From d69cd4400e166e519d38c1c5be690638605b8931 Mon Sep 17 00:00:00 2001
From: Richard Patel <terorie@alphakevin.club>
Date: Sat, 9 Feb 2019 16:46:36 +0100
Subject: [PATCH] Use fasthttp.PipelineClient

---
 config.go    |  4 ----
 crawl.go     | 50 ++++++++++++++++++++++++++++++++++----------------
 main.go      |  2 +-
 scheduler.go |  3 +++
 worker.go    | 10 ++++++++--
 5 files changed, 46 insertions(+), 23 deletions(-)

diff --git a/config.go b/config.go
index b8c15ef..1ef5e6c 100644
--- a/config.go
+++ b/config.go
@@ -194,10 +194,6 @@ func readConfig() {
 
 	config.UserAgent = viper.GetString(ConfUserAgent)
 
-	setDialTimeout(viper.GetDuration(ConfDialTimeout))
-
-	setTimeout(viper.GetDuration(ConfTimeout))
-
 	config.JobBufferSize = viper.GetInt(ConfJobBufferSize)
 
 	config.Verbose = viper.GetBool(ConfVerbose)
diff --git a/crawl.go b/crawl.go
index 3aef115..5ee17f0 100644
--- a/crawl.go
+++ b/crawl.go
@@ -3,6 +3,7 @@ package main
 import (
 	"bytes"
 	"crypto/tls"
+	"github.com/spf13/viper"
 	"github.com/terorie/od-database-crawler/ds/redblackhash"
 	"github.com/terorie/od-database-crawler/fasturl"
 	"github.com/valyala/fasthttp"
@@ -15,24 +16,33 @@ import (
 	"time"
 )
 
-var client = fasthttp.Client {
-	TLSConfig: &tls.Config{
-		InsecureSkipVerify: true,
-	},
+var tlsConfig = tls.Config {
+	InsecureSkipVerify: true,
 }
 
-func setDialTimeout(d time.Duration) {
-	client.Dial = func(addr string) (net.Conn, error) {
-		return fasthttp.DialTimeout(addr, d)
+func newHTTPClient(url *fasturl.URL) *fasthttp.PipelineClient {
+	var isTLS bool
+	switch url.Scheme {
+	case fasturl.SchemeHTTP:
+		isTLS = false
+	case fasturl.SchemeHTTPS:
+		isTLS = true
+	}
+
+	return &fasthttp.PipelineClient {
+		MaxConns:     viper.GetInt(ConfWorkers),
+		Addr:         url.Host,
+		IsTLS:        isTLS,
+		TLSConfig:    &tlsConfig,
+		ReadTimeout:  viper.GetDuration(ConfTimeout),
+		WriteTimeout: viper.GetDuration(ConfTimeout) / 2,
+		Dial: func(addr string) (conn net.Conn, e error) {
+			return fasthttp.DialTimeout(addr, viper.GetDuration(ConfDialTimeout))
+		},
 	}
 }
 
-func setTimeout(d time.Duration) {
-	client.ReadTimeout = d
-	client.WriteTimeout = d / 2
-}
-
-func GetDir(j *Job, f *File) (links []fasturl.URL, err error) {
+func (w *WorkerContext) GetDir(j *Job, f *File) (links []fasturl.URL, err error) {
 	f.IsDir = true
 	f.Name = path.Base(j.Uri.Path)
 
@@ -45,7 +55,7 @@ func GetDir(j *Job, f *File) (links []fasturl.URL, err error) {
 	res := fasthttp.AcquireResponse()
 	defer fasthttp.ReleaseResponse(res)
 
-	err = client.Do(req, res)
+	err = w.client.Do(req, res)
 	fasthttp.ReleaseRequest(req)
 
 	if err != nil {
@@ -123,6 +133,14 @@ func ParseDir(body []byte, baseUri *fasturl.URL) (links []fasturl.URL, err error
 					continue
 				}
 
+				if strings.HasSuffix(link.Path, ".php") {
+					continue
+				}
+
+				if strings.Contains(link.Path, "/cgi-bin/") {
+					continue
+				}
+
 				links = append(links, link)
 			}
 		}
@@ -131,7 +149,7 @@ func ParseDir(body []byte, baseUri *fasturl.URL) (links []fasturl.URL, err error
 	return
 }
 
-func GetFile(u fasturl.URL, f *File) (err error) {
+func (w *WorkerContext) GetFile(u fasturl.URL, f *File) (err error) {
 	f.IsDir = false
 	u.Path = path.Clean(u.Path)
 	f.Name = path.Base(u.Path)
@@ -148,7 +166,7 @@ func GetFile(u fasturl.URL, f *File) (err error) {
 	res.SkipBody = true
 	defer fasthttp.ReleaseResponse(res)
 
-	err = client.Do(req, res)
+	err = w.client.Do(req, res)
 	fasthttp.ReleaseRequest(req)
 
 	if err != nil {
diff --git a/main.go b/main.go
index e2d67ac..11c06b9 100644
--- a/main.go
+++ b/main.go
@@ -35,7 +35,7 @@ var serverCmd = cobra.Command {
 }
 
 var crawlCmd = cobra.Command {
-	Use: "crawl",
+	Use: "crawl <url>",
 	Short: "Crawl an URL",
 	Long: "Crawl the URL specified.\n" +
 		"Results will not be uploaded to the database,\n" +
diff --git a/scheduler.go b/scheduler.go
index 9abe491..a87bb27 100644
--- a/scheduler.go
+++ b/scheduler.go
@@ -25,6 +25,9 @@ func Schedule(c context.Context, remotes <-chan *OD) {
 		logrus.WithField("url", remote.BaseUri.String()).
 			Info("Starting crawler")
 
+		// Create HTTP client
+		remote.WCtx.Prepare()
+
 		// Collect results
 		results := make(chan File)
 
diff --git a/worker.go b/worker.go
index 118e28f..2bedf21 100644
--- a/worker.go
+++ b/worker.go
@@ -3,6 +3,7 @@ package main
 import (
 	"github.com/beeker1121/goque"
 	"github.com/sirupsen/logrus"
+	"github.com/valyala/fasthttp"
 	"math"
 	"sort"
 	"strings"
@@ -18,6 +19,11 @@ type WorkerContext struct {
 	Queue *BufferedQueue
 	lastRateLimit time.Time
 	numRateLimits int
+	client *fasthttp.PipelineClient
+}
+
+func (w *WorkerContext) Prepare() {
+	w.client = newHTTPClient(&w.OD.BaseUri)
 }
 
 func (w *WorkerContext) Worker(results chan<- File) {
@@ -91,7 +97,7 @@ func (w *WorkerContext) DoJob(job *Job, f *File) (newJobs []Job, err error) {
 	if len(job.Uri.Path) == 0 { return }
 	if job.Uri.Path[len(job.Uri.Path)-1] == '/' {
 		// Load directory
-		links, err := GetDir(job, f)
+		links, err := w.GetDir(job, f)
 		if err != nil {
 			if !isErrSilent(err) {
 				logrus.WithError(err).
@@ -141,7 +147,7 @@ func (w *WorkerContext) DoJob(job *Job, f *File) (newJobs []Job, err error) {
 		}
 	} else {
 		// Load file
-		err := GetFile(job.Uri, f)
+		err := w.GetFile(job.Uri, f)
 		if err != nil {
 			if !isErrSilent(err) {
 				logrus.WithError(err).