diff options
| author | ArthurHoaro <arthur@hoa.ro> | 2020-10-13 12:26:55 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2020-10-13 12:26:55 +0200 |
| commit | 458b6b9918ec27154dd45416947bb93bedb97109 (patch) | |
| tree | c1c565def0a4fffac5d0556794451c49fc4d52e4 /tests/bookmark | |
| parent | 543b16b4f4bbde4e9857490e2175e44b4d941eb3 (diff) | |
| parent | 2cd0509b503332b1989f06da45d569d4d2929be5 (diff) | |
| download | Shaarli-458b6b9918ec27154dd45416947bb93bedb97109.tar.gz Shaarli-458b6b9918ec27154dd45416947bb93bedb97109.tar.zst Shaarli-458b6b9918ec27154dd45416947bb93bedb97109.zip | |
Merge pull request #1540 from ArthurHoaro/fix/metadata-regexes
Improve regex to extract HTML metadata (title, description, etc.)
Diffstat (limited to 'tests/bookmark')
| -rw-r--r-- | tests/bookmark/LinkUtilsTest.php | 89 |
1 files changed, 89 insertions, 0 deletions
diff --git a/tests/bookmark/LinkUtilsTest.php b/tests/bookmark/LinkUtilsTest.php index ef00b92f..29941c8c 100644 --- a/tests/bookmark/LinkUtilsTest.php +++ b/tests/bookmark/LinkUtilsTest.php | |||
| @@ -94,8 +94,78 @@ class LinkUtilsTest extends TestCase | |||
| 94 | public function testHtmlExtractExistentNameTag() | 94 | public function testHtmlExtractExistentNameTag() |
| 95 | { | 95 | { |
| 96 | $description = 'Bob and Alice share cookies.'; | 96 | $description = 'Bob and Alice share cookies.'; |
| 97 | |||
| 98 | // Simple one line | ||
| 97 | $html = '<html><meta>stuff2</meta><meta name="description" content="' . $description . '"/></html>'; | 99 | $html = '<html><meta>stuff2</meta><meta name="description" content="' . $description . '"/></html>'; |
| 98 | $this->assertEquals($description, html_extract_tag('description', $html)); | 100 | $this->assertEquals($description, html_extract_tag('description', $html)); |
| 101 | |||
| 102 | // Simple OpenGraph | ||
| 103 | $html = '<meta property="og:description" content="' . $description . '">'; | ||
| 104 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 105 | |||
| 106 | // Simple reversed OpenGraph | ||
| 107 | $html = '<meta content="' . $description . '" property="og:description">'; | ||
| 108 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 109 | |||
| 110 | // ItemProp OpenGraph | ||
| 111 | $html = '<meta itemprop="og:description" content="' . $description . '">'; | ||
| 112 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 113 | |||
| 114 | // OpenGraph without quotes | ||
| 115 | $html = '<meta property=og:description content="' . $description . '">'; | ||
| 116 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 117 | |||
| 118 | // OpenGraph reversed without quotes | ||
| 119 | $html = '<meta content="' . $description . '" property=og:description>'; | ||
| 120 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 121 | |||
| 122 | // OpenGraph with noise | ||
| 123 | $html = '<meta tag1="content1" property="og:description" tag2="content2" content="' . | ||
| 124 | $description . '" tag3="content3">'; | ||
| 125 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 126 | |||
| 127 | // OpenGraph reversed with noise | ||
| 128 | $html = '<meta tag1="content1" content="' . $description . '" ' . | ||
| 129 | 'tag3="content3" tag2="content2" property="og:description">'; | ||
| 130 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 131 | |||
| 132 | // OpenGraph multiple properties start | ||
| 133 | $html = '<meta property="unrelated og:description" content="' . $description . '">'; | ||
| 134 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 135 | |||
| 136 | // OpenGraph multiple properties end | ||
| 137 | $html = '<meta property="og:description unrelated" content="' . $description . '">'; | ||
| 138 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 139 | |||
| 140 | // OpenGraph multiple properties both end | ||
| 141 | $html = '<meta property="og:unrelated1 og:description og:unrelated2" content="' . $description . '">'; | ||
| 142 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 143 | |||
| 144 | // OpenGraph multiple properties both end with noise | ||
| 145 | $html = '<meta tag1="content1" property="og:unrelated1 og:description og:unrelated2" '. | ||
| 146 | 'tag2="content2" content="' . $description . '" tag3="content3">'; | ||
| 147 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 148 | |||
| 149 | // OpenGraph reversed multiple properties start | ||
| 150 | $html = '<meta content="' . $description . '" property="unrelated og:description">'; | ||
| 151 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 152 | |||
| 153 | // OpenGraph reversed multiple properties end | ||
| 154 | $html = '<meta content="' . $description . '" property="og:description unrelated">'; | ||
| 155 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 156 | |||
| 157 | // OpenGraph reversed multiple properties both end | ||
| 158 | $html = '<meta content="' . $description . '" property="og:unrelated1 og:description og:unrelated2">'; | ||
| 159 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 160 | |||
| 161 | // OpenGraph reversed multiple properties both end with noise | ||
| 162 | $html = '<meta tag1="content1" content="' . $description . '" tag2="content2" '. | ||
| 163 | 'property="og:unrelated1 og:description og:unrelated2" tag3="content3">'; | ||
| 164 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 165 | |||
| 166 | // Suggestion from #1375 | ||
| 167 | $html = '<meta property="og:description" name="description" content="' . $description . '">'; | ||
| 168 | $this->assertEquals($description, html_extract_tag('description', $html)); | ||
| 99 | } | 169 | } |
| 100 | 170 | ||
| 101 | /** | 171 | /** |
| @@ -105,6 +175,25 @@ class LinkUtilsTest extends TestCase | |||
| 105 | { | 175 | { |
| 106 | $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>'; | 176 | $html = '<html><meta>stuff2</meta><meta name="image" content="img"/></html>'; |
| 107 | $this->assertFalse(html_extract_tag('description', $html)); | 177 | $this->assertFalse(html_extract_tag('description', $html)); |
| 178 | |||
| 179 | // Partial meta tag | ||
| 180 | $html = '<meta content="Brief description">'; | ||
| 181 | $this->assertFalse(html_extract_tag('description', $html)); | ||
| 182 | |||
| 183 | $html = '<meta property="og:description">'; | ||
| 184 | $this->assertFalse(html_extract_tag('description', $html)); | ||
| 185 | |||
| 186 | $html = '<meta tag1="content1" property="og:description">'; | ||
| 187 | $this->assertFalse(html_extract_tag('description', $html)); | ||
| 188 | |||
| 189 | $html = '<meta property="og:description" tag1="content1">'; | ||
| 190 | $this->assertFalse(html_extract_tag('description', $html)); | ||
| 191 | |||
| 192 | $html = '<meta tag1="content1" content="Brief description">'; | ||
| 193 | $this->assertFalse(html_extract_tag('description', $html)); | ||
| 194 | |||
| 195 | $html = '<meta content="Brief description" tag1="content1">'; | ||
| 196 | $this->assertFalse(html_extract_tag('description', $html)); | ||
| 108 | } | 197 | } |
| 109 | 198 | ||
| 110 | /** | 199 | /** |