From c070c634630fe911610c1b178ea7026d07fd7abb Mon Sep 17 00:00:00 2001
From: Alexandre Tuleu <alexandre.tuleu.2005@polytechnique.org>
Date: Sun, 17 Jan 2016 19:05:30 +0100
Subject: [PATCH] Adds HTTP Get rate limitation on cache

We don't want to have a large bandwidth on bedetheque.com, so the pace
at which we perform GET request is limited by a maximal number of
request over a window (per example no more than 10 request over 10
second)

If more request are required, the request is simply paused until we go
back within the limit ;). Go rulez !
---
 album_description_cache.go      | 30 ++++++++++++---------
 album_description_cache_test.go |  3 ++-
 rate_limited_getter.go          | 47 +++++++++++++++++++++++++++++++++
 3 files changed, 67 insertions(+), 13 deletions(-)
 create mode 100644 rate_limited_getter.go

diff --git a/album_description_cache.go b/album_description_cache.go
index f421f26..b19791e 100644
--- a/album_description_cache.go
+++ b/album_description_cache.go
@@ -4,7 +4,6 @@ import (
 	"encoding/json"
 	"fmt"
 	"io"
-	"net/http"
 	"os"
 	"path"
 	"path/filepath"
@@ -21,17 +20,17 @@ import (
 type AlbumDescriptionCache struct {
 	basepath string
 	lock     lockfile.Lockfile
-
+	getter   HTTPGetter
 	// time to live of the cache, data which is older than this TTL will be automatically removed
 	TTL time.Duration
 }
 
 // NewAlbumDescriptionCache is creating a new album description at specified location
-func NewAlbumDescriptionCache(path string) (*AlbumDescriptionCache, error) {
+func NewAlbumDescriptionCache(path string, maxRequest uint, window time.Duration) (*AlbumDescriptionCache, error) {
 	res := &AlbumDescriptionCache{
 		basepath: path,
-
-		TTL: 3 * 31 * 24 * time.Hour, // 3 Months
+		getter:   NewRateLimitedGetter(maxRequest, window),
+		TTL:      3 * 31 * 24 * time.Hour, // 3 Months
 	}
 	var err error
 	res.lock, err = lockfile.New(filepath.Join(path, "global.lock"))
@@ -79,7 +78,7 @@ func linkFromSelection(s *goquery.Selection) Link {
 
 func (c *AlbumDescriptionCache) fetchAndCache(ID AlbumID) (*AlbumDescription, error) {
 	URL := path.Join("www.bedetheque.com", fmt.Sprintf("BD--%d.html", ID))
-	resp, err := http.Get("http://" + URL)
+	resp, err := c.getter.Get("http://" + URL)
 	if err != nil {
 		return nil, err
 	}
@@ -107,7 +106,7 @@ func (c *AlbumDescriptionCache) fetchAndCache(ID AlbumID) (*AlbumDescription, er
 			return
 		}
 
-		resp, err := http.Get(src)
+		resp, err := c.getter.Get(src)
 		if err != nil {
 			errors <- err
 			return
@@ -224,13 +223,20 @@ func (c *AlbumDescriptionCache) fetchAndCache(ID AlbumID) (*AlbumDescription, er
 		}
 	}
 
-	if len(errorList) == 0 {
-		return res, nil
+	if len(errorList) != 0 {
+		return nil, fmt.Errorf("Could not parse description from http://%s:%s",
+			URL,
+			strings.Join(append([]string{""}, errorList...), "\n * "))
 	}
 
-	return nil, fmt.Errorf("Could not parse description from http://%s:%s",
-		URL,
-		strings.Join(append([]string{""}, errorList...), "\n * "))
+	f, err := os.Create(c.albumPath(ID))
+	if err != nil {
+		return res, err
+	}
+	defer f.Close()
+
+	enc := json.NewEncoder(f)
+	return res, enc.Encode(res)
 }
 
 // GetDescription retrieves from the cache or either from www.bedetheque.com the AlbumDescription of an album
diff --git a/album_description_cache_test.go b/album_description_cache_test.go
index 4b2ad21..06022d8 100644
--- a/album_description_cache_test.go
+++ b/album_description_cache_test.go
@@ -3,6 +3,7 @@ package main
 import (
 	"io/ioutil"
 	"os"
+	"time"
 
 	. "gopkg.in/check.v1"
 )
@@ -30,7 +31,7 @@ func (s *AlbumDescriptionCacheSuite) TestCanFetchCache(c *C) {
 	c.Assert(err, IsNil)
 	defer os.RemoveAll(tmpdir)
 
-	cache, err := NewAlbumDescriptionCache(tmpdir)
+	cache, err := NewAlbumDescriptionCache(tmpdir, 10, 10*time.Second)
 	c.Assert(err, IsNil)
 
 	for ID, expected := range data {
diff --git a/rate_limited_getter.go b/rate_limited_getter.go
new file mode 100644
index 0000000..c766573
--- /dev/null
+++ b/rate_limited_getter.go
@@ -0,0 +1,47 @@
+package main
+
+import (
+	"net/http"
+	"time"
+)
+
+// An HTTPGetter can GET over HTTP
+type HTTPGetter interface {
+	Get(URL string) (*http.Response, error)
+}
+
+type simpleHTTPGetter struct{}
+
+// NewHTTPGetter returns the simplest object that can GET over HTTP
+func NewHTTPGetter() HTTPGetter {
+	return &simpleHTTPGetter{}
+}
+
+func (g *simpleHTTPGetter) Get(URL string) (*http.Response, error) {
+	return http.Get(URL)
+}
+
+type rateLimitedHTTPGetter struct {
+	tokens chan bool
+	window time.Duration
+}
+
+// NewRateLimitedGetter returns an HTTPGetter that is limited by a
+// maximal amount of request over a time window
+func NewRateLimitedGetter(maxRequest uint, window time.Duration) HTTPGetter {
+	return &rateLimitedHTTPGetter{
+		tokens: make(chan bool, maxRequest),
+		window: window,
+	}
+}
+
+func (g *rateLimitedHTTPGetter) Get(URL string) (*http.Response, error) {
+	g.tokens <- true
+	defer func() {
+		go func() {
+			time.Sleep(g.window)
+			<-g.tokens
+		}()
+	}()
+	return http.Get(URL)
+}