def test_cell_values_finder(self):
cell_values_results = [
([CellValue(3143), CellValue(2525)], [
CellPosition(col=2, row=2),
CellPosition(col=2, row=5),
CellPosition(col=2, row=7)
]),
([CellValue('Общий итог'), CellValue()], [
CellPosition(col=6, row=0),
CellPosition(col=6, row=1),
CellPosition(col=5, row=5),
CellPosition(col=0, row=11),
CellPosition(col=0, row=12),
CellPosition(col=1, row=12),
CellPosition(col=2, row=12),
CellPosition(col=3, row=12),
CellPosition(col=4, row=12),
CellPosition(col=5, row=12),
CellPosition(col=6, row=12),
]),
]
finder = AllCellPositionsFinder(df=self.duplicates_df)
for cell_values, results in cell_values_results:
finder.value_finder = ExactValuesFinder(cell_values=cell_values)
finder_results = finder.get_all_positions()
self._check_results(expected_result=results,
finder_result=finder_results)
def test_startwith(self):
patterns_results = [
('*', [
CellPosition(col=0, row=1),
CellPosition(col=0, row=2),
CellPosition(col=0, row=3),
CellPosition(col=0, row=4),
CellPosition(col=0, row=5),
CellPosition(col=0, row=6),
CellPosition(col=0, row=7),
CellPosition(col=0, row=8),
CellPosition(col=0, row=9),
CellPosition(col=0, row=10),
CellPosition(col=0, row=11),
CellPosition(col=0, row=12),
CellPosition(col=0, row=13),
CellPosition(col=0, row=14),
CellPosition(col=0, row=15),
]),
('НЕБИВОЛОЛ-',
[CellPosition(col=0, row=132),
CellPosition(col=0, row=133)]),
('102', []),
# TODO ДОбавить ещё проверки. Вариант котороый есть ив текстовой ячейке и числовой
]
finder = AllCellPositionsFinder(df=self.simple_df)
for value, results in patterns_results:
finder.value_finder = StartWithFinder(cell_value=CellValue(value))
finder_results = finder.get_all_positions()
self._check_results(expected_result=results,
finder_result=finder_results)
class ByValueNeighborhoodsLTRBFilterDF(FilterDfAbstract):
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.cell_pos_finder = AllCellPositionsFinder(df=self._df)
def _filter(self, start_cell_value: CellValue,
start_cell_neighbors: List[NeighborCell],
end_cell_value: CellValue,
end_cell_neighbors: List[NeighborCell]):
_start_position = CellPosition()
for start_position in self.cell_pos_finder.get_position():
neighbors_on_places = all([
neighbor.is_neighbor(start_position)
for neighbor in start_cell_neighbors
])
if neighbors_on_places:
_start_position = start_position
break
_end_position = CellPosition()
for end_position in self.cell_pos_finder.get_position():
neighbors_on_places = all([
neighbor.is_neighbor(end_position)
for neighbor in start_cell_neighbors
])
if neighbors_on_places:
_end_position = end_position
break
res_df = self._filter_df_ltrb(_start_position, _end_position)
return res_df
def test_endwith(self):
patterns_results = [
('2,5МГ+6,25МГ №30',
[CellPosition(col=0, row=1),
CellPosition(col=0, row=3)]),
('25', []),
# TODO ДОбавить ещё проверки. Вариант котороый есть ив текстовой ячейке и числовой
]
finder = AllCellPositionsFinder(df=self.duplicates_df)
for value, results in patterns_results:
finder.value_finder = EndWithFinder(cell_value=CellValue(value))
finder_results = finder.get_all_positions()
self._check_results(expected_result=results,
finder_result=finder_results)
def test_regex_cell_value_pattern(self):
patterns_results = [
('.*КАНОН.*\d{4}.*', [CellPosition(col=0, row=5)]),
('.*АРИТЕЛ ПЛЮС(?!.*2,5).*', [
CellPosition(col=0, row=2),
CellPosition(col=0, row=7),
CellPosition(col=0, row=10)
]),
]
finder = AllCellPositionsFinder(df=self.duplicates_df)
for pattern, results in patterns_results:
finder.value_finder = RegexFinder(cell_value=CellValue(pattern))
finder_results = finder.get_all_positions()
self._check_results(expected_result=results,
finder_result=finder_results)
def test_cell_values_finder_with_neighbors(self):
cell_values_neighbors_results = [
([CellValue(3143), CellValue(2525)], [
NeighborCell(df=self.duplicates_df,
cell_value=CellValue(4690),
cell_offset=CellOffset(col=-1, row=0))
], [CellPosition(col=2, row=5)]),
]
finder = AllCellPositionsFinder(df=self.duplicates_df)
for cell_values, neighbors, results in cell_values_neighbors_results:
finder.value_finder = ExactValuesFinder(cell_values=cell_values)
finder.neighbors_container = NeighborsContainer(
neighbors=neighbors)
finder_results = finder.get_all_positions()
self._check_results(expected_result=results,
finder_result=finder_results)
def test_cell_value_finder_with_offset(self):
cell_values_results = [
(CellValue('SKU'), CellOffset(col=1,
row=0), [CellPosition(col=1,
row=0)]),
(CellValue('SKU'), CellOffset(col=-10, row=-10), []),
(CellValue('Общий итог'), CellOffset(col=0, row=1), [
CellPosition(col=6, row=1),
