Bdtyktl

How to use grep or any other tool to get a specific value in an output

In the below output I need to get the value 255.00 in the line with Minimum: 255.00 (1.0000)

A pattern like: Channel Statistics:s+Gray:s+Minimum: +([d.]+)

Image: test.tif

  Format: TIFF (Tagged Image File Format)

  Geometry: 2525x1785

  Class: DirectClass

  Type: bilevel

  Depth: 1 bits-per-pixel component

  Channel Depths:

    Gray:     1 bits

  Channel Statistics:

    Gray:

      Minimum:                   255.00 (1.0000)

      Maximum:                   255.00 (1.0000)

      Mean:                      255.00 (1.0000)

      Standard Deviation:          0.00 (0.0000)

  Filesize: 581

  Interlace: No

  Orientation: Unknown

  Background Color: white

  Border Color: #DFDFDF

  Matte Color: #BDBDBD

Using perl, you can do the following. It captures the numeric value after minimum: inside the block Channel Statistics: and prints it:

perl -0 -ne '/Channel Statistics:s+Gray:s+Minimum:h+([d.]+)/ && print $1,"n"' file

Output: (for given example)

255.00

Explanation:

-0      # specifies the input record separator. If there are no digits, the null character is the separator. The whole file is read in a single string.

-n      # Iterate over the file

-e      # execute the command line

Regex:

/                           # regex delimiter

    Channel Statistics:     # literally

    s+                     # 1 or more any kind of spaces

    Gray:                   # literally

    s+                     # 1 or more any kind of spaces

    Minimum:                # literally

    h+                     # 1 or more horizontal spaces

    (                       # start group 1

        [d.]+              # 1 or more digit or dot

    )                       # end group

/                           # regex delimiter

With sed

sed -rn 's/^s+Minimum:s+([0-9.]+).+$/1/p' image.data

In slow-mo:

• 
  -r tells sed we use th extened regexp "syntax"


• 
  -n tells sed to not print non-matching lines


• 
  s/^s+Minimum:s+([0-9.]+).+$/1/ matches your target line and replaces it with just the value you are after


• 
  p tells sed to print the result

If you need to disambiguate by taking in account the content of previous lines, it is just slightly more complicated:

sed -r ':a;N;$!ba; s/^.*Gray:s*ns+Minimum:s+([0-9.]+).+$/1/' image.data

where:

• 
  :a;N;$!ba; is a loop in the sed language that loads the whole file at once


• 
  -n is no longer necessary since there are no other lines we don't want to print


• the final p is no longer necessary either since we don't use -n