icd/screenshot-service
2
0
Fork 0
Code Issues 18 Pull Requests Projects 1 Releases Wiki Activity

Add metadata to screenshots

Browse Source
This commit is contained in:
Kuba Orlik 2022-06-17 11:44:05 +02:00
parent d06426e22f
commit a96a68517b
2 changed files with 10 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
Docker/annotate_header.py
View File

@ -6,6 +6,7 @@ import os
import sys import sys
import pytesseract import pytesseract
import uuid import uuid
import json
from pytesseract import Output from pytesseract import Output
from PIL import Image, ImageDraw, ImageFont from PIL import Image, ImageDraw, ImageFont
@ -14,7 +15,8 @@ output_file_relative = sys.argv[
] # this is also the existing source screenshot to annotate. It will be updated in-place ] # this is also the existing source screenshot to annotate. It will be updated in-place
output_file = "/opt/static/" + output_file_relative output_file = "/opt/static/" + output_file_relative
needles = sys.argv[2:] domain = sys.argv[2]
needles = sys.argv[3:]
base_url = os.getenv("BASE_URL") base_url = os.getenv("BASE_URL")
@ -36,8 +38,8 @@ with Image.open(output_file) as im:
os.remove(cropped_filename) os.remove(cropped_filename)
draw = ImageDraw.Draw(im) draw = ImageDraw.Draw(im)
n_boxes = len(d["level"]) n_boxes = len(d["level"])
print(needles)
font = ImageFont.truetype("/usr/share/fonts/noto/NotoSansDisplay-Medium.ttf", 48) font = ImageFont.truetype("/usr/share/fonts/noto/NotoSansDisplay-Medium.ttf", 48)
found_needles = []
for i in range(n_boxes): for i in range(n_boxes):
(x, y, w, h, text) = ( (x, y, w, h, text) = (
d["left"][i], d["left"][i],
@ -48,7 +50,7 @@ with Image.open(output_file) as im:
) )
for [needle, comment] in partition(needles, 2): for [needle, comment] in partition(needles, 2):
if needle.lower() in text.lower(): if needle.lower() in text.lower():
print("needle", needle, text, x, y) found_needles.append(needle)
# modify y so it's aligned not with the top of the text, but with the midline # modify y so it's aligned not with the top of the text, but with the midline
y = y + h / 2 y = y + h / 2
radius = 30 radius = 30
@ -78,4 +80,7 @@ with Image.open(output_file) as im:
font=font, font=font,
) )
im.save(output_file, "PNG") im.save(output_file, "PNG")
print('{"new_file": "' + base_url + "/static/" + output_file_relative + "\"}") print(json.dumps({"new_file":
{"url": base_url + "/static/" + output_file_relative,
"domain": domain,
"found_headers": found_needles}}))

2
Docker/run-analysis.sh
View File

@ -53,7 +53,7 @@ while IFS= read -r DOMAIN; do
filename="$ID/${index}.png" filename="$ID/${index}.png"
scrot "/opt/static/$filename" scrot "/opt/static/$filename"
grab "searching $DOMAIN" grab "searching $DOMAIN"
BASE_URL="$BASE_URL" python annotate_header.py "$filename" \ BASE_URL="$BASE_URL" python annotate_header.py "$filename" "$DOMAIN" \
"set-cookie" "identyfikator internetowy z cookie" \ "set-cookie" "identyfikator internetowy z cookie" \
"Cookie" "identyfikator internetowy z cookie" \ "Cookie" "identyfikator internetowy z cookie" \
"Referer" "Część mojej historii przeglądania" & "Referer" "Część mojej historii przeglądania" &