Split string into multiple rows using SQL in SAP HANA
In this post I would like to describe options for splitting concatenated string into multiple rows using SQL in SAP HANA. Depending on which version of HANA system you are working you can choose the one which is valid.
Creating custom function (when SAP HANA system version is 2.0 SP02 or lower)
If your HANA system is of version 2.0 SP2 or older, then there is no in-built function for splitting strings, that’s why you will need to create custom function.
You have the comma separated string ‘A1,A2,A3,A4,A5,A6’ which you want to display as multiple rows. Below there is a code for table function splitting comma separated string into multiple lines.
FUNCTION "_SYS_BIC"."TMP::TF_SPLIT_STRING" ( INPUT_STRING VARCHAR(5000) ) RETURNS TABLE ( "OUTPUT_SPLIT" VARCHAR(5000) ) LANGUAGE SQLSCRIPT SQL SECURITY INVOKER AS BEGIN DECLARE COUNTER INT := 1; SPLIT_VALUES = SELECT SUBSTR_BEFORE(:INPUT_STRING,',') SINGLE_VAL FROM DUMMY; SELECT SUBSTR_AFTER(:INPUT_STRING,',') || ',' INTO INPUT_STRING FROM DUMMY; WHILE( LENGTH(:INPUT_STRING) > 0 ) DO SPLIT_VALUES = SELECT SUBSTR_BEFORE(:INPUT_STRING,',') SINGLE_VAL FROM DUMMY UNION SELECT SINGLE_VAL FROM :SPLIT_VALUES; SELECT SUBSTR_AFTER(:INPUT_STRING,',') INTO INPUT_STRING FROM DUMMY; END WHILE; RETURN SELECT SINGLE_VAL AS "OUTPUT_SPLIT" FROM :SPLIT_VALUES; END
*If your string is separated by different delimiter simply replace
',' with other symbol.
Now let’s query the function for the string which we need to split:
DO BEGIN DECLARE TEST_STRING VARCHAR(50) := 'A,B,C,D,E,F,G,H'; SELECT "OUTPUT_SPLIT" FROM "_SYS_BIC"."TMP::TF_SPLIT_STRING" (:TEST_STRING); END
Using SQLSCRIPT built-in Library (recommended when SAP HANA version is 2.0 SP03 or higher)
Version 2.0 SP03 of SAP HANA offers new enhancement of SQLScript library – SQLSCRIPT_STRING. Functions within this library enables easy way of splitting string into multiple rows. Additionally developer has more flexibility to define splitting rules.
You have the comma separated string
which you want to display as multiple rows.
Use following code:
I. Explicitly declare usage of SQLSCRIPT_STRING library. Assign alias for the library (in this example name of the alias is “LIBRARY”, but you can assign any other name)
II. Explicitly declare output table
III. Use SPLIT_TO_TABLE function. Combine it with the alias assigned in point I. As input for SPLIT_TO_TABLE function use string which you want to split, and after comma define delimiter (in this example comma is delimiter). At the end you need to assign output of that function to the table variable defined in point II.
IV. Query the table variable
After running following the query string will be splitted into multiple rows:
DO BEGIN USING SQLSCRIPT_STRING AS LIBRARY; DECLARE TEST_OUTPUT TABLE(RESULT NVARCHAR(5000)); DECLARE TEST_STRING VARCHAR(50) := 'A,B,C,D,E,F,G,H'; TEST_OUTPUT = LIBRARY:SPLIT_TO_TABLE(:TEST_STRING,','); SELECT * FROM :TEST_OUTPUT; END
There is a string which combines first and last name, phone and the address:
'Sarah Blake, 98 921 29 30, 270 Sycamore Street Brookfield, WI 53045, US'
You want to split this string into three values: full name, phone, address.
Use following code:
DO BEGIN USING SQLSCRIPT_STRING AS LIBRARY; DECLARE TEST_OUTPUT TABLE(RESULT NVARCHAR(5000)); DECLARE TEST_STRING VARCHAR(100) := 'Sarah Blake, 98 921 29 30, 270 Sycamore Street Brookfield, WI 53045, US'; TEST_OUTPUT = LIBRARY:SPLIT_TO_TABLE(:TEST_STRING,',',2); SELECT * FROM :TEST_OUTPUT; END
Here I used third optional parameter (MAXSPLIT) which is available for SPLIT_TO_TABLE and assigned value of 2. This way I will limit the split to only first two comma occurrences (subsequent commas will not be considered).
Besides defining delimiter for split, you can also specify maximum number of splits. Function takes first n-number of delimiter occurences, and remaining part is being displayed in the last row. As you can see in the string from the example there are 4 commas, so if this parameter would not be specified, then in output there would be 5 rows.
There is a string which combines number, date and time. It’s separated by #, DATE, TIME string:
You want to split this string into three strings: number, date, time.
