Update: support cnocr server mode and optimise output

README.md

@@ -18,6 +18,10 @@
 
 ## 版本
 
+### 4.9.2
+
+1. 支持 CNOCR 基于 FastAPI 的 HTTP [服务模式](https://github.com/breezedeus/cnocr#http%E6%9C%8D%E5%8A%A1)。
+
 ### 4.9.1
 
 1. 修复新版 CNOCR 的输出问题（因没有 mac 设备，只更新了源码）。

src/config.py

@@ -2,18 +2,14 @@
 @Description: Capture than OCR - Variable
 @Author: Chandler Lu
 @Date: 2020-03-09 20:32:15
-LastEditTime: 2021-01-07 17:29:41
+LastEditTime: 2022-09-12 17:50:31
 '''
 # -*- coding: UTF-8 -*-
 import os
 
 # Key - Alfred
 BAIDU_API_KEY = os.environ['baidu_api_key']
 BAIDU_SECRET_KEY = os.environ['baidu_secret_key']
-if os.environ['baidu_language_type'] == '':
-    BAIDU_LANGUAGE_TYPE = 'CHN_ENG'
-else:
-    BAIDU_LANGUAGE_TYPE = os.environ['baidu_language_type']
 TENCENT_SECRET_ID = os.environ['tencent_secret_id']
 TENCENT_SECRET_KEY = os.environ['tencent_secret_key']
 GOOGLE_ACCESS_TOKEN = os.environ['google_access_token']
@@ -24,12 +20,11 @@
 MATHPIX_APP_KEY = os.environ['mathpix_app_key']
 
 # Key - Quicker
-
 # BAIDU_API_KEY = 'rmMynojL9KapDOikDTgKlImy'
 # BAIDU_SECRET_KEY = '3QKoI1E56u16tEMdwBnpXSPNezdoZWFD'
 # TENCENT_SECRET_ID = ''
 # TENCENT_SECRET_KEY = ''
-# # GOOGLE_ACCESS_TOKEN = ''
+# GOOGLE_ACCESS_TOKEN = ''
 # GOOGLE_POST_REFERER = ''
 # GOOGLE_HTTP_PROXY = ''
 # CAIYUN_TRANSLATE_TOKEN = '3975l6lr5pcbvidl6jl2'
@@ -46,6 +41,10 @@
 CAIYUN_TRANSLATE_API = 'http://api.interpreter.caiyunai.com/v1/translator'
 MATHPIX_API = 'https://api.mathpix.com/v3/text'
 
+# CNOCR Control
+CNOCR_SERVE = 0
+CNOCR_API = 'http://127.0.0.1:8501/ocr'
+
 # Tencent Control
 TENCENT_SERVICE = 'ocr'
 TENCENT_ACTION = 'GeneralFastOCR'
@@ -54,6 +53,13 @@
 TENCENT_CONTENT_TYPE = 'application/json; charset=utf-8'
 
 # Baidu Control
+try:
+    if os.environ['baidu_language_type'] == '':
+        BAIDU_LANGUAGE_TYPE = 'CHN_ENG'
+    else:
+        BAIDU_LANGUAGE_TYPE = os.environ['baidu_language_type']
+except KeyError:
+    BAIDU_LANGUAGE_TYPE = 'CHN_ENG'
 BAIDU_OCR_SPACING_OFFSET = 8
 BAIDU_OCR_SPACING_VARIANCE = 15
 BAIDU_OCR_WIDTH_OFFSET = 50

src/ocr.py

@@ -3,7 +3,7 @@
 @version: 4.9.1
 @Author: Chandler Lu
 @Date: 2019-11-26 23:52:36
-LastEditTime: 2022-08-24 08:07:18
+LastEditTime: 2022-09-12 17:41:48
 '''
 # -*- coding: UTF-8 -*-
 import sys
@@ -60,12 +60,22 @@ def convert_image_base64(pic_path):
 
 
 def cnocr_ocr(pic_path):
-    from cnocr import CnOcr
-    ocr = CnOcr(det_model_name='naive_det')
-    res = ocr.ocr(pic_path)
+    if c.CNOCR_SERVE == 0:
+        from cnocr import CnOcr
+        ocr = CnOcr(det_model_name='naive_det')
+        res = ocr.ocr(pic_path)
+    elif c.CNOCR_SERVE == 1:
+        try:
+            req = requests.post(
+                c.CNOCR_API, timeout=1.5, files={'image': (pic_path, open(pic_path, 'rb'))})
+        except requests.Timeout:
+            print('Request timeout!', end='')
+            sys.exit(0)
+        res = req.json()['results']
     for i in res:
         print(i['text'], end='')
 
+
 '''
 Baidu OCR
 '''
@@ -268,8 +278,12 @@ def mathpix_ocr(pic_path):
             })
         )
         if (response.status_code == 200):
-            response_json = response.json()['latex_styled']
-            print(response_json, end='')
+            if 'error' in response.json():
+                print(response.json()['error'], end='')
+            elif 'latex_styled' in response.json():
+                print(response.json()['latex_styled'], end='')
+            else:
+                print(response.json()['text'], end='')
         else:
             print('Request failed!', end='')
     except requests.exceptions.ConnectionError:
@@ -495,9 +509,9 @@ def remove_pic(pic_path):
  ________
 < rabbit >
  --------
-        \   ^__^
-         \  (oo)\_______
-            (__)\       )\/\
+        \\   ^__^
+         \\  (oo)\\_______
+            (__)\\       )\\/\
                 ||----w |
                 ||     ||
 '''