grep: extract content between prefix and suffix - grep

Question

I've a file content like this:
Listening for transport dt_socket at address: 8000
------------------------------------------------------------
🔥 ^[[1m HAPI FHIR^[[22m 5.4.0 - Command Line Tool
------------------------------------------------------------
Process ID : 21719#psgd
Max configured JVM memory (Xmx) : 3.2GB
Detected Java version : 11.0.7
------------------------------------------------------------
^[[32m2021-07-01^[[0;39m ^[[1;32m12:27:40.79^[[0;39m ^[[37m[main]^[[0;39m ^[[37mWARN ^[[0;39m ^[[1;34mo.f.c.i.s.c.ClassPathScanner^[[0;39m ^[[1;37mUnable to resolve location classpath:db/migration. Note this warning will become an error in Flyway 7.
^[[0;39m^[[32m2021-07-01^[[0;39m ^[[1;32m12:27:42.641^[[0;39m ^[[37m[main]^[[0;39m ^[[37mWARN ^[[0;39m ^[[1;34mo.f.c.i.s.c.ClassPathScanner^[[0;39m ^[[1;37mUnable to resolve location classpath:db/migration. Note this warning will become an error in Flyway 7.
^[[0;39m^[[32m2021-07-01^[[0;39m ^[[1;32m12:27:44.693^[[0;39m ^[[37m[main]^[[0;39m ^[[37mINFO ^[[0;39m ^[[1;34mc.u.f.j.m.t.InitializeSchemaTask^[[0;39m ^[[1;37m3_3_0.20180115.0: Initializing ORACLE_12C schema for HAPI FHIR
^[[0;39m^[[32m2021-07-01^[[0;39m ^[[1;32m12:27:44.848^[[0;39m ^[[37m[main]^[[0;39m ^[[37mINFO ^[[0;39m ^[[1;34mc.u.f.j.m.t.BaseTask^[[0;39m ^[[1;37m3_3_0.20180115.0: SQL "create sequence SEQ_BLKEXCOL_PID start with 1 increment by 50" returned 0
^[[0;39m^[[32m2021-07-01^[[0;39m ^[[1;32m12:27:44.918^[[0;39m ^[[37m[main]^[[0;39m ^[[37mINFO ^[[0;39m ^[[1;34mc.u.f.j.m.t.BaseTask^[[0;39m ^[[1;37m3_3_0.20180115.0: SQL "
create sequence SEQ_BLKEXCOLFILE_PID start with 1 increment by 50" returned 0
^[[0;39m^[[32m2021-07-01^[[0;39m ^[[1;32m12:27:47.573^[[0;39m ^[[37m[main]^[[0;39m ^[[37mINFO ^[[0;39m ^[[1;34mc.u.f.j.m.t.BaseTask^[[0;39m ^[[1;37m3_3_0.20180115.0: SQL "
create table HFJ_BINARY_STORAGE_BLOB (
BLOB_ID varchar2(200 char) not null,
BLOB_DATA blob not null,
CONTENT_TYPE varchar2(100 char) not null,
BLOB_HASH varchar2(128 char),
PUBLISHED_DATE timestamp not null,
RESOURCE_ID varchar2(100 char) not null,
BLOB_SIZE number(10,0),
primary key (BLOB_ID)
)" returned 0
I need to extract only content inside between SQL " and " returned 0 trimming all whitespaces.
Any ideas?
I've tried to reduce problem using:
$ echo 'sdf SQL" sdf sdf" returned 0' | grep 's/SQL"\(.*\)" returned 0/\1/' -
But it's getting empty.
My expected output is:
create sequence SEQ_BLKEXCOL_PID start with 1 increment by 50;
create sequence SEQ_BLKEXCOLFILE_PID start with 1 increment by 50;
create table HFJ_BINARY_STORAGE_BLOB (
BLOB_ID varchar2(200 char) not null,
BLOB_DATA blob not null,
CONTENT_TYPE varchar2(100 char) not null,
BLOB_HASH varchar2(128 char),
PUBLISHED_DATE timestamp not null,
RESOURCE_ID varchar2(100 char) not null,
BLOB_SIZE number(10,0),
primary key (BLOB_ID)
);
I've tried to perform:
cat test.log | sed -E 's/.* SQL"(.*)" returned 0/\1/'
It's returning me all file content...
Using awk, it returns empty:
$ awk -v RS='SQL "[[:space:]]+?\n\n+.*returned 0' '
RT{
gsub(/^SQL "\n+|\n+$/,"",RT)
sub(/" returned 0[[:space:]]+?\n*$/,"",RT)
print RT";"
}
' test.log

