Search/text.go

259 lines
7.5 KiB
Go

package main
import (
"encoding/json"
"fmt"
"html/template"
"log"
"net/http"
"time"
)
var textSearchEngines []SearchEngine
var textResultsChan = make(chan []TextSearchResult)
func init() {
textSearchEngines = []SearchEngine{
{Name: "Google", Func: wrapTextSearchFunc(PerformGoogleTextSearch), Weight: 1},
{Name: "LibreX", Func: wrapTextSearchFunc(PerformLibreXTextSearch), Weight: 2},
{Name: "Brave", Func: wrapTextSearchFunc(PerformBraveTextSearch), Weight: 2},
{Name: "DuckDuckGo", Func: wrapTextSearchFunc(PerformDuckDuckGoTextSearch), Weight: 5},
// {Name: "SearXNG", Func: wrapTextSearchFunc(PerformSearXNGTextSearch), Weight: 2}, // Uncomment when implemented
}
}
func HandleTextSearch(w http.ResponseWriter, query, safe, lang string, page int) {
startTime := time.Now()
cacheKey := CacheKey{Query: query, Page: page, Safe: safe == "true", Lang: lang, Type: "text"}
combinedResults := getTextResultsFromCacheOrFetch(cacheKey, query, safe, lang, page)
hasPrevPage := page > 1 // dupe
//displayResults(w, combinedResults, query, lang, time.Since(startTime).Seconds(), page, hasPrevPage, hasNextPage)
// Prefetch next and previous pages
go prefetchPage(query, safe, lang, page+1)
if hasPrevPage {
go prefetchPage(query, safe, lang, page-1)
}
elapsedTime := time.Since(startTime)
tmpl, err := template.New("text.html").Funcs(funcs).ParseFiles("templates/text.html")
if err != nil {
log.Printf("Error parsing template: %v", err)
http.Error(w, "Internal Server Error", http.StatusInternalServerError)
return
}
data := struct {
Results []TextSearchResult
Query string
Page int
Fetched string
LanguageOptions []LanguageOption
CurrentLang string
HasPrevPage bool
HasNextPage bool
NoResults bool
}{
Results: combinedResults,
Query: query,
Page: page,
Fetched: fmt.Sprintf("%.2f seconds", elapsedTime.Seconds()),
LanguageOptions: languageOptions,
CurrentLang: lang,
HasPrevPage: page > 1,
HasNextPage: len(combinedResults) >= 50,
NoResults: len(combinedResults) == 0,
}
err = tmpl.Execute(w, data)
if err != nil {
log.Printf("Error executing template: %v", err)
http.Error(w, "Internal Server Error", http.StatusInternalServerError)
}
}
func getTextResultsFromCacheOrFetch(cacheKey CacheKey, query, safe, lang string, page int) []TextSearchResult {
cacheChan := make(chan []SearchResult)
var combinedResults []TextSearchResult
go func() {
results, exists := resultsCache.Get(cacheKey)
if exists {
log.Println("Cache hit")
cacheChan <- results
} else {
log.Println("Cache miss")
cacheChan <- nil
}
}()
select {
case results := <-cacheChan:
if results == nil {
combinedResults = fetchTextResults(query, safe, lang, page)
if len(combinedResults) > 0 {
resultsCache.Set(cacheKey, convertToSearchResults(combinedResults))
}
} else {
textResults, _, _ := convertToSpecificResults(results)
combinedResults = textResults
}
case <-time.After(2 * time.Second):
log.Println("Cache check timeout")
combinedResults = fetchTextResults(query, safe, lang, page)
if len(combinedResults) > 0 {
resultsCache.Set(cacheKey, convertToSearchResults(combinedResults))
}
}
return combinedResults
}
func prefetchPage(query, safe, lang string, page int) {
cacheKey := CacheKey{Query: query, Page: page, Safe: safe == "true", Lang: lang, Type: "text"}
if _, exists := resultsCache.Get(cacheKey); !exists {
log.Printf("Page %d not cached, caching now...", page)
pageResults := fetchTextResults(query, safe, lang, page)
if len(pageResults) > 0 {
resultsCache.Set(cacheKey, convertToSearchResults(pageResults))
}
} else {
log.Printf("Page %d already cached", page)
}
}
func fetchTextResults(query, safe, lang string, page int) []TextSearchResult {
var results []TextSearchResult
for _, engine := range textSearchEngines {
log.Printf("Using search engine: %s", engine.Name)
searchResults, duration, err := engine.Func(query, safe, lang, page)
updateEngineMetrics(&engine, duration, err == nil)
if err != nil {
log.Printf("Error performing search with %s: %v", engine.Name, err)
continue
}
results = append(results, validateResults(searchResults)...)
