Question

OK So i am confused (obviously)

I’m trying to return rows (from Oracle) where a text field contains a complete word, not just the substring.

a simple example is the word ‘I’.

Show me all rows where the string contains the word ‘I’, but not simply where ‘I’ is a substring somewhere as in '%I%'

so I wrote what i thought would be a simple regex:

select REGEXP_INSTR(upper(description), '\bI\b') from mytab;

expecting that I should be detected with word boundaries. I get no results (or rather the result 0 for each row.

what i expect:

• ‘I am the Administrator’ -> 1
• ‘I’m the administrator’ -> 0
• ‘Am I the administrator’ -> 1
• ‘It is the infamous administrator’ -> 0
• ‘The adminisrtrator, tis I’ -> 1

isn’t the /b supposed to find the contained string by word boundary?

tia

Answer 1

I believe that \b is not supported by your flavor of regex :

http://download.oracle.com/docs/cd/B19306_01/appdev.102/b14251/adfns_regexp.htm#i1007670

Therefore you could do something like :

(^|\s)word(\s|$)

To at least ensure that your “word” is separated by some whitespace or it’s the whole string.