add filename and page

app.py

@@ -17,7 +17,7 @@ s3 = boto3.client(
 # S3 bucket and keys
 BUCKET       = "doccano-processed"
 #INIT_KEY     = "gradio/initial_data_train.json"
-INIT_KEY = "gradio/refugee_train_initial_data.json"
+INIT_KEY = "gradio/refugee_train_initial_data_v2.json"
 #VALID_PREFIX = "validated_records/"
 VALID_PREFIX = "refugee_train_validated/"
 
@@ -152,6 +152,8 @@ def create_demo() -> gr.Blocks:
         prog      = gr.Slider(0, dynamic_dataset.len-1, value=0, step=1, label="Example #", interactive=False)
         inp_box   = gr.HighlightedText(label="Sentence", interactive=True)
         status    = gr.Checkbox(label="Validated?", value=False, interactive=False)
+        filename_disp = gr.Markdown(label="Filename")    # NEW: shows current filename
+        page_disp     = gr.Markdown(label="Page")        # NEW: shows current page number
         gr.Markdown(
             "[📖 Entity Tag Guide](https://huggingface.co/spaces/rafmacalaba/datause-annotation/blob/main/guidelines.md)"
         )
@@ -165,10 +167,21 @@ def create_demo() -> gr.Blocks:
             validate_btn = gr.Button("✅ Validate")
             skip_next = gr.Button("⏭️ Next Unvalidated")
 
+        # def load_example(idx):
+        #     rec  = validated_store.get(idx, dynamic_dataset.example(idx))
+        #     segs = prepare_for_highlight(rec)
+        #     return segs, rec.get("validated", False), idx
+
         def load_example(idx):
             rec  = validated_store.get(idx, dynamic_dataset.example(idx))
             segs = prepare_for_highlight(rec)
-            return segs, rec.get("validated", False), idx
+            return (
+                segs,
+                rec.get("validated", False),
+                idx,
+                rec.get("filename", ""),    # <-- returns filename for filename_disp
+                f"Page {rec.get('page', '')}"  # <-- returns page for page_disp
+            )
 
         def update_example(highlighted, idx: int):
             # grab the record