Use following code:
DO BEGIN USING SQLSCRIPT_STRING AS LIBRARY; DECLARE TEST_OUTPUT TABLE(RESULT NVARCHAR(5000)); DECLARE TEST_STRING VARCHAR(50) := '10000000123#20190101DATE115737TIME'; TEST_OUTPUT = LIBRARY:SPLIT_REGEXPR_TO_TABLE(:TEST_STRING, '\#|[A-Z]*[A-Z]') ; SELECT * FROM :TEST_OUTPUT WHERE "RESULT" != ''; END
Here I used function SPLIT_REGEXPR_TO_TABLE. It can be combined with regular expressions, which gives user even more options for defining logic for splitting strings. Regular expression from the example splits string after each occurencs of # symbol or uppercase alphabet string.
SAP HANA SPS 02 introduced built-in libraries giving SQL developers new functions. In SPS 03 new librarary SQLSCRIPT_STRING has been added which contains multiple functions for manipulating with strings. If you want to find out more about SQLSCRIPT_STRING library, check SQLScript reference by SAP.
If your system is running on older version of SAP HANA, as workaround you can develop custom function as described in the post.
In my next post you can check how to Split table column value into multiple rows using SQL in SAP HANA
Good explanation how to split a string. My situation though is I have a table with a column, whose values are a comma separated list. How to leverage this SPLIT function on a table column and generate a separate row per split value?
Header_1 | Header_2
A | 1,2,3
B | 4,5
Header_1 | Header_2
A | 1
A | 2
A | 3
B | 4
B | 5
Sorry for late reply. I've just published a new post which describes your scenario (link added at the end of this post).
First I would like to thank you for sharing the possibilities of "SQLSCRIPT_STRING" library .
I tried to use it for one of my use cases , but I came across insufficient privileges error . I tried to find the authorizations required for Using libraries in procedure , but could not find documentation related to that . Could you help me with authorizations required to use libraries in procedures.
What is your SAP HANA system version? Is it 2.0 SP03 or higher ?
HANA System version is 2.0 SP03
Do you have "PUBLIC" role assigned to your user? I think this should be sufficient to run the library functions (they persist in SYS schema).
Firstly, thanks a lot for showing us how to use SQLSCRIPT_STRING, it was a very helpful document.
I have a requirement where I have an input string as shown below, where data is been consumed with column and its respective data using Tilda and Pipe delimiters. Can you please guide how split the data using the new functionality.
~COLUMN_1~2F-8S~COLUMN_2~SEC- WATER ~COLUMN_3~2SF-S-M-10~COLUMN_4~10~COLUMN_5~55.91~COLUMN_6~20127.60~COLUMN_7~ TRUE ~|
~COLUMN_1~2F-8S~COLUMN_2~SEC- WATER ~COLUMN_3~2SF-S-P-10~COLUMN_4~10~COLUMN_5~10.84~COLUMN_6~3902.40~COLUMN_7~ TRUE ~|
Need to show the data as below
Note: I am using 2.0 SPS 3
You can use SPLIT_REGEXPR function. Here is an example (it includes first two strings which you provided):
In future if you have specific question, raise it through SAP community rather than in comments 🙂
Very useful and detailed instructions
Please consider that your TF from Scenario 1 doesnt work it gets only one value without split-string.
Therefore I had to adjust it:
I am trying to pass multiple values from a Calculation View for each Input parameter to a Table Function. This table function internally calls a Calculation view which has these multiple input parameters mapped. Since Table function input parameters cannot take multiple values, as per the post above, since I am in hana 2 sp03, I thought of using SQLSCRIPT_STRING inside the table function to split the input parameter values.
When I add the below statement after the BEGIN in the Table function and when I activate it, I get the following error.
USING SQLSCRIPT_STRING AS LIBRARY;
Repository: Encountered an error in repository runtime extension;Model inconsistency. Using library is not supported in current SQLScript configuration
Can I not use this inside a Table function?
When I give it in the SQL console, it works.
USING SQLSCRIPT_STRING AS LIBRARY;
DECLARE TEST_OUTPUT TABLE(RESULT NVARCHAR(5000));
DECLARE TEST_STRING VARCHAR(50) := 'N100,N102';
TEST_OUTPUT = LIBRARY:SPLIT_TO_TABLE(:TEST_STRING,',');
SELECT * FROM :TEST_OUTPUT;
Hi Lalitha, did you resolve this issue?
I like the way you explained the function, however the solution itself is unfortunately not good because the performance is very poor.
I did some tests and below you can see comparison of your solution (SPLIT_STRING_01) with another solution. It shows, that the solution SPLIT_STRING_03 is twice faster:
Thanks for that analysis!
Anyway I think that Library split function is definitely the best.
I try to do a "Where-used" on BW4 and hence want to read Scripted HANA Views from "_SYS_REPO"."ACTIVE_OBJECT" column CDATA.
As this is a LOB column it may contain (and often does) > 5000 Characters.
When using SQLSCRIPT_STRING, I do get an error.
Is there a good way to split the LOB string independent of length?
Thanks and regards,
I attached the usage of built-in SQLSCRIPT_STRING library in AMDP.
Off course the table function created here is useless, but allows to show case the invocation in AMDP.
SELECT RESULT AS MEMBER FROM SQLSCRIPT_STRING:SPLIT_TO_TABLE(String, ',', -1)
Data Preview on CDS Table Function
CDS Table Function