def detect_text(self, file):
ocr_time = timeutil.TimeElapsed()
words_list = []
tick = datetime.now()
"""Detects text in the file."""
client = vision.ImageAnnotatorClient()
sys.path.append('../util')
# sys.path.append('..\\util')
# print(sys.path)
# [START vision_python_migration_text_detection]
with io.open(file, 'rb') as image_file:
content = image_file.read()
image = vision.types.Image(content=content)
response = client.text_detection(image=image)
texts = response.text_annotations
# for text in texts:
# print("{}".format(text.description))
# sysutil.copy2clip(text.description)
# pyperclip.copy(text.description)
# break;
index = 0
for text in texts:
index += 1
# 방법 1: vertext 정보를 list로 변환하여 return
# print(text.description, text.bounding_poly.vertices)
# print(type(text.description), type(text.bounding_poly.vertices))
# print('"{} [{:>15s}]"'.format(index, text.description))
vertices = [[int(vertex.x), int(vertex.y)]
for vertex in text.bounding_poly.vertices]
temp = [text.description, vertices]
# 방법 2: 직접 generator object를 return, 사용할 때 철가 필요
# print(text.description, text.bounding_poly.vertices)
# print(type(text.description), type(text.bounding_poly.vertices))
# print('"{} [{:>15s}]"'.format(index, text.description))
# vertices = ('[{},{}]'.format(int(vertex.x), int(vertex.y))
# for vertex in text.bounding_poly.vertices)
words_list.append(temp)
print("Time to spend for OCR 인식", ocr_time.getelapsed())
return words_list
def load_data(self, filename):
self.filename = filename
loadingtime = timeutil.TimeElapsed()
with open(self.filename, 'r', encoding='utf8') as words_f:
# file에서 읽어서 loading하는 부분은 실 환경에 따라 변경됨.
for line in words_f:
# print('[', line, ']')
data = line.replace('\n', '')
if data.__len__() == 0:
continue
data = data.replace('\t', ' ')
wordslist = data.split(' ')
print(wordslist)
for word in wordslist:
DataLoader.wordsset.add(word)
print("Time to spend for loading", loadingtime.getelapsed())
print("Words count", len(DataLoader.wordsset))
def detect_text(path):
ocrtimer = timeutil.TimeElapsed()
"""Detects text in the file."""
client = vision.ImageAnnotatorClient()
with io.open(path, 'rb') as image_file:
content = image_file.read()
image = vision.types.Image(content=content)
# image = vision.types.Image()
response = client.text_detection(image=image)
texts = response.text_annotations
print('Type(texts):', type(texts))
print('Texts:', texts)
print('--------------------')
print("\nOCR elapsed: ", ocrtimer.getelapsed())
print('\nAll Data:')
# ---- 전체 단어와 position을 나열한다
count = 0
for text in texts:
newtext = text.description.replace('\n', ' ')
print('{:>2} Desc:[{:>20s}],'.format(count, newtext), end='')
vertices = ([
'({},{})'.format(vertex.x, vertex.y)
for vertex in text.bounding_poly.vertices
])
# print('\n\tvertices:', vertices, ' Type:', type(vertices))
print('bounds: {}'.format(','.join(vertices)))
count = count + 1
# ---- 해당하는 단어의 position을 가져온다
print('--------------------')
for text in texts:
if text.description == '당도':
for vertex in text.bounding_poly.vertices:
print('Found....', vertex)
import sys
sys.path.insert(1, '/Users/andrew/Documents/py_basic/util')
import timeutil
def loadDataToSet(data):
dataset = set()
dataset.add('123')
dataset.add('abc')
return dataset
if __name__ == '__main__':
loadtimer = timeutil.TimeElapsed()
