Decimal numbers and letters in the same collums: round numbers


 
Thread Tools Search this Thread
Top Forums UNIX for Beginners Questions & Answers Decimal numbers and letters in the same collums: round numbers
# 1  
Old 12-27-2016
Decimal numbers and letters in the same collums: round numbers

Hi!

I found and then adapt the code for my pipeline...

Code:
awk -F"," -vOFS="," '{printf "%0.2f %0.f\n",$2,$4}' xxx > yyy

I add -F"," -vOFS="," (for input and output as csv file) and I change the columns and the number of decimal...

It works but I have also some problems... here my columns

Code:
Ubuntum, Bash version: 4.3.46

Code:
awk -F"," -vOFS="," '{printf "%0.2f %0.f\n",$2,$4}' xxx > yyy

xxx (input csv file) = I have over 50 columns and over 5000 lines... mix columns (numbers, letters, word, alphanumeric)

Code:
.112      ,    0.25         ,    0.48      ,    .112
5.232    ,    0.5852     ,    10.25    ,    10.25
8.455    ,    nd            ,    10         ,    8.455
n.d.       ,    10.4558   ,    aa_a     ,    n3d5
-8.55     ,    -12.458    ,    80.985  ,    -8.55

yyy (output file), it is space and not "," (csv)

Code:
0.25 0
0.59 10
0.00 8
10.46 0
-12.46 -9

1) I would need a csv file as output... without replace the "space" to "," using sed or other commands...

2) When I apply the code the alphanumeric letters (ie: n3d5) or only letter/words (nd) become 0.00 (column 2, line 3) and 0 (column 4, line 4). They must be as before, alphanumeric letters or only letter/words

Thanks a lot!!!


Moderator's Comments:
Mod Comment Please use CODE tags as required by forum rules!

Last edited by RudiC; 12-27-2016 at 03:54 PM.. Reason: Added CODE tags.
# 2  
Old 12-27-2016
Hello echo manolis,

Could you please try following and let me know if this helps you.
Code:
awk -F, 'function check(var){var=var~/[[:digit:]]/?"%0.2f":(var~/[[:alpha:]]/?"%s":"");return var} {gsub(/[[:space:]]+,[[:space:]]+/,",",$0);a=check($2);b=check($4);printf(a "," b"\n",$2,$4)}' OFS=","   Input_file

Output will be as follows.
Code:
0.25,0.11
0.59,10.25
nd,8.46
10.46,n3d5
-12.46,-8.55

EDIT: Adding a non-one liner form of solution too successfully now.
Code:
awk -F, 'function check(var){
                                var=var~/[[:digit:]]/?"%0.2f":(var~/[[:alpha:]]/?"%s":"");
                                return var
                            }
                            {
                                gsub(/[[:space:]]+,[[:space:]]+/,",",$0);
                                a=check($2);
                                b=check($4);
                                printf(a "," b"\n",$2,$4)
                            }
        '    Input_file

Thanks,
R. Singh

Last edited by RavinderSingh13; 12-28-2016 at 02:16 PM.. Reason: Adding a non-one liner form of solution too successfully now.
# 3  
Old 12-27-2016
Not too far off RavinderSingh13's proposal, but with some added felxibility rg. field width, still e.g. some error checking missing :
Code:
awk -F, 'function FMT(PAR) {return "%" (PAR~/[^0-9. -]/?"*s":"0.*f")} {gsub(/ /,_); printf FMT($2) FS FMT($4) RS, 2, $2, 0, $4}' file
0.25,0
0.59,10
nd,8
10.46,n3d5
-12.46,-9

This User Gave Thanks to RudiC For This Post:
# 4  
Old 12-27-2016
Well done, but this:
Quote:
Originally Posted by RavinderSingh13
Code:
                                a=check($2);
                                b=check($3);

Is perhaps a typo. Shouldn't it read:

Code:
                                a=check($2);
                                b=check($4);

You might also want to further simplify your program by immediately printing inside the function, without returning anything:

