fix-uid-bug-and-add-debug-section (#23)

- fix uid reload and show log files (6e92592af2988e14cab1358ac455a313679bef52)
- fix typo (97c6f8e64fe737eb05234ccd746d062163c7fbe9)
- add closing files (a2810d675bbc1566b5a57b99774675ba1dc2d527)
- clean up code (94e80555ec10c4ab120d5abaa3f16508d55330ee)


Co-authored-by: zcy <ZeroCommand@users.noreply.huggingface.co>

app.py

@@ -3,6 +3,7 @@ import gradio as gr
 
 from app_leaderboard import get_demo as get_demo_leaderboard
 from app_text_classification import get_demo as get_demo_text_classification
+from app_debug import get_demo as get_demo_debug
 from run_jobs import start_process_run_job, stop_thread
 
 try:
@@ -11,6 +12,8 @@ try:
             get_demo_text_classification(demo)
         with gr.Tab("Leaderboard"):
             get_demo_leaderboard()
+        with gr.Tab("Logs(Debug)"):
+            get_demo_debug(demo)
 
     start_process_run_job()
 

app_debug.py

@@ -0,0 +1,37 @@
+import gradio as gr
+import pipe
+from os import listdir
+from os.path import isfile, join
+LOG_PATH = "./tmp"
+CONFIG_PATH = "./cicd/configs"
+
+def get_accordions_of_files(path, files):
+    components = []
+    for file in files:
+        with gr.Row():
+            with gr.Accordion(label=file, open=False):
+                with gr.Row():
+                    with open(join(path, file), "r") as f:
+                        gr.Markdown(f.read())
+    return components
+
+def get_accordions_of_log_files():
+    log_files = [f for f in listdir(LOG_PATH) if isfile(join(LOG_PATH, f)) and f.endswith("_log")]
+    return get_accordions_of_files(LOG_PATH, log_files)
+
+def get_accordions_of_config_files():
+    config_files = [f for f in listdir(CONFIG_PATH) if isfile(join(CONFIG_PATH, f)) and f.endswith(".yaml")]
+    return get_accordions_of_files(CONFIG_PATH, config_files)
+
+def get_demo(demo):
+    with gr.Row():
+        # check if jobs is an attribute of pipe
+        if hasattr(pipe, "jobs"):
+            gr.Markdown(f"current jobs in queue: {len(pipe.jobs)}")
+    with gr.Accordion(label="Config Files", open=False):
+        config_accordion = get_accordions_of_config_files()
+        demo.load(get_accordions_of_config_files, outputs=config_accordion, every=1)
+    with gr.Accordion(label="Log Files", open=False):
+        log_accordions = get_accordions_of_log_files()
+        demo.load(get_accordions_of_log_files, outputs=log_accordions, every=1)
+                

app_text_classification.py

@@ -27,11 +27,10 @@ CONFIG_PATH = "./config.yaml"
 
 
 def get_demo(demo):
-    uid = uuid.uuid4()
     with gr.Row():
         gr.Markdown(INTRODUCTION_MD)
         uid_label = gr.Textbox(
-            label="Evaluation ID:", value=uid, visible=False, interactive=False
+            label="Evaluation ID:", value=uuid.uuid4, visible=False, interactive=False
         )
     with gr.Row():
         model_id_input = gr.Textbox(
@@ -70,19 +69,28 @@ def get_demo(demo):
 
     with gr.Accordion(label="Model Wrap Advance Config (optional)", open=False):
         run_local = gr.Checkbox(value=True, label="Run in this Space")
-        use_inference = read_inference_type(uid) == "hf_inference_api"
-        run_inference = gr.Checkbox(value=use_inference, label="Run with Inference API")
+        run_inference = gr.Checkbox(value="False", label="Run with Inference API")
+        @gr.on(triggers=[uid_label.change], inputs=[uid_label], outputs=[run_inference])
+        def get_run_mode(uid):
+            return (
+                gr.update(value=read_inference_type(uid) == "hf_inference_api" and not run_local.value)
+            )
         inference_token = gr.Textbox(value="", label="HF Token for Inference API", visible=False, interactive=True)
 
