def __process_matches(matches: list) -> Result:
result = Result()
for match in matches:
result.merge_with({
'rules': [match.rule],
'strings': match.strings,
'score': match.meta['score']
})
return result
def process(self, temp_path: str, real_path: str) -> Result:
result = Result()
path = real_path.replace(self.git_root, '')
if path in self.changed_files:
result.score = self.UNCOMMITTED_FILES_WEIGHTING
result.rules = ['Git_Uncommitted_Changes']
else:
result.score = self.COMMITTED_FILES_WEIGHTING
return result
def getCompanies(self, name: str = None):
"""
Retrieves one or many items from database depending on the inclusion of a filter parameter
"""
try:
if name is None:
companies = self.database.getCompanies({})
else:
companies = self.database.getCompanies({"Company Name": name})
return Result(200, 'successful', companies)
except Exception as ex:
self.logger.error(ex)
return Result(400, 'An Error Occurred.', [])
def process(self, temp_path: str, real_path: str) -> Result:
result = Result()
wp_path = real_path.replace(self.wp_root, '')
if wp_path not in self.wp_checksums:
return result
elif self.wp_checksums[wp_path] != self.__get_file_checksum(real_path):
result.rules.append('Hash_Verification_Failure')
result.score = self.HASH_SCORE_WEIGHTING
else:
result.rules.append('Hash_Verification_Success')
result.score = -self.HASH_SCORE_WEIGHTING
return result
def category(title):
news_title_list = []
news_title_list.append(title)
data = text2matrix(news_title_list)
# 使用 MNB 分类模型对新闻进行分类
news_category_list = classfiy_predict(model_path='mnb.model', data=data)
return Result(items=news_category_list[0]).to_json()
def news(offset, limit=20):
result = {}
items = []
category = request.args.get('category')
keyword = request.args.get('keyword')
if offset < 0 or limit <= 0:
result = Result(code=Result.FAIL, msg='offset或limit参数有误！')
return result.to_json()
# 筛选字段，过滤 _id(ObjectId) 避免序列化异常
column = {'_id': 0}
# 组装查询条件，无法拼接出带模糊查询的 sql，pymongo 就是个垃圾！！！
# condition = {}
# if category:
# condition['news_web_category'] = category
# if keyword:
# regex = {'$regex': '.*{}.*'.format(keyword)}
# condition['new_title'] = regex
# print(condition)
try:
if not category and not keyword:
items = list(mongo.collection.find({}, column).sort('news_datetime', -1).skip(offset).limit(limit))
count = mongo.collection.find({}, column).count()
result = PageResult(items, count=count)
if category and not keyword:
items = list(mongo.collection.find({'news_web_category': category}, column).sort('news_datetime', -1).skip(
offset).limit(limit))
count = mongo.collection.find({'news_web_category': category}, column).count()
result = PageResult(items, count=count)
if not category and keyword:
items = list(
mongo.collection.find({'news_title': re.compile(keyword, re.IGNORECASE)}, column).sort('news_datetime',
-1).skip(
offset).limit(limit))
count = mongo.collection.find({'news_title': re.compile(keyword, re.IGNORECASE)}, column).count()
result = PageResult(items, count=count)
if keyword and category:
items = list(
mongo.collection.find({'news_web_category': category, 'news_title': re.compile(keyword, re.IGNORECASE)},
column).sort('news_datetime', -1).skip(offset).limit(limit))
count = mongo.collection.find(
{'news_web_category': category, 'news_title': re.compile(keyword, re.IGNORECASE)}, column).count()
result = PageResult(items, count=count)
except Exception as e:
result = Result(code=Result.FAIL, msg=e)
return result.to_json()
def scan(path_list: List[str], config: Any) -> Dict[str, Result]:
"""
For each of the given file paths, scan it for malicious PHP
patterns and assign it a score based on how suspicious it seems.
"""
processors = get_processors(config)
num_targets = -1
scan_count = 0
suspicious_count = 0
results = {}
if config.show_progress:
path_list = list(path_list)
num_targets = len(path_list)
log.info("{} files found. Scanning...".format(num_targets))
last_update = time.time()
for path in path_list:
scan_count += 1
if os.path.isdir(path):
continue
if not validate_file(path):
log.debug(path + " doesn't look like a PHP file, skipping.")