CellPosition(col=0, row=12),
]),
(CellValue('Общий итог'), CellOffset(col=0, row=2),
[CellPosition(col=6, row=2)]),
]
finder = AllCellPositionsFinder(df=self.duplicates_df)
for cell_value, cell_offset, results in cell_values_results:
finder.value_finder = ExactValueFinder(cell_value=cell_value)
finder.cell_offset_action = CellOffsetAction(
cell_offset=cell_offset)
finder_results = finder.get_all_positions()
self._check_results(expected_result=results,
finder_result=finder_results)
def test_cell_value_finder_with_neighbors(self):
cell_values_neighbors_results = [
(CellValue(2525), [
NeighborCell(df=self.duplicates_df,
cell_value=CellValue('Общий итог'),
cell_offset=CellOffset(col=-2, row=6))
], [CellPosition(col=2, row=5)]),
(CellValue(0), [
NeighborCell(df=self.duplicates_df,
cell_value=CellValue(),
cell_offset=CellOffset(col=0, row=-1))
], [CellPosition(col=5, row=6)]),
(CellValue('Общий итог'), [
NeighborCell(df=self.duplicates_df,
cell_value=CellValue(),
cell_offset=CellOffset(col=0, row=1))
], [CellPosition(col=6, row=0),
CellPosition(col=0, row=11)]),
(CellValue('Общий итог'), [
NeighborCell(df=self.duplicates_df,
cell_value=CellValue(''),
cell_offset=CellOffset(col=0, row=1))
], [CellPosition(col=6, row=0),
CellPosition(col=0, row=11)]),
(CellValue('Общий итог'), [
NeighborCell(df=self.duplicates_df,
cell_value=CellValue(),
cell_offset=CellOffset(col=0, row=1)),
NeighborCell(df=self.duplicates_df,
cell_value=CellValue(),
cell_offset=CellOffset(col=1, row=1)),
], [CellPosition(col=0, row=11)]),
(CellValue('Общий итог'), [
NeighborCell(df=self.duplicates_df,
cell_value=CellValue(),
cell_offset=CellOffset(col=0, row=1)),
NeighborCell(df=self.duplicates_df,
cell_value=CellValue(''),
cell_offset=CellOffset(col=1, row=1)),
], [CellPosition(col=0, row=11)]),
(CellValue('Общий итог'), [
NeighborCell(df=self.duplicates_df,
cell_value=CellValue(''),
cell_offset=CellOffset(col=0, row=1)),
NeighborCell(df=self.duplicates_df,
cell_value=CellValue(''),
cell_offset=CellOffset(col=1, row=1)),
], [CellPosition(col=0, row=11)]),
(CellValue('Общий итог'), [
NeighborCell(df=self.duplicates_df,
cell_value=CellValue(''),
cell_offset=CellOffset(col=0, row=1)),
NeighborCell(df=self.duplicates_df,
cell_value=CellValue(),
cell_offset=CellOffset(col=1, row=1)),
], [CellPosition(col=0, row=11)]),
]
finder = AllCellPositionsFinder(df=self.duplicates_df)
for cell_value, neighbors, results in cell_values_neighbors_results:
finder.value_finder = ExactValueFinder(cell_value=cell_value)
finder.neighbors_container = NeighborsContainer(
neighbors=neighbors)
finder_results = finder.get_all_positions()
self._check_results(expected_result=results,
finder_result=finder_results)
def test_all_positions(self):
cell_values_results = [
(CellValue(4302),
[CellPosition(col=1, row=1),
CellPosition(col=1, row=3)]),
(CellValue(2256),
[CellPosition(col=1, row=8),
CellPosition(col=6, row=8)]),
(CellValue('Общий итог'),
[CellPosition(col=6, row=0),
CellPosition(col=0, row=11)]), (CellValue(-999), []),
(CellValue('Qwerty'), []),
(CellValue(''), [
CellPosition(col=6, row=1),
CellPosition(col=5, row=5),
CellPosition(col=0, row=12),
CellPosition(col=1, row=12),
CellPosition(col=2, row=12),
CellPosition(col=3, row=12),
CellPosition(col=4, row=12),
CellPosition(col=5, row=12),
CellPosition(col=6, row=12)
]),
(CellValue(0), [
CellPosition(col=3, row=1),
CellPosition(col=4, row=1),
CellPosition(col=5, row=1),
CellPosition(col=3, row=2),
CellPosition(col=4, row=2),
CellPosition(col=5, row=2),
CellPosition(col=3, row=3),
CellPosition(col=4, row=3),
CellPosition(col=5, row=3),
CellPosition(col=2, row=4),
CellPosition(col=3, row=4),
CellPosition(col=4, row=4),
CellPosition(col=5, row=4),
CellPosition(col=3, row=5),
CellPosition(col=4, row=5),
CellPosition(col=2, row=6),
CellPosition(col=3, row=6),
CellPosition(col=4, row=6),
CellPosition(col=5, row=6),
CellPosition(col=4, row=7),
CellPosition(col=5, row=7),
CellPosition(col=2, row=8),
CellPosition(col=3, row=8),
CellPosition(col=4, row=8),
CellPosition(col=5, row=8),
CellPosition(col=2, row=9),
CellPosition(col=3, row=9),
CellPosition(col=5, row=9),
CellPosition(col=2, row=10),
CellPosition(col=3, row=10),
CellPosition(col=4, row=10),
CellPosition(col=5, row=10),
])
]
finder = AllCellPositionsFinder(df=self.duplicates_df)
for cell_value, results in cell_values_results:
finder.value_finder = ExactValueFinder(cell_value=cell_value)
finder_results = finder.get_all_positions()
self._check_results(expected_result=results,
finder_result=finder_results)
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.cell_pos_finder = AllCellPositionsFinder(df=self._df)