Extract Data from a large string to table using SQL - sql

Question

I'm trying to extract data strings from a long string using SQL query.
the string "'35522':{'item_id':'35522','sku':'deded','RowTotal':37.5,'qty':2"
I am trying to create a loop query that extracts data from the string.
The desired output is a table with columns (item_id,sku, RowTotal,qty) and each line will be extracted from the string above in the relevant column.
Im trying to create a function that will do that but currentley im not close.
Can you please assist me with ideas?
DECLARE #String VARCHAR(4000) =
'{:{item_id:35522,sku:deep-line-elixir,RowTotal:37.5,qty:2},
:{item_id:35527,sku:self-care-pamper-pack,RowTotal:158,qty:2},
:{item_id:35531,sku:neck-chest-rejuvenating-serum,RowTotal:21.87,qty:1},
:{item_id:35534,sku:pm-recovery-night-cream,RowTotal:23.75,qty:1},couponCode:,itemsQty:6}"'
DECLARE #b VARBINARY(4000) = CONVERT(varbinary(4000),#string)
DECLARE #StartPos int=9
DECLARE #Len tinyint=13
;WITH C (Orig,Startpos,Value) AS (
SELECT #b,#StartPos,CONVERT(VARCHAR,SUBSTRING(#b,#StartPos,#Len))
UNION ALL
SELECT #b,C.Startpos+#Len,CONVERT(VARCHAR,SUBSTRING(#b,C.StartPos+#Len,#Len)) FROM C
WHERE C.Startpos+#Len < = LEN(#b)
)
select C.Value from c where c.value like 'item%'
This is my code so far

Answer 1

The following approach splits the data into rows using STRING_SPLIT before replacing the additional characters and using case expressions to match field values. The temporary table is used to generate a row_num which can be used to order values when generating the group_num used to group the related values across multiple rows into a single related row
DECLARE #SampleString VARCHAR(4000) =
'{:{item_id:35522,sku:deep-line-elixir,RowTotal:37.5,qty:2},
:{item_id:35527,sku:self-care-pamper-pack,RowTotal:158,qty:2},
:{item_id:35531,sku:neck-chest-rejuvenating-serum,RowTotal:21.87,qty:1},
:{item_id:35534,sku:pm-recovery-night-cream,RowTotal:23.75,qty:1},couponCode:,itemsQty:6}"';
create table #temp_values (id int identity(1,1), value VARCHAR(200) );
insert into #temp_values (value) SELECT value FROM STRING_SPLIT(REPLACE(REPLACE(#SampleString,'{',''),'}',''),',');
WITH split_data AS (
SELECT id,value FROM #temp_values
),
extracted_data_raw AS (
SELECT
id as row_num,
CASE
WHEN value LIKE '%item_id%' THEN 1
WHEN value LIKE '%sku:%' THEN 2
WHEN value LIKE '%RowTotal:%' THEN 3
WHEN value LIKE '%qty:%' AND value NOT LIKE '%itemsQty%' THEN 4
END as type_num,
CASE
WHEN value LIKE '%item_id%' THEN TRIM(REPLACE(value,':item_id:',''))
END as item_id,
CASE WHEN value LIKE '%sku:%' THEN TRIM(REPLACE(value,'sku:','')) END as sku,
CASE WHEN value LIKE '%RowTotal:%' THEN TRIM(REPLACE(value,'RowTotal:','')) END as RowTotal,
CASE WHEN value LIKE '%qty:%' AND value NOT LIKE '%itemsQty%' THEN TRIM(REPLACE(value,'qty:','')) END as qty,
value
FROM split_data
)
,extracted_data_clean AS (
SELECT
MAX(item_id) as item_id,
MAX(sku) as sku,
MAX(RowTotal) as RowTotal,
MAX(qty) as qty
FROM (
SELECT
ROW_NUMBER() OVER (PARTITION BY type_num ORDER BY row_num) group_num,
item_id,
sku,
RowTotal,
qty
FROM
extracted_data_raw
WHERE
type_num IS NOT NULL
) t
GROUP BY group_num
)
select * from extracted_data_clean
Outputs:
item_id
sku
RowTotal
qty
35522
deep-line-elixir
37.5
2
35527
self-care-pamper-pack
158
2
35531
neck-chest-rejuvenating-serum
21.87
1
35534
pm-recovery-night-cream
23.75
1
Working Demo DB Fiddle
Edit 1:
You also referenced "'35522':{'item_id':'35522','sku':'deded','RowTotal':37.5,'qty':2" in your question. The following should work for both samples while also converting to respective data types:
DECLARE #SampleString VARCHAR(4000) =
'{"35522":{item_id:35522,sku:deep-line-elixir,RowTotal:37.5,qty:2},
"35522":{item_id:35527,sku:self-care-pamper-pack,RowTotal:158,qty:2},
"35522":{item_id:35531,sku:neck-chest-rejuvenating-serum,RowTotal:21.87,qty:1},
"35522":{item_id:35534,sku:pm-recovery-night-cream,RowTotal:23.75,qty:1},couponCode:,itemsQty:6}"';
create table #temp_values (id int identity(1,1), value VARCHAR(200) );
insert into #temp_values (value) SELECT value FROM STRING_SPLIT(REPLACE(REPLACE(#SampleString,'{',''),'}',''),',');
WITH split_data AS (
SELECT id,value FROM #temp_values
),
extracted_data_raw AS (
SELECT
id as row_num,
CASE
WHEN value LIKE '%item_id%' THEN 1
WHEN value LIKE '%sku:%' THEN 2
WHEN value LIKE '%RowTotal:%' THEN 3
WHEN value LIKE '%qty:%' AND value NOT LIKE '%itemsQty%' THEN 4
END as type_num,
CASE
WHEN value LIKE '%item_id%' THEN CONVERT(INT,TRIM(
REPLACE(SUBSTRING(
value,CHARINDEX('item_id',value),LEN(value)
),'item_id:','')
))
END as item_id,
CASE WHEN value LIKE '%sku:%' THEN TRIM(REPLACE(value,'sku:','')) END as sku,
CASE WHEN value LIKE '%RowTotal:%' THEN CONVERT(NUMERIC(10,2),TRIM(REPLACE(value,'RowTotal:',''))) END as RowTotal,
CASE WHEN value LIKE '%qty:%' AND value NOT LIKE '%itemsQty%' THEN CONVERT(NUMERIC(10,2),TRIM(REPLACE(value,'qty:',''))) END as qty,
value
FROM split_data
)
,extracted_data_clean AS (
SELECT
MAX(item_id) as item_id,
MAX(sku) as sku,
MAX(RowTotal) as RowTotal,
MAX(qty) as qty
FROM (
SELECT
ROW_NUMBER() OVER (PARTITION BY type_num ORDER BY row_num) group_num,
item_id,
sku,
RowTotal,
qty
FROM
extracted_data_raw
WHERE
type_num IS NOT NULL
) t
GROUP BY group_num
)
select * from extracted_data_clean;
drop table #temp_values;
Working Demo DB Fiddle
Let me know if this works for you.