Grep output with multiple Colors?

问题:

Is there a elegant method in bash of running grep against a text file with 2 or more patterns, and each pattern that matches is output in a different color?

So a line that matches on MALE and AUGUST would out put MALE in blue and AUGUST in orange? I am open to the use of sed, awk, grep and crayons or other...

回答1:

You can cascade greps with different colors by specifying --color=always and using the regular expression 'foo|$' to pass all lines.

For example:

tail -f myfwlog | GREP_COLOR='01;36' egrep --color=always 'ssh|$' | GREP_COLOR='01;31' egrep -i --color=always 'drop|deny|$'

If you want the entire line to be highlighted, update your regular expression accordingly:

.... GREP_COLOR='01;31' egrep -i --color=always '^.*drop.*$|^.*deny.*$|$'

回答2:

grep is a regular expression matcher, not a syntax highlighter :). You'll have to use multiple invocations of grep, using a different value of GREP_COLOR for each.

GREP_COLOR="1;32" grep foo file.txt | GREP_COLOR="1;36" grep bar

That would highlight "foo" and "bar" in different colors in lines that match both. I don't think there is a (simple) way to handle all occurrences of either pattern, short of merging the output stream of two independent calls:

{ GREP_COLOR="1;32" grep foo file.txt
  GREP_COLOR="1;36" grep bar file.txt
} | ...

which will obviously look different than if there were a way to assign a separate color to each regular expression.

---

You can use awk to substitute each match with itself wrapped in the correct control code.

 echo "foo bar" | awk '{ gsub("bar", "\033[1;33m&\033[0m");
                         gsub("foo", "\033[1;36m&\033[0m"); print }'

In each line, you globally replace anything matching the given regular expression with itself (&) wrapped in the ANSI escape sequences for desired color (which grep --color does for you). After processing all of the possible matches, you need to explicitly print the line.

回答3:

If you want something out of the box, you're probably looking for hhighlighter.

Here's an example:

Take a look. It's incredibly useful for coloring words in different colors automatically. It's an impressive project that's built on top of ack.

回答4:

I had the need to highlight strings in text files, and did not find a way that suited my needs, so I wrote a simple C program to colorize strings matching regular expressions.

You can download the source code from GitHub :

git clone http://github.com/mbornet-hl/hl

then :

cd hl/cr; make

Here is the usage :

hl: version 1.21
Usage: hl [-h|-eidD][-E][-rgybmcwRGYBMCW] regexp ...
  -h : help
  -v : version
  -u : do not bufferize output on stdout
  -e : extended regular expressions
  -i : ignore case
  -E : print on stderr
  -r : red
  -g : green
  -y : yellow
  -b : blue
  -m : magenta
  -c : cyan
  -w : white
  -R : red     (reverse video)
  -G : green   (reverse video)
  -Y : yellow  (reverse video)
  -B : blue    (reverse video)
  -M : magenta (reverse video)
  -C : cyan    (reverse video)
  -W : white   (reverse video)
  -d : debug
  -D : display regular expressions

To colorize the word "red" in red, and the word "blue" in blue, you just have to type in the following command :

echo "red blue red blue" | hl -r red -b blue

and here is an example to highlight the output of the ifconfig command :

hl -ei -m '^(eth|(vir)?br|vnet)[0-9.]*:[0-9]+\>'             \
       -b '^(eth|(vir)?br|vnet)[0-9.]*\.[0-9]+\>'            \
       -c '([0-9a-f]{2}:){5}[0-9a-f]{2}'                     \
       -g '\<UP\>|\<RUNNING\>|([0-9]{1,3}\.){3}[0-9]{1,3}\>' \
       -y '^(eth|(vir)?br|vnet)[0-9.:]*\>'                   \
       -W '[0-9a-f]{4}::[0-9a-f]{4}\:[0-9a-f]{4}:[0-9a-f]{4}:[0-9a-f]{4}' \
       -r ' (errors|dropped|overruns):[^0][0-9]*'

The order in which you specify the colors is important : if a string matches several regular expressions, the 1st one as a higher priority than the 2nd one, the 2nd one as a higher priority than the 3rd one, and so on ...

Hope This Helps.

回答5:

Try the wonderful rpen http://github.com/rtulke/rpen

require egrep or grep

cp rpen.py /usr/local/bin/rpen chmod 777 /usr/local/bin/rpen

then try

ps xau |rpen Ss "S+" "\?\?"

回答6:

So here's the version(s) using sed. Let's say you want the word "FAILED" to be colored red, that would be:

sed 's/\(ERROR\)/\o033[31m\1\o033[39m/'

To have the whole line containing the word "FAILED" marked in red:

sed 's/\(.*ERROR.*\)/\o033[31m\1\o033[39m/'

To have multiple words marked red:

sed 's/\(ERROR\|FAILED\)/\o033[31m\1\o033[39m/'

To have multiple words and their wohle lines marked in red:

sed 's/\(.*FAILED.*\|.*ERROR.*\)/\o033[31m\1\o033[39m/'

To have multiple colors (ERROR=red/FAILED=blue)

sed -e 's/\(ERROR\)/\o033[31m\1\o033[39m/' -e 's/\(FAILED\)/\o033[34m\1\o033[39m/'

To use all this over ssh:

ssh user@host -t "tail -n 1024 -f /some/log/file | sed --unbuffered -e 's/\(.*ERROR.*\|.*FAILED.*\)/\o033[31m\1\o033[39m/' -e 's/\(.*postfix.*\|.*dhcpd.*\)/\o033[32m\1\o033[39m/'"

And for not having to type this everytime, just declare it as a function in your bashrc.

For other colors look up ANSI codes.

回答7:

How about this? You can do it with grep ONLY! Tricks here are (1) you take /OR/ of the words you are searching; (2) you use -A and -B options with large enough numbers (larger then the line count of your file).

echo "you make it" > tmp
echo "you take it" >> tmp
echo "you cake it" >> tmp
echo "you bake it" >> tmp
echo "you wake it" >> tmp

GREP_COLOR='1;32' grep -P "(take|cake|bake)" --color=always tmp | GREP_COLOR='1;33' grep -P "(cake|bake)" --color=always -A10000 -B10000 | GREP_COLOR='1;34' grep -P "(bake)" --color=always -A10000 -B10000 

回答8:

cxpgrep is right for this purpose - highlight patterns in different colors.

cxpgrep 'PAT1|PAT2|...' FILE_OR_DIRECTORY

回答9:

My answer is below:

tailf logfile awk '/MALE/' {print "\033[34m" $0 "\033[39m"}  '/MALE/' {print "\033[33m" $0 "\033[39m"}