func (ud *urlDriver) Cache(u *urls.Url) error {
for _, s := range badUrlStrings {
if strings.Index(u.Url, s) != -1 {
return fmt.Errorf("Url contains bad substring '%s'.", s)
}
}
res, err := http.Get(u.Url)
if err != nil {
return err
}
defer res.Body.Close()
if res.StatusCode != 200 {
return fmt.Errorf("Received non-200 response '%s' from server.",
res.Status)
}
// 1 << 22 == 4MB
if res.ContentLength > 1<<22 {
return fmt.Errorf("Response too large (%d MB) to cache safely.",
res.ContentLength/1024/1024)
}
u.CachedAs = ud.Encode(u.Url)
fh, err := os.OpenFile(util.JoinPath(*urlCacheDir, u.CachedAs),
os.O_WRONLY|os.O_CREATE|os.O_TRUNC, os.FileMode(0600))
defer fh.Close()
if err != nil {
return err
}
if _, err := io.Copy(fh, res.Body); err != nil {
return err
}
u.CacheTime = time.Now()
u.MimeType = res.Header.Get("Content-Type")
if _, err := ud.Upsert(bson.M{"url": u.Url}, u); err != nil {
return err
}
return nil
}
"os"
"strings"
"time"
)
const driverName string = "urls"
const shortenPath string = "/s/"
const cachePath string = "/c/"
const autoShortenLimit int = 120
var badUrlStrings = []string{
"4chan",
}
var urlCacheDir *string = flag.String("url_cache_dir",
util.JoinPath(os.Getenv("HOME"), ".sp0rkle"),
"Path to store cached content under.")
type urlDriver struct {
*urls.UrlCollection
// Remember the last url seen on a per-channel basis
lastseen map[string]bson.ObjectId
l logging.Logger
}
func UrlDriver(db *db.Database, l logging.Logger) *urlDriver {
return &urlDriver{
UrlCollection: urls.Collection(db, l),
lastseen: make(map[string]bson.ObjectId),
l: l,
}