From 21f6c21de4fad1f8e466cb949054555f1c39fb9b Mon Sep 17 00:00:00 2001
From: Alexandre Tuleu <alexandre.tuleu.2005@polytechnique.org>
Date: Thu, 14 Jan 2016 14:15:01 +0100
Subject: [PATCH] Makes the sanitazition test pass

---
 bd.go      | 32 +++++++++++++++++++++++++++++---
 bd_test.go |  8 ++++----
 2 files changed, 33 insertions(+), 7 deletions(-)

diff --git a/bd.go b/bd.go
index 8875c2e..14e8ee7 100644
--- a/bd.go
+++ b/bd.go
@@ -1,6 +1,13 @@
 package main
 
-import "time"
+import (
+	"regexp"
+	"strings"
+	"time"
+
+	"golang.org/x/text/transform"
+	"golang.org/x/text/unicode/norm"
+)
 
 // An AlbumState describe the state of an Album
 type AlbumState int
@@ -35,8 +42,27 @@ type Album struct {
 	PrintDate    time.Time
 }
 
-func sanitizeTitleString(title string) string {
-	return ""
+var endDelim = regexp.MustCompile(` \(.*\)\z`)
+var wordBoundaries = regexp.MustCompile(`[^[:alnum:]]+`)
+var punctuation = regexp.MustCompile(`[!?\.:;,]`)
+
+func sanitizeTitleString(title string, removeEndParent bool) string {
+	// first sanitize accuented characters.
+	isOk := func(r rune) bool {
+		return r < 32 || r >= 127
+	}
+	// The isOk filter is such that there is no need to chain to norm.NFC
+	t := transform.Chain(norm.NFKD, transform.RemoveFunc(isOk))
+	// This Transformer could also trivially be applied as an io.Reader
+	// or io.Writer filter to automatically do such filtering when reading
+	// or writing data anywhere.
+	title, _, _ = transform.String(t, title)
+	//Now we remove all punctuation
+	if removeEndParent == true {
+		title = endDelim.ReplaceAllString(title, "")
+	}
+
+	return strings.Trim(wordBoundaries.ReplaceAllString(punctuation.ReplaceAllString(title, ""), "-"), "-")
 }
 
 func (*Album) GetBedethequeComURI() string {
diff --git a/bd_test.go b/bd_test.go
index 3125964..fe80a87 100644
--- a/bd_test.go
+++ b/bd_test.go
@@ -16,18 +16,18 @@ func (s *AlbumSuite) TestSanitazation(c *C) {
 	data := map[string]string{
 		"Le cycle de l'eau - I": "Le-cycle-de-l-eau-I",
 		"Nef des fous (La)":     "Nef-des-fous",
-		"Oiseau noir (L')":      "Oiseau-Noir",
+		"Oiseau noir (L')":      "Oiseau-noir",
 		"Foo":                   "Foo",
 		"Nuit de l'étoile (La)":         "Nuit-de-l-etoile",
 		"Mon Père saigne l'Histoire":    "Mon-Pere-saigne-l-Histoire",
 		"Les disparus d'apostrophes !":  "Les-disparus-d-apostrophes",
-		"Eden - It's an Endless World!": "Eden-it-s-an-Endless-World",
+		"Eden - It's an Endless World!": "Eden-It-s-an-Endless-World",
 		"100.000 femmes (Les)":          "100000-femmes",
 	}
 
 	for title, expected := range data {
-		san := sanitizeTitleString(title)
-		c.Check(san, Equals, expected)
+		san := sanitizeTitleString(title, true)
+		c.Check(san, Equals, expected, Commentf("Processing '%s'", title))
 	}
 }