dataset = loadDataToSet('copy2clip is working')
print("loading....", loadtimer.getelapsed())
count = 0
testdate = ['abc', 'def']
for word in testdate:
if word in dataset:
count = count + 1
print("searching....", loadtimer.getelapsed(), ",", count)
'''
1안
DB에 있는 단어, 조사를 완성하여 set을 구성
OCR로 인식된 단어를 set에서 검색
2안
OCR로 인식된 단어로 set을 구성
def load_data(self, fname):
self.filename = fname
loadingtime = timeutil.TimeElapsed()
num_not_necessary = 0
name_of_sheet = '05.31'
# df = pd.read_excel(filename, sheet_name='Sheet1')
df = pd.read_excel(self.filename,
sheet_name=name_of_sheet,
header=num_not_necessary)
count = 0
# col_lists = ['공통', '공통\n(실증필요단어)', '의류', '속옷', '패션잡화', '가구', '침구/침장', '가전', '스포츠/레저', '식품', '주방용품', \
# '이미용\n(실증필요단어)', '생활/잡화', '유아동용품', '문화/서비스', '보석/장신구', '건강', '이미용']
col_lists = df.columns
# print('col_lists ==>', col_lists)
# ---- column들의 길이를 가져온다 ----
# maxColumnLenghts = []
# for col in range(len(df.columns)):
# maxColumnLenghts.append(max(df.iloc[:,col].astype(str).apply(len)))
# print('Max Column Lengths ', maxColumnLenghts)
if num_not_necessary is not None:
max_index = df.__len__() - num_not_necessary
else:
max_index = df.__len__()
for col in col_lists:
print('-------------', col, '------------------')
for i in range(0, max_index):
cell_value = df.at[i, col]
if str(cell_value).replace(' ', '') == 'nan':
continue
# 값을 보정 (특히 숫자인 경우 필요)
if cell_value == 1:
cell_value = '100%'
print('{:3>d} {}'.format(count, cell_value))
DataLoader.wordsset.add(cell_value)
count = count + 1
count = self.add_word_with_postposition(cell_value, count)
# with open(self.filename, 'r', encoding='utf8') as words_f:
# # file에서 읽어서 loading하는 부분은 실 환경에 따라 변경됨.
# for line in words_f:
# # print('[', line, ']')
# data = line.replace('\n', '')
# if data.__len__() == 0:
# continue
#
# data = data.replace('\t', ' ')
# wordslist = data.split(' ')
# print(wordslist)
#
# for word in wordslist:
# DataLoader.wordsset.add(word)
print("Time to spend for 금지어 로딩", loadingtime.getelapsed())
print("Words count", len(DataLoader.wordsset))
# ---- 전체 단어와 position을 나열한다
count = 0
for text in texts:
newtext = text.description.replace('\n', ' ')
print('{:>2} Desc:[{:>20s}],'.format(count, newtext), end='')
vertices = ([
'({},{})'.format(vertex.x, vertex.y)
for vertex in text.bounding_poly.vertices
])
# print('\n\tvertices:', vertices, ' Type:', type(vertices))
print('bounds: {}'.format(','.join(vertices)))
count = count + 1
# ---- 해당하는 단어의 position을 가져온다
print('--------------------')
for text in texts:
if text.description == '당도':
for vertex in text.bounding_poly.vertices:
print('Found....', vertex)
if __name__ == '__main__':
wholetimer = timeutil.TimeElapsed()
# detect_text('/Volumes/USB3-64/Image/10167389_1.jpg')
detect_text(
'/Volumes/USB3-64/Image/3. 10164315 쁘띠프루티 울트라 모이스처라이징 페이셜 마스크팩.jpg'
)
print("\n\nTotal elapsed: ", wholetimer.getelapsed())
def main():
global loader
STX = 2
# 엑셀 파일에 있는 금칙어를 로드,
loader = dataloader.DataLoader()
loader.load_data('/Users/andrew/Downloads/pWords.xlsx')
# 형태소분석기를 로드
tokenizer = analyser.Analyser()
# TCP/IP 구성 정보를 로드
global tcp_config
global conn
global interval
try:
with open('tcp_config.json', encoding='utf-8') as json_file:
tcp_config = json.load(json_file)
except FileNotFoundError:
print("No File exists...")
