Unix/Linux Go Back    



Help in awk/bash

Shell Programming and Scripting




Kindly Note - This is a Single User Post by Forum Member Don Cragun Regarding:
Help in awk/bash.
Please Follow The Primary Link Above to View the Full Discussion.

   
Old Unix and Linux 12-28-2012   -   Original Discussion by bioinfo
Don Cragun's Unix or Linux Image
Don Cragun Don Cragun is offline Forum Staff  
Administrator
 
Join Date: Jul 2012
Last Activity: 12 December 2017, 6:43 AM EST
Location: San Jose, CA, USA
Posts: 10,769
Thanks: 590
Thanked 3,769 Times in 3,217 Posts
Quote:
Originally Posted by bioinfo View Post
Thanks for the reply.
Can you please explain it somewhat.

Thanks again.

Code:
1  echo "awk with rounded values"
2  awk ' FNR == NR {v[sprintf("%.2f", $2)]; next}
3  sprintf("%.2f", $4) in v {print $0, FILENAME}' 1.txt [a-h].txt
4
5  echo "awk with truncated values"
6  awk '
7  function trunc(val) {
8          split(val, a, /[.]/)
9          return a[1] "." substr(a[2] "00", 1, 2)
10 }
11 FNR == NR {v[trunc($2)]; next}
12 trunc($4) in v {print $0, FILENAME}' 1.txt [a-h].txt

I have added line numbers to aid in this discussion, but note that the line numbers cannot appear in the script when you run it.

Also note that I have added an awk next command to lines 2 and 11. With the given sample data it won't make any difference, but with other data or with different fields being checked, it could be important.

In the suggestion on lines 1-3, the sprint("%.2f", arg) converts the string specified by arg to a floating point value and produces a string that represents that floating point value rounded to two digits after the decimal point. Line two uses that to create an array with indices that are the rounded floating point values of the second field ($2) in the first input file (lines where the record number within the file [FNR] is equal to the line number of all records read by awk [NR]).

(The next command I added here causes awk to skip to the next record instead of checking whether or not any remaining commands in the script should be executed. Without the next, the next line will process lines from all input files. It doesn't affect processing here because there is no field 4 in file one. The empty field 4 will be converted to 0.00 and none of the strings in the second field in the 1.txt will be converted to 0.00.)

Line 3 tests whether the same conversion used in line 2 produces a string that is an index in the array v (index in array evaluates to TRUE if index if is an index in the array named array. So, if $4 (rounded to two decimal places) in any of the files after the 1st file match $2 (rounded to two decimal places) in the first file, the print command will be run printing the current input line ($0) and the name of the file containing the line (FILENAME).

The 1.txt [a-h].txt on lines 3 and 12 specifies the eleven input files to be processed by these awk scripts.

The suggestion on lines 5-12 uses the same logic as the 1st suggestion but truncates the strings to two decimal places instead of rounding to two decimal places. Since the truncation logic is more complex than the single function call to sprint() used to perform the rounding, I wrote a function (lines 7-10) to convert the string to a string representing a floating point value with two decimal places.

The split() on line 8 creates an array of one or two elements with the first element containing all of the characters before the "." and the second element containing all of the characters after the ".". If there is no "." in the input value, the first element of the array will contain the entire input string and the second element of the array will not be set (and when referenced will act as an empty string). The return command on line 9 returns a string that is the concatenation of the first element in the array, a decimal point, and the 1st two characters of the concatenation of the second element of the array followed by "00". (The concatenation with "00" takes care of cases where field 2 in the first file or field 4 in the remaining files have an integer value with no decimal point and the case where the input field has a period but there are less than two digits after the decimal point.)

The logic on lines 11 and 12 is the same as the logic on lines 2 and 3.
The Following User Says Thank You to Don Cragun For This Useful Post:
bioinfo (12-29-2012)