1<?php
2
3class init_clean_id_test extends DokuWikiTest
4{
5
6    /** @inheritDoc */
7    function teardown() : void
8    {
9        global $cache_cleanid;
10        $cache_cleanid = array();
11    }
12
13    /**
14     * DataProvider
15     *
16     * @return Generator|array
17     * @see testCleanId
18     */
19    public function provideTestData()
20    {
21        // input, expected, optional options array
22        $tests = [
23            ['page', 'page'],
24            ['pa_ge', 'pa_ge'],
25            ['pa%ge', 'pa_ge'],
26            ['pa#ge', 'pa_ge'],
27            ['pàge', 'page'],
28            ['pagĖ', 'page'],
29            ['pa$%^*#ge', 'pa_ge'],
30            ['*page*', 'page'],
31            ['ښ', 'ښ'],
32            ['päge', 'paege'],
33            ['foo bar', 'foo_bar'],
34            ['PÄGÖ', 'paegoe'],
35            ['Faß', 'fass'],
36            ['ښ侧化并곦  β', 'ښ侧化并곦_β'],
37            ['page:page', 'page:page'],
38            ['page;page', 'page:page'],
39            ['page:page 1.2', 'page:page_1.2'],
40            ['page._#!', 'page'],
41            ['._#!page', 'page'],
42            ['page._#!page', 'page._page'],
43            ['ns._#!:page', 'ns:page'],
44            ['ns:._#!page', 'ns:page'],
45            ['ns._#!ns:page', 'ns._ns:page'],
46            ['ns_:page', 'ns:page'],
47            ['page...page', 'page...page'],
48            ['page---page', 'page---page'],
49            ['page___page', 'page_page'],
50            ['page_-.page', 'page_-.page'],
51            [':page', 'page'],
52            [':ns:page', 'ns:page'],
53            ['page:', 'page'],
54            ['ns:page:', 'ns:page'],
55            ['Łł', 'll'],
56
57            // use-slash handling
58            ['page/page', 'page_page', ['useslash' => 0]],
59            ['page/page', 'page:page', ['useslash' => 1]],
60
61            // different sep-char
62            ['pa-ge', 'pa-ge', ['sepchar' => '-']],
63            ['pa%ge', 'pa-ge', ['sepchar' => '-']],
64
65            // no deaccenting
66            ['pàge', 'pàge', ['deaccent' => 0]],
67            ['pagĖ', 'pagė', ['deaccent' => 0]],
68            ['pagĒēĔĕĖėĘęĚě', 'pagēēĕĕėėęęěě', ['deaccent' => 0]],
69            ['ښ', 'ښ', ['deaccent' => 0]],
70            ['ښ侧化并곦ঝഈ', 'ښ侧化并곦ঝഈ', ['deaccent' => 0]],
71
72            // romanize
73            ['pàge', 'page', ['deaccent' => 2]],
74            ['pagĖ', 'page', ['deaccent' => 2]],
75            ['pagĒēĔĕĖėĘęĚě', 'pageeeeeeeeee', ['deaccent' => 2]],
76            ['ښ', 'ښ', ['deaccent' => 2]],
77            ['ښ侧化并곦ঝഈ', 'ښ侧化并곦ঝഈ', ['deaccent' => 2]],
78
79            // deaccent and force ascii
80            ['pàge', 'page', ['deaccent' => 1, 'ascii' => true]],
81            ['pagĖ', 'page', ['deaccent' => 1, 'ascii' => true]],
82            ['pagĒēĔĕĖėĘęĚě', 'pageeeeeeeeee', ['deaccent' => 1, 'ascii' => true]],
83            ['ښ', '', ['deaccent' => 1, 'ascii' => true]],
84            ['ښ侧化并곦ঝഈ', '', ['deaccent' => 1, 'ascii' => true]],
85
86            // romanize and force ascii
87            ['pàge', 'page', ['deaccent' => 2, 'ascii' => true]],
88            ['pagĖ', 'page', ['deaccent' => 2, 'ascii' => true]],
89            ['pagĒēĔĕĖėĘęĚě', 'pageeeeeeeeee', ['deaccent' => 2, 'ascii' => true]],
90            ['ښ', '', ['deaccent' => 2, 'ascii' => true]],
91            ['ښ侧化并곦ঝഈ', '', ['deaccent' => 2, 'ascii' => true]],
92        ];
93
94        foreach ($tests as $test) {
95            // defaults
96            $sepchar = isset($test[2]['sepchar']) ? $test[2]['sepchar'] :  '_';
97            $deaccent = isset($test[2]['deaccent']) ? $test[2]['deaccent'] : 1;
98            $ascii = isset($test[2]['ascii']) ? $test[2]['ascii'] : false;
99
100            // unless set, test both useslash settings
101            if (isset($test[2]['useslash'])) {
102                yield([$test[0], $test[1], $ascii, $sepchar, $deaccent, $test[2]['useslash']]);
103            } else {
104                yield([$test[0], $test[1], $ascii, $sepchar, $deaccent, 0]);
105                yield([$test[0], $test[1], $ascii, $sepchar, $deaccent, 1]);
106            }
107        }
108    }
109
110    /**
111     * @dataProvider provideTestData
112     * @param string $input
113     * @param string $expected
114     * @param bool $ascii
115     * @param string $sepchar
116     * @param int $deaccent
117     * @param int $useslash
118     */
119    function testCleanId($input, $expected, $ascii, $sepchar, $deaccent, $useslash)
120    {
121        // set dokuwiki defaults
122        global $conf;
123        $conf['sepchar'] = $sepchar;
124        $conf['deaccent'] = $deaccent;
125        $conf['useslash'] = $useslash;
126
127        $result = cleanID($input, $ascii);
128        $this->assertEquals($expected, $result);
129    }
130
131
132
133    function test_caching_ascii()
134    {
135        global $conf;
136        $conf['deaccent'] = 0;
137        $this->assertEquals('pàge', cleanID('pàge', false));
138        $this->assertEquals('page', cleanID('pàge', true));
139
140        $this->assertEquals('page', cleanID('pagĖ', true));
141        $this->assertEquals('pagė', cleanID('pagĖ', false));
142    }
143
144}
145