Update app.py

app.py

@@ -12,7 +12,7 @@ from tqdm.autonotebook import tqdm
 from pinecone import Pinecone, ServerlessSpec
 pc = Pinecone(api_key=os.getenv('PINECONE_API_KEY'))
 
-index_name = "prorata-postman-ds-128"
+index_name = "prorata-postman-ds-128-v2"
 index = pc.Index(index_name)
 
 
@@ -75,30 +75,101 @@ def get_matches_reranked(q, k=20):
 
     return matches_colbertv2
 
+def filter_matches(matches_colbertv2, score_thr=0.0):
+    matches_colbertv2_f = []
+    url_to_chunk_l = {}
+
+    for idx, match in enumerate(matches_colbertv2):
+        if match['colbertv2_score'] > score_thr:
+            _url = match['metadata']['url']
+            if not _url in url_to_chunk_l:
+                url_to_chunk_l[_url] = []
+                url_to_chunk_l[_url].append(match)
+
+                matches_colbertv2_f.append(match)
+
+    return matches_colbertv2_f
+
+style_str = """
+    <style>
+    .doc-title {
+        /* font-family: cursive, sans-serif; */
+        font-family: Optima, sans-serif;
+        width: 100%;
+        display: inline-block;
+        font-size: 2em;
+        font-weight: bolder;
+        padding-top: 20px;
+        /* font-style: italic; */
+    }
+    .doc-url {
+        /* font-family: cursive, sans-serif; */
+        font-size: 1em;
+        padding-left: 40px;
+        padding-bottom: 10px;
+        /* font-weight: bolder; */
+        /* font-style: italic; */
+    }
+    .doc-text {
+        /* font-family: cursive, sans-serif; */
+        font-family: Optima, sans-serif;
+        font-size: 1.5em;
+        padding-left: 40px;
+        padding-bottom: 20px;
+        /* font-weight: bolder; */
+        /* font-style: italic; */
+    }
+    .doc-title > img {
+        width: 22px;
+        height: 22px;
+        border-radius: 50%;
+        overflow: hidden;
+        background-color: transparent;
+        display: inline-block;
+        vertical-align: middle;
+    }
+    .doc-title > score {
+        font-family: Optima, sans-serif;
+        font-weight: normal;
+        float: right;
+    }
+    </style>
+"""
 
 import gradio as gr
 from io import StringIO
+from urllib.parse import urlparse
 
 def output_chunks_reranked(msg):
     matches_colbertv2 = get_matches_reranked(msg, k=20)
-
+    matches_colbertv2 = filter_matches(matches_colbertv2, score_thr=0.55)
     _out = StringIO()
+
+    if not matches_colbertv2:
+        print(style_str, file=_out)
+        print(f"<div>", file=_out)
+        print(f"<div class=\"doc-title\">No sources relevant to this target were found.</div>", file=_out)
+        print(f"</div>", file=_out)
+        return _out.getvalue()
+
     for idx, match in enumerate(matches_colbertv2):
-        print(f"{idx+1}: {match['colbertv2_score']:.2f}", end='', file=_out)
-        print(textwrap.fill(match['metadata']['text'], initial_indent='       ', subsequent_indent='          ', width=100), file=_out)
-        print(file=_out)
+        print(style_str, file=_out)
+        print(f"<div>", file=_out)
+        favicon = f"<img src=\"https://www.google.com/s2/favicons?sz=128&amp;domain={urlparse(match['metadata']['url']).netloc}\"/>"
+        print(f"<div class=\"doc-title\">{favicon}&nbsp&nbsp;{match['metadata']['title']}<score>{match['colbertv2_score']:.2f}</score></div>", file=_out)
+        print(f"<div class=\"doc-url\"><a href=\"{match['metadata']['url']}\"  target=\"_blank\">{match['metadata']['url']}</a></div>", file=_out)
+        # print(f"    (Score: {match['colbertv2_score']:.2f})", file=_out)
+        print(f"<div class=\"doc-text\">{match['metadata']['text']}</div>", file=_out)
+        print(f"</div>", file=_out)
 
     return _out.getvalue()
 
-
 with gr.Blocks() as demo:
     msg = gr.Textbox(label='Target')
-    results_box = gr.Textbox(label='Matches', lines=30, autoscroll=False)
+    # results_box = gr.Textbox(label='Matches', lines=30, autoscroll=False)
+    results_box = gr.HTML(label='Matches')
 
     msg.submit(output_chunks_reranked, msg, results_box, queue=False)
-    
-    demo.queue()
-    demo.launch(share=True)
 
     
 if __name__ == "__main__":