Problem About Caculating Wide-char Strings' Character Count

it's like this, i have a temporary table such as
create table temp_table (str varchar(50))
and i have a data table
create table data_table (str varchar(20))

now i import my data(in which there is some corrupted lines) into the temporary table, they should be all ansi-character strings and no more than 20 characters, but now some wrong datas in which there are wide-characters are mixed in. as the result of these wide-characters, the corrupted strings each takes over 20 bytes, but i can't filter them out, as when i enter in "len(str)", the sql server returns character counts, instead of byte counts, i thought this should only happen when i was using a unicode date type!(e.g. nvarchar). but now the server also behaves like this on those ansi date types. it seems all string manipulating functions refering string length behaves like this.

so when i am trying to run:
insert into data_table select str from temp_table where len(str) <= 20
or
insert into data_table select left(str,20) from temp_table

it will always end up with a string truncating error
String or binary data would be truncated. The statement has been terminated

So now my problem is how to get the count of byte, but character, of a string containing wide-characters?


i'm using sql server 2005 standard version with sp2