def redact_words(self, refs: List[str]):
def word_path(ref: str) -> str:
# Remove leading '#'.
return ref[1:]
for ref in refs:
r = word_path(ref)
word = resolve_pointer(self.ocr_result, r)
word["text"] = first_char(word["text"])
set_pointer(self.ocr_result, r, word)
def redact_lines(self, refs: List[str]):
def line_path(ref: str) -> str:
end = ref.find("/word")
# Remove leading '#' and trailing word path.
return ref[1:end]
for ref in refs:
r = line_path(ref)
line = resolve_pointer(self.ocr_result, r)
tokens = line["text"].split(' ')
word_id = int(ref.split('/')[-1])
tokens[word_id] = first_char(tokens[word_id])
line["text"] = ' '.join(tokens)
set_pointer(self.ocr_result, r, line)
def redact_page_results(self, refs: Set[str]):
def add_analyze_layer(elem: str) -> str:
return elem.replace('#/', '#/analyzeResult/')
page_results = self.ocr_result["analyzeResult"]["pageResults"]
for page_result in page_results:
tables: List[dict] = page_result["tables"]
for table in tables:
cells: List[dict] = table["cells"]
for cell in cells:
elements: List[str] = cell["elements"]
for elem_id, element in enumerate(elements):
full_elem = add_analyze_layer(element)
if full_elem in refs:
tokens = cell["text"].split(' ')
tokens[elem_id] = first_char(tokens[elem_id])
cell["text"] = ' '.join(tokens)
def redact(self):
for label in self.fott_label.labels:
if len(self.labels_to_redact) == 0 or label.label in self.labels_to_redact:
for entity in label.value:
entity.text = first_char(entity.text)
def test_first_char_empty(self) -> None:
text = ""
actual = first_char(text)
assert "" == actual
def test_first_char_diacritics(self) -> None:
text = "Anaïs, Noël, Sørina, François, Mátyás, Agnès, Fañch, Reiß"
actual = first_char(text)
assert "Aaaaa, Aaaa, Aaaaaa, Aaaaaaaa, Aaaaaa, Aaaaa, Aaaaa, Aaaa" == actual
def test_first_char_price(self) -> None:
text = "$3000.00"
actual = first_char(text)
assert "$0000.00" == actual
def test_first_char_date(self) -> None:
text = "1900/01/01"
actual = first_char(text)
assert "0000/00/00" == actual
def test_first_char_Apple(self) -> None:
text = "Apple"
actual = first_char(text)
assert "Aaaaa" == actual