
Tax({"value": t[0], "rate": t[1]}, page_n=page_n, value_key="value", rate_key="rate")
for t in taxes]
self.orientation = Orientation({"value": orientation}, value_key="value", page_n=page_n)
self.total_tax = Amount({"value": total_tax}, value_key="value", page_n=page_n)
self.total_excl = Amount({"value": total_excl}, value_key="value", page_n=page_n)
# Invoke Document constructor
super(Receipt, self).__init__(input_file)
# Run checks
self._checklist()
# Reconstruct extra fields
self._reconstruct()