     with gr.Accordion(label="Scanner Advance Config (optional)", open=False):
-        selected = read_scanners(uid)
-        # currently we remove data_leakage from the default scanners
-        # Reason: data_leakage barely raises any issues and takes too many requests
-        # when using inference API, causing rate limit error
-        scan_config = selected + ["data_leakage"]
         scanners = gr.CheckboxGroup(
-            choices=scan_config, value=selected, label="Scan Settings", visible=True
+            label="Scan Settings", visible=True
         )
+        @gr.on(triggers=[uid_label.change], inputs=[uid_label], outputs=[scanners])
+        def get_scanners(uid):
+            selected = read_scanners(uid)
+            # currently we remove data_leakage from the default scanners
+            # Reason: data_leakage barely raises any issues and takes too many requests
+            # when using inference API, causing rate limit error
+            scan_config = selected + ["data_leakage"]
+            return (gr.update(
+                    choices=scan_config, value=selected, label="Scan Settings", visible=True
+                ))
 
     with gr.Row():
         run_btn = gr.Button(
@@ -98,7 +106,7 @@ def get_demo(demo):
 
     dataset_id_input.change(
         check_dataset_and_get_config, 
-        inputs=[dataset_id_input, uid_label], outputs=[dataset_config_input]
+        inputs=[dataset_id_input], outputs=[dataset_config_input]
     )
 
     dataset_config_input.change(

io_utils.py

@@ -1,6 +1,5 @@
 import os
 import subprocess
-import gradio as gr
 import yaml
 
 import pipe
@@ -28,6 +27,7 @@ def read_scanners(uid):
     with open(get_yaml_path(uid), "r") as f:
         config = yaml.load(f, Loader=yaml.FullLoader)
         scanners = config.get("detectors", [])
+    f.close()
     return scanners
 
 
@@ -37,9 +37,12 @@ def write_scanners(scanners, uid):
         config = yaml.load(f, Loader=yaml.FullLoader)
         if config:
             config["detectors"] = scanners
+    f.close()
     # save scanners to detectors in yaml
     with open(get_yaml_path(uid), "w") as f:
         yaml.dump(config, f, Dumper=Dumper)
+    f.close()
+
 
 
 # read model_type from yaml file
@@ -48,6 +51,7 @@ def read_inference_type(uid):
     with open(get_yaml_path(uid), "r") as f:
         config = yaml.load(f, Loader=yaml.FullLoader)
         inference_type = config.get("inference_type", "")
+    f.close()
     return inference_type
 
 
@@ -62,9 +66,11 @@ def write_inference_type(use_inference, inference_token, uid):
             config["inference_type"] = "hf_pipeline"
             # FIXME: A quick and temp fix for missing token
             config["inference_token"] = ""
+    f.close()
     # save inference_type to inference_type in yaml
     with open(get_yaml_path(uid), "w") as f:
         yaml.dump(config, f, Dumper=Dumper)
+    f.close()
 
 
 
@@ -75,6 +81,7 @@ def read_column_mapping(uid):
         config = yaml.load(f, Loader=yaml.FullLoader)
         if config:
             column_mapping = config.get("column_mapping", dict())
+    f.close()
     return column_mapping
 
 
@@ -82,6 +89,7 @@ def read_column_mapping(uid):
 def write_column_mapping(mapping, uid):
     with open(get_yaml_path(uid), "r") as f:
         config = yaml.load(f, Loader=yaml.FullLoader)
+    f.close()
 
     if config is None:
         return
@@ -92,6 +100,8 @@ def write_column_mapping(mapping, uid):
 
     with open(get_yaml_path(uid), "w") as f:
         yaml.dump(config, f, Dumper=Dumper)
+    f.close()
+
 
 
 # convert column mapping dataframe to json
@@ -114,6 +124,7 @@ def get_logs_file(uid):
 def write_log_to_user_file(id, log):
     with open(f"./tmp/{id}_log", "a") as f:
         f.write(log)
+    f.close()
 
 
 def save_job_to_pipe(id, job, lock):

text_classification.py

@@ -171,7 +171,6 @@ def infer_output_label_column(
             str(i): id2label_mapping[label]
             for i, label in zip(id2label.keys(), dataset_labels)
         }
-    # print('>>>>> column_mapping >>>>>', column_mapping)
 
     id2label_df = pd.DataFrame(
         {

text_classification_ui_helpers.py

@@ -23,7 +23,7 @@ HF_SPACE_ID = "SPACE_ID"
 HF_WRITE_TOKEN = "HF_WRITE_TOKEN"
 
 
-def check_dataset_and_get_config(dataset_id, uid):
+def check_dataset_and_get_config(dataset_id):
     try:
         # write_column_mapping(None, uid)  # reset column mapping
         configs = datasets.get_dataset_config_names(dataset_id)