From 5df34b5165fee209b28cd9f268673f2a4fec06d9 Mon Sep 17 00:00:00 2001
From: llaffont <llaffont@afi-sa.fr>
Date: Tue, 18 Oct 2016 17:16:03 +0200
Subject: [PATCH] hotline #43215 fix title normalization
---
Linse.php | 17 +++--------------
Linse/Converter.php | 6 +++++-
Linse/Iso5426Converter.php | 2 +-
tests/LinseTest.php | 24 ++++++++++++++++++++++++
4 files changed, 33 insertions(+), 16 deletions(-)
diff --git a/Linse.php b/Linse.php
index de15a55..0d3fc00 100644
--- a/Linse.php
+++ b/Linse.php
@@ -23,7 +23,6 @@
class Linse {
protected $_converter;
- private $articles; // Articles rejetes
private $inclu; // Mots inclus
private $exclu; // Mots vides
private $pluriel; // Règles des pluriels
@@ -67,8 +66,6 @@ class Linse {
public function __construct() {
- $this->articles = ['L\'','LE ','LA ','LES ','UN ','UNE '];
-
$this->inclu = ['AN','AS','OR','U2','AI','LU','XO','DO','RE','MI','FA','SI','AC','DC','XX','B','C','D','E','F','G','H','I','J','K','M','P','Q','R','S','T','V','W','X','Y','Z','L','YU','UT','LI','OC','PI','ZU','WU','TO','OZ','ZZ','XX', 'PC', 'DS'];
$this->exclu = ['L','LE','LA','LES','UN','UNE','LES','DES','MES','TES','CES'];
@@ -126,17 +123,9 @@ class Linse {
public function codeAlphaTitre($titre) {
- $titre = $this->alphaMaj($titre);
- foreach($this->articles as $article) {
- $lg = strlen($article);
- if(strLeft($titre, $lg)==$article) {
- $titre = strMid($titre,$lg,256);
- break;
- }
- }
-
- $titre = $this->alphaMaj($titre);
- return $titre;
+ return preg_replace('/^((L|LE|LA|LES|UN|UNE)\s)/',
+ '',
+ $this->alphaMaj($titre));
}
diff --git a/Linse/Converter.php b/Linse/Converter.php
index f71b6a3..26ae5a5 100644
--- a/Linse/Converter.php
+++ b/Linse/Converter.php
@@ -39,6 +39,10 @@ class Linse_Converter {
public function toIndexable($data) {
- return trim(str_replace($this->_ascii_map, $this->_ascii_to_uppercase, $data));
+ return preg_replace('/\s+/',
+ ' ',
+ trim(str_replace($this->_ascii_map,
+ $this->_ascii_to_uppercase,
+ $data)));
}
}
\ No newline at end of file
diff --git a/Linse/Iso5426Converter.php b/Linse/Iso5426Converter.php
index 8133961..c49e8f6 100644
--- a/Linse/Iso5426Converter.php
+++ b/Linse/Iso5426Converter.php
@@ -22,7 +22,7 @@
class Linse_Iso5426Converter extends Linse_Converter {
public function toIndexable($data) {
- $data = str_replace([chr(136), chr(137)], '', $data); // Les delimiteurs d'article bnf
+ $data = preg_replace('/\210.+\211/', '', $data); // Les delimiteurs d'article bnf NSB NSE
$result = '';
$len = strlen($data);
diff --git a/tests/LinseTest.php b/tests/LinseTest.php
index a3ef518..939cb42 100644
--- a/tests/LinseTest.php
+++ b/tests/LinseTest.php
@@ -70,4 +70,28 @@ class LinseTest extends PHPUnit_Framework_TestCase {
public function ethShouldBeHandled() {
$this->assertEquals('D', $this->_model->alphaMaj('ð'));
}
+
+
+ public function titlesAndAlpha() {
+ return [
+ [ 'Norvege, Islande', 'NORVEGE ISLANDE' ],
+ [ 'L\'ame divisee' , 'AME DIVISEE' ],
+ [ 'Le velo de Jojo', 'VELO DE JOJO' ],
+ [ 'Les tontons flingueurs', 'TONTONS FLINGUEURS' ],
+ [ 'La voiture de la maison', 'VOITURE DE LA MAISON' ],
+ [ 'Un dimanche', 'DIMANCHE' ],
+ [ 'Un voyage a la mer', 'VOYAGE A LA MER' ],
+ [ chr(136) . 'Joseph Canteloube. ' . chr(137) . 'Anthologie des chants populaires franco-canadiens . [A 1 v.]',
+ 'ANTHOLOGIE DES CHANTS POPULAIRES FRANCO-CANADIENS'],
+ ];
+ }
+
+ /**
+ * @dataProvider titlesAndAlpha
+ * @test
+ */
+ public function alphaTitleShouldAnswer($title, $alpha) {
+ $this->assertEquals($alpha,
+ $this->_model->beForIso5426()->codeAlphaTitre($title));
+ }
}
\ No newline at end of file
--
GitLab