Add gradio app support for module (#2132)

* add gradio app support for module

* add gradio app support for module

* add gradio app support for module

* fix

* add ernie_vilg demo

* upgrade

* update module version

modules/image/text_to_image/ernie_vilg/README.md

@@ -400,6 +400,9 @@ DiscoDiffusion Prompt 技巧资料：https://docs.google.com/document/d/1l8s7uS2
       image = Image.open(BytesIO(base64.b64decode(result)))
       image.save('result_{}.png'.format(i))
 
+- ### gradio app 支持
+  从paddlehub 2.3.1开始支持使用链接 http://127.0.0.1:8866/gradio/ernie_vilg 在浏览器中访问ernie_vilg的gradio app。
+
 
 ## 六、更新历史
 
@@ -415,6 +418,9 @@ DiscoDiffusion Prompt 技巧资料：https://docs.google.com/document/d/1l8s7uS2
 
   移除分辨率参数，移除默认 AK 和 SK
 
+* 1.3.0
+  新增对gradio app的支持
+
   ```shell
-  $ hub install ernie_vilg == 1.2.0
+  $ hub install ernie_vilg == 1.3.0
   ```

modules/text/lexical_analysis/jieba_paddle/README.md

@@ -207,6 +207,10 @@
 
   - 关于PaddleHub Serving更多信息参考：[服务部署](../../../../docs/docs_ch/tutorial/serving.md)
 
+- ## gradio app 支持
+  从paddlehub 2.3.1开始支持使用链接 http://127.0.0.1:8866/gradio/jieba_paddle 在浏览器中访问jieba_paddle的gradio app。
+
+
 
