7.6 support more file type and fix some problems

.idea/whale-anno.iml

@@ -1,7 +1,9 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <module type="PYTHON_MODULE" version="4">
   <component name="NewModuleRootManager">
-    <content url="file://$MODULE_DIR$" />
+    <content url="file://$MODULE_DIR$">
+      <excludeFolder url="file://$MODULE_DIR$/be/app/projects" />
+    </content>
     <orderEntry type="inheritedJdk" />
     <orderEntry type="sourceFolder" forTests="false" />
   </component>

be/.gitignore

@@ -1,3 +1,6 @@
 .idea
 __pycache__
 app/projects/*
+
+../.idea
+

be/app/api/v1/__init__.py

@@ -1,5 +1,5 @@
 from flask import Blueprint
-from app.api.v1 import index, project, files, anno
+from ...api.v1 import index, project, files, anno
 
 
 def create_blueprint_v1():

be/app/api/v1/files.py

@@ -70,6 +70,7 @@ def get_json():
         for item in anno_data:
             item_dict = {}
             item_dict['file'] = item['fileName']
+
             item_dict['text'] = read_txt_file(PROJECT_PATH.format(project_name)+'/'+item['fileName'])
             # 这一步来去掉anno.json中的isSmall
             for entity in item['annoDetails']:
@@ -94,12 +95,16 @@ def get_json():
     # 使用send_from_directory 或者使用send_file时要特别注意文件的路径，路径不对的话会报404
     # 本线默认目录时app下，所以不需要再加/app了，所以不能用PROJECT_PATH
     # 2.创建response对象返回数据
-    # response = make_response(send_from_directory(DOWNLOAD_FILE_LOCATION.format(project_name),filename='result.json', as_attachment=True))
+    #使用response可以将result.json再删除掉
-    # response.headers["Content-disposition"] = 'attachment; filename=result.json'
+    response = make_response(send_from_directory('', filename=DOWNLOAD_FILE_LOCATION.format(project_name),
-    # return response
+                                                 as_attachment=True))
+    response.headers["Content-disposition"] = 'attachment; filename={}_result.json'.format(project_name)
+    # print(PROJECT_PATH.format(project_name)+'/result.json')
+    os.remove(PROJECT_PATH.format(project_name)+'/result.json')
+    return response
 
     # 3. 直接使用send from directory 返回json文件
     # return send_from_directory('', filename=DOWNLOAD_FILE_LOCATION.format(project_name),as_attachment=True)
 
     # 4. 使用send file 返回json文件
-    return send_file(DOWNLOAD_FILE_LOCATION.format(project_name), as_attachment=True, attachment_filename=project_name+'_result.json')
+    # return send_file(DOWNLOAD_FILE_LOCATION.format(project_name), as_attachment=True, attachment_filename=project_name+'_result.json')

be/app/api/v1/project.py

@@ -1,3 +1,5 @@
+import shutil
+
 from ...config.setting import PROJECT_NAME, PROJECT_TYPE, ENTITY_TYPES, PROJECTS, PROJECT_PATH, PROJECT_CONFIG_PATH, \
     ANNO_OUTPUT_PATH
 from ...libs.redprint import RedPrint
@@ -14,23 +16,34 @@ from ...entities.entities import AnnoContents
 
 api = RedPrint('project')
 
+
 @api.route('/get_zipped_data', methods=['POST'])
 def get_zipped_data():
-    # time.sleep(0.02)
     ret_info = ReturnInfo()
     try:
         project_name = request.form.get('projectName')
-        print(project_name)
 
         upload_file = request.files['file']
         file_path = os.path.join(PROJECT_PATH.format(project_name), upload_file.filename)
         target_path = PROJECT_PATH.format(project_name)
-        # print(file_path)
         upload_file.save(file_path)
-        print(target_path)
+        file_type = file_path.split('.')[-1]
+        # print("zipped data received: .{} form".format(file_type))
         unzip_file(file_path, target_path)
         os.remove(file_path)
 