continue
results[path] = Result(config.score_alert_threshold)
log.debug("Scanning " + path)
temp_path = preprocess(path, TMP_DIR)
for processor in processors:
if processor.ready():
results[path].merge_with(processor.process(temp_path, path))
if results[path].is_suspicious():
suspicious_count += 1
if config.show_progress and time.time(
) > last_update + PROGRESS_UPDATE_PERIOD:
last_update = time.time()
print_progress(scan_count, num_targets, suspicious_count,
config.start_time)
os.unlink(temp_path)
if config.priority < 3:
sleep((3 - config.priority) * 0.05)
return results
def measure_tree_width(image_path, trunk_corners):
result = Result() # 保存当前图片的分割结果，初始化为失败
result.set_image_path(image_path)
result.set_time(0)
if not os.path.exists(image_path):
result.set_info(InfoEnum.IMAGE_NOT_EXIST)
return result
if trunk_corners is None:
result.set_info(InfoEnum.BAD_MANUAL_ANNO)
return result
result.set_trunk_left_top(trunk_corners['left_top'])
result.set_trunk_left_bottom(trunk_corners['left_bottom'])
result.set_trunk_right_top(trunk_corners['right_top'])
result.set_trunk_left_bottom(trunk_corners['right_bottom'])
im = cv2.imread(image_path) # 加载图片
im, resize_ratio = resize_image_with_ratio(im) # 调整尺寸
# 检测标定物
calibrator = get_calibrator(im, 0, DEBUG and SHOW)
if calibrator is None:
result.set_info(InfoEnum.CALIBRATOR_DET_FAILED)
return result
# 在结果中保存激光点坐标
# resize坐标 -> 原始坐标
org_calibrate_pts = []
for calibrate_pt in calibrator.get_calibrate_points():
org_calibrate_pt = recover_coordinate(calibrate_pt, resize_ratio)
org_calibrate_pts.append(org_calibrate_pt)
result.set_calibrate_points(org_calibrate_pts)
# 计算树径
trunk_left_top = resize_coordinate(trunk_corners['left_top'], resize_ratio)
trunk_left_bottom = resize_coordinate(trunk_corners['left_bottom'],
resize_ratio)
trunk_right_top = resize_coordinate(trunk_corners['right_top'],
resize_ratio)
trunk_right_bottom = resize_coordinate(trunk_corners['right_bottom'],
resize_ratio)
l_line = Edge(trunk_left_top, trunk_left_bottom)
r_line = Edge(trunk_right_top, trunk_right_bottom)
alpha = angle(l_line.vec(), r_line.vec())
if alpha < 5:
pixel_dis_l2r = parallel_distance(l_line, r_line)
pixel_dis_r2l = parallel_distance(r_line, l_line)
if pixel_dis_r2l is not None and pixel_dis_r2l is not None:
pixel_width = (pixel_dis_l2r + pixel_dis_r2l) / 2.0
RP_ratio = calibrator.RP_ratio()
real_width = (pixel_width * RP_ratio)
result.set_info(InfoEnum.SUCCESS)
result.set_width(real_width)
result.set_conf(1.0)
else:
result.set_info(InfoEnum.BAD_MANUAL_ANNO)
else:
result.set_info(InfoEnum.BAD_MANUAL_ANNO)
return result
def measure_all(image_path_list):
"""
为一批图像测算树径
:param image_path_list: 图像路径列表
:return: 树径测算结果列表
"""
if image_path_list is None or len(image_path_list) == 0:
# 输入不合法
return {'results': []}
seg_count = 0 # 计数分割操作的次数
results_all = [] # 保存所有结果
image_num = len(image_path_list)
for i, im_path in enumerate(image_path_list):
im_id = im_path.split('/')[-1].split('.')[0]
result = Result() # 当前图片测量结果
result.set_image_path(im_path)
results_all.append(result.get_result())
time_start = time.time()
if DEBUG:
print('-' * 30)
print('[%d/%d]: %s' % (image_num, i + 1, im_path.split('/')[-1]))
if not os.path.exists(im_path):
# 图像不存在
result.set_info(InfoEnum.IMAGE_NOT_EXIST)
continue
im_org = cv2.imread(im_path) # 加载图片
im, resize_ratio = resize_image_with_ratio(im_org) # 调整尺寸
# step1: 检测标定物
calibrator = get_calibrator(im, im_id, DEBUG and SHOW)
if calibrator is None:
result.set_info(InfoEnum.CALIBRATOR_DET_FAILED)
continue
# 检查站位
calibrator_ratio = calibrator.get_pixel_scale() / im.shape[0]
if calibrator_ratio > 0.25:
result.set_info(InfoEnum.STAND_TOO_CLOSE)
return result
# 在结果中保存标定物坐标
# resize坐标 -> 原始坐标
org_calibrate_pts = []
for calibrate_pt in calibrator.get_calibrate_points():
