A bracket range sounds simple until the collation decides what sits between a and f. Test it once so a bad code never reaches your table.
Finding Apostrophes in SQL Server Text: Straight and Curly
Finding apostrophes in SQL Server text takes one trick and one trap. The trick is doubling the quote inside a LIKE pattern. The trap is that text can hold a curly apostrophe, and the usual search skips it. This post tests both on a table of eight shop names.
Special Characters in a LIKE Pattern: Which Ones Are Wildcards
Special characters in a LIKE pattern can match the wrong rows, because three symbols mean more than themselves. This post tests every common symbol on one small table. It shows how to search for a percent sign, an underscore and a bracket, and why a space in the pattern counts.
![Four LIKE wildcards with a real example each: percent matches Baker and Brooks for B percent, underscore finds Baker, Patel and Taylor for underscore a percent, [A-K] returns five names, and [^A-K] returns nine](https://blog.sqlauthority.com/wp-content/uploads/2026/10/wildcard-basics-in-sql-server-how-like-patterns-work-cover.webp)
Wildcard Basics in SQL Server: How LIKE Patterns Work
Wildcard basics for SQL Server: how LIKE works with %, _, sets and ranges, shown on one small table of last names. It also covers the mistakes with equal signs and sets, case and accents in ranges, and matching a wildcard literally. Every result comes from SQL Server 2025.
REGEXP_REPLACE Text Cleanup: Replacing Tags Is Not Parsing HTML
REGEXP_REPLACE can remove controlled tag-like text, but that is not the same as parsing HTML. Test quoted brackets, surviving script text and word boundaries before deciding whether the replacement fits the input.
Rows That Look Identical but Are Not: Finding Hidden Characters
Expose hidden characters with DATALENGTH and UNICODE, clean known whitespace safely, and add a constraint to prevent recurrence.
FOR XML PATH: Decode Escaped Text Through TYPE
FOR XML PATH escapes reserved characters. Use TYPE and scalar extraction to recover text instead of manually replacing entity spellings.