+        # Put files in the folder directly under the project directory
+        for item in os.listdir(target_path):
+            # If item is a folder,copy files in item to project folder,then remove this folder
+            folder_path = target_path + '/' + item
+            if os.path.isdir(folder_path):
+                for file in os.listdir(folder_path):
+                    shutil.copy(folder_path + '/' + file, folder_path + '/../')
+                    # Here  to deal with chinese encode in module zipfile and rarfile
+                    if file_type == 'zip':
+                        os.rename(folder_path + '/../' + file,
+                                  folder_path + '/../' + file.encode('cp437').decode('GBK'))
+                shutil.rmtree(folder_path)
     except Exception as e:
         print(e)
         ret_info.errCode = 404

be/app/app.py

@@ -2,7 +2,7 @@ from flask import Flask
 
 
 def create_app():
-    app = Flask(__name__, static_folder='../../fe/dist/', static_url_path='/')
+    app = Flask(__name__, static_folder='../../fe/dist/',template_folder='../../fe/dist/', static_url_path='/')
 
     app.config.from_object('app.config.setting')
     app.config.from_object('app.config.secure')

be/app/libs/tools.py

@@ -1,6 +1,14 @@
 import os
 import json
 import zipfile
+import rarfile
+import py7zr
+import tarfile
+import gzip
+from os import rename
+from os import listdir
+from shutil import move
+
 
 def make_dir(path):
     folder = os.path.exists(path)
@@ -31,16 +39,33 @@ def read_txt_file(file_path):
 
 def unzip_file(zip_src, dst_dir):
     """
-    解压zip文件
+    解压数据文件
-    :param zip_src: zip文件的全路径
+    :param zip_src: 压缩包文件的全路径
-    :param dst_dir: 要解压到的目的文件夹
+    :param dst_dir: 要解压到的项目文件夹
     :return:
     """
-    r = zipfile.is_zipfile(zip_src)
+
-    if r:
+    if zipfile.is_zipfile(zip_src):
         fz = zipfile.ZipFile(zip_src, "r")
         for file in fz.namelist():
             fz.extract(file, dst_dir)
-    else:
+        return "unzip .zip file success"
-        return "请上传zip类型压缩文件"
+    elif rarfile.is_rarfile(zip_src):
+        fr = rarfile.RarFile(zip_src, "r")
+        print(fr.namelist())
+        for file in fr.namelist():
+            fr.extract(file, dst_dir)
+        return "unzip .rar file success"
+    elif py7zr.is_7zfile(zip_src):
+        f7z = py7zr.SevenZipFile(zip_src, "r")
+        f7z.extractall(path=dst_dir)
+        f7z.close()
+    elif tarfile.is_tarfile(zip_src):
 
+        ft = tarfile.TarFile(zip_src, "r")
+        print(ft.getnames())
+        for file in ft.getnames():
+            ft.extract(file,dst_dir)
+
+    else:
+        return "请上传.zip .rar .tar .7z格式的文件"

