def markup_response_codes(page: Page, list_id: str,
line_number: int) -> List[ResponseCode]:
# Get line in the page.
line_bb = page.get_line_bb(line_number,
utils.get_list_dir(list_id),
padding=100,
right_half=True)
markup_roi = page.get_roi(line_bb)
# Iterate through and mark each scan code.
response_codes = []
question_number = 1
while True:
print("Please mark each response code in survey question %d." %
question_number)
bounding_box = markup_roi.markup()
text = markup_roi.find_text(bounding_box)
# sometimes OCR picks up stray symbols, get rid of them.
text = ''.join(ch for ch in text if ch.isalnum())
bounding_box = bounding_box.update_coordinate_system(line_bb.top_left)
roi = page.get_roi(bounding_box)
response_code = ResponseCode(bounding_box, question_number, text)
print("Extracted scan code: \"%s\"" % response_code.value)
while True:
print("Is this correct? [y|n]")
yes_no = input().lower()
if yes_no == "y":
break
else:
print("Please enter the correct response code: ")
response_code.value = input()
response_codes.append(response_code)
print(
r"""Hit enter (no input) to mark another response in the same survey question.
Enter 'n' to move to the next survey question. Enter 'q' to finish. [enter|n|q]"""
)
next_step = input().lower()
if next_step == "n":
question_number += 1
elif next_step == "q":
break
return response_codes
def handle_missing_page_id(aligned_page: Page, original_page: Page,
list_id: str, id_bounding_box: BoundingBox,
page_number: int) -> Tuple[bool, Page]:
# to check if it's a homography issue, see if the list ID is visible on the raw page
test_id = original_page.get_list_id(id_bounding_box)
if test_id == list_id:
print('Homography error on page {}, using uncorrected page instead.'.
format(page_number))
return True, original_page
# didn't find on raw page, ask the user to confirm the ID
id_area = original_page.get_roi(id_bounding_box)
id_area = id_area.add_border(top=30)
# annotate the response image
question = "ID: {}? (y|n)".format(list_id)
# keep looping until the y or n key is pressed
while True:
# display the image and wait for a keypress
key = id_area.show(title="review", message=question, resize=False)
# if the 'y' key is pressed, user approves
if key == ord("y"):
cv2.destroyAllWindows()
return True, original_page
# if the 'n' key is pressed, user rejects
elif key == ord("n"):
cv2.destroyAllWindows()
break
return False, aligned_page
def check_num_barcodes(page: Page, list_dir: str, num_scanned_barcodes: int,
results_stats) -> None:
# Manually loop and count barcodes
num_actual_barcodes = 0
for line_number in range(1, utils.MAX_BARCODES_ON_PAGE + 1):
line_bb = page.get_line_bb(line_number, list_dir)
# extract the barcode portion
line_bb.top_left.x = line_bb.bottom_right.x - 700
barcode_roi = page.get_roi(line_bb).invert()
BARCODE_EXISTS_THRESHOLD = 20000 # if a barcode exists in the area it averages 29k black pixels.
if barcode_roi.numWhitePixels() > BARCODE_EXISTS_THRESHOLD:
num_actual_barcodes += 1
else:
break # we have likely reached the end of the page.
if num_actual_barcodes < num_scanned_barcodes:
print(
"Something went wrong with the image alignment! Cannot accurately count missed barcodes."
)
elif num_actual_barcodes > num_scanned_barcodes:
results_stats[
"num_missed_barcodes"] += num_actual_barcodes - num_scanned_barcodes
def create_error_image(page: Page, barcode_coords: BoundingBox,
first_response_coords: BoundingBox) -> Image:
full_response_bounding_box = get_response_including_barcode(
barcode_coords, first_response_coords, page.size)
error_image = page.get_roi(full_response_bounding_box)
return error_image