Fix save_inference_model bug

paddlehub/compat/module/module_v1.py

@@ -98,6 +98,9 @@ class ModuleV1(object):
         log.logger.info('{} pretrained paramaters loaded by PaddleHub'.format(num_param_loaded))
 
     def _load_extra_info(self):
+        if not 'extra_info' in self.desc:
+            return
+
         for key, value in self.desc.extra_info.items():
             self.__dict__['get_{}'.format(key)] = value
 
@@ -108,7 +111,7 @@ class ModuleV1(object):
     def _load_model(self):
         model_path = os.path.join(self.directory, 'model')
         exe = paddle.static.Executor(paddle.CPUPlace())
-        self.program, _, _ = paddle.static.load_inference_model(model_path, executor=exe)
+        self.program, _, _ = paddle.fluid.io.load_inference_model(model_path, executor=exe)
 
         # Clear the callstack since it may leak the privacy of the creator.
         for block in self.program.blocks:
@@ -240,6 +243,9 @@ class ModuleV1(object):
     def assets_path(self):
         return os.path.join(self.directory, 'assets')
 
+    def get_name_prefix(self):
+        return self.desc.name_prefix
+
     @property
     def is_runnable(self):
         '''
@@ -247,3 +253,29 @@ class ModuleV1(object):
         `hub run` command.
         '''
         return self.default_signature != None
+
+    def save_inference_model(self,
+                             dirname: str,
+                             model_filename: str = None,
+                             params_filename: str = None,
+                             combined: bool = False):
+        if hasattr(self, 'processor'):
+            if hasattr(self.processor, 'save_inference_model'):
+                return self.processor.save_inference_model(dirname, model_filename, params_filename, combined)
+
+        if combined:
+            model_filename = '__model__' if not model_filename else model_filename
+            params_filename = '__params__' if not params_filename else params_filename
+
+        place = paddle.CPUPlace()
+        exe = paddle.static.Executor(place)
+
+        feed_dict, fetch_dict, program = self.context(for_test=True, trainable=False)
+        paddle.fluid.io.save_inference_model(
+            dirname=dirname,
+            main_program=program,
+            executor=exe,
+            feeded_var_names=[var.name for var in list(feed_dict.values())],
+            target_vars=list(fetch_dict.values()),
+            model_filename=model_filename,
+            params_filename=params_filename)