exit('socket configuration exception')
host = tcp_config['hostname']
port = tcp_config['port']
interval = tcp_config['interval']
error_cnt = 0
while True:
conn = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
try:
conn.connect((host, port))
except ConnectionRefusedError:
conn.close()
time.sleep(10)
continue
while True:
try:
content = conn.recv(200)
except socket.error as e:
print("error while receiving :: " + str(e), e.errno)
if e.errno == errno.EPIPE:
conn.close()
break
else:
raise
# exit("terminating")
except:
print("error 2 while receiving :: ")
print(errno)
break
line_time = timeutil.TimeElapsed()
if len(content) == 0:
error_cnt += 1
if error_cnt > 3:
error_cnt = 0
break
if content[0] != STX:
print('wrong data from the server..')
continue
actual_data = content[2:]
print('[' + actual_data.decode(encoding='cp949') + ']')
# 일반로직에 따라수신한 문장의 단어별 일따라 금지어 존재 검색
pword_list = check_pword(actual_data.decode(encoding='cp949'))
if len(pword_list) > 1:
print(pword_list, len(pword_list))
# 복합명사 처리를 위해 속기기록을 형태소 분석하여 명사만 가져온다
nouns_list = tokenizer.get_noun_tokens(
actual_data.decode(encoding='cp949'))
for noun in nouns_list:
print('Noun', noun)
print("Time spent to analyse line: ", line_time.getelapsed())
print('------------------------------------')
import sys
import os
# sys.path.append('/Users/andrew/Documents/py_basic/util')
sys.path.insert(1, '/Users/andrew/Documents/py_basic/util')
# sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), '../..')))
# print(sys.path)
import timeutil
a = timeutil.TimeElapsed()
for i in range(10000):
print(".", end='')
print()
print("elapsed....", a.getelapsed())
for i in range(2000):
print(i, end='')
print()
print("elapsed2....", a.getelapsed())
def main():
# 엑셀 파일에 있는 금칙어를 로드,
loader = dataloader.DataLoader()
loader.load_data(
'/Users/andrew/Documents/RPA_신세계TV쇼핑/SSG닷컴_RM_키워드(금칙어)_해제 완료.xlsx'
)
totaltime = timeutil.TimeElapsed()
print('-----------------------------------------------')
# GCV 오류 발생한 단어에 대한 dictionary를 로드
gcv_error_data = {}
try:
with open('gcv_error_dic.json', encoding='utf-8') as json_file:
gcv_error_data = json.load(json_file)
except FileNotFoundError:
print("No File exists...")
print('-----------------------------------------------')
# 상품 이미지 파일을 GCV를 이용하여 텍스트 인식
hndlr = textDetectionUsingGCV.OCRHandler()
text_lists = hndlr.detect_text(
'/Volumes/USB3-64/Image/1. 10160920 다나한 인삼잎 보윤 수분크림.jpg'
)
# text_lists = hndlr.detect_text('/Volumes/USB3-64/Image/2. 10160902 참존 마유 골든컴플렉스 2종 세트.png')
# text_lists = hndlr.detect_text('/Volumes/USB3-64/Image/3. 10164315 쁘띠프루티 울트라 모이스처라이징 페이셜 마스크팩.jpg')
# text_lists = hndlr.detect_text('/Volumes/USB3-64/Image/4. 10164318 쁘띠프루티 카밍 앤 브라이트 페이셜 마스크팩.png')
# text_lists = hndlr.detect_text('/Volumes/USB3-64/Image/5. 10164572 참존 인텐시브 골드 앰플.jpg')
# text_lists = hndlr.detect_text('/Volumes/USB3-64/Image/6. 10166107 일동제약 퍼스트랩 프로바이오틱 마스크팩.png')
# text_lists = hndlr.detect_text('/Volumes/USB3-64/Image/6. 10166107 일동제약-1.png')
# text_lists = hndlr.detect_text('/Volumes/USB3-64/Image/6. 10166107 일동제약-2.png')
# text_lists = hndlr.detect_text('/Volumes/USB3-64/Image/6. 10166107 일동제약-3.png')
# text_lists = hndlr.detect_text('/Volumes/USB3-64/Image/7. 10167823 이오 에브리원 솝 코코넛레몬.jpg')
# text_lists = hndlr.detect_text('/Volumes/USB3-64/Image/8. 10060298 티레이저 기기 + 크림.jpg')
print('-----------------------------------------------')
