Extract n characters after a word in linux including new line

I need to cut n number of characters after a specific word from a text file in linux . The tricky part here is the n number of characters is spread across two lines. Almost all of the solutions given for the similar scenario only performs this kind of string extraction within the same line. For example, I have this entry in a text file like below:

I. 2023/06/02 17:57:58. Connection to server 'xxx' as user 'xxx' has been faded out (closed).
I. 2023/06/02 18:01:58. ...... connected to server 'xxx' as user 'xxx'.
I. 2023/06/02 18:30:02. Connection to server 'xxx' as user 'xxx' has been faded out (closed).
E. 2023/06/02 18:38:01. ERROR #9027 DIST(103 xxx.xxx) - /generic/gtr/tdext.c(719)
        A begin with transaction id =  'c7f0f1' @ '00000000' P3 '0000000000dd' c '87' ^eWH '0000' WH 'dd' c '87' ^eW>T '0000000000' g '00000000' and local_qid = 0000000000000000000000000000
000000000000000000000000000000721fda00360005 was seen from site xxx.xxwhile a transaction was still active.

From this I would like to extract only the entries after "local_qid", i.e. 0000000000000000000000000000
000000000000000000000000000000721fda00360005 and display in single line. Extracting this into a new variable or a text file into single line is also fine.

Could some one please shed some light if this is possible?

P.S.: It is also possible that the whole set of numbers after local_qid will be in the same line as well.

Thanks in advance to all the experts in the group!!

I have tried sed, awk, grep and almost all of them do this only till the line where they find the match with the word given (local_qid), output is either like 000000000000000000000000000 (or) if I use sed/awk command to exlcude local_qid then the output is 000000000000000000000000000000721fda00360005.

Here's a simple awk solution:

awk '
  /local_qid/{gsub(/.*local_qid = /,"");id=$0;next}
  id!=""{print id,$1;id=""}
'

It first finds a line with "local_qid", extracts the id and stores it in a variable; then it proceeds to the next line. If the id variable contains a value, the id and the first field of the line is printed; and then the id variable is reset.

The second condition is only true in the lines immediately following the "local_qid" lines.

Solution using TXR.

$ txr data.txr data
local_qid = 0000000000000000000000000000000000000000000000000000000000721fda00360005
local_qid = 0000000000000000000000000000000000000000000000000000000000721fda00360006

I augmented the data to have a wrapped and unwrapped instance of the line:

$ cat dataI. 2023/06/02 17:57:58. Connection to server 'xxx' as user 'xxx' has been faded out (closed).
I. 2023/06/02 18:01:58. ...... connected to server 'xxx' as user 'xxx'.
I. 2023/06/02 18:30:02. Connection to server 'xxx' as user 'xxx' has been faded out (closed).
E. 2023/06/02 18:38:01. ERROR #9027 DIST(103 xxx.xxx) - /generic/gtr/tdext.c(719)
        A begin with transaction id =  'c7f0f1' @ '00000000' P3 '0000000000dd' c '87' ^eWH '0000' WH 'dd' c '87' ^eW>T '0000000000' g '00000000' and local_qid = 0000000000000000000000000000
000000000000000000000000000000721fda00360005 was seen from site xxx.xxwhile a transaction was still active.
I. 2023/06/03 17:57:58. Connection to server 'xxx' as user 'xxx' has been faded out (closed).
I. 2023/06/03 18:01:58. ...... connected to server 'xxx' as user 'xxx'.
I. 2023/06/03 18:30:02. Connection to server 'xxx' as user 'xxx' has been faded out (closed).
E. 2023/06/03 18:38:01. ERROR #9027 DIST(103 xxx.xxx) - /generic/gtr/tdext.c(719)
        A begin with transaction id =  'c7f0f2' @ '00000000' P3 '0000000000dd' c '87' ^eWH '0000' WH 'dd' c '87' ^eW>T '0000000000' g '00000000' and local_qid = 0000000000000000000000000000000000000000000000000000000000721fda00360006 was seen from site xxx.xxwhile a transaction was still active.

The code uses the freeform directive with a two-line window. freeform helps to reduce the number of cases that line-oriented TXR matching code has to deal with, in situations where there is arbitrary breaking and such:

@(repeat)
@  (freeform "" 2)
        A begin with @(skip) local_qid = @qid was seen @(skip)
@  (do (put-line `local_qid = @qid`))
@(end)

@(freeform "" 2) says, "match the following pattern(s) using an alternate version of the input stream in which two lines are combined into one, using nothing as the separator". After that we just match for the desired line with the local_qid as if it were one unwrapped line.

freeform has no permanent effect on the shape of the input; the alteration is only seen by pattern code under its scope.

freeform uses a lazy string data structure, so it can be used without a numeric limit argument in situations in which the number of lines that need unwrapping is not easily known in advance. (In those situations you have to be careful not to force an instantiation of the entire remainder of a large file as a single string: only match what is needed.) freeform has the property that material from the combined line that was not consumed by pattern matching magically turns back into individual lines again, so multi-line logic can continue in the same scope.

One potential option using awk:

cat file.txt
I. 2023/06/02 17:57:58. Connection to server 'xxx' as user 'xxx' has been faded out (closed).
I. 2023/06/02 18:01:58. ...... connected to server 'xxx' as user 'xxx'.
I. 2023/06/02 18:30:02. Connection to server 'xxx' as user 'xxx' has been faded out (closed).
E. 2023/06/02 18:38:01. ERROR #9027 DIST(103 xxx.xxx) - /generic/gtr/tdext.c(719)
        A begin with transaction id =  'c7f0f1' @ '00000000' P3 '0000000000dd' c '87' ^eWH '0000' WH 'dd' c '87' ^eW>T '0000000000' g '00000000' and local_qid = 0000000000000000000000000000
000000000000000000000000000000721fda00360005 was seen from site xxx.xxwhile a transaction was still active.


awk 'BEGIN{RS="\n\n"} {for (i=1; i<=NF; i++) {if ($i == "local_qid") {print $(i + 2) ($(i + 3) ~ /[0-9]{1,}/ ? $(i + 3) : "")}}}' file.txt
0000000000000000000000000000000000000000000000000000000000721fda00360005

This changes the Record Separator (RS) from one newline ("\n", i.e. read in and process each line, one-by-one) to two newlines (i.e. read in all of the lines until "\n\n" is seen). You can then locate the string "local_qid" and print the field after it (the actual local_qid) and conditionally print the field after that if it is comprised of one or more digits (i.e. if the local_qid extends over to the next line).