Code:
# based on the program by RavinderSingh13
awk -F, 'function myprint(var){
                                if( var~/[[:digit:]]/ ) 
                                    printf("%0.2f", var);
                                if( var~/[[:alpha:]]/ ) 
                                    printf("%s", var);
                            }
                            {
                                gsub( /[[:space:]]+,[[:space:]]+/, ",", $0 );
                                myprint( $2 );
                                printf( "," );
                                myprint( $4 );
                                printf( "\n" );
                            }
        '    Input_file

If you know you have only to differentiate between "[[:digit:]]" and "[[:alpha:]]" you can further simplify myprint() to use if ... else instead of separate ifs.

bakunin
This User Gave Thanks to bakunin For This Post:
# 5  
Old 12-27-2016
I'm afraid there's a logic flaw in the format determination in above posts: if the field in question contains ANY non-numeric character, use "%s" else use "%f" with an adaptable width. So, the (var ~ /[[:digit:]]/) doesn't lead nowhere - although it doesn't hurt, either, in above constellations. Had you reversed the sequence of the two tests, the "n3d5" field would have received the wrong format string.
These 2 Users Gave Thanks to RudiC For This Post:
# 6  
Old 12-27-2016
Quote:
Originally Posted by RudiC
if the field in question contains ANY non-numeric character, use "%s" else use "%f" with an adaptable width.
Darn, you are right. I hate it when you do that to me. ;-)

The regexp can be repaired perhaps: (var ~ /^-*[0-9.]*[0-9]+$/).

Testing for [[:alpha:]] alone will also not do the trick because of "misformed" numbers like: "123.456.789", which should not be treated as numbers despite passing the [[:digit:]]-test. Thinking about it it is not possible to determine "number or not" based on characters alone:

Code:
-123.456  # is a number
123-456.  # is not


I hope this helps.

bakunin

Last edited by bakunin; 12-27-2016 at 06:19 PM..
# 7  
Old 12-27-2016
If I understand the requirements correctly, I don't think bakunin's code work with string like n3d5. Since both alpha and numeric characters are present, it prints both a number and a string:
Code:
0.25,0.11
0.59,10.25
nd,8.46
10.46,0.00n3d5
-12.46,-8.55

and like RavinderSingh13's code, there aren't supposed to be any digits after the decimal point when converting field #4.

I came up with the following before seeing RudiC's code. It uses similar logic but approaches it a little bit differently:
Code:
awk -F, -v OFS=, '
function format(field, digits_after_radix) {
	gsub(/^ *| *$/, "", $field)
	return sprintf("%"($field~/[[:alpha:]]/?"s":"0."digits_after_radix"f"),
		    $field)
}
{	print format(2, 2), format(4, 0)
}' xxx > yyy

Note also that this code strips leading and trailing <space>s from the fields before printing them, but preserves any internal blanks in alphanumeric strings. R. Singh's code strips leading and trailing whitespace characters from field #2, but only leading whitespace characters from field #4 (with an input file that contains four fields). RudiC's code strips all <space> characters from both fields. And, bakunin's code strips the same characters that R. Singh's code strips. With your sample data, none of this matters, but if your real data contains trailing whitespace characters in field #4 or if field #2 or #4 contains spaces in the middle of the field, you'll need to adjust whatever code you choose to give the results you want.

If xxx contains:
Code:
.112      ,    0.25         ,    0.48      ,    .112
5.232    ,    0.5852     ,    10.25    ,    10.25
8.455    ,    nd            ,    10         ,    8.455
n.d.       ,    10.4558   ,    aa_a     ,    n3d5
-8.55     ,    -12.458    ,    80.985  ,    -8.55
-8.55     ,    not a number    ,    80.985  ,    Not 1 number only

the above code produces the output:
Code:
0.25,0
0.59,10
nd,8
10.46,n3d5
-12.46,-9
not a number,Not 1 number only

With the awk on macOS Sierra version 10.12.2, RavinderSingh13's code with the above input file, produces the output:
Code:
0.25,0.11
0.59,10.25
nd,8.46
10.46,n3d5
-12.46,-8.55
not a number,0.00

which I do not understand. I would have expected the n3d5 in the 4th output line to be 0.00 instead since (like the last field on the last line) that field contains both digits and alphabetics.
These 2 Users Gave Thanks to Don Cragun For This Post:
Login or Register to Ask a Question

