Automated deployment from GitHub (source commit: e5098b)

scripts/demo.py

@@ -23,7 +23,7 @@ SB_BUCKET_NAME = get_env_variable("SB_BUCKET_NAME")
 SB_SCHEMA_NAME = get_env_variable("SB_SCHEMA_NAME")
 SB_TABLE_NAME = get_env_variable("SB_TABLE_NAME")
 
-example_files = load_examples()
+example_files = load_examples(exclude_empty=True)
 print(f"Loaded {len(example_files)} example files from dataset:")
 
 # --- Image Processing Functions ---

xournalpp_htr/xio.py

@@ -101,7 +101,7 @@ def get_temporary_filename() -> Path:
     return output_file_tmp_noOCR
 
 
-def load_examples():
+def load_examples(exclude_empty: bool = False):
     if not huggingface_hub_available:
         raise ImportError(
             "The `huggingface_hub` package is required to load the example data."
@@ -117,6 +117,12 @@ def load_examples():
     data_dir = Path(local_dir) / "data"
 
     # Collect all matching file paths recursively
-    file_paths = [str(f) for f in data_dir.rglob("*") if f.suffix.lower() in extensions]
+    file_paths = sorted(
+        [str(f) for f in data_dir.rglob("*") if f.suffix.lower() in extensions]
+    )
+
+    # Remove empty files if requested
+    if exclude_empty:
+        file_paths = [fp for fp in file_paths if "empty" not in Path(fp).stem.lower()]
 
     return file_paths