 ## 五、更新历史
 
@@ -218,6 +222,9 @@
 
   移除 fluid api
 
+* 1.1.0
+  新增对gradio app的支持
+
   - ```shell
-    $ hub install jieba_paddle==1.0.1
+    $ hub install jieba_paddle==1.1.0
     ```

modules/text/lexical_analysis/jieba_paddle/module.py

@@ -14,7 +14,7 @@ from paddlehub.module.module import serving
 
 @moduleinfo(
     name="jieba_paddle",
-    version="1.0.1",
+    version="1.1.0",
     summary=
     "jieba_paddle is a chineses tokenizer using BiGRU base on the PaddlePaddle deeplearning framework. More information please refer to https://github.com/fxsjy/jieba.",
     author="baidu-paddle",
@@ -54,6 +54,24 @@ class JiebaPaddle(hub.Module):
 
         return seg_list
 
+    def create_gradio_app(self):
+        import gradio as gr
+
+        def inference(text):
+            results = self.cut(sentence=text)
+            return results
+
+        title = "jieba_paddle"
+        description = "jieba_paddle is a word segmentation model based on paddlepaddle deep learning framework."
+
+        examples = [['今天是个好日子']]
+        app = gr.Interface(inference,
+                           "text", [gr.outputs.Textbox(label="words")],
+                           title=title,
+                           description=description,
+                           examples=examples)
+        return app
+
     def check_dependency(self):
         """
         Check jieba tool dependency.

modules/text/lexical_analysis/lac/module.py

@@ -11,18 +11,18 @@ import os
 
 import numpy as np
 import six
+from paddle.inference import Config
+from paddle.inference import create_predictor
+
 from .custom import Customization
 from .processor import load_kv_dict
 from .processor import parse_result
 from .processor import word_to_ids
-from paddle.inference import Config
-from paddle.inference import create_predictor
-
-from paddlehub.utils.utils import sys_stdin_encoding
-from paddlehub.utils.parser import txt_parser
 from paddlehub.module.module import moduleinfo
 from paddlehub.module.module import runnable
 from paddlehub.module.module import serving
+from paddlehub.utils.parser import txt_parser
+from paddlehub.utils.utils import sys_stdin_encoding
 
 
 class DataFormatError(Exception):
@@ -40,6 +40,7 @@ class DataFormatError(Exception):
     author_email="paddle-dev@baidu.com",
     type="nlp/lexical_analysis")
 class LAC:
+
     def __init__(self, user_dict=None):
         """
         initialize with the necessary elements
@@ -66,8 +67,8 @@ class LAC:
         """
         predictor config setting
         """
-        model = self.default_pretrained_model_path+'.pdmodel'
-        params = self.default_pretrained_model_path+'.pdiparams'
+        model = self.default_pretrained_model_path + '.pdmodel'
+        params = self.default_pretrained_model_path + '.pdiparams'
         cpu_config = Config(model, params)
         cpu_config.disable_glog_info()
         cpu_config.disable_gpu()

paddlehub/serving/app_compat.py

@@ -12,19 +12,30 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-
-import traceback
+import socket
+import threading
 import time
+import traceback
+from multiprocessing import Process
+from threading import Lock
 
-from flask import Flask, request
+import requests
+from flask import Flask
+from flask import redirect
+from flask import request
+from flask import Response
 
 from paddlehub.serving.model_service.base_model_service import cv_module_info
 from paddlehub.serving.model_service.base_model_service import nlp_module_info
 from paddlehub.serving.model_service.base_model_service import v2_module_info
-from paddlehub.utils import utils, log
+from paddlehub.utils import log
+from paddlehub.utils import utils
 
 filename = 'HubServing-%s.log' % time.strftime("%Y_%m_%d", time.localtime())
 
+_gradio_apps = {}  # Used to store all launched gradio apps
+_lock = Lock()  # Used to prevent parallel requests to launch a server twice
+
 
 def package_result(status: str, msg: str, data: dict):
     '''
@@ -55,6 +66,54 @@ def package_result(status: str, msg: str, data: dict):
     return {"status": status, "msg": msg, "results": data}
 
 
+def create_gradio_app(module_info: dict):
+    '''
+    Create a gradio app and launch a server for users.
+    Args:
+        module_info(dict): Module info include module name, method name and
+                            other info.
+    Return:
+        int: port number, if server has been successful.
+
+    Exception:
+        Raise a exception if server can not been launched.
+    '''
+    module_name = module_info['module_name']
+    port = None
+    with _lock:
+        if module_name not in _gradio_apps:
+            try:
+                serving_method = getattr(module_info["module"], 'create_gradio_app')
+            except Exception:
+                raise RuntimeError('Module {} is not supported for gradio app.'.format(module_name))
+
+            def get_free_tcp_port():
+                tcp = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+                tcp.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEPORT, 1)
+                tcp.bind(('localhost', 0))
+                addr, port = tcp.getsockname()
+                tcp.close()
+                return port
+
+            port = get_free_tcp_port()
+            app = serving_method()
+            process = Process(target=app.launch, kwargs={'server_port': port})
+            process.start()
+
+            def check_alive():
+                nonlocal port
+                while True:
+                    try:
+                        requests.get('http://localhost:{}/'.format(port))
+                        break
+                    except Exception:
+                        time.sleep(1)
+
+            check_alive()
+            _gradio_apps[module_name] = port
+    return port
+
+
 def predict_v2(module_info: dict, input: dict):
     '''
 
@@ -159,6 +218,47 @@ def create_app(init_flag: bool = False, configs: dict = None):
         results = predict_v2(module_info, inputs)
         return results
 
+    @app_instance.route('/gradio/<module_name>', methods=["GET", "POST"])
+    def gradio_app(module_name: str):
+        if module_name in v2_module_info.modules:
+            module_info = v2_module_info.get_module_info(module_name)
+            module_info['module_name'] = module_name
+        else:
+            msg = "Module {} is not supported for gradio app.".format(module_name)
+            return package_result("111", msg, "")
+        create_gradio_app(module_info)
+        return redirect("/gradio/{}/app".format(module_name), code=302)
+
+    @app_instance.route("/gradio/<module_name>/<path:path>", methods=["GET", "POST"])
+    def request_gradio_app(module_name: str, path: str):
+        '''
+        Gradio app server url interface. We route urls for gradio app to gradio server.
+
+        Args:
+            module_name(str): Module name for gradio app.
+            path(str): All resource path from gradio server.
+
+        Returns:
+            Any thing from gradio server.
+        '''
+        port = _gradio_apps[module_name]
+        if path == 'app':
+            proxy_url = request.url.replace(request.host_url + 'gradio/{}/app'.format(module_name),
+                                            'http://localhost:{}/'.format(port))
+        else:
+            proxy_url = request.url.replace(request.host_url + 'gradio/{}/'.format(module_name),
+                                            'http://localhost:{}/'.format(port))
+        resp = requests.request(method=request.method,
+                                url=proxy_url,
+                                headers={key: value
+                                         for (key, value) in request.headers if key != 'Host'},
+                                data=request.get_data(),
+                                cookies=request.cookies,
+                                allow_redirects=False)
+        headers = [(name, value) for (name, value) in resp.raw.headers.items()]
+        response = Response(resp.content, resp.status_code, headers)
+        return response
+
     return app_instance
 
 

requirements.txt

@@ -17,3 +17,4 @@ tqdm
 visualdl >= 2.0.0
 # gunicorn not support windows
 gunicorn >= 19.10.0; sys_platform != "win32"
+gradio