Delays and logs refactored as middlewares.

This commit is contained in:
Musab Gültekin
2019-06-20 09:54:30 +03:00
parent 514fe2e8d2
commit f88b88986c
2 changed files with 31 additions and 32 deletions

View File

@ -5,7 +5,10 @@ import (
"github.com/PuerkitoBio/goquery"
"github.com/geziyor/geziyor/internal"
"log"
"math/rand"
"os"
"runtime/debug"
"time"
)
// RequestMiddleware called before requests made.
@ -15,6 +18,11 @@ type RequestMiddleware func(g *Geziyor, r *Request)
// ResponseMiddleware called after request response receive
type ResponseMiddleware func(g *Geziyor, r *Response)
func init() {
log.SetOutput(os.Stdout)
rand.Seed(time.Now().UnixNano())
}
// recoverMiddleware recovers scraping being crashed.
// Logs error and stack trace
func recoverMiddleware() {
@ -51,6 +59,22 @@ func defaultHeadersMiddleware(g *Geziyor, r *Request) {
r.Header = internal.SetDefaultHeader(r.Header, "User-Agent", g.Opt.UserAgent)
}
// delayMiddleware delays requests
func delayMiddleware(g *Geziyor, r *Request) {
if g.Opt.RequestDelayRandomize {
min := float64(g.Opt.RequestDelay) * 0.5
max := float64(g.Opt.RequestDelay) * 1.5
time.Sleep(time.Duration(rand.Intn(int(max-min)) + int(min)))
} else {
time.Sleep(g.Opt.RequestDelay)
}
}
// logMiddleware logs requests
func logMiddleware(g *Geziyor, r *Request) {
log.Println("Fetching: ", r.URL.String())
}
// parseHTMLMiddleware parses response if response is HTML
func parseHTMLMiddleware(g *Geziyor, r *Response) {
if !g.Opt.ParseHTMLDisabled && r.isHTML() {