color the search result when the input is paragraph not a single line - bash

Question

I am trying to print the contents of the whole file with highlighted search string.
For a simple file where Record is equal to a single line then I can do this easily using:
grep --color=auto "myseacrchpattern" inputfile
Here my records are in form of paragraphs not single line. Example:
CREATE TABLE mytable ( id SERIAL,
name varchar(20),
cost int );
CREATE TABLE notmytable ( id SERIAL,
name varchar(20),
cost int );
If I do use grep for keyword : "notmytable",it will give me colored output but only that line is printed.
grep --color=auto 'notmytable' inputfile
CREATE TABLE notmytable ( id SERIAL, # <-- "notmytable" is in red but its not the whole query
I need something like this :
CREATE TABLE notmytable ( id SERIAL, # <---"notmytable" is in red
name varchar(20),
cost int );
I can print the desired paragraph with awk or perl but how to color it :
awk -v RS=';' -v ORS=';\n' '/notmytable/' inputfile
CREATE TABLE notmytable ( id SERIAL,
name varchar(20),
cost int );
OR perl :
perl -00lne 'print $_ if /notmytable/' inputfile
CREATE TABLE notmytable ( id SERIAL,
name varchar(20),
cost int );

Answer 1

perl "-MTerm::ANSIColor qw(:constants)" -00lnE'
next if not /notmytable/;
for (split "\n") { /notmytable/ ? say RED $_, RESET : say }
' input
The :constants tag provides RED and such. There are other ways, see Term::ANSIColor
Note that there has to be some duplicate searching, since we need to first identify the paragraph, but then print only that line in color while printing others normally.
If only the pattern need be colored, double-parsing isn't needed (and it's far easier and nicer)
perl -MTerm::ANSIColor -00lnE'say if s/(notmytable)/colored($1,"red")/eg' input

Answer 2

If a string matching the desired regexp is found then surround it by the appropriate characters to change its color and print the record containing it:
$ awk -v RS=';' -v ORS=';\n' 'gsub(/notmytable/,"<RED>&</RED>")' file
CREATE TABLE <RED>notmytable</RED> ( id SERIAL,
name varchar(20),
cost int );
Just change <RED> and </RED> to the escape sequences for that color, e.g.
awk -v RS=';' -v ORS=';\n' 'gsub(/notmytable/,"\033[31m&\033[0m")' file
or if you don't want to hard-code those color values you could do:
awk -v RS=';' -v ORS=';\n' -v red="$(tput setaf 1)" -v nrm="$(tput sgr0)" 'gsub(/notmytable/,red"&"nrm)' file
BTW if you have blank lines between all records you'll probably find using -v RS= -v ORS='\n\n' works better for you than -v RS=';' -v ORS=';\n'.

Answer 3

Combine your awk or perl solution with grep:
perl -00lne 'print $_ if /notmytable/' input|grep -C1000 notmytable
The -C1000 option makes grep keep 1000 lines of surrounding context around the matching line, actually turning grep into a mere colorizer rather than line selector.
You can wrap it into a bash function:
function paragrep() { perl -00lne 'print $_ if /'"$1"/ "$2"|grep -C1000 "$1"; }
Usage example:
$ paragrep notmytable input
CREATE TABLE notmytable ( id SERIAL, # <---"notmytable" is in red
name varchar(20),
cost int );
$