// If results are found, break out of the loop
if len(results) > 0 {
break
}
}
// If no results found after trying all engines
if len(results) == 0 {
log.Printf("No text results found for query: %s, trying other nodes", query)
results = tryOtherNodesForTextSearch(query, safe, lang, page, []string{hostID})
}
return results
}
func validateResults(searchResults []SearchResult) []TextSearchResult {
var validResults []TextSearchResult
// Remove anything that is missing a URL or Header
for _, result := range searchResults {
textResult := result.(TextSearchResult)
if textResult.URL != "" || textResult.Header != "" {
validResults = append(validResults, textResult)
}
}
return validResults
}
func wrapTextSearchFunc(f func(string, string, string, int) ([]TextSearchResult, time.Duration, error)) func(string, string, string, int) ([]SearchResult, time.Duration, error) {
return func(query, safe, lang string, page int) ([]SearchResult, time.Duration, error) {
textResults, duration, err := f(query, safe, lang, page)
if err != nil {
return nil, duration, err
}
searchResults := make([]SearchResult, len(textResults))
for i, result := range textResults {
searchResults[i] = result
}
return searchResults, duration, nil
}
}
func tryOtherNodesForTextSearch(query, safe, lang string, page int, visitedNodes []string) []TextSearchResult {
for _, nodeAddr := range peers {
if contains(visitedNodes, nodeAddr) {
continue // Skip nodes already visited
}
results, err := sendTextSearchRequestToNode(nodeAddr, query, safe, lang, page, visitedNodes)
if err != nil {
log.Printf("Error contacting node %s: %v", nodeAddr, err)
continue
}
if len(results) > 0 {
return results
}
}
return nil
}
func sendTextSearchRequestToNode(nodeAddr, query, safe, lang string, page int, visitedNodes []string) ([]TextSearchResult, error) {
visitedNodes = append(visitedNodes, nodeAddr)
searchParams := struct {
Query string `json:"query"`
Safe string `json:"safe"`
Lang string `json:"lang"`
Page int `json:"page"`
ResponseAddr string `json:"responseAddr"`
VisitedNodes []string `json:"visitedNodes"`
}{
Query: query,
Safe: safe,
Lang: lang,
Page: page,
ResponseAddr: fmt.Sprintf("http://localhost:%d/node", config.Port),
VisitedNodes: visitedNodes,
}
msgBytes, err := json.Marshal(searchParams)
if err != nil {
return nil, fmt.Errorf("failed to marshal search parameters: %v", err)
}
msg := Message{
ID: hostID,
Type: "search-text",
Content: string(msgBytes),
}
err = sendMessage(nodeAddr, msg)
if err != nil {
return nil, fmt.Errorf("failed to send search request to node %s: %v", nodeAddr, err)
}
// Wait for results
select {
case res := <-textResultsChan:
return res, nil
case <-time.After(20 * time.Second):
return nil, fmt.Errorf("timeout waiting for results from node %s", nodeAddr)
}
}
func handleTextResultsMessage(msg Message) {
var results []TextSearchResult
err := json.Unmarshal([]byte(msg.Content), &results)
if err != nil {
log.Printf("Error unmarshalling text results: %v", err)
return
}
log.Printf("Received text results: %+v", results)
// Send results to textResultsChan
go func() {
textResultsChan <- results
}()
}