mysql select query ignoring inner spaces - mysql

Question

Banging me head against the wall with this one.
I have table containing postcodes and street names and I have another table where Houses are listed for sale ( where the Street name is missing) and I am tryin to get the Street name for each post code.
The problem is that table 1 stores the postcode without the space and table 2 which I am trying to update stores the post code with the space.
So in table 1 the postcode is stored as "l249pb" and table 2 it is stored as "l24 9pb".
Now if the post codes where both stored in exactly the same format i.e without the space I would expect this query to work:
UPDATE Table1
INNER JOIN Table2 ON ( Table1.PostCode = Table2.PostCode )
SET Table1.StreetName = Table2.StreetName
I have tried this but it wont work :
UPDATE Table1
INNER JOIN Table2 ON ( Table1.PostCode = REPLACE(Table2.PostCode,' ',''))
SET Table1.StreetName = Table2.StreetName
can anyone tell me how to check for a match ignoring spaces ( like a trim but removing every space )
Many thanks for any help you can offer.

Answer 1

With the data you've given your UPDATE runs just fine. Probably the whitespaces you see are not actually spaces, but something else, e.g. non-breaking spaces, tabs etc.
After normal SPACE, the next most common white spaces (which are not line breaks) are CHARACTER TABULATION (ie. horizontal tab) and NO-BREAK SPACE. You could use CHAR(9) and CHAR(160), respectively, to reference them in your query.
It also might be possible that your table viewer application shows line breaks as a space for brevity, so if replacing space, tab and nbsp isn't enough, try replacing those, too.
If you really need to replace all white space characters… Unfortunately there is no "white space wildcard" to use in MySQL. Technically, you could make a monster REPLACE(REPLACE(REPLACE(REPLACE…-call, which, in the end, would replace all whitespace characters with ''. For example, to replace every THREE-PER-EM SPACE, first look for its Unicode code point (U+2004), then you can replace its occurences e.g. with:
REPLACE(PostCode, CHAR(0x2004 using ucs2), '')
There is a hackish shortcut to this: if you are sure that your data should contain only Latin-1 characters and no ? (question mark), you could CONVERT() the string first as latin1, which replaces all characters with overflowing code as ?and then replace all ? as '':
REPLACE(CONVERT(PostCode using latin1), '?', '')
This can be useful in one-off, manual queries, but for continuing use, better replace the characters explicitly.
But first you should check your data input sanitizer/validator, so future records won't be such a mess. Perhaps you could consider running a bulk replace to normalize the data on PostCode column(s), if possible, before even trying to do your join query. Legacy systems with legacy data only get worse over time.