ail-framework/bin/lib/objects/Ocrs.py

337 lines
11 KiB
Python
Raw Permalink Normal View History

#!/usr/bin/env python3
# -*-coding:UTF-8 -*
import os
import sys
from datetime import datetime
from io import BytesIO
from PIL import Image
from PIL import ImageDraw
from pymisp import MISPObject
sys.path.append(os.environ['AIL_BIN'])
##################################
# Import Project packages
##################################
from lib.objects.abstract_daterange_object import AbstractDaterangeObject, AbstractDaterangeObjects
from lib.ConfigLoader import ConfigLoader
from packages import Date
# from lib import Language
# from lib.data_retention_engine import update_obj_date, get_obj_date_first
from flask import url_for
config_loader = ConfigLoader()
r_cache = config_loader.get_redis_conn("Redis_Cache")
r_object = config_loader.get_db_conn("Kvrocks_Objects")
baseurl = config_loader.get_config_str("Notifications", "ail_domain")
IMAGE_FOLDER = config_loader.get_files_directory('images')
config_loader = None
# SET x1,y1:x2,y2:x3,y3:x4,y4:extracted_text
class Ocr(AbstractDaterangeObject):
"""
AIL Message Object. (strings)
"""
def __init__(self, id):
super(Ocr, self).__init__('ocr', id)
def exists(self):
return r_object.exists(f'ocr:{self.id}')
def get_content(self, r_type='str'):
"""
Returns content
"""
global_id = self.get_global_id()
content = r_cache.get(f'content:{global_id}')
if not content:
dict_content = {}
for extracted in r_object.smembers(f'ocr:{self.id}'):
extracted = extracted.split(':', 4)
x, y = extracted[0].split(',', 1)
# get text line, y +- 20
rounded_y = round(int(y) / 20) * 20
if rounded_y not in dict_content:
dict_content[rounded_y] = []
dict_content[rounded_y].append((int(x), int(y), extracted[-1]))
content = ''
new_line = True
l_key = sorted(dict_content.keys())
for key in l_key:
dict_content[key] = sorted(dict_content[key], key=lambda c: c[0])
for text in dict_content[key]:
if new_line:
content = f'{content}{text[2]}'
new_line = False
else:
content = f'{content} {text[2]}'
content = f'{content}\n'
new_line = True
# Set Cache
if content:
global_id = self.get_global_id()
r_cache.set(f'content:{global_id}', content)
r_cache.expire(f'content:{global_id}', 300)
if r_type == 'str':
return content
elif r_type == 'bytes':
if content:
return content.encode()
def get_date(self): # TODO
return Date.get_today_date_str()
def get_source(self): # TODO
"""
Returns source/feeder name
"""
return 'ocr'
# l_source = self.id.split('/')[:-2]
# return os.path.join(*l_source)
def get_basename(self): # TODO
return 'ocr'
def get_language(self):
languages = self.get_languages()
if languages:
return languages.pop()
else:
return None
def get_link(self, flask_context=False):
if flask_context:
url = url_for('correlation.show_correlation', type=self.type, id=self.id)
else:
url = f'{baseurl}/correlation/show?type={self.type}&id={self.id}'
return url
def get_svg_icon(self):
return {'style': 'fas', 'icon': '\uf065', 'color': 'yellow', 'radius': 5}
def get_image_path(self):
rel_path = os.path.join(self.id[0:2], self.id[2:4], self.id[4:6], self.id[6:8], self.id[8:10], self.id[10:12], self.id[12:])
filename = os.path.join(IMAGE_FOLDER, rel_path)
return os.path.realpath(filename)
def get_misp_object(self): # TODO
obj = MISPObject('instant-message', standalone=True)
obj_date = self.get_date()
if obj_date:
obj.first_seen = obj_date
else:
self.logger.warning(
f'Export error, None seen {self.type}:{self.subtype}:{self.id}, first={obj_date}')
# obj_attrs = [obj.add_attribute('first-seen', value=obj_date),
# obj.add_attribute('raw-data', value=self.id, data=self.get_raw_content()),
# obj.add_attribute('sensor', value=get_ail_uuid())]
obj_attrs = []
for obj_attr in obj_attrs:
for tag in self.get_tags():
obj_attr.add_tag(tag)
return obj
# options: set of optional meta fields
def get_meta(self, options=None, translation_target=''):
"""
:type options: set
"""
if options is None:
options = set()
meta = self._get_meta(options=options)
meta['tags'] = self.get_tags()
meta['content'] = self.get_content()
# optional meta fields
if 'investigations' in options:
meta['investigations'] = self.get_investigations()
if 'link' in options:
meta['link'] = self.get_link(flask_context=True)
if 'icon' in options:
meta['svg_icon'] = self.get_svg_icon()
if 'img' in options:
meta['img'] = self.draw_bounding_boxs()
if 'map' in options:
meta['map'] = self.get_img_map_coords()
if 'language' in options:
meta['language'] = self.get_language()
if 'translation' in options and translation_target:
if meta.get('language'):
source = meta['language']
else:
source = None
