ClickHouse/tests/queries/0_stateless/01746_extract_text_from_html.reference

107 lines
4.0 KiB
Plaintext
Raw Normal View History

2021-02-27 19:22:38 +00:00
-- { echo }
SELECT extractTextFromHTML('');
SELECT extractTextFromHTML(' ');
SELECT extractTextFromHTML(' ');
SELECT extractTextFromHTML('Hello');
Hello
SELECT extractTextFromHTML('Hello, world');
Hello, world
SELECT extractTextFromHTML('Hello, world');
Hello, world
SELECT extractTextFromHTML(' Hello, world');
Hello, world
SELECT extractTextFromHTML(' Hello, world ');
Hello, world
SELECT extractTextFromHTML(' \t Hello,\rworld \n ');
Hello, world
SELECT extractTextFromHTML('Hello<world');
Hello
SELECT extractTextFromHTML('Hello < world');
Hello
SELECT extractTextFromHTML('Hello > world');
Hello > world
SELECT extractTextFromHTML('Hello<world>');
Hello
SELECT extractTextFromHTML('Hello<>world');
Helloworld
SELECT extractTextFromHTML('Hello<!>world');
Helloworld
SELECT extractTextFromHTML('Hello<!->world');
Helloworld
SELECT extractTextFromHTML('Hello<!-->world');
Helloworld
SELECT extractTextFromHTML('Hello<!--->world');
Helloworld
SELECT extractTextFromHTML('Hello<!---->world');
Helloworld
SELECT extractTextFromHTML('Hello <!-- --> World');
Hello World
SELECT extractTextFromHTML('Hello<!-- --> World');
Hello World
SELECT extractTextFromHTML('Hello<!-- -->World');
HelloWorld
SELECT extractTextFromHTML('Hello <!-- -->World');
Hello World
SELECT extractTextFromHTML('Hello <u> World</u>');
Hello World
SELECT extractTextFromHTML('Hello <u>World</u>');
Hello World
SELECT extractTextFromHTML('Hello<u>World</u>');
HelloWorld
SELECT extractTextFromHTML('Hello<u> World</u>');
Hello World
SELECT extractTextFromHTML('<![CDATA[ \t Hello,\rworld \n ]]>');
\t Hello,\rworld \n
SELECT extractTextFromHTML('Hello <![CDATA[Hello\tworld]]> world!');
Hello Hello\tworld world!
SELECT extractTextFromHTML('Hello<![CDATA[Hello\tworld]]>world!');
HelloHello\tworldworld!
SELECT extractTextFromHTML('Hello <![CDATA[Hello <b>world</b>]]> world!');
Hello Hello <b>world</b> world!
SELECT extractTextFromHTML('<![CDATA[<sender>John Smith</sender>]]>');
<sender>John Smith</sender>
SELECT extractTextFromHTML('<![CDATA[<sender>John <![CDATA[Smith</sender>]]>');
<sender>John <![CDATA[Smith</sender>
SELECT extractTextFromHTML('<![CDATA[<sender>John <![CDATA[]]>Smith</sender>]]>');
<sender>John <![CDATA[Smith]]>
SELECT extractTextFromHTML('<![CDATA[<sender>John ]]><![CDATA[Smith</sender>]]>');
<sender>John Smith</sender>
SELECT extractTextFromHTML('<![CDATA[<sender>John ]]> <![CDATA[Smith</sender>]]>');
<sender>John Smith</sender>
SELECT extractTextFromHTML('<![CDATA[<sender>John]]> <![CDATA[Smith</sender>]]>');
<sender>John Smith</sender>
SELECT extractTextFromHTML('<![CDATA[<sender>John ]]>]]><![CDATA[Smith</sender>]]>');
<sender>John ]]>Smith</sender>
SELECT extractTextFromHTML('Hello<script>World</script> goodbye');
Hello goodbye
SELECT extractTextFromHTML('Hello<script >World</script> goodbye');
Hello goodbye
SELECT extractTextFromHTML('Hello<scripta>World</scripta> goodbye');
HelloWorld goodbye
SELECT extractTextFromHTML('Hello<script type="text/javascript">World</script> goodbye');
Hello goodbye
SELECT extractTextFromHTML('Hello<style type="text/css">World</style> goodbye');
Hello goodbye
SELECT extractTextFromHTML('Hello<script:p>World</script:p> goodbye');
HelloWorld goodbye
SELECT extractTextFromHTML('Hello<script:p type="text/javascript">World</script:p> goodbye');
HelloWorld goodbye
SELECT extractTextFromHTML('Hello<style type="text/css">World <!-- abc --> </style> goodbye');
Hello goodbye
SELECT extractTextFromHTML('Hello<style type="text/css">World <!-- abc --> </style \n > goodbye');
Hello goodbye
SELECT extractTextFromHTML('Hello<style type="text/css">World <!-- abc --> </ style> goodbye');
Hello
SELECT extractTextFromHTML('Hello<style type="text/css">World <!-- abc --> </stylea> goodbye');
Hello
SELECT extractTextFromHTML('Hello<style type="text/css">World <![CDATA[</style>]]> </stylea> goodbye');
Hello]]> goodbye
SELECT extractTextFromHTML('Hello<style type="text/css">World <script>abc</script> </stylea> goodbye');
Hello
SELECT extractTextFromHTML('Hello<style type="text/css">World <script>abc</script> </style> goodbye');
Hello goodbye