thevar1able/ClickHouse
1
0
Fork 0
mirror of https://github.com/ClickHouse/ClickHouse.git synced 2024-12-11 17:02:25 +00:00
Code Issues Packages Projects Releases Wiki Activity

Tests and documentation

Browse Source
This commit is contained in:
Alexey Milovidov 2021-02-28 00:55:56 +03:00
parent 2ac673b12a
commit 0ab4afeeed
3 changed files with 98 additions and 21 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
tests/queries/0_stateless/01674_htm_xml_coarse_parse.reference
View File

@ -2,7 +2,7 @@
Here is CDTATA.
This is a white space test.
This is a complex test. <script type="text/javascript">Hello, world</script> world <style> hello
This is a complex test.<script type="text/javascript">Hello, world</script> world <style> hello
hello, world
hello, world

107
tests/queries/0_stateless/01746_extract_text_from_html.reference
View File

@ -1,53 +1,120 @@
-- { echo }
SELECT extractTextFromHTML('');
SELECT extractTextFromHTML(' ');
SELECT extractTextFromHTML(' ');
SELECT extractTextFromHTML('Hello');
Hello
SELECT extractTextFromHTML('Hello, world');
Hello, world
SELECT extractTextFromHTML('Hello, world');
Hello, world
Hello, world
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
Helloworld
Helloworld
Helloworld
Helloworld
Helloworld
Helloworld
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('Hello<!--->world');
Hello world
SELECT extractTextFromHTML('Hello<!---->world');
Hello world
SELECT extractTextFromHTML('Hello <!-- --> World');
Hello World
SELECT extractTextFromHTML('Hello<!-- --> World');
Hello World
HelloWorld
SELECT extractTextFromHTML('Hello<!-- -->World');
Hello World
SELECT extractTextFromHTML('Hello <!-- -->World');
Hello World
SELECT extractTextFromHTML('Hello <u> World</u>');
Hello World
HelloWorld
SELECT extractTextFromHTML('Hello <u>World</u>');
Hello World
SELECT extractTextFromHTML('Hello<u>World</u>');
Hello World
SELECT extractTextFromHTML('Hello<u> World</u>');
Hello World
SELECT extractTextFromHTML('<![CDATA[ \t Hello,\rworld \n ]]>');
\t Hello,\rworld \n
Hello Hello\tworld world!
HelloHello\tworldworld!
Hello Hello <b>world</b> world!
SELECT extractTextFromHTML('Hello <![CDATA[Hello\tworld]]> world!');
HelloHello\tworld world!
SELECT extractTextFromHTML('Hello<![CDATA[Hello\tworld]]>world!');
HelloHello\tworld world!
SELECT extractTextFromHTML('Hello <![CDATA[Hello <b>world</b>]]> world!');
HelloHello <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>
<sender>John <![CDATA[Smith]]>
SELECT extractTextFromHTML('<![CDATA[<sender>John <![CDATA[]]>Smith</sender>]]>');
<sender>John <![CDATA[ Smith ]]>
SELECT extractTextFromHTML('<![CDATA[<sender>John ]]><![CDATA[Smith</sender>]]>');
<sender>John 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>JohnSmith</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
HelloWorld goodbye
SELECT extractTextFromHTML('Hello<scripta>World</scripta> goodbye');
Hello World 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
HelloWorld goodbye
HelloWorld goodbye
SELECT extractTextFromHTML('Hello<script:p>World</script:p> goodbye');
Hello World goodbye
SELECT extractTextFromHTML('Hello<script:p type="text/javascript">World</script:p> goodbye');
Hello World 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
SELECT extractTextFromHTML('Hello<style type="text/css">World <![CDATA[</style>]]> </style> goodbye');
Hello goodbye
Hello
SELECT extractTextFromHTML('Hello<style type="text/css">World <![CDAT[</style>]]> </style> goodbye');
Hello ]]> goodbye
SELECT extractTextFromHTML('Hello<style type="text/css">World <![endif]--> </style> 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
SELECT extractTextFromHTML('<![CDATA[]]]]><![CDATA[>]]>');
]]>
SELECT extractTextFromHTML('
<img src="pictures/power.png" style="margin-bottom: -30px;" />
<br><span style="padding-right: 10px; font-size: 10px;">xkcd.com</span>
</div>
');
xkcd.com

10
tests/queries/0_stateless/01746_extract_text_from_html.sql
View File

@ -58,5 +58,15 @@ SELECT extractTextFromHTML('Hello<style type="text/css">World <!-- abc --> </sty
SELECT extractTextFromHTML('Hello<style type="text/css">World <![CDATA[</style>]]> </stylea> goodbye');
SELECT extractTextFromHTML('Hello<style type="text/css">World <![CDATA[</style>]]> </style> goodbye');
SELECT extractTextFromHTML('Hello<style type="text/css">World <![CDAT[</style>]]> </style> goodbye');
SELECT extractTextFromHTML('Hello<style type="text/css">World <![endif]--> </style> goodbye');
SELECT extractTextFromHTML('Hello<style type="text/css">World <script>abc</script> </stylea> goodbye');
SELECT extractTextFromHTML('Hello<style type="text/css">World <script>abc</script> </style> goodbye');
SELECT extractTextFromHTML('<![CDATA[]]]]><![CDATA[>]]>');
SELECT extractTextFromHTML('
<img src="pictures/power.png" style="margin-bottom: -30px;" />
<br><span style="padding-right: 10px; font-size: 10px;">xkcd.com</span>
</div>
');