feat: Add SearXNG search provider and settings
This commit is contained in:
parent
138e41e122
commit
c8c71f69cc
@ -65,6 +65,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "Besøg websitet nævnt i samtalen"
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "SearXNG URL"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -65,6 +65,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "Die in der Nachricht erwähnte Website besuchen"
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "SearXNG-URL"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -36,8 +36,8 @@
|
||||
"generateTitle": {
|
||||
"label": "Generate Title using AI"
|
||||
},
|
||||
"ollamaStatus" :{
|
||||
"label":"Enable or disable Ollama connection status check"
|
||||
"ollamaStatus": {
|
||||
"label": "Enable or disable Ollama connection status check"
|
||||
}
|
||||
},
|
||||
"sidepanelRag": {
|
||||
@ -65,6 +65,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "Visit the website mentioned in the message"
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "SearXNG URL"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -65,6 +65,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "Visita el sitio web mencionado en el mensaje"
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "URL de SearXNG"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -62,6 +62,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "مراجعه به وب سایت ذکر شده در پیام"
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "آدرس SearXNG"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -65,6 +65,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "Visitez le site web mentionné dans le message"
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "URL SearXNG"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -65,6 +65,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "Visita il sito web menzionato nel messaggio"
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "URL SearXNG"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -68,6 +68,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "メッセージに記載されたウェブサイトを訪問してください"
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "SearXNG URL"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -68,6 +68,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "메시지에 언급된 웹사이트 방문"
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "SearXNG URL"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -68,6 +68,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "സന്ദേശത്തിൽ പറയുന്ന വെബ്സൈറ്റ് സന്ദർശിക്കുക."
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "SearXNG URL"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -65,6 +65,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "Besøk nettstedet nevnt i samtalen"
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "SearXNG URL"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -65,6 +65,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "Visitar o site mencionado na mensagem"
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "URL do SearXNG"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -66,6 +66,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "Посетите веб-сайт, указанный в сообщении."
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "URL-адрес SearXNG"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -65,6 +65,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "Besök webbplatsen som nämns i meddelandet"
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "SearXNG URL"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -65,6 +65,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "Відвідати веб-сайт, згаданий у повідомленні"
