def form_table_cell(bounding_box, form_word):
model = _models.FormTableCell(
text="Cell", row_index=3, column_index=4, row_span=2, column_span=3, bounding_box=bounding_box[0],
confidence=0.7, is_header=True, is_footer=False, page_number=3, field_elements=[form_word[0]]
)
model_repr = "FormTableCell(text=Cell, row_index=3, column_index=4, row_span=2, column_span=3, bounding_box={}, confidence=0.7, " \
"is_header=True, is_footer=False, page_number=3, field_elements=[{}])".format(bounding_box[1], form_word[1])[:1024]
assert repr(model) == model_repr
return model, model_repr
def test_form_table_to_dict(self):
table = _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),
],
)
d = table.to_dict()
final = {
"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},
],
}
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
