2020-10-31 11:02:59 +00:00
|
|
|
package workers
|
|
|
|
|
|
|
|
import (
|
|
|
|
"fmt"
|
|
|
|
"github.com/deletescape/goop/internal/utils"
|
|
|
|
"github.com/valyala/fasthttp"
|
|
|
|
"io/ioutil"
|
|
|
|
"os"
|
|
|
|
"regexp"
|
|
|
|
"sync"
|
|
|
|
"time"
|
|
|
|
)
|
|
|
|
|
|
|
|
var refRegex = regexp.MustCompile(`(?m)(refs(/[a-zA-Z0-9\-\.\_\*]+)+)`)
|
|
|
|
|
2020-10-31 12:38:17 +00:00
|
|
|
func FindRefWorker(c *fasthttp.Client, queue chan string, baseUrl, baseDir string, wg *sync.WaitGroup) {
|
2020-10-31 11:02:59 +00:00
|
|
|
wg.Add(1)
|
|
|
|
defer wg.Done()
|
|
|
|
var ctr int
|
|
|
|
for {
|
|
|
|
select {
|
2020-10-31 12:38:17 +00:00
|
|
|
case path := <-queue:
|
|
|
|
if path == "" {
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
ctr = 0
|
2020-10-31 11:02:59 +00:00
|
|
|
uri := utils.Url(baseUrl, path)
|
|
|
|
code, body, err := c.Get(nil, uri)
|
|
|
|
fmt.Printf("[-] Fetching %s [%d]\n", uri, code)
|
|
|
|
if err != nil {
|
|
|
|
fmt.Fprintf(os.Stderr, "error: %s\n", err)
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
if code == 200 {
|
|
|
|
if utils.IsHtml(body) {
|
|
|
|
fmt.Printf("warning: %s appears to be an html file, skipping\n", uri)
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
if err := utils.CreateParentFolders(utils.Url(baseDir, path)); err != nil {
|
|
|
|
fmt.Fprintf(os.Stderr, "error: %s\n", err)
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
if err := ioutil.WriteFile(utils.Url(baseDir, path), body, os.ModePerm); err != nil {
|
|
|
|
fmt.Fprintf(os.Stderr, "error: %s\n", err)
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
|
|
|
|
for _, ref := range refRegex.FindAll(body, -1) {
|
2020-10-31 12:38:17 +00:00
|
|
|
queue <- utils.Url(".git", string(ref))
|
2020-10-31 11:02:59 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
default:
|
|
|
|
// TODO: get rid of dirty hack somehow
|
|
|
|
if ctr >= graceTimes {
|
|
|
|
return
|
|
|
|
}
|
|
|
|
ctr++
|
|
|
|
time.Sleep(gracePeriod)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|