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "SearXNG URL-адреса"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -68,6 +68,11 @@
|
||||
},
|
||||
"visitSpecificWebsite": {
|
||||
"label": "访问消息中提到的网站。"
|
||||
},
|
||||
"searxng": {
|
||||
"url": {
|
||||
"label": "SearXNG 网址"
|
||||
}
|
||||
}
|
||||
},
|
||||
"system": {
|
||||
|
@ -3,7 +3,7 @@ import { getSearchSettings, setSearchSettings } from "@/services/search"
|
||||
import { SUPPORTED_SERACH_PROVIDERS } from "@/utils/search-provider"
|
||||
import { useForm } from "@mantine/form"
|
||||
import { useQuery, useQueryClient } from "@tanstack/react-query"
|
||||
import { Select, Skeleton, Switch, InputNumber } from "antd"
|
||||
import { Select, Skeleton, Switch, InputNumber, Input } from "antd"
|
||||
import { useTranslation } from "react-i18next"
|
||||
|
||||
export const SearchModeSettings = () => {
|
||||
@ -14,7 +14,9 @@ export const SearchModeSettings = () => {
|
||||
isSimpleInternetSearch: false,
|
||||
searchProvider: "",
|
||||
totalSearchResults: 0,
|
||||
visitSpecificWebsite: false
|
||||
visitSpecificWebsite: false,
|
||||
searxngURL: "",
|
||||
searxngJSONMode: false
|
||||
}
|
||||
})
|
||||
|
||||
@ -62,6 +64,23 @@ export const SearchModeSettings = () => {
|
||||
/>
|
||||
</div>
|
||||
</div>
|
||||
{form.values.searchProvider === "searxng" && (
|
||||
<>
|
||||
<div className="flex sm:flex-row flex-col space-y-4 sm:space-y-0 sm:justify-between">
|
||||
<span className="text-gray-700 dark:text-neutral-50">
|
||||
{t("generalSettings.webSearch.searxng.url.label")}
|
||||
</span>
|
||||
<div>
|
||||
<Input
|
||||
placeholder="https://searxng.example.com"
|
||||
className="w-full mt-4 sm:mt-0 sm:w-[200px]"
|
||||
required
|
||||
{...form.getInputProps("searxngURL")}
|
||||
/>
|
||||
</div>
|
||||
</div>
|
||||
</>
|
||||
)}
|
||||
<div className="flex sm:flex-row flex-col space-y-4 sm:space-y-0 sm:justify-between">
|
||||
<span className="text-gray-700 dark:text-neutral-50 ">
|
||||
{t("generalSettings.webSearch.searchMode.label")}
|
||||
|
@ -62,20 +62,45 @@ export const setTotalSearchResults = async (totalSearchResults: number) => {
|
||||
await storage.set("totalSearchResults", totalSearchResults.toString())
|
||||
}
|
||||
|
||||
export const getSearxngURL = async () => {
|
||||
const searxngURL = await storage.get("searxngURL")
|
||||
return searxngURL || ""
|
||||
}
|
||||
|
||||
export const isSearxngJSONMode = async () => {
|
||||
const searxngJSONMode = await storage.get<boolean>("searxngJSONMode")
|
||||
return searxngJSONMode ?? false
|
||||
}
|
||||
|
||||
export const setSearxngJSONMode = async (searxngJSONMode: boolean) => {
|
||||
await storage.set("searxngJSONMode", searxngJSONMode)
|
||||
}
|
||||
|
||||
export const setSearxngURL = async (searxngURL: string) => {
|
||||
await storage.set("searxngURL", searxngURL)
|
||||
}
|
||||
|
||||
export const getSearchSettings = async () => {
|
||||
const [isSimpleInternetSearch, searchProvider, totalSearchResult, visitSpecificWebsite] =
|
||||
const [isSimpleInternetSearch, searchProvider, totalSearchResult, visitSpecificWebsite,
|
||||
searxngURL,
|
||||
searxngJSONMode
|
||||
] =
|
||||
await Promise.all([
|
||||
getIsSimpleInternetSearch(),
|
||||
getSearchProvider(),
|
||||
totalSearchResults(),
|
||||
getIsVisitSpecificWebsite()
|
||||
getIsVisitSpecificWebsite(),
|
||||
getSearxngURL(),
|
||||
isSearxngJSONMode()
|
||||
])
|
||||
|
||||
return {
|
||||
isSimpleInternetSearch,
|
||||
searchProvider,
|
||||
totalSearchResults: totalSearchResult,
|
||||
visitSpecificWebsite
|
||||
visitSpecificWebsite,
|
||||
searxngURL,
|
||||
