添加图片压缩，html 也对应压缩

Dockerfile

@@ -13,9 +13,10 @@ RUN sed -i 's/security.ubuntu/mirrors.aliyun/g' /etc/apt/sources.list && \
     apt-get -qq -y install git python3-pip && \
     apt-get clean && \
     rm -rf /var/lib/apt/lists/* && \
-    pip3 install flask requests
+    pip3 install flask requests pillow bs4 lxml -i https://pypi.tuna.tsinghua.edu.cn/simple 
 
 RUN git clone https://github.com/ider-zh/api-pdf2htmlex.git
 
+# COPY $PWD/* api-pdf2htmlex/
 EXPOSE 5000
 CMD python3 api-pdf2htmlex/run.py

Pipfile

@@ -1,10 +1,14 @@
 [[source]]
-url = "https://pypi.org/simple"
+url = "https://pypi.tuna.tsinghua.edu.cn/simple"
 verify_ssl = true
 name = "pypi"
 
 [packages]
 flask = "*"
+bs4 = "*"
+pillow = "*"
+lxml = "*"
+reuqests = "*"
 
 [dev-packages]
 requests = "*"

docker-compose.yml

@@ -0,0 +1,54 @@
+version: '2.2'
+services:
+  # pdf2htmlex_large:
+  #   build: .
+  #   #image: ider/api-pdf2htmlex
+  #   environment:
+  #     - PDF2PDF=http://pdfconv:5000/pdf2pdf
+  #     - POOL_SIZE=9
+  #   ports:
+  #     - 5001:5000
+  #   restart: always
+  #   mem_limit: 30G
+  #   memswap_limit: 800G
+  #   container_name: 'pdf2htmlEX_L'
+  #   links:
+  #     - pdfconv
+
+  # pdf2htmlex_small:
+  #   build: .
+  #   #image: ider/api-pdf2htmlex
+  #   environment:
+  #     - POOL_SIZE=10
+  #   expose:
+  #     - 5000
+  #   restart: always
+  #   mem_limit: 5G
+  #   memswap_limit: 5G
+  #   scale: 5
+
+  # pdf2htmlex-lb:
+  #   image: 'dockercloud/haproxy:latest'
+  #   volumes:
+  #     - /var/run/docker.sock:/var/run/docker.sock
+  #   links:
+  #     - pdf2htmlex_small
+  #   restart: always
+  #   ports:
+  #     - 5000:80
+
+  # pdfconv:
+  #   image: ider/libreoffice
+  #   restart: always
+
+  pdf2htmlex_small:
+    build: .
+    #image: ider/api-pdf2htmlex
+    environment:
+      - POOL_SIZE=10
+    ports:
+      - 5000:5000
+    restart: always
+    mem_limit: 5G
+    memswap_limit: 5G
+    scale: 1

file_compress.py

@@ -0,0 +1,61 @@
+'''
+@Version: 0.0.1
+@Author: ider
+@Date: 2019-08-10 16:55:38
+@LastEditors: ider
+@LastEditTime: 2019-08-10 18:09:52
+@Description: 压缩 图片和 html
+'''
+from bs4 import BeautifulSoup
+import codecs
+from PIL import Image
+import io
+import os
+
+def compress_base64_img(data):
+    img_data = codecs.decode(data.encode('utf8'),'base64')
+    img = Image.open(io.BytesIO(img_data))
+    newio = io.BytesIO()
+    img.save(newio,format='webp',quality=10,lossless=False)
+    newio.seek(0)
+    return codecs.encode(newio.read(),'base64').decode('utf8')
+
+
+def compress(file_path):
+    if file_path.endswith('.png'):
+        return compress_png(file_path)
+    elif file_path.endswith('.html'):
+        return compress_html(file_path)
+    return file_path
+
+def compress_png(file_path):
+    with open(file_path,'rb')as f:
+        im = Image.open(f)
+        new_file_path = file_path.rstrip('png') + 'webp'
+        with open(new_file_path,'wb')as fp:
+            im.save(fp,format='webp',quality=10,lossless=False)
+    return new_file_path
+
+def compress_html(file_path):
+    with open(file_path,'rb')as f:
+        html = f.read()
+    soup = BeautifulSoup(html,'lxml')
+    flag = 0
+    for img in soup.select('img'):
+        data = img.attrs.get('src')
+        if not data or not data.startswith('data:image'):
+            continue
+        if data.startswith('data:image/webp;base64'):
+            continue
+        line = data.split(',')
+        data = compress_base64_img(line[1])
+        img.attrs['src'] = 'data:image/webp;base64,'+data
+        flag = 1
+    if flag == 1:
+        with open(file_path,'wt')as f:
+            f.write(str(soup))
+    return file_path
+
+if __name__ == '__main__':
+    print(compress('/tmp/cccc/img-3-3.png'))
+    print(compress('/tmp/cccc/temp.html'))

pdf2htmlEX

@@ -27,6 +27,8 @@ def remote_convert(src, dst, argv):
             with zipfile.ZipFile(fp, 'r') as myzip:
                 myzip.extractall(dst)
             return
+        else:
+            print(req.text)
     else:
         print('remote api is not available!')
         # raise Exception('remote api is not available! %s'%req.text)