Address HTML viewer bug when doing sycamore_crawler_http_sort_all

openai-proxy/py_proxy/proxy.py

@@ -140,6 +140,15 @@ def proxy():
         download_name = download_name
     )
 
+@app.route('/viewHtml/<path:arg>', methods=['GET', 'OPTIONS'])
+def proxy_html(arg):
+    if request.method == 'OPTIONS':
+        return optionsResp('GET')
+    path = "/" + arg
+    if "/../" in path:  # Prevent filesystem snooping
+        return "invalid path"
+    return send_file(path, mimetype='text/html')
+
 def make_openai_call(messages, model_id="gpt-4"):
     response = openai.ChatCompletion.create(
       model=model_id,

ui/src/ConversationList.tsx

@@ -1,7 +1,7 @@
 import React, { Dispatch, SetStateAction, useState, useEffect, useRef } from 'react';
 import { ActionIcon, createStyles, Loader, Navbar, Text, useMantineTheme, rem, Center, Container, Group, Anchor, TextInput } from '@mantine/core';
 import { Settings } from './Types'
-import { createConversation, deleteConversation, getConversations, is2dot12plus } from './OpenSearch';
+import { createConversation, deleteConversation, getConversations } from './OpenSearch';
 import { IconChevronRight, IconMessagePlus, IconTrash } from '@tabler/icons-react';
 import { useHover } from '@mantine/hooks';
 const useStyles = createStyles((theme) => ({

ui/src/Doclist.tsx

@@ -14,6 +14,10 @@ const DocumentItem = ({ document }: { document: SearchResultDocument }) => {
             console.log("You clicked: ", document)
             localStorage.setItem('pdfDocumentMetadata', dataString);
             window.open('/viewPdf');
+        } else if (document.hasAbsoluteUrl()) {
+            window.open(document.url);
+        } else if (document.url.startsWith("/")) {
+            window.open("/viewHtml" + document.url);
         } else {
             window.open(document.url);
         }

ui/src/Types.tsx

@@ -21,6 +21,10 @@ export class SearchResultDocument {
         }
         return false;
     }
+
+    hasAbsoluteUrl() {
+        return this.url.search(/^[a-z]+:\/\//i) >= 0;
+    }
 }
 export class UserChat {
     id: string = "";