From 793c2b2a704b6038c70cd5d005ef184424ebcf78 Mon Sep 17 00:00:00 2001
From: Unbewohnte <me@unbewohnte.su>
Date: Mon, 23 Jan 2023 16:55:23 +0300
Subject: [PATCH] HUGE FIX: partially fixed abnormal memory consumption via
 actually closing http connections; Side-effect: no more what I thought was
 "bandwidth-throttling"

---
 Makefile             |  2 +-
 src/main.go          | 15 ++++++++++-----
 src/web/requests.go  | 15 ++++++++-------
 src/worker/worker.go |  3 +++
 4 files changed, 22 insertions(+), 13 deletions(-)

diff --git a/Makefile b/Makefile
index 528ea23..553ec79 100644
--- a/Makefile
+++ b/Makefile
@@ -20,7 +20,7 @@ WINDIR64:=$(WINDIR)_x64
 DARWINDIR64:=$(DARWINDIR)_x64
 
 
-all: clean
+all:
 	cd $(SRCDIR) && go build && mv $(EXE) ..
 
 test: all
diff --git a/src/main.go b/src/main.go
index 204ac5c..212b956 100644
--- a/src/main.go
+++ b/src/main.go
@@ -24,6 +24,8 @@ import (
 	"fmt"
 	"io"
 	"log"
+	"net/http"
+	_ "net/http/pprof"
 	"net/url"
 	"os"
 	"os/signal"
@@ -37,7 +39,7 @@ import (
 	"unbewohnte/wecr/worker"
 )
 
-const version = "v0.2.1"
+const version = "v0.2.2"
 
 const (
 	defaultConfigFile           string = "conf.json"
@@ -80,9 +82,8 @@ func init() {
 	// set log output
 	logger.SetOutput(os.Stdout)
 
-	// and work around random log prints by /x/net library
+	// make default http logger silent
 	log.SetOutput(io.Discard)
-	log.SetFlags(0)
 
 	// parse and process flags
 	flag.Parse()
@@ -137,6 +138,10 @@ func init() {
 
 	// global path to output file
 	outputFilePath = filepath.Join(workingDirectory, *outputFile)
+
+	go func() {
+		http.ListenAndServe(":8000", nil)
+	}()
 }
 
 func main() {
@@ -321,8 +326,8 @@ func main() {
 	defer outputFile.Close()
 
 	// prepare channels
-	jobs := make(chan web.Job, conf.Workers*5)
-	results := make(chan web.Result, conf.Workers*5)
+	jobs := make(chan web.Job, conf.Workers)
+	results := make(chan web.Result, conf.Workers)
 
 	// create initial jobs
 	for _, initialPage := range conf.InitialPages {
diff --git a/src/web/requests.go b/src/web/requests.go
index abe6e66..90e6d83 100644
--- a/src/web/requests.go
+++ b/src/web/requests.go
@@ -27,10 +27,8 @@ import (
 
 // Get page data coming from url with optional user agent and timeout
 func GetPage(url string, userAgent string, timeOutMs uint64) ([]byte, error) {
-	// client := &http.Client{}
-	// client.CheckRedirect = http.DefaultClient.CheckRedirect
-	// client.Transport = http.DefaultClient.Transport
-	// client.Timeout = time.Duration(timeOutMs)
+	http.DefaultClient.CloseIdleConnections()
+	http.DefaultClient.Timeout = time.Duration(timeOutMs * uint64(time.Millisecond))
 
 	req, err := http.NewRequest("GET", url, nil)
 	if err != nil {
@@ -38,19 +36,18 @@ func GetPage(url string, userAgent string, timeOutMs uint64) ([]byte, error) {
 	}
 	req.Header.Set("User-Agent", userAgent)
 
-	// response, err := client.Do(req)
 	response, err := http.DefaultClient.Do(req)
 	if err != nil {
 		return nil, err
 	}
 	defer response.Body.Close()
 
-	responseBody, err := io.ReadAll(response.Body)
+	pageData, err := io.ReadAll(response.Body)
 	if err != nil {
 		return nil, err
 	}
 
-	return responseBody, nil
+	return pageData, nil
 }
 
 // Fetch file from url and save to file at filePath
@@ -65,11 +62,13 @@ func FetchFile(url string, userAgent string, timeOutMs uint64, filePath string)
 		return err
 	}
 	req.Header.Set("User-Agent", userAgent)
+	req.Close = true
 
 	response, err := client.Do(req)
 	if err != nil {
 		return nil
 	}
+	response.Close = true
 	defer response.Body.Close()
 
 	file, err := os.Create(filePath)
@@ -80,5 +79,7 @@ func FetchFile(url string, userAgent string, timeOutMs uint64, filePath string)
 
 	_, _ = io.Copy(file, response.Body)
 
+	client.CloseIdleConnections()
+
 	return nil
 }
diff --git a/src/worker/worker.go b/src/worker/worker.go
index e93007a..40769e7 100644
--- a/src/worker/worker.go
+++ b/src/worker/worker.go
@@ -222,6 +222,7 @@ func (w *Worker) Work() {
 					}
 				}
 			}
+			pageLinks = nil
 		}()
 
 		// process and output result
@@ -336,6 +337,8 @@ func (w *Worker) Work() {
 		if savePage {
 			w.savePage(pageURL, pageData)
 		}
+		pageData = nil
+		pageURL = nil
 
 		// sleep before the next request
 		time.Sleep(time.Duration(w.Conf.Requests.RequestPauseMs * uint64(time.Millisecond)))