words_found_list = []
# 인식된 각 텍스트를 금칙어에 있는지 확인
count = 1
for curr_word in text_lists[1:]:
# 1. single 단어를 검색한다
result = loader.find_word(curr_word[DESC])
if result == FOUND:
print("금지어 발견 -->", curr_word[DESC], curr_word[VERTEX])
words_found_list.append(curr_word)
count = count + 1
continue
# 미발견 시 복합단어 처리 위해, 그러나 다음 단어가 없으면 더 이상 복합 단어 처리는 하지 않음
if count + 1 >= len(text_lists):
count = count + 1
continue
next_word = text_lists[count + 1]
# single 단어 검색 안되면, 인접 단어와 조합하여 검색. 그러나, 인접하지 않으면 처리하지 않음
front_p2_x = curr_word[VERTEX][P2][X]
front_p2_y = curr_word[VERTEX][P2][Y]
back_p1_x = next_word[VERTEX][P1][X]
back_p1_y = next_word[VERTEX][P1][Y]
if abs(front_p2_x - back_p1_x) > 10 or abs(front_p2_y -
back_p1_y) > 10:
count = count + 1
continue
# 2. single 단어 검색 안되면, 인접 단어와 space 넣어 조합하여 검색
combined_word = curr_word[DESC] + ' ' + next_word[DESC]
result = loader.find_word(combined_word)
if result == FOUND:
print("복합 금지어 발견 1-->", combined_word)
count = count + 1
continue
# 3. single 단어 검색 안되면, 인접 단어와 space 없이 조합하여 검색
combined_word = curr_word[DESC] + next_word[DESC]
# print(combined_word, len(combined_word))
result = loader.find_word(combined_word)
if result == FOUND:
print("복합 금지어 발견 2-->", combined_word)
count = count + 1
continue
# 4. 오류 사전에서 combined word에 대한 대체명사를 검색하여 있으면 대체명사로 검색
try:
replaced_word = gcv_error_data[combined_word]
# print('대체어: [', combined_word, '->', replaced_word, ']')
except KeyError:
replaced_word = ''
if replaced_word != '':
result = loader.search_in_dict(replaced_word)
if result:
print("대체어 금지어 발견 3-->", curr_word[DESC], curr_word[VERTEX])
count = count + 1
continue
# 5. combined word 길이가 4음절 이상이면, ngram 분리하여 각각 검색
grams = [2, 3, 4, 5]
if len(combined_word) > 3:
for gram in grams:
start = 0
end_pos = len(combined_word) + 1
for end in range(gram, end_pos):
substr = combined_word[start:end]
start = start + 1
# print('NGRAM 단어:', substr)
result = loader.search_in_dict(substr)
if result == FOUND:
print("부분 금지어 발견 4-->", curr_word[DESC],
curr_word[VERTEX])
break
count = count + 1
print("Time spent for 금지어 발견: ", totaltime.getelapsed())
print('-----------------------------------------------')
# print('length : ', length)
# print('width : ', width)
# print('height : ', height)
# print('fps : ', fps)
# cap = cv2.VideoCapture('BotInsight_M7_L3.mp4')
# cap = cv2.VideoCapture('/Volumes/NO NAME/20191021 영상기술검수 에러반려 예시/video black/N2018010300026_기술검수반려 [M]제왕수산 갈치 뷰티_갈치조림 (수정).MXF')
cap = cv2.VideoCapture('/Volumes/NO NAME/20191021 영상기술검수 에러반려 예시/video black/N2018010800080_기술검수반려 [M]AHC 아이크림 시즌6_인포.MXF')
if not cap.isOpened():
print("already open..")
print("so, exit..")
exit(0)
timer = timeutil.TimeElapsed()
count = 0
frametimer = timeutil.TimeElapsed()
framepos = 0.0
stop_flag = False
while True:
if stop_flag:
break
# 읽을 frame 위치를 지정
cap.set(cv2.CAP_PROP_POS_FRAMES, framepos)
ret, frame = cap.read()ㅜ
if not ret:
break