Answer 1

This can be done using custom RS in gnu-awk that splits data on SQL "..." text block and then inside action block it extracts text between quotes without leading space.
awk -v RS=' SQL "[^"]+"' 'RT {
gsub(/^[^"]*"[[:space:]]*|"[^"]*$/, "", RT); print RT ";"}' file.sql
create sequence SEQ_BLKEXCOL_PID start with 1 increment by 50;
create sequence SEQ_BLKEXCOLFILE_PID start with 1 increment by 50;
create table HFJ_BINARY_STORAGE_BLOB (
BLOB_ID varchar2(200 char) not null,
BLOB_DATA blob not null,
CONTENT_TYPE varchar2(100 char) not null,
BLOB_HASH varchar2(128 char),
PUBLISHED_DATE timestamp not null,
RESOURCE_ID varchar2(100 char) not null,
BLOB_SIZE number(10,0),
primary key (BLOB_ID)
);

Answer 2

With GNU awk, with your shown samples, please try following code once.
awk -v RS='SQL "[[:space:]]*\n\n+.*returned 0' '
RT{
gsub(/^SQL "\n+|\n+$/,"",RT)
sub(/" returned 0[[:space:]]*\n*$/,"",RT)
print RT";"
}
' Input_file
Explanation: Simple explanation would be, setting RS as SQL "[[:space:]]+?\n\n+.*returned 0 for awk program, removing not required strings like SQL " with new lines and returned 0 at last of value before printing it.
Explanation of regex is as follows: match SQL followed by space " followed by 1 or more spaces(optional) followed by 1 or more new lines till returned 0 here.

Answer 3

If you have gnu grep then you can use this PCRE regex:
grep -oPz ' SQL "\K[^"]+' file.sql
create sequence SEQ_BLKEXCOL_PID start with 1 increment by 50
create sequence SEQ_BLKEXCOLFILE_PID start with 1 increment by 50
create table HFJ_BINARY_STORAGE_BLOB (
BLOB_ID varchar2(200 char) not null,
BLOB_DATA blob not null,
CONTENT_TYPE varchar2(100 char) not null,
BLOB_HASH varchar2(128 char),
PUBLISHED_DATE timestamp not null,
RESOURCE_ID varchar2(100 char) not null,
BLOB_SIZE number(10,0),
primary key (BLOB_ID)
)
Explanation:
' SQL ": Search SQL " text
\K: Reset matched info
[^"]+: Match 1+ of any characters that are not "
To get formatting as desired (in comments) use this grep + sed (gnu) solution:
grep -oPzZ ' SQL "\K[^"]+' file.sql |
sed -E '$s/$/\n/; s/\x0/;/; s/^[[:blank:]]+//'
create sequence SEQ_BLKEXCOL_PID start with 1 increment by 50;
create sequence SEQ_BLKEXCOLFILE_PID start with 1 increment by 50;
create table HFJ_BINARY_STORAGE_BLOB (
BLOB_ID varchar2(200 char) not null,
BLOB_DATA blob not null,
CONTENT_TYPE varchar2(100 char) not null,
BLOB_HASH varchar2(128 char),
PUBLISHED_DATE timestamp not null,
RESOURCE_ID varchar2(100 char) not null,
BLOB_SIZE number(10,0),
primary key (BLOB_ID)
);