Previous Thread | Next Thread

10 More Discussions You Might Find Interesting

1. Shell Programming and Scripting

[FUN] Numbers to Roman letters/num

Heyas Just a little fun script (code block) i'd like to share for fun. #/bin/bash # roman.sh # # Function # num2roman() { # NUM # Returns NUM in roman letters # input=$1 # input num output="" # Clear output string len=${#input} # Initial length to count down ... (9 Replies)
Discussion started by: sea
9 Replies

2. UNIX for Dummies Questions & Answers

sed - extract a group of Letters/numbers

I have a file with hundreds of lines in it. I wanted to extract anything that matches the following: KR followed by 4 digits: example KR1201 cat list | sed "s///g" Is the closest I've come, and obviously it is not what I want. This would remove all of the items that I want and leave me... (2 Replies)
Discussion started by: newbie2010
2 Replies

3. Shell Programming and Scripting

Sorting mixed numbers and letters

Hello, I have a file such as this: chr1 chr2 chr1 chr2 chr3 chr10 chr4 chr5 chrz chr1AI want to sort it, I use this command: sort -k1 -th -n testfilebut I get this output, how can I fix this? chr1 chr1 chr10 chr1A chr2 chr2 (3 Replies)
Discussion started by: Homa
3 Replies

4. Shell Programming and Scripting

awk : match only the pattern string , not letters or numbers after that.

Hi Experts, I am finding difficulty to get exact match: file OPERATING_SYSTEM=HP-UX LOOPBACK_ADDRESS=127.0.0.1 INTERFACE_NAME="lan3" IP_ADDRESS="10.53.52.241" SUBNET_MASK="255.255.255.192" BROADCAST_ADDRESS="" INTERFACE_STATE="" DHCP_ENABLE=0 INTERFACE_NAME="lan3:1"... (6 Replies)
Discussion started by: rveri
6 Replies

5. Shell Programming and Scripting

reducing values in columns with both numbers and letters

Hi, I columns with both number and letters however i need the number 4 trimmed off the lines that have 3 numbers in them so it just because the 2 preceding numbers only For example V25QG2-K18QG-V25CG2 L26HG-L17HA-L26CG I434QD1-L19HB2-I434CD1 I434QD1-A31QB-I434CD1 ... (7 Replies)
Discussion started by: olifu02
7 Replies

6. Shell Programming and Scripting

sed command, look for numbers following letters

If I have a set of strings, C21 F231 H42 1C10 1F113 and I want to isolate the ints following the char, what would the sed string be to find numbers after letters? If I do, *, I will get numbers after letters, but I am looking to do something like, sed 's/*/\t*/g' this will give me... (14 Replies)
Discussion started by: LMHmedchem
14 Replies

7. Shell Programming and Scripting

Regarding decimal numbers

Hello... I am new to unix and I am wondering if in a C-shell script , Are we supposed to use only whole numbers........ for example..if a program needs to calculate the average of some numbers........ @ avg = (($1 +$2 + $3)/3)) is returning a whole number.........How can a decimal be... (7 Replies)
Discussion started by: ravindra22
7 Replies

8. Shell Programming and Scripting

decimal numbers

Hi friends How can I use "for loop" for decimal numbers? ex: 0.1 < x < 0.6 I used this commands but does'nt work. LIMIT=0.6 for ((x=0.1; x<=LIMIT; x++)) do - - - done Many thanks (1 Reply)
Discussion started by: snow
1 Replies

9. UNIX for Dummies Questions & Answers

Help! scrolling numbers and letters

Hello all I am a unix newbie.... I have a sun netra t1 and it is freaking out I am connected to it through a console port, and it is just spitting out a ton on numbers and letters like below its just keeps going and going. I have tried rebooting it and I cannot get it back to any kind of a... (1 Reply)
Discussion started by: intraining11
1 Replies

10. UNIX for Dummies Questions & Answers

Letters, Numbers or Alphanumerical

How do I check if a variable consisted of letters, numbers or both letters and numbers? For example, I have a variable $X and I want to print "1" if it contains only letters, "2" if it contains only numbers and "3" if it contains both (2 Replies)
Discussion started by: sleepster
2 Replies
Login or Register to Ask a Question