def create_non_spaced_string(self, text, diff_tuples, size_filo,
search_range_filo):
PADDING_CHAR = '¦'
# pad values because of filos
text_padded = Random.append_pad_values(text, size_filo, PADDING_CHAR)
text_split = list(text_padded)
current_chars_filo = Ranged_Filo(size_filo, search_range_filo, True)
filo_mid_index = current_chars_filo.get_middle_index()
final_text = ""
for char_index, char in enumerate(text_split):
current_chars_filo.push(char)
# if current middle char is ' ' and there is a diff tuple for that, don't push it to final string
current_tuple = current_chars_filo.get_middle_items(True, True)
current_middle_char = current_tuple[filo_mid_index]
its_a_diff_tuple = False
for diff_tuple_index, diff_tuple in enumerate(diff_tuples):
if current_tuple == diff_tuple:
diff_tuples[
diff_tuple_index] = "done" # mark this tuple as corrected
its_a_diff_tuple = True
break # escape inner loop
if current_middle_char is not PADDING_CHAR: # do not append padded chars
if not its_a_diff_tuple and current_middle_char is not None:
final_text += current_middle_char
return final_text
def add_linebreaks(self, previous_line, current_line, previous_line_index,
sd_line_index, line_heigth_info):
MODE = 'TAKE_CURRENT_LINE_DIST'
if previous_line is None:
return None
if MODE is 'TAKE_CURRENT_LINE_DIST':
MARGIN = 0 # tolerance margin
current_lh_info = line_heigth_info[sd_line_index]
(xp_start, yp_start, xp_stop, yp_stop) = previous_line.coordinates
(xc_start, yc_start, xc_stop, yc_stop) = current_line.coordinates
y_dist = yc_start - yp_stop
if y_dist <= 0:
return None
line_distance = current_lh_info.get_line_distance()
y_times = (y_dist + MARGIN) / line_distance
y_times_absolute = TypeCasts.round_to_int(y_times)
if y_times_absolute > 0:
generated_text = Random.append_pad_values(
"", y_times_absolute, "\n")
return generated_text
else:
return None
self.cpr.print("Undefined case reached shouldn't happen")
return None
def compare_ocr_strings_hamming(ocr_string1,
ocr_string2,
pad_difference=True):
if pad_difference is True:
len_str1 = len(ocr_string1)
len_str2 = len(ocr_string2)
if len_str1 > len_str2:
ocr_string2 = Random.append_pad_values(ocr_string2,
len_str1 - len_str2)
elif len_str2 > len_str1:
ocr_string1 = Random.append_pad_values(ocr_string1,
len_str2 - len_str1)
# print("Do Hammingdist ",ocr_string1," to " ,ocr_string2)
result = distpkg.hamming(ocr_string1, ocr_string2)
return result