38 lines
1.1 KiB
Python
38 lines
1.1 KiB
Python
# test with:
|
|
|
|
# docker image build -t headless-fox Docker && docker run -v $PWD/static:/opt/static -ti headless-fox python annotate_header.py screenshot.png 1 "content-type" "etag"
|
|
|
|
import os
|
|
import sys
|
|
import pytesseract
|
|
import uuid
|
|
from pytesseract import Output
|
|
from PIL import Image, ImageDraw
|
|
|
|
print(sys.argv)
|
|
|
|
image = sys.argv[1]
|
|
index = sys.argv[2]
|
|
needles = sys.argv[3:]
|
|
|
|
# print(d)
|
|
|
|
with Image.open(image) as im:
|
|
x_offset = 2054
|
|
y_offset = 313
|
|
cropped = im.crop((x_offset, y_offset, 2875, 1558))
|
|
cropped_filename = "/opt/static/" + uuid.uuid4().hex + ".png"
|
|
cropped.save(cropped_filename)
|
|
d = pytesseract.image_to_data(cropped_filename, output_type=Output.DICT)
|
|
os.remove(cropped_filename);
|
|
draw = ImageDraw.Draw(im)
|
|
n_boxes = len(d['level'])
|
|
print(needles)
|
|
for i in range(n_boxes):
|
|
(x, y, w, h, text) = (d['left'][i], d['top'][i], d['width'][i], d['height'][i], d['text'][i])
|
|
for needle in needles:
|
|
if needle.lower() in text.lower():
|
|
print("needle", needle, text, x, y)
|
|
draw.line((x_offset, y + y_offset, x_offset + 200, y + y_offset), fill=128)
|
|
im.save("/opt/static/output.png", "PNG")
|