meta['translation'] = self.translate(content=meta.get('content'), source=source, target=translation_target)
if 'language' in options:
meta['language'] = self.get_language()
return meta
def get_objs_container(self):
objs_containers = set()
# chat
objs_containers.add(self.get_first_correlation('chat'))
subchannel = self.get_first_correlation('chat-subchannel')
if subchannel:
objs_containers.add(subchannel)
thread = self.get_first_correlation('chat-thread')
if thread:
objs_containers.add(thread)
return objs_containers
def create_coord_str(self, bbox):
c1, c2, c3, c4 = bbox
x1, y1 = c1
x2, y2 = c2
x3, y3 = c3
x4, y4 = c4
return f'{int(x1)},{int(y1)}:{int(x2)},{int(y2)}:{int(x3)},{int(y3)}:{int(x4)},{int(y4)}'
def _unpack_coord(self, coord):
return coord.split(',', 1)
def get_coords(self):
coords = []
for extracted in r_object.smembers(f'ocr:{self.id}'):
coord = []
bbox = extracted.split(':', 4)[:-1]
for c in bbox:
x, y = self._unpack_coord(c)
coord.append((int(x), int(y)))
coords.append(coord)
return coords
def get_img_map_coords(self):
coords = []
for extracted in r_object.smembers(f'ocr:{self.id}'):
extract = extracted.split(':', 4)
x1, y1 = self._unpack_coord(extract[0])
x2, y2 = self._unpack_coord(extract[1])
x3, y3 = self._unpack_coord(extract[2])
x4, y4 = self._unpack_coord(extract[3])
coords.append((f'{x1},{y1},{x2},{y2},{x3},{y3},{x4},{y4}', extract[4]))
return coords
def edit_text(self, coordinates, text, new_text, new_coordinates=None):
pass
def add_text(self, coordinates, text):
val = f'{coordinates}:{text}'
return r_object.sadd(f'ocr:{self.id}', val)
def remove_text(self, val):
return r_object.srem(f'ocr:{self.id}', val)
def update_correlation(self, date=None):
if date:
self.add(date, None)
image_correl = self.get_obj_correlations('image', '', self.id)
for obj_type in image_correl:
if obj_type != 'ocr':
for obj_raw in image_correl[obj_type]:
obj_subtype, obj_id = obj_raw.split(':', 1)
self.add_correlation(obj_type, obj_subtype, obj_id)
def create(self, extracted_texts, tags=[]):
# r_object.sadd(f'{self.type}:all', self.id)
created = False
for extracted in extracted_texts:
bbox, text = extracted
if len(text) > 1:
str_coords = self.create_coord_str(bbox)
self.add_text(str_coords, text)
created = True
if created:
# Correlations
self._copy_from('image', self.id)
self.update_correlation()
self.add_correlation('image', '', self.id)
for tag in tags:
self.add_tag(tag)
return self.id
# # WARNING: UNCLEAN DELETE /!\ TEST ONLY /!\
def delete(self):
r_object.delete(f'ocr:{self.id}')
def draw_bounding_boxs(self):
img = Image.open(self.get_image_path()).convert("RGBA")
draw = ImageDraw.Draw(img)
for bbox in self.get_coords():
c1, c2, c3, c4 = bbox
2024-04-09 12:22:11 +00:00
draw.line((tuple(c1), tuple(c2)), fill="yellow", width=2)
draw.line((tuple(c2), tuple(c3)), fill="yellow", width=2)
draw.line((tuple(c3), tuple(c4)), fill="yellow", width=2)
draw.line((tuple(c4), tuple(c1)), fill="yellow", width=2)
# img.show()
buff = BytesIO()
img.save(buff, "PNG")
return buff.getvalue()
def create(obj_id, detections, tags=[]):
obj = Ocr(obj_id)
if not obj.exists():
obj_id = obj.create(detections, tags=tags)
if obj_id:
return obj
# TODO preload languages
def extract_text(image_path, languages, threshold=0.2):
import easyocr
reader = easyocr.Reader(languages, verbose=False)
texts = reader.readtext(image_path)
# print(texts)
extracted = []
for bbox, text, score in texts:
if score > threshold:
extracted.append((bbox, text))
return extracted
def get_ocr_languages():
return {'af', 'ar', 'as', 'az', 'be', 'bg', 'bh', 'bs', 'cs', 'cy', 'da', 'de', 'en', 'es', 'et', 'fa', 'fr', 'ga', 'hi', 'hr', 'hu', 'id', 'is', 'it', 'ja', 'kn', 'ko', 'ku', 'la', 'lt', 'lv', 'mi', 'mn', 'mr', 'ms', 'mt', 'ne', 'nl', 'no', 'oc', 'pi', 'pl', 'pt', 'ro', 'ru', 'sk', 'sl', 'sq', 'sr', 'sv', 'sw', 'ta', 'te', 'th', 'tl', 'tr', 'ug', 'uk', 'ur', 'uz', 'vi', 'zh'}
def sanityze_ocr_languages(languages, ocr_languages=None):
langs = set()
if not ocr_languages:
ocr_languages = get_ocr_languages()
for lang in languages:
if lang in ocr_languages:
if lang == 'zh':
langs.add('ch_sim')
elif lang == 'sr':
langs.add('rs_latin')
else:
langs.add(lang)
return langs
class Ocrs(AbstractDaterangeObjects):
"""
OCR Objects
"""
def __init__(self):
super().__init__('ocr', Ocr)
def sanitize_id_to_search(self, name_to_search):
return name_to_search # TODO
#### API ####
def api_get_ocr(obj_id, translation_target=None):
ocr = Ocr(obj_id)
if not ocr.exists():
return {"status": "error", "reason": "Unknown ocr"}, 404
meta = ocr.get_meta({'content', 'icon', 'img', 'language', 'link', 'map', 'translation'}, translation_target=translation_target)
return meta, 200