searxngJSONMode
|
||||
}
|
||||
}
|
||||
|
||||
@ -83,17 +108,23 @@ export const setSearchSettings = async ({
|
||||
isSimpleInternetSearch,
|
||||
searchProvider,
|
||||
totalSearchResults,
|
||||
visitSpecificWebsite
|
||||
visitSpecificWebsite,
|
||||
searxngJSONMode,
|
||||
searxngURL
|
||||
}: {
|
||||
isSimpleInternetSearch: boolean
|
||||
searchProvider: string
|
||||
totalSearchResults: number
|
||||
visitSpecificWebsite: boolean
|
||||
searxngURL: string
|
||||
searxngJSONMode: boolean
|
||||
}) => {
|
||||
await Promise.all([
|
||||
setIsSimpleInternetSearch(isSimpleInternetSearch),
|
||||
setSearchProvider(searchProvider),
|
||||
setTotalSearchResults(totalSearchResults),
|
||||
setIsVisitSpecificWebsite(visitSpecificWebsite)
|
||||
setIsVisitSpecificWebsite(visitSpecificWebsite),
|
||||
setSearxngJSONMode(searxngJSONMode),
|
||||
setSearxngURL(searxngURL)
|
||||
])
|
||||
}
|
||||
|
@ -14,5 +14,9 @@ export const SUPPORTED_SERACH_PROVIDERS = [
|
||||
{
|
||||
label: "Brave",
|
||||
value: "brave"
|
||||
},
|
||||
{
|
||||
label: "Searxng",
|
||||
value: "searxng"
|
||||
}
|
||||
]
|
145
src/web/search-engines/searxng.ts
Normal file
145
src/web/search-engines/searxng.ts
Normal file
@ -0,0 +1,145 @@
|
||||
import { urlRewriteRuntime } from "~/libs/runtime"
|
||||
import { cleanUrl } from "~/libs/clean-url"
|
||||
import { getSearxngURL, isSearxngJSONMode, getIsSimpleInternetSearch, totalSearchResults } from "@/services/search"
|
||||
import { pageAssistEmbeddingModel } from "@/models/embedding"
|
||||
import type { Document } from "@langchain/core/documents"
|
||||
import { RecursiveCharacterTextSplitter } from "langchain/text_splitter"
|
||||
import { MemoryVectorStore } from "langchain/vectorstores/memory"
|
||||
import { PageAssistHtmlLoader } from "~/loader/html"
|
||||
import {
|
||||
defaultEmbeddingChunkOverlap,
|
||||
defaultEmbeddingChunkSize,
|
||||
defaultEmbeddingModelForRag,
|
||||
getOllamaURL
|
||||
} from "~/services/ollama"
|
||||
|
||||
interface SearxNGJSONResult {
|
||||
title: string
|
||||
url: string
|
||||
content: string
|
||||
}
|
||||
|
||||
interface SearxNGJSONResponse {
|
||||
results: SearxNGJSONResult[]
|
||||
}
|
||||
|
||||
export const searxngSearch = async (query: string) => {
|
||||
const searxngURL = await getSearxngURL()
|
||||
if (!searxngURL) {
|
||||
throw new Error("SearXNG URL not configured")
|
||||
}
|
||||
|
||||
const isJSONMode = await isSearxngJSONMode()
|
||||
const results = isJSONMode
|
||||
? await searxngJSONSearch(searxngURL, query)
|
||||
: await searxngWebSearch(searxngURL, query)
|
||||
|
||||
const TOTAL_SEARCH_RESULTS = await totalSearchResults()
|
||||
const searchResults = results.slice(0, TOTAL_SEARCH_RESULTS)
|
||||
|
||||
const isSimpleMode = await getIsSimpleInternetSearch()
|
||||
|
||||
if (isSimpleMode) {
|
||||
await getOllamaURL()
|
||||
return searchResults.map((result) => {
|
||||
return {
|
||||
url: result.link,
|
||||
content: result.content
|
||||
}
|
||||
})
|
||||
}
|
||||
|
||||
const docs: Document<Record<string, any>>[] = []
|
||||
for (const result of searchResults) {
|
||||
const loader = new PageAssistHtmlLoader({
|
||||
html: "",
|
||||
url: result.link
|
||||
})
|
||||
|
||||
const documents = await loader.loadByURL()
|
||||
documents.forEach((doc) => {
|
||||
docs.push(doc)
|
||||
})
|
||||
}
|
||||
|
||||
const ollamaUrl = await getOllamaURL()
|
||||
const embeddingModel = await defaultEmbeddingModelForRag()
|
||||
const ollamaEmbedding = await pageAssistEmbeddingModel({
|
||||
model: embeddingModel || "",
|
||||
baseUrl: cleanUrl(ollamaUrl)
|
||||
})
|
||||
|
||||
const chunkSize = await defaultEmbeddingChunkSize()
