4.14 awk -- A Programming Language

4.14 awk -- A Programming Language

V. Aho, Peter W. Weinberger, and Brian W. Kernighan) is an interpretive programming language for processing text files. Of all 6 possible combinations of the three initials, awk seems most appropriate. At times the language seems awkward (to say the least). It is also very powerful for text processing. It has many of the same features as "C".

The AWK language is a good text-processing language. It has more features and capabilities than sed, but less than Perl.

The book: The AWK programming Language is a good awk reference. It was the source for parts of this lab.

Also look at: UNIX System V Release 3 Programmers Guide, Chapter 4 AWK.

The program can be placed on the command line or in a file. The program can read and write files named within the program. You can also pass arguments to the program from the command line.

Invoking awk one of two ways, the first places the program on the command line, the second places the awk program in a file.

awk 'program text' [file...] 
awk -f program-file [file...] 

An example of a two statement awk program on the command line follows. The program prints all lines with either "Arizona" or "72".

username@gort ~ $ awk '/Arizona/;/72/' /home/cis137/data 

Many "useful" awk programs are only 1 to 4 lines long! It may be one of the easiest ways to rearrange the order of fields in a file or do a summary report. Some examples of short awk programs:

1. Print the number of input lines:

   END { print NR }
   

   This could have been done with wc

   wc -l
   

2. Print the 40th line:

   NR == 40
   

3. Print the lines 40-50:

   NR >= 40 && NR <= 50
   

4. Print the last field of every line:

   { print $NF }
   

   NF is the number of fields. $NF is the last field in the record.
5. Print the last field of the last line:

           { field = $NF }
   END     { print field  }
   

   field is a variable.
6. Print lines with more than 4 fields:

   NF > 4
   

7. Print the word count for the file:

          { words = words + NF }
   END    { print words        }
   

   This could have been done with wc

   wc -w
   

8. Print lines with "Alaska":

   /Alaska/
   

9. Count the lines with "Alaska":

   /Alaska/    { count++      }
   END         { print count  }
   

10. Find the largest third field:

    BEGIN       { max = -1000000  }
    $3 > max    { max = $3        }
    END         { print max  }
    

11. Find the largest third field and save the line, too:

    BEGIN       { max = -1000000           }
    $3 > max    { max = $3; savedline = $0 }
    END         { print savedline          }
    

12. Print every line that has less than 4 fields:

    NF < 4  
    

13. Print every line shorter than 50 characters:

    length < 50  
    

14. Print the fields in reverse order:

    { for ( j=NF; j>0; j-- ) printf("%s ", $j) 
      printf("\n")
    }
    

15. Total up all of the fields in a file:

        { for ( j=1; j<=NF; j++ ) total += $j }
    END { print( total ) }
    

16. Renumber a file:

           { $1 = "" ; print NR, $0 }  
    

    The "$1 = """ removes the old line number. To renumber by tens:

           { $1 = "" ; print 10*NR, $0 }  
    

The format of the awk program is:

 
pattern  { action } 
pattern  { action } 
pattern  { action } 

This can be repeated many times in an awk program.

• If pattern is blank the action is done for all lines.
• If action is blank the entire line is printed.

Two special patterns exist: BEGIN and END.

The action for BEGIN is processed prior to the first line of the input. This can be used for printing headers and setting field and record separators. The action for END is processed at the end of the input. This can be used for report summaries.

In a given record, the input fields are referred to as $1, $2, $3 ...

The entire record is referred to as $0.

1. What awk command can you use to print out all lines in args.c that have printf?
   

   ---

   
   
   

   ---

2. What awk command will print out all lines in args.c that don't have a plus sign?
   

   ---

   
   
   

   ---

3. What awk command will print out all lines in args.c that are longer than 20 characters?
   

   ---

   
   
   

   ---

4. What awk command will print out lines in args.c that don't have a "(" and are longer than 10 characters?
   

   ---

   
   
   

   ---

5. Using the input file
   http://uml.lt.tucson.az.us/hl2.2006-spring/files/data.

   Arizona Phoenix 6285295 Tucson 32
   California Sacramento 38593635 Eureka 12
   Oregon Salem 12345234 Portland 45
   Washington Olympia 6549872 Seattle 36
   Illinois Springfield 6759346 Chicago 14
   Maine Augusta 456923 Lewiston 23
   Texas Austin 23967433 Houston 26
   
   

   Write an awk program that reverses the order of the fields. Note that this file always has 5 fields. This does not need to be as complicated as the example in the AWK book.
   

   ---

   
   
   

   ---

6. Write an awk program that prints the average of the third field in data file.
   

   ---

   
   
   

   ---

7. Write an awk program that removes the first field and prints only those lines where the third field is greater than 15 million. (Again, use the data).
   

   ---

   
   
   

   ---

4.14 awk -- A Programming Language