SQL解析NVARCHAR字段

Question

I am loading data from Excels into database on SQL Server 2008. There is one column which is in nvarchar data type. This field contains the data as

Text text text text text    text text text text text.
(ABC-2010-4091, ABC-2011-0586,     ABC-2011-0587, ABC-2011-0604)     
Text text text text text    text text text text text. 
(ABC-2011-0562,     ABC-2011-0570, ABC-2011-0575, ABC-2011-0588)    

so its text with many sentences of this kind.

For each row I need to get the data ABC-####-#### , respectivelly I only need the last part. So eg for ABC-2010-4091 I need to obtain 4091 . This number I will need to join to other table. I guess it would be enough to get the last parts of the format ABC-####-#### , then I should be able to handle the request.

So the example of given above, the result should be 4091, 0586, 0587, 0604, 0562, 0570, 0575, 0588 in the row instead of the whole nvarchar value field.

Is this possible somehow? The text in the nvarchar field differ, but the text format (ABC-####-####) I want to work with is still the same. Only the count of characters for the last part may vary so its not only 4 numbers, but could be 5 or more.

What is the best approach to get these data? Should I parse it in SSIS or on the SQL server side with SQL Query? And how?

I am aware this is though task. I appreciate every help or advice how to deal with this. I have not tried anything yet as I do not know where to start. I read articles about SQL parsing, but I want to ask for best approach to deal with this task.

Answer 1

Stackoverflow is about programming.

Sit down and start programming.

Ok, seriously. That is string parsing and the last part in brackets with multiple fields means no bulk import, it is not a standard CSV file.

Either you use SSIS in SQL Server and program the parsing there or.... you write a program for that.

String maniupation in SQL is the worst part of the language and I would avoid it.

So, yes, sit down and program a routine. Probable the fastest way.

Answer 2

If I understand correctly, "ABS-####-####" will be the value coming through in the column and the numeric part is variable in length.

If that is the case, maybe this will work.

Use a "Derived Column" transformation. Lets say we call "ABC-####-####" = Column1

SUBSTRING("Column1",(FINDSTRING("Column1","-",2)+1),LEN(Column1)-(FINDSTRING("Column1","-",2)))

If I am not mistaken, that should give you the last # values in a new column no matter how long that value is.

HTH

Answer 3

I have worked this problem out with the following guides:

Split Multi Value Column into Multiple Records
&
Remove Multiple Spaces with Only One Space