add filters

app.py

@@ -30,7 +30,7 @@ def embed_figures(html_body, resources):
         html_body = html_body.replace(key, img_tag)
     return html_body
 
-def parse_notebook():
+def parse_notebook(filter_options):
     reset_tmp_folder()
 
     found_notebook = False
@@ -42,8 +42,16 @@ def parse_notebook():
         notebook_string = notebook_data["text"]
         notebook_id = notebook_data["id"].split("/")[-1]
 
-        if len(notebook_string)>1 * 1024 * 1024:
+        if filter_options == "none":
             found_notebook = True
+        elif filter_options == ">1MB":
+            if len(notebook_string)>1 * 1024 * 1024:
+                found_notebook = True
+        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:
+                found_notebook = true
 
     out_path = os.path.join(TMP_DIR, notebook_id)
     
@@ -60,12 +68,14 @@ def parse_notebook():
 
 with gr.Blocks() as demo:
     gr.Markdown("# Kaggle Notebooks")
+    filter_options = gr.Radio(["none", "has outputs", ">1MB"], 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()
     html = gr.HTML("")
 
     
-    button.click(fn=parse_notebook, inputs=[], outputs=[html, file])
-    demo.load(fn=parse_notebook, inputs=[], outputs=[html, file])
+    button.click(fn=parse_notebook, inputs=[filter_options], outputs=[html, file])
+    demo.load(fn=parse_notebook, inputs=[filter_options], outputs=[html, file])
 
 demo.launch()