move input files into temporary directory

app.py

@@ -137,7 +137,8 @@ def launch(self):
                         params = params + pipeline_params
                         btn_run.click(fn=self.whisper_inf.transcribe_file,
                                       inputs=params,
-                                      outputs=[tb_indicator, files_subtitles])
+                                      outputs=[tb_indicator, files_subtitles]).then(
+                            fn=lambda value: gr.update(value=None), inputs=input_file, outputs=input_file)
                         btn_openfolder.click(fn=lambda: self.open_folder("outputs"), inputs=None, outputs=None)
 
                     with gr.TabItem(_("Youtube")):  # tab2

modules/whisper/base_transcription_pipeline.py

@@ -11,6 +11,8 @@
 import gc
 from copy import deepcopy
 import time
+import shutil
+import tempfile
 
 from modules.uvr.music_separator import MusicSeparator
 from modules.utils.paths import (WHISPER_MODELS_DIR, DIARIZATION_MODELS_DIR, OUTPUT_DIR, DEFAULT_PARAMETERS_CONFIG_PATH,
@@ -203,6 +205,26 @@ def run(self,
         total_elapsed_time = time.time() - start_time
         return result, total_elapsed_time
 
+    def isGradioCachePath(self, in_path: str):
+        """
+        Check if given path is a gradio chache path
+
+        Parameters
+        ----------
+        in_path
+            the path to check
+
+        Returns
+        ----------
+        bool: True if it is a Gradio cache path, False otherwise
+        """
+        path = os.path.normpath(in_path)
+        path = path.split(os.sep)
+        for p in path:
+            if p.lower() == "gradio":
+                return True
+        return False
+
     def transcribe_file(self,
                         files: Optional[List] = None,
                         input_folder_path: Optional[str] = None,
@@ -258,37 +280,48 @@ def transcribe_file(self,
             if files and isinstance(files[0], gr.utils.NamedString):
                 files = [file.name for file in files]
 
+
+            progress(0, desc="getting input files")
             files_info = {}
-            for file in files:
-                transcribed_segments, time_for_task = self.run(
-                    file,
-                    progress,
-                    file_format,
-                    add_timestamp,
-                    *pipeline_params,
-                )
-
-                file_name, file_ext = os.path.splitext(os.path.basename(file))
-                if save_same_dir and input_folder_path:
-                    output_dir = os.path.dirname(file)
+            with tempfile.TemporaryDirectory(dir=".", prefix="_tmp_input") as tmp_folder:
+                tmp_files = []
+                for f in files:
+                    if self.isGradioCachePath(f):
+                        # likely a gradio cached file, so move it
+                        tmp_files.append(shutil.move(f, tmp_folder))
+                    else:
+                        # likely no gradio cached file, use the original one
+                        tmp_files.append(f)
+                for file in tmp_files:
+                    transcribed_segments, time_for_task = self.run(
+                        file,
+                        progress,
+                        file_format,
+                        add_timestamp,
+                        *pipeline_params,
+                    )
+
+                    file_name, file_ext = os.path.splitext(os.path.basename(file))
+                    if save_same_dir and input_folder_path:
+                        output_dir = os.path.dirname(file)
+                        subtitle, file_path = generate_file(
+                            output_dir=output_dir,
+                            output_file_name=file_name,
+                            output_format=file_format,
+                            result=transcribed_segments,
+                            add_timestamp=add_timestamp,
+                            **writer_options
+                        )
+
                     subtitle, file_path = generate_file(
-                        output_dir=output_dir,
+                        output_dir=self.output_dir,
                         output_file_name=file_name,
                         output_format=file_format,
                         result=transcribed_segments,
                         add_timestamp=add_timestamp,
                         **writer_options
                     )
-
-                subtitle, file_path = generate_file(
-                    output_dir=self.output_dir,
-                    output_file_name=file_name,
-                    output_format=file_format,
-                    result=transcribed_segments,
-                    add_timestamp=add_timestamp,
-                    **writer_options
-                )
-                files_info[file_name] = {"subtitle": read_file(file_path), "time_for_task": time_for_task, "path": file_path}
+                    files_info[file_name] = {"subtitle": read_file(file_path), "time_for_task": time_for_task, "path": file_path}
 
             total_result = ''
             total_time = 0