def form_line(bounding_box, form_word):
model = _models.FormLine(text="Word Word",
bounding_box=bounding_box[0],
words=[form_word[0], form_word[0]],
page_number=1)
model_repr = "FormLine(text=Word Word, bounding_box={}, words=[{}, {}], page_number=1)".format(
bounding_box[1], form_word[1], form_word[1])[:1024]
assert repr(model) == model_repr
return model, model_repr
def form_line(bounding_box, form_word):
appearance = Appearance(style=Style(name="other", confidence=1.0))
model = _models.FormLine(text="Word Word",
bounding_box=bounding_box[0],
words=[form_word[0], form_word[0]],
page_number=1,
appearance=appearance)
model_repr = "FormLine(text=Word Word, bounding_box={}, words=[{}, {}], page_number=1, kind=line, appearance={})".format(
bounding_box[1], form_word[1], form_word[1], appearance)[:1024]
assert repr(model) == model_repr
return model, model_repr
def test_form_line_to_dict(self):
form_line = _models.FormLine(
text="sample line",
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
words=[
_models.FormWord(
text="sample",
confidence=0.92,
page_number=1,
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
),
_models.FormWord(
text="line",
confidence=0.92,
page_number=1,
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
),
],
page_number=2,
appearance=_models.TextAppearance(
style_name="other", style_confidence=0.90
),
)
d = form_line.to_dict()
final = {
"text": "sample line",
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
"words": [
{
"text": "sample",
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
"confidence": 0.92,
"page_number": 1,
"kind": "word",
},
{
"text": "line",
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
"confidence": 0.92,
"page_number": 1,
"kind": "word",
},
],
"page_number": 2,
"kind": "line",
"appearance": {"style_name": "other", "style_confidence": 0.90},
}
assert d == final
def test_form_page_to_dict(self):
form_page = _models.FormPage(
page_number=1,
text_angle=180.0,
width=5.5,
height=8.0,
unit="pixel",
tables= [
_models.FormTable(
page_number=2,
cells=[
_models.FormTableCell(
text="info",
row_index=1,
column_index=3,
row_span=1,
column_span=2,
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
confidence=0.87,
is_header=False,
is_footer=True,
page_number=1,
field_elements=[
_models.FormWord(
text="word",
confidence=0.92,
page_number=1,
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
),
]
)
],
row_count=10,
column_count=5,
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
),
],
lines=[_models.FormLine(
text="sample line",
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
words=[
_models.FormWord(
text="sample",
confidence=0.92,
page_number=1,
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
),
_models.FormWord(
text="line",
confidence=0.92,
page_number=1,
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
),
],
page_number=2,
appearance=_models.TextAppearance(
style_name="other", style_confidence=0.90
),
),
],
selection_marks=[_models.FormSelectionMark(
text="checkbox",
state="selected",
confidence=0.92,
page_number=1,
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
),
],
)
d = form_page.to_dict()
final = {
"page_number": 1,
"text_angle": 180.0,
"width": 5.5,
"height": 8.0,
"unit": "pixel",
"tables": [
{"cells": [
{
"text": "info",
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
"row_index": 1,
"column_index": 3,
"row_span": 1,
"column_span": 2,
"confidence": 0.87,
"is_header": False,
"is_footer": True,
"page_number": 1,
"field_elements": [
{
"text": "word",
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
"confidence": 0.92,
"page_number": 1,
"kind": "word",
}
],
},
],
"page_number": 2,
"row_count": 10,
"column_count": 5,
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
},
],
"lines": [{
"text": "sample line",
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
"words": [
{
"text": "sample",
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
"confidence": 0.92,
"page_number": 1,
"kind": "word",
},
{
"text": "line",
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
"confidence": 0.92,
"page_number": 1,
"kind": "word",
},
],
"page_number": 2,
"kind": "line",
"appearance": {"style_name": "other", "style_confidence": 0.90},
}],
"selection_marks": [{
"text": "checkbox",
"state": "selected",
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
"confidence": 0.92,
"page_number": 1,
"kind": "selectionMark",
}],
}
assert d == final
def test_recognized_form_to_dict(self):
form = _models.RecognizedForm(
form_type="test_form",
form_type_confidence="0.84",
model_id="examplemodel123",
page_range=_models.FormPageRange(1, 1),
fields={
"example": _models.FormField(
value_type="phoneNumber",
label_data=_models.FieldData(
text="phone",
page_number=1,
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
),
value_data=_models.FieldData(
text="55554444",
page_number=1,
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
),
name="phone",
value="55554444",
confidence=0.99,
)
},
pages=[_models.FormPage(
page_number=1,
text_angle=180.0,
width=5.5,
height=8.0,
unit="pixel",
lines=[_models.FormLine(
text="sample line",
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
words=[
_models.FormWord(
text="sample",
confidence=0.92,
page_number=1,
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
),
_models.FormWord(
text="line",
confidence=0.92,
page_number=1,
bounding_box=[
_models.Point(1427.0, 1669.0),
_models.Point(1527.0, 1669.0),
_models.Point(1527.0, 1698.0),
_models.Point(1427.0, 1698.0),
],
),
],
page_number=2,
appearance=_models.TextAppearance(
style_name="other", style_confidence=0.90
),
)],
)
]
)
d = form.to_dict()
final = {
"form_type": "test_form",
"form_type_confidence": "0.84",
"model_id": "examplemodel123",
"page_range": {"first_page_number": 1, "last_page_number": 1},
"fields": {
"example": {
"value_type": "phoneNumber",
"label_data": {
"text": "phone",
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
"page_number": 1,
"field_elements": []
},
"value_data": {
"text": "55554444",
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
"page_number": 1,
"field_elements": []
},
"name": "phone",
"value": "55554444",
"confidence": 0.99,
}
},
"pages": [{
"page_number": 1,
"text_angle": 180.0,
"width": 5.5,
"height": 8.0,
"unit": "pixel",
"lines": [{
"text": "sample line",
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
"words": [
{
"text": "sample",
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
"confidence": 0.92,
"page_number": 1,
"kind": "word",
},
{
"text": "line",
"bounding_box": [
{"x": 1427.0, "y": 1669.0},
{"x": 1527.0, "y": 1669.0},
{"x": 1527.0, "y": 1698.0},
{"x": 1427.0, "y": 1698.0},
],
"confidence": 0.92,
"page_number": 1,
"kind": "word",
},
],
"page_number": 2,
"kind": "line",
"appearance": {"style_name": "other", "style_confidence": 0.90},
}],
"selection_marks": [],
"tables": [],
}],
}
assert d == final
