server side favicon retrieving and caching

internal/api/handler.go

@@ -34,6 +34,7 @@ func NewHandler(cfg config.ConfigInstance) http.Handler {
 	mux.HandleFunc("GET", "/v1/schema/{filename...}", v1.GetSchemaFile)
 	mux.HandleFunc("GET", "/v1/stats", useCfg(cfg, v1.Stats))
 	mux.HandleFunc("GET", "/v1/stats/ws", useCfg(cfg, v1.StatsWS))
+	mux.HandleFunc("GET", "/v1/favicon/{alias}", auth.RequireAuth(v1.GetFavIcon))
 	return mux
 }
 

internal/api/v1/favicon.go

@@ -0,0 +1,200 @@
+package v1
+
+import (
+	"bytes"
+	"context"
+	"errors"
+	"io"
+	"net/http"
+	"strings"
+	"sync"
+	"time"
+
+	"github.com/PuerkitoBio/goquery"
+	"github.com/vincent-petithory/dataurl"
+	U "github.com/yusing/go-proxy/internal/api/v1/utils"
+	"github.com/yusing/go-proxy/internal/logging"
+	route "github.com/yusing/go-proxy/internal/route/types"
+)
+
+type content struct {
+	header http.Header
+	data   []byte
+	status int
+}
+
+func newContent() *content {
+	return &content{
+		header: make(http.Header),
+	}
+}
+
+func (c *content) Header() http.Header {
+	return c.header
+}
+
+func (c *content) Write(data []byte) (int, error) {
+	c.data = append(c.data, data...)
+	return len(data), nil
+}
+
+func (c *content) WriteHeader(statusCode int) {
+	c.status = statusCode
+}
+
+// GetFavIcon returns the favicon of the route
+//
+// Returns:
+//   - 200 OK: if icon found
+//   - 400 Bad Request: if alias is empty or route is not HTTPRoute
+//   - 404 Not Found: if route or icon not found
+//   - 500 Internal Server Error: if internal error
+//   - others: depends on route handler response
+func GetFavIcon(w http.ResponseWriter, req *http.Request) {
+	alias := req.PathValue("alias")
+	if alias == "" {
+		U.RespondError(w, U.ErrMissingKey("alias"), http.StatusBadRequest)
+		return
+	}
+	r := listRoute(alias)
+	if r == nil {
+		http.NotFound(w, req)
+		return
+	}
+	switch r := r.(type) {
+	case route.HTTPRoute:
+		var icon []byte
+		var status int
+		var errMsg string
+
+		homepage := r.RawEntry().Homepage
+		if homepage != nil && homepage.Icon != nil {
+			if homepage.Icon.IsRelative {
+				icon, status, errMsg = findIcon(r, req, homepage.Icon.Value)
+			} else {
+				icon, status, errMsg = getIconAbsolute(homepage.Icon.Value)
+			}
+		} else {
+			// try extract from "link[rel=icon]"
+			icon, status, errMsg = findIcon(r, req, "/")
+		}
+		if status != http.StatusOK {
+			http.Error(w, errMsg, status)
+			return
+		}
+		U.WriteBody(w, icon)
+	default:
+		http.Error(w, "bad request", http.StatusBadRequest)
+	}
+}
+
+// cache key can be absolute url or route name.
+var (
+	iconCache   = make(map[string][]byte)
+	iconCacheMu sync.RWMutex
+)
+
+func loadIconCache(key string) (icon []byte, ok bool) {
+	iconCacheMu.RLock()
+	icon, ok = iconCache[key]
+	iconCacheMu.RUnlock()
+	return
+}
+
+func storeIconCache(key string, icon []byte) {
+	iconCacheMu.Lock()
+	defer iconCacheMu.Unlock()
+	iconCache[key] = icon
+}
+
+func getIconAbsolute(url string) ([]byte, int, string) {
+	icon, ok := loadIconCache(url)
+	if ok {
+		return icon, http.StatusOK, ""
+	}
+
+	resp, err := U.Get(url)
+	if err != nil {
+		storeIconCache(url, nil)
+		logging.Error().Err(err).
+			Str("url", url).
+			Msg("failed to get icon")
+		return nil, http.StatusBadGateway, "connection error"
+	}
+
+	defer resp.Body.Close()
+	icon, err = io.ReadAll(resp.Body)
+	if err != nil {
+		// storeIconCache(url, nil) // can retry
+		logging.Error().Err(err).
+			Str("url", url).
+			Msg("failed to read icon")
+		return nil, http.StatusInternalServerError, "internal error"
+	}
+
+	storeIconCache(url, icon)
+	return icon, http.StatusOK, ""
+}
+
+func findIcon(r route.HTTPRoute, req *http.Request, path string) (icon []byte, status int, errMsg string) {
+	key := r.TargetName()
+	icon, ok := loadIconCache(key)
+	if ok {
+		if icon == nil {
+			return nil, http.StatusNotFound, "icon not found"
+		}
+		return icon, http.StatusOK, ""
+	}
+
+	icon, status, errMsg = findIconSlow(r, req, path)
+	// set even if error (nil)
+	storeIconCache(key, icon)
+	return
+}
+
+func findIconSlow(r route.HTTPRoute, req *http.Request, path string) (icon []byte, status int, errMsg string) {
+	c := newContent()
+	ctx, cancel := context.WithTimeoutCause(req.Context(), 3*time.Second, errors.New("favicon request timeout"))
+	defer cancel()
+	newReq := req.WithContext(ctx)
+	newReq.URL.Path = path
+	newReq.URL.RawPath = path
+	newReq.URL.RawQuery = ""
+	newReq.RequestURI = path
+	r.ServeHTTP(c, newReq)
+	if c.status != http.StatusOK {
+		return nil, c.status, "upstream error: " + http.StatusText(c.status)
+	}
+	// return icon data
+	if path != "/" {
+		return c.data, http.StatusOK, ""
+	}
+	// try extract from "link[rel=icon]" from path "/"
+	doc, err := goquery.NewDocumentFromReader(bytes.NewBuffer(c.data))
+	if err != nil {
+		return nil, http.StatusInternalServerError, "internal error"
+	}
+	ele := doc.Find("link[rel=icon]").First()
+	if ele.Length() == 0 {
+		return nil, http.StatusNotFound, "icon not found"
+	}
+	href := ele.AttrOr("href", "")
+	if href == "" {
+		return nil, http.StatusNotFound, "icon not found"
+	}
+	// https://en.wikipedia.org/wiki/Data_URI_scheme
+	if strings.HasPrefix(href, "data:image/") {
+		dataURI, err := dataurl.DecodeString(href)
+		if err != nil {
+			logging.Error().Err(err).
+				Str("route", r.TargetName()).
+				Msg("failed to decode favicon")
+			return nil, http.StatusInternalServerError, "internal error"
+		}
+		return dataURI.Data, http.StatusOK, ""
+	}
+	if href[0] != '/' {
+		return getIconAbsolute(href)
+	}
+	return findIconSlow(r, req, href)
+}