Feat: Add Wikipedia search engine (#478)

* feat: add Wikipedia search engine

* wikipedia

* make format

src/config/tools.py

@@ -13,6 +13,7 @@ class SearchEngine(enum.Enum):
     DUCKDUCKGO = "duckduckgo"
     BRAVE_SEARCH = "brave_search"
     ARXIV = "arxiv"
+    WIKIPEDIA = "wikipedia"
 
 
 # Tool configuration

src/tools/search.py

@@ -5,9 +5,17 @@ import logging
 import os
 from typing import List, Optional
 
-from langchain_community.tools import BraveSearch, DuckDuckGoSearchResults
+from langchain_community.tools import (
+    BraveSearch,
+    DuckDuckGoSearchResults,
+    WikipediaQueryRun,
+)
 from langchain_community.tools.arxiv import ArxivQueryRun
-from langchain_community.utilities import ArxivAPIWrapper, BraveSearchWrapper
+from langchain_community.utilities import (
+    ArxivAPIWrapper,
+    BraveSearchWrapper,
+    WikipediaAPIWrapper,
+)
 
 from src.config import SearchEngine, SELECTED_SEARCH_ENGINE
 from src.config import load_yaml_config
@@ -24,6 +32,7 @@ LoggedTavilySearch = create_logged_tool(TavilySearchResultsWithImages)
 LoggedDuckDuckGoSearch = create_logged_tool(DuckDuckGoSearchResults)
 LoggedBraveSearch = create_logged_tool(BraveSearch)
 LoggedArxivSearch = create_logged_tool(ArxivQueryRun)
+LoggedWikipediaSearch = create_logged_tool(WikipediaQueryRun)
 
 
 def get_search_config():
@@ -76,5 +85,19 @@ def get_web_search_tool(max_search_results: int):
                 load_all_available_meta=True,
             ),
         )
+    elif SELECTED_SEARCH_ENGINE == SearchEngine.WIKIPEDIA.value:
+        wiki_lang = search_config.get("wikipedia_lang", "en")
+        wiki_doc_content_chars_max = search_config.get(
+            "wikipedia_doc_content_chars_max", 4000
+        )
+        return LoggedWikipediaSearch(
+            name="web_search",
+            api_wrapper=WikipediaAPIWrapper(
+                lang=wiki_lang,
+                top_k_results=max_search_results,
+                load_all_available_meta=True,
+                doc_content_chars_max=wiki_doc_content_chars_max,
+            ),
+        )
     else:
         raise ValueError(f"Unsupported search engine: {SELECTED_SEARCH_ENGINE}")