Files
page-assist/src/web/web.ts

132 lines
3.9 KiB
TypeScript
Raw Normal View History

2024-03-23 14:44:05 +05:30
import { getWebSearchPrompt } from "~/services/ollama"
2024-04-28 00:36:33 +05:30
import { webGoogleSearch } from "./search-engines/google"
import { webDuckDuckGoSearch } from "./search-engines/duckduckgo"
import { getIsVisitSpecificWebsite, getSearchProvider } from "@/services/search"
2024-04-28 00:36:33 +05:30
import { webSogouSearch } from "./search-engines/sogou"
import { webBraveSearch } from "./search-engines/brave"
import { getWebsiteFromQuery, processSingleWebsite } from "./website"
import { searxngSearch } from "./search-engines/searxng"
2024-12-13 20:03:52 +05:30
import { braveAPISearch } from "./search-engines/brave-api"
2025-02-01 11:22:12 +05:30
import { webBaiduSearch } from "./search-engines/baidu"
2025-02-14 18:17:12 +08:00
import { searchIod } from "./iod"
import type { IodRegistryEntry } from "~/types/iod"
const getHostName = (url: string) => {
2024-03-31 15:57:56 +05:30
try {
const hostname = new URL(url).hostname
return hostname
} catch (e) {
return ""
}
}
const searchWeb = (provider: string, query: string) => {
switch (provider) {
case "duckduckgo":
return webDuckDuckGoSearch(query)
2024-04-28 00:36:33 +05:30
case "sogou":
return webSogouSearch(query)
case "brave":
return webBraveSearch(query)
case "searxng":
return searxngSearch(query)
2024-12-13 20:03:52 +05:30
case "brave-api":
return braveAPISearch(query)
2025-02-01 11:22:12 +05:30
case "baidu":
return webBaiduSearch(query)
2024-03-31 15:57:56 +05:30
default:
return webGoogleSearch(query)
}
}
2025-02-14 18:17:12 +08:00
export const getSystemPromptForWeb = async (
query: string,
keywords: string[] = [],
webSearch = true,
iodSearch = false
) => {
2024-03-31 15:57:56 +05:30
try {
const websiteVisit = getWebsiteFromQuery(query)
2025-02-14 18:17:12 +08:00
let webSearchResults: {
url: any
content: string
}[] = []
2025-02-14 18:17:12 +08:00
// let search_results_web = ""
2025-02-14 18:17:12 +08:00
if (webSearch) {
const isVisitSpecificWebsite = await getIsVisitSpecificWebsite()
2025-02-14 18:17:12 +08:00
if (isVisitSpecificWebsite && websiteVisit.hasUrl) {
const url = websiteVisit.url
const queryWithoutUrl = websiteVisit.queryWithouUrls
webSearchResults = await processSingleWebsite(url, queryWithoutUrl)
} else {
const searchProvider = await getSearchProvider()
webSearchResults = await searchWeb(searchProvider, query)
}
2025-02-14 18:17:12 +08:00
// search_results_web = webSearchResults
// .map(
// (result, idx) =>
// `<result source="${result.url}" id="${idx}">${result.content}</result>`
// )
// .join("\n")
}
let iodSearchResults: IodRegistryEntry[] = []
2025-02-14 18:17:12 +08:00
// let search_results_iod = ""
if (iodSearch) {
iodSearchResults = await searchIod(query, keywords)
// search_results_iod = iodSearchResults
// .map(
// (result, idx) =>
// `<result source="${result.url}" id="${idx}">${result.content}</result>`
// )
// .join("\n")
}
2025-02-15 13:02:24 +08:00
const iod_search_results = iodSearchResults.map((res) => ({
2025-02-14 18:17:12 +08:00
url: `${res.doId}: ${res.name}`,
content: res.description
2025-02-15 13:02:24 +08:00
})).map(
(result, idx) =>
`<result source="${result.url}" id="${idx}">${result.content}</result>`
)
.join("\n");
console.log("iod_search_result:"+iod_search_results);
const web_search_results = webSearchResults.map(
(result, idx) =>
`<result source="${result.url}" id="${idx}">${result.content}</result>`
)
.join("\n");
const search_results = (iodSearch?"<数联网搜索结果>"+iod_search_results+"</数联网搜索结果>":"")
+ (webSearch?"<万维网搜索结果>"+web_search_results+"</万维网搜索结果>":"");
2024-03-31 15:57:56 +05:30
const current_date_time = new Date().toLocaleString()
2024-03-31 15:57:56 +05:30
const system = await getWebSearchPrompt()
2024-03-31 15:57:56 +05:30
const prompt = system
.replace("{current_date_time}", current_date_time)
.replace("{search_results}", search_results)
2024-03-31 15:57:56 +05:30
return {
prompt,
2025-02-14 18:17:12 +08:00
webSources: webSearchResults.map((result) => {
return {
2024-03-31 15:57:56 +05:30
url: result.url,
name: getHostName(result.url),
type: "url"
}
2025-02-14 18:17:12 +08:00
}),
iodSources: iodSearchResults,
2024-03-31 15:57:56 +05:30
}
} catch (e) {
console.error(e)
return {
prompt: "",
source: []
}
2024-03-31 15:57:56 +05:30
}
}