ansxor.ca
forked from
jollywhoppers.com/witchsky.app
Bluesky app fork with some witchin' additions 💫
Sitemap Handlers for bsky.app (#9525)
* implement sitemap handlers for users
* ensure compressed payload is passed through
* improved handling
* reverse header order
* add the sitemap to robots.txt
authored by
Jim Calabro
and committed by
GitHub
f7e61056
1fded4e9
+72
bskyweb/cmd/bskyweb/server.go
+72
bskyweb/cmd/bskyweb/server.go
···
9
9
"encoding/json"
10
10
"errors"
11
11
"fmt"
12
+
"io"
12
13
"io/fs"
14
+
"log/slog"
13
15
"net/http"
14
16
"net/netip"
15
17
"net/url"
···
40
42
cfg *Config
41
43
42
44
ipccClient http.Client
45
+
46
+
// sitemapClient is used for fetching sitemaps from the appview. It has
47
+
// DisableCompression set to true so that gzipped responses are passed
48
+
// through without being decompressed.
49
+
sitemapClient http.Client
43
50
}
44
51
45
52
type Config struct {
···
114
121
TLSClientConfig: &tls.Config{
115
122
InsecureSkipVerify: true,
116
123
},
124
+
},
125
+
},
126
+
sitemapClient: http.Client{
127
+
Transport: &http.Transport{
128
+
MaxIdleConns: 100,
129
+
MaxIdleConnsPerHost: 10,
130
+
IdleConnTimeout: 90 * time.Second,
131
+
TLSHandshakeTimeout: 10 * time.Second,
132
+
ForceAttemptHTTP2: true,
133
+
DisableCompression: true,
117
134
},
118
135
},
119
136
}
···
336
353
337
354
// ipcc
338
355
e.GET("/ipcc", server.WebIpCC)
356
+
357
+
// sitemap handlers
358
+
e.GET("/sitemap/users.xml.gz", server.handleSitemapUsersIndex)
359
+
e.GET("/sitemap/users/*", server.handleSitemapUsersSubpage)
339
360
340
361
if linkHost != "" {
341
362
linkUrl, err := url.Parse(linkHost)
···
753
774
}
754
775
return c.JSON(200, outResponse)
755
776
}
777
+
778
+
func (srv *Server) handleSitemapUsersIndex(c echo.Context) error {
779
+
url := fmt.Sprintf("%s/external/sitemap/users.xml.gz", srv.cfg.appviewHost)
780
+
return srv.serveSitemapRequest(c, url, "user index")
781
+
}
782
+
783
+
func (srv *Server) handleSitemapUsersSubpage(c echo.Context) error {
784
+
path := c.Param("*")
785
+
url := fmt.Sprintf("%s/external/sitemap/users/%s", srv.cfg.appviewHost, path)
786
+
return srv.serveSitemapRequest(c, url, "user subpage")
787
+
}
788
+
789
+
func (srv *Server) serveSitemapRequest(c echo.Context, url, sitemapType string) error {
790
+
req, err := http.NewRequest(http.MethodGet, url, nil)
791
+
if err != nil {
792
+
slog.Error("failed to construct sitemap request", "err", err, "type", sitemapType)
793
+
return c.String(http.StatusInternalServerError, "Internal Server Error")
794
+
}
795
+
796
+
resp, err := srv.sitemapClient.Do(req)
797
+
if err != nil {
798
+
slog.Error("failed to send sitemap request to appview", "err", err, "type", sitemapType)
799
+
return c.String(http.StatusInternalServerError, "Internal Server Error")
800
+
}
801
+
defer resp.Body.Close()
802
+
803
+
if resp.StatusCode != http.StatusOK {
804
+
buf, err := io.ReadAll(resp.Body)
805
+
if err != nil {
806
+
slog.Error("failed to read sitemap error response body", "err", err)
807
+
}
808
+
809
+
slog.Error("invalid sitemap response code",
810
+
"err", err,
811
+
"type", sitemapType,
812
+
"code", resp.StatusCode,
813
+
"body", string(buf),
814
+
)
815
+
return c.String(http.StatusInternalServerError, "Internal Server Error")
816
+
}
817
+
818
+
c.Response().Header().Set("Content-Type", "application/xml")
819
+
c.Response().Header().Set("Content-Encoding", "gzip")
820
+
c.Response().WriteHeader(resp.StatusCode)
821
+
822
+
if _, err = io.Copy(c.Response().Writer, resp.Body); err != nil {
823
+
slog.Error("failed to copy sitemap response body to client", "err", err, "type", sitemapType)
824
+
}
825
+
826
+
return nil
827
+
}