There is no reasonable way to save a document as corrupted as what you posted, but assuming you replace the>and similar characters in the text with their related entities, e.g.:> ;, you can put the document you want to accept into an appropriate library, such asDomDocumentwhich will handle the rest.

$input = <<<_E_ < div class='test' >1 > 0 is < b >true and apples >>> bananas< / div > _E_; $input = preg_replace([ '#<\s+#', '#loadHTML($input, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD); var_dump($d->saveHTML());

Output:

string(80) "
1 > 0 is true and apples >>> bananas
"

This regular expression is also valid:

It divides the valid part in the HTML tag into four parts and replaces the remaining parts (spaces) with them.

Regex101 Demo

/(]*\S)\s*(>)/g

• ( - Capture the opening angle bracket (section 1)
• \s*- matches any whitespace
• (\/?)- Capturing optional backslashes (Part 2)
• \s*- matches any space after a backslash
• ([^]*\S)- captures content within tags without trailing spaces (section 3)
• \s*- Matches spaces after the content and before the closing angle bracket
• (>)- Capture right angle bracket (section 4)

const reg = /(<)\s*(\/?)\s*([^<>]*\S)\s*(>)/g const str = "< div class='test' >1 > 0 is < b >true< / b > and apples >>> bananas< / div >" const newStr = str.replace(reg, ""); console.log(newStr);