org_calibrate_pt = recover_coordinate(calibrate_pt, resize_ratio)
org_calibrate_pts.append(org_calibrate_pt)
result.set_calibrate_points(org_calibrate_pts)
# step2: 覆盖标定物，避免标定物影响分割
im_cover = calibrator.cover_calibrator(im)
if DEBUG:
visualize_image(im_cover,
'img_cover',
im_id=im_id,
show=DEBUG and SHOW)
# 参考标定物切割图片
# 图片块尺寸由小到大迭代尝试
crop_params = [2, 4]
for j, n_crop in enumerate(crop_params):
im_patch = calibrator.crop_image(im_cover, n_dis_w=n_crop)
patch_h, patch_w, _ = im_patch.shape
if DEBUG:
print('H:%d, W:%d' % (patch_h, patch_w))
visualize_image(im_patch,
'patch_%d' % n_crop,
im_id=im_id,
show=DEBUG and SHOW)
# step4: 分割树干
if im_patch.shape[0] > config.NET_MAX_WIDTH and im_patch.shape[
1] > config.NET_MAX_WIDTH:
# 待分割的目标尺寸太大
result.set_info(InfoEnum.TRUNK_TOO_THICK)
break
# 交互式分割
trunk_mask = segment_trunk_int(im_patch,
calibrator.positive_pts(),
None,
im_id=seg_count)
seg_count += 1
if DEBUG:
visualize_image(trunk_mask,
'trunk_mask',
im_id=im_id,
show=DEBUG and SHOW)
# step5: 计算树径
trunk = Trunk(trunk_mask)
if DEBUG:
visualize_image(trunk.contour_mask,
'trunk_contour',
im_id=im_id,
show=DEBUG and SHOW)
if not trunk.is_seg_succ():
# 初步估计分割是否成功
result.set_info(InfoEnum.TRUNK_EDGE_UNCLEAR)
if DEBUG:
print(InfoEnum.TRUNK_EDGE_UNCLEAR)
continue
else:
# 计算实际树径
RP_ratio = calibrator.RP_ratio() # 缩放因子
shot_distance = calibrator.shot_distance() # 拍摄距离
trunk_width, seg_conf, patch_trunk_corners = trunk.real_width_v2(
shot_distance, RP_ratio)
if trunk_width > 0:
# 置信度：分割置信度 x 标定置信度
conf = seg_conf * calibrator.get_conf()
time_end = time.time()
time_consume = int(time_end - time_start)
# 图片块坐标 -> resize图片坐标
trunk_left_top = calibrator.recover_coordinate(
patch_trunk_corners['left_top'])
trunk_right_top = calibrator.recover_coordinate(
patch_trunk_corners['right_top'])
trunk_left_bottom = calibrator.recover_coordinate(
patch_trunk_corners['left_bottom'])
trunk_right_bottom = calibrator.recover_coordinate(
patch_trunk_corners['right_bottom'])
# resize坐标 -> 原图坐标
trunk_left_top = recover_coordinate(
trunk_left_top, resize_ratio)
trunk_left_bottom = recover_coordinate(
trunk_left_bottom, resize_ratio)
trunk_right_top = recover_coordinate(
trunk_right_top, resize_ratio)
trunk_right_bottom = recover_coordinate(
trunk_right_bottom, resize_ratio)
result.set_width(trunk_width)
result.set_conf(conf)
result.set_trunk_left_top(trunk_left_top)
result.set_trunk_right_top(trunk_right_top)
result.set_trunk_left_bottom(trunk_left_bottom)
result.set_trunk_right_bottom(trunk_right_bottom)
result.set_info(InfoEnum.SUCCESS)
result.set_time(time_consume)
if DEBUG:
print('Trunk width: %.2f CM (%.2f).' %
(trunk_width / 10.0, conf))
pts = org_calibrate_pts + \
[trunk_left_top, trunk_left_bottom, trunk_right_top, trunk_right_bottom]
show_pts(im_org, pts, im_id=im_id, show=DEBUG and SHOW)
break
else:
result.set_info(InfoEnum.TRUNK_EDGE_UNCLEAR)
if DEBUG:
print('Error is too large.')
output = {'results': results_all}
return output
# path = '/home/hadoop/Desktop/test_all'
# filelist = os.listdir(path)
# right = 0
# for file in filelist:
# src_name = file.split('.')[0]
# result = num_recog.main(os.path.join(path,file))
# print src_name,result
# if int(src_name) == int(result):
# right += 1
# else:
# continue
# right_rate = float(right)/len(filelist)
# print ('rate is {}%'.format(right_rate*100))
if __name__ == "__main__":
result = Result()
img_type = sys.argv[1]
img_file = sys.argv[2]
# img_type = "file"
# img_file = "D:/machine_learning/recognization/insurance_doc/1.jpg"
try:
num_recog = Number_recognition()
predict_data = num_recog.main(img_file)
data = {'captcha': predict_data}
result.code = code.SUCCESS
result.msg = "success"
result.data = data
except Exception, e:
exstr = traceback.format_exc()
logging.info('#####系统异错误error:%s', exstr)
result.code = code.FAIL