Extract substring from string in postgresql between an end delimiter and its previous motif-CodePudding

i'm really struggling on this one :

I have some extradata in a postgresql db in which a number matters.

Exemple:

{"code_partenaire":"welcome","sous_statut_abc":"1","statut_alpha":"blabla"}
{"code_partenaire":"code33","sous_statut_def":"1782","statut_alpha":"blablabla"}
{"code_partenaire":"222","sous_statut_ghei":"17","statut_alpha":"blablaa"}

I would like to extract the character between ","statut_alpha" and its previous ":"

Because there is the number i'm looking for.

Do you have an idea ?

Output should be

1
1782
17

Thanks for your help :)

CodePudding user response：

For the output you given , the substring is tedious process , with the conditions (1) before : ","statut_alpha" for this used the reverse string and find the last position of ":" (2) after the last ":" before ","statut_alpha"

create table substdata (datac varchar(300));
insert into substdata values ('{"code_partenaire":"welcome","sous_statut_abc":"1","statut_alpha":"blabla"}');
insert into substdata values ('{"code_partenaire":"code33","sous_statut_def":"1782","statut_alpha":"blablabla"}');
insert into substdata values ('{"code_partenaire":"222","sous_statut_ghei":"17","statut_alpha":"blablaa"}');



select    
  substring 
  (datac,0,
  POSITION('","statut_alpha"' IN datac)) 
,
  length( substring 
  (datac,0,
  POSITION('","statut_alpha"' IN datac)) ), (
  length( substring 
  (datac,0,
  POSITION('","statut_alpha"' IN datac)) )
   -  
 pOSITION('":"' in 
 reverse(
  substring 
  (datac,0,
  POSITION('","statut_alpha"' IN datac))
  ))) as minus 
  , 
substring (substring 
  (datac,0,
  POSITION('","statut_alpha"' IN datac)) 
  ,
  
  (
  length( substring 
  (datac,0,
  POSITION('","statut_alpha"' IN datac)) )
   -  
 pOSITION('":"' in 
 reverse(
  substring 
  (datac,0,
  POSITION('","statut_alpha"' IN datac))
  ))) 2,( length( substring 
  (datac,0,
  POSITION('","statut_alpha"' IN datac)) ))) as finresult 
 from substdata

OUTPUT:

substring   length  minus   finresult
{"code_partenaire":"welcome","sous_statut_abc":"1   49  47  1
{"code_partenaire":"code33","sous_statut_def":"1782     51  46  1782
{"code_partenaire":"222","sous_statut_ghei":"27     47  44  17

if the first condition is always in the third position in the string

you can use the unest with ordinally give rownumber and filter with it

select row_number() over(),t from (
  select
 row_number() over() as id ,t
  from substdata ,
  unnest(string_to_array(
  substring 
  (datac,0,
  POSITION('","statut_alpha"' IN datac)), '":"'))  
  WITH ORDINALITY As T ) as drt 

WHERE (id % 3) = 0

OUTPUT 2 :

row_number  t
1   1
2   1782
3   17

CodePudding user response：

Found it out :)

split_part(extradata,'"',8)