be/app/projects/demo5/anno.json

@@ -1 +0,0 @@
-[{"fileName": "test10.txt", "annoDetails": [], "isAnno": true}, {"fileName": "test12.txt", "annoDetails": [{"name": "业于北京工业大", "type": "person1", "start": 3, "end": 10}], "isAnno": true}, {"fileName": "test6.txt", "annoDetails": [], "isAnno": true}, {"fileName": "test4.txt", "annoDetails": [], "isAnno": true}, {"fileName": "test2.txt", "annoDetails": [{"name": "小明毕业于北京工业", "type": "person1", "start": 0, "end": 9, "isSmall": false}, {"name": "小明毕业于北京", "type": "person1", "start": 0, "end": 7, "isSmall": true}, {"name": "小明毕业于北", "type": "person1", "start": 0, "end": 6, "isSmall": true}, {"name": "小明毕业于", "type": "person1", "start": 0, "end": 5, "isSmall": true}, {"name": "明毕业于北京工业", "type": "person1", "start": 1, "end": 9, "isSmall": true}, {"name": "明毕业于北京工", "type": "person1", "start": 1, "end": 8, "isSmall": true}, {"name": "明毕业于北京", "type": "person1", "start": 1, "end": 7, "isSmall": true}, {"name": "毕业于北京工业", "type": "person1", "start": 2, "end": 9, "isSmall": true}, {"name": "毕业于北京", "type": "person1", "start": 2, "end": 7, "isSmall": true}, {"name": "毕业于北", "type": "person1", "start": 2, "end": 6, "isSmall": true}, {"name": "毕业", "type": "person1", "start": 2, "end": 4}, {"name": "毕", "type": "person1", "start": 2, "end": 3, "isSmall": true}, {"name": "业于北京工业大", "type": "person1", "start": 3, "end": 10, "isSmall": true}, {"name": "业于北京", "type": "person1", "start": 3, "end": 7, "isSmall": true}, {"name": "于北京工", "type": "person1", "start": 4, "end": 8, "isSmall": true}, {"name": "于北京", "type": "person1", "start": 4, "end": 7, "isSmall": true}, {"name": "北京工", "type": "person1", "start": 5, "end": 8, "isSmall": true}, {"name": "京工", "type": "person1", "start": 6, "end": 8, "isSmall": true}, {"name": "工业大", "type": "person1", "start": 7, "end": 10, "isSmall": true}], "isAnno": true}, {"fileName": "test3.txt", "annoDetails": [{"name": "小明", "type": "person1", "start": 0, "end": 2}, {"name": "明毕业于北京", "type": "organiztion", "start": 1, "end": 7, "isSmall": false}, {"name": "京工业", "type": "123", "start": 6, "end": 9, "isSmall": true}, {"name": "业大学", "type": "location", "start": 8, "end": 11, "isSmall": true}], "isAnno": true}, {"fileName": "test11.txt", "annoDetails": [{"name": "北", "type": "person1", "start": 5, "end": 6, "isSmall": false}], "isAnno": true}, {"fileName": "test5.txt", "annoDetails": [{"name": "工业大学", "type": "person1", "start": 7, "end": 11}], "isAnno": true}, {"fileName": "test1.txt", "annoDetails": [{"name": "弟们 雄起", "type": "person1", "start": 11, "end": 16}], "isAnno": true}]

be/app/projects/demo5/config.json

@@ -1 +0,0 @@
-{"projectName": "demo5", "projectType": "命名实体识别", "entityTypes": "[{\"type\":\"person1\",\"color\":\"#e61490\"},{\"type\":\"location\",\"color\":\"#0aab8a\"},{\"type\":\"organiztion\",\"color\":\"#2770cd\"},{\"type\":\"123\",\"color\":\"#1c7a82\"}]"}

be/app/projects/demo5/test1.txt

@@ -1 +0,0 @@
-hello 你好啊 兄弟们 雄起

be/app/projects/demo5/test10.txt

@@ -1 +0,0 @@
-小明毕业于北京工业大学

be/app/projects/demo5/test11.txt

@@ -1 +0,0 @@
-小明毕业于北京工业大学

be/app/projects/demo5/test12.txt

@@ -1 +0,0 @@
-小明毕业于北京工业大学

be/app/projects/demo5/test2.txt

@@ -1 +0,0 @@
-小明毕业于北京工业大学

be/app/projects/demo5/test3.txt

@@ -1 +0,0 @@
-小明毕业于北京工业大学

be/app/projects/demo5/test4.txt

@@ -1 +0,0 @@
-小明毕业于北京工业大学

be/app/projects/demo5/test5.txt

@@ -1 +0,0 @@
-小明毕业于北京工业大学

be/app/projects/demo5/test6.txt

@@ -1 +0,0 @@
-小明毕业于北京工业大学

be/app/projects/demo5/test7.txt

@@ -1 +0,0 @@
-小明毕业于北京工业大学

be/app/projects/demo5/test8.txt

@@ -1 +0,0 @@
-小明毕业于北京工业大学

be/app/projects/demo5/test9.txt

@@ -1 +0,0 @@
-小明毕业于北京工业大学

be/run.py

@@ -1,6 +1,6 @@
 # 入口程序文件
 from app.app import create_app
-
+from flask import render_template
 app = create_app()
 
 # host='0.0.0.0'
@@ -13,5 +13,15 @@ app = create_app()
 # 9060是端口
 # v1是版本
 # index是首页
+
+
+@app.route('/')
+def index():
+    '''
+    using this route to load index page
+    '''
+    return render_template('/index.html')
+
+
 if __name__ == '__main__':
     app.run(host='127.0.0.1', port=9060, debug=True)