Linux Extract Matched Text Field from File

Question

I have a file which has many lines of the format:

bc("STG1/Phone") = {type=bana_pub; cbb=12.354; abb=0.0}`

I'm looking to extract cbb=12.354;. Currently, I'm doing the following:

cat input_file.txt | grep cbb | awk -F " " '{ print $4 }'`

The problem is that my approach is location specific i.e. assumes it's always 4th field. How do I extract text of the form cbb= knowing after the = it could be any length and the semi-colon ; is optional. The only guarantee I have is that the term cbb=12.354; will be surrounded by whitespace if that helps. The file in future may be of the format:

bc("STG1/Phone") = {type=bana_pub; cbb=12.354; abb=0.0}
bc("STG1/Phone") = {type=bana_pub;  abb=0.0; cbb=12.354}

My gut tells me regex is probably the way to go, but I generally try and avoid it if I can as I prefer simple matching tools (which I understand better).

Thanks in anticipation for your help.

Answer 1

Solution:

grep -Eo 'cbb=[^;}]+'

Let's test it:

$ grep -Eo 'cbb=[^;}]+' <<<'bc("STG1/Phone") = {type=bana_pub; cbb=12.354; abb=0.0}`'
$ cbb=12.354

Explanation:

When you use ... | grep cbb | ... you're using basic regex. Advanced regex isn't so complicated.

Option -E is for advanced regex, useful for don't escape some metacharacters. -o is for print just what grep matches instead the whole line.

The regex cbb=[^;}]+would be the same for any other cmd, not just grep.

cbb= is a fixed string, no metacharacter there (c followed by b etc)

[^;}]+ square brackets delimit a character set in a single position. A caret at the beginning means negated character set. The plus sign means one or more character. This way it will match any character, at least one, until it finds a ; or }

Here's a good link to learn more about regexes: https://www.regular-expressions.info/characters.html

Answer 2

This works and is position-independent:

grep cbb input_file.txt | awk -F "cbb=" '{ print $2 }'| awk -F ";" '{print "cbb=" $1}'

First it selects only lines containing cbb, then uses the string cbb= as separator and finally uses ; as field separator adding the string cbb= to the final result.

Answer 3

You can also use sed (since sed is called only once, should be faster)

sed -n 's/^.*\(cbb=[0-9\.]*\).*$/\1/p' sample.txt

Where sample.txt is your input file. Check only for numerical ([0-9.]) to address a possible issue with optional semicolon.

Answer 4

In this case, grep is the right tool for the job. However, I thought I'd add:

• Perl

  perl -lane 'print $1 if /(cbb=[^;}]+)/' input_file.txt
  

• AWK

  awk 'match($0,/cbb=[^;}]+/,m) {print m[0]}' input_file.txt
  

• Sed

  sed -rn 's/.*(cbb=[^;}]+).*/\1/p' input_file.txt
  

---

Credits to Paulo for understanding what the OP meant with:

after the = it could be any length and the semi-colon ; is optional. The only guarantee I have is that the term cbb=12.354; will be surrounded by whitespace