fix: revert to old compression method for parquet file & now show 10 first emails subjects

app.py

@@ -17,9 +17,8 @@ title = "Confidential forensics tool with ChatGPT"
 examples = ["Who is Phillip Allen?", "What the project in Austin is about?", "Give me more details about the real estate project"]
 
 index = GPTSimpleVectorIndex.load_from_disk('email.json')
-
-dat_fr = pd.DataFrame({"Documents loaded": ["email.json", "email_200.parquet"]})
 df = pd.read_parquet("email_200.parquet")
+df2 = df["subject"].replace([None], "<no title>").head(10).to_frame()
 
 def search_emails(opt, message, n=3):
     "Outputs the top n emails that match the most the pattern"
@@ -34,7 +33,7 @@ def search_emails(opt, message, n=3):
             message_tmp = df.sort_values('similarities', ascending=False).head(n)
             message_tmp = [(row.file, row.body, row.similarities) for index, row in message_tmp.iterrows()]
             for msg in message_tmp:
-                message += f"{msg[0]}\nContent: {msg[1].strip()}\n{msg[2]}\n\n"
+                message += f"Mail ID: {msg[0]}\nContent: {msg[1].strip()}\nSimilarity score: {msg[2]}\n\n"
         except Exception as e:
             message = "An error occured when handling your query, please try again."
             print(e)
@@ -78,9 +77,10 @@ with gr.Blocks(title=title) as demo:
     # """ + title + """
     """)
     dat = gr.Dataframe(
-            value=dat_fr,
+            value=df2,
             max_cols=1,
             max_rows=4,
+            title="Documents loaded",
             overflow_row_behaviour="paginate",
     )
     btn_upload = gr.UploadButton("Upload a new document...", file_types=["text"])

email_200.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:059c095bad18d8c299ab701ca19ac1d938e82bba28af7d6e9a333bd20a35586a
-size 3098469
+oid sha256:689f185923970ee7a8da61eb63ae2116d75af0a473dd2735c0848bc87b505135
+size 8670196