add filters

app.py

@@ -34,23 +34,30 @@ def parse_notebook(filter_options):
     reset_tmp_folder()
 
     found_notebook = False
-    counter = 0
     while not found_notebook:
-        counter+=1
-        print(counter, filter_options)
         notebook_data = next(ds_iter)
         notebook_string = notebook_data["text"]
         notebook_id = notebook_data["id"].split("/")[-1]
 
         if filter_options == "none":
             found_notebook = True
+        elif filter_options == ">10MB":
+            if len(notebook_string)>10 * 1024 * 1024:
+                found_notebook = True
         elif filter_options == ">1MB":
             if len(notebook_string)>1 * 1024 * 1024:
                 found_notebook = True
+        elif filter_options == ">100KB":
+            if len(notebook_string)>100 * 1024:
+                found_notebook = True
+        elif filter_options == ">10KB":
+            if len(notebook_string)>10 * 1024:
+                found_notebook = True
+        # too slow
         elif filter_options == "has outputs":
             notebook_parsed = nbformat.reads(notebook_string, as_version=4)
             (notebook_body, resources) = html_exporter.from_notebook_node(notebook_parsed)
-            if len(resources)>0:
+            if len(resources["outputs"])>0:
                 found_notebook = True
 
     out_path = os.path.join(TMP_DIR, notebook_id)
@@ -68,7 +75,7 @@ def parse_notebook(filter_options):
 
 with gr.Blocks() as demo:
     gr.Markdown("# Kaggle Notebooks")
-    filter_options = gr.Radio(["none", "has outputs", ">1MB"], value="none", label="Notebook filters", info="A lot of notebooks are short or have the outputs stripped - filters help finding interesting ones."),
+    filter_options = gr.Radio(["none",">10KB", ">100KB", ">1MB", ">10MB"], value="none", label="Notebook filters", info="A lot of notebooks are short or have the outputs stripped - filters help finding interesting ones."),
 
     button = gr.Button("Show next!")
     file = gr.File()