|
||||
const chunkOverlap = await defaultEmbeddingChunkOverlap()
|
||||
const textSplitter = new RecursiveCharacterTextSplitter({
|
||||
chunkSize,
|
||||
chunkOverlap
|
||||
})
|
||||
|
||||
const chunks = await textSplitter.splitDocuments(docs)
|
||||
const store = new MemoryVectorStore(ollamaEmbedding)
|
||||
await store.addDocuments(chunks)
|
||||
|
||||
return store
|
||||
}
|
||||
|
||||
const searxngJSONSearch = async (baseURL: string, query: string) => {
|
||||
const searchURL = `${cleanUrl(baseURL)}?q=${encodeURIComponent(query)}&format=json`
|
||||
|
||||
const abortController = new AbortController()
|
||||
setTimeout(() => abortController.abort(), 20000)
|
||||
|
||||
try {
|
||||
const response = await fetch(searchURL, {
|
||||
signal: abortController.signal,
|
||||
headers: {
|
||||
'Accept': 'application/json'
|
||||
}
|
||||
})
|
||||
|
||||
if (!response.ok) {
|
||||
throw new Error(`SearXNG search failed: ${response.statusText}`)
|
||||
}
|
||||
|
||||
const data = await response.json() as SearxNGJSONResponse
|
||||
|
||||
return data.results.map(result => ({
|
||||
title: result.title,
|
||||
link: result.url,
|
||||
content: result.content
|
||||
}))
|
||||
} catch (error) {
|
||||
console.error('SearXNG JSON search failed:', error)
|
||||
return []
|
||||
}
|
||||
}
|
||||
|
||||
const searxngWebSearch = async (baseURL: string, query: string) => {
|
||||
const searchURL = `${cleanUrl(baseURL)}?q=${encodeURIComponent(query)}`
|
||||
|
||||
await urlRewriteRuntime(cleanUrl(searchURL), "searxng")
|
||||
|
||||
const abortController = new AbortController()
|
||||
setTimeout(() => abortController.abort(), 10000)
|
||||
|
||||
try {
|
||||
const htmlString = await fetch(searchURL, {
|
||||
signal: abortController.signal
|
||||
}).then(response => response.text())
|
||||
|
||||
const parser = new DOMParser()
|
||||
const doc = parser.parseFromString(htmlString, "text/html")
|
||||
|
||||
const searchResults = Array.from(doc.querySelectorAll("article.result")).map(result => {
|
||||
const title = result.querySelector("h3")?.textContent?.trim()
|
||||
const link = result.querySelector("a.url_header")?.getAttribute("href")
|
||||
const content = result.querySelector("p.content")?.textContent?.trim()
|
||||
return { title, link, content }
|
||||
}).filter(result => result.title && result.link && result.content)
|
||||
|
||||
return searchResults
|
||||
} catch (error) {
|
||||
console.error('SearXNG web search failed:', error)
|
||||
return []
|
||||
}
|
||||
}
|
@ -5,6 +5,7 @@ import { getIsVisitSpecificWebsite, getSearchProvider } from "@/services/search"
|
||||
import { webSogouSearch } from "./search-engines/sogou"
|
||||
import { webBraveSearch } from "./search-engines/brave"
|
||||
import { getWebsiteFromQuery, processSingleWebsite } from "./website"
|
||||
import { searxngSearch } from "./search-engines/searxng"
|
||||
|
||||
const getHostName = (url: string) => {
|
||||
try {
|
||||
@ -23,6 +24,8 @@ const searchWeb = (provider: string, query: string) => {
|
||||
return webSogouSearch(query)
|
||||
case "brave":
|
||||
return webBraveSearch(query)
|
||||
case "searxng":
|
||||
return searxngSearch(query)
|
||||
default:
|
||||
return webGoogleSearch(query)
|
||||
}
|
||||
@ -47,11 +50,11 @@ export const getSystemPromptForWeb = async (query: string) => {
|
||||
|
||||
} else {
|
||||
const searchProvider = await getSearchProvider()
|
||||
//@ts-ignore
|
||||
search = await searchWeb(searchProvider, query)
|
||||
}
|
||||
|
||||
|
||||
|
||||
const search_results = search
|
||||
.map(
|
||||
(result, idx) =>
|
||||
|
Loading…
x
Reference in New Issue
Block a user