sed command to swap characters

The name of the pictureThe name of the pictureThe name of the pictureClash Royale CLAN TAG#URR8PPP











up vote
5
down vote

favorite
1












My input file layout is: mm/dd/yyyy,hh:mm,other fields

I need to format it as: yyyy-mm-dd hh:mm:00,other fields



sample input:



01/02/1998,09:30,0.4571,0.4613,0.4529,0.4592,6042175
01/02/1998,09:45,0.4592,0.4613,0.4529,0.4571,9956023
01/02/1998,10:00,0.4571,0.4613,0.455,0.4613,8939555
01/02/1998,10:15,0.4613,0.4697,0.4571,0.4697,12823627
01/02/1998,10:30,0.4676,0.4969,0.4613,0.4906,28145145


sample output:



1998-01-02 09:30:00,0.4571,0.4613,0.4529,0.4592,6042175
etc...


I tried to use:



sed -r 's/(^[0-9][0-9])(/[0-9][0-9]/)(/[0-9][0-9][0-9][0-9],)/312/g









share|improve this question



























    up vote
    5
    down vote

    favorite
    1












    My input file layout is: mm/dd/yyyy,hh:mm,other fields

    I need to format it as: yyyy-mm-dd hh:mm:00,other fields



    sample input:



    01/02/1998,09:30,0.4571,0.4613,0.4529,0.4592,6042175
    01/02/1998,09:45,0.4592,0.4613,0.4529,0.4571,9956023
    01/02/1998,10:00,0.4571,0.4613,0.455,0.4613,8939555
    01/02/1998,10:15,0.4613,0.4697,0.4571,0.4697,12823627
    01/02/1998,10:30,0.4676,0.4969,0.4613,0.4906,28145145


    sample output:



    1998-01-02 09:30:00,0.4571,0.4613,0.4529,0.4592,6042175
    etc...


    I tried to use:



    sed -r 's/(^[0-9][0-9])(/[0-9][0-9]/)(/[0-9][0-9][0-9][0-9],)/312/g









    share|improve this question

























      up vote
      5
      down vote

      favorite
      1









      up vote
      5
      down vote

      favorite
      1






      1





      My input file layout is: mm/dd/yyyy,hh:mm,other fields

      I need to format it as: yyyy-mm-dd hh:mm:00,other fields



      sample input:



      01/02/1998,09:30,0.4571,0.4613,0.4529,0.4592,6042175
      01/02/1998,09:45,0.4592,0.4613,0.4529,0.4571,9956023
      01/02/1998,10:00,0.4571,0.4613,0.455,0.4613,8939555
      01/02/1998,10:15,0.4613,0.4697,0.4571,0.4697,12823627
      01/02/1998,10:30,0.4676,0.4969,0.4613,0.4906,28145145


      sample output:



      1998-01-02 09:30:00,0.4571,0.4613,0.4529,0.4592,6042175
      etc...


      I tried to use:



      sed -r 's/(^[0-9][0-9])(/[0-9][0-9]/)(/[0-9][0-9][0-9][0-9],)/312/g









      share|improve this question















      My input file layout is: mm/dd/yyyy,hh:mm,other fields

      I need to format it as: yyyy-mm-dd hh:mm:00,other fields



      sample input:



      01/02/1998,09:30,0.4571,0.4613,0.4529,0.4592,6042175
      01/02/1998,09:45,0.4592,0.4613,0.4529,0.4571,9956023
      01/02/1998,10:00,0.4571,0.4613,0.455,0.4613,8939555
      01/02/1998,10:15,0.4613,0.4697,0.4571,0.4697,12823627
      01/02/1998,10:30,0.4676,0.4969,0.4613,0.4906,28145145


      sample output:



      1998-01-02 09:30:00,0.4571,0.4613,0.4529,0.4592,6042175
      etc...


      I tried to use:



      sed -r 's/(^[0-9][0-9])(/[0-9][0-9]/)(/[0-9][0-9][0-9][0-9],)/312/g






      text-processing sed awk regular-expression perl






      share|improve this question















      share|improve this question













      share|improve this question




      share|improve this question








      edited Nov 18 at 10:04









      Rui F Ribeiro

      38.2k1475123




      38.2k1475123










      asked Jun 2 '15 at 11:58









      Karthik Appigatla

      284




      284




















          6 Answers
          6






          active

          oldest

          votes

















          up vote
          6
          down vote



          accepted










          sed -e 's/(..)/(..)/(....),(.....),(.*)/3-1-2 4:00,5/'


          Edited to include the input from the comments below:



          sed -e 's#(..).(..).(....),(.....),#3-1-2 4:00,#'





          share|improve this answer






















          • It's not necessary to capture (.*) just to add it to the end (5) of the replacement string
            – glenn jackman
            Jun 2 '15 at 14:39






          • 1




            @MikeS - he doesn't need to match slashes - they're matched. He could just use another .
            – mikeserv
            Jun 2 '15 at 14:46






          • 1




            This to me looks the simplest, though I like chaos' solution in another answer as well. One minor change I would make: When you need a lot of backslashes in the search pattern, the first character after the 's' can define something else as a separator. E.g.: sed -e 's#(..)/(..)/(....),(.....),(.*)#3-2-1 4:00,5#' (thanks @mikeserv: my text was confusing. Fixed)
            – Mike S
            Jun 2 '15 at 15:46











          • On second thought, my bad, jhilmer, if you're of the feminine persuasion. Whatever persuades you, it's a good sed script - but glenn's right about your trailing capture - it's not doing you any favors. I would drop it.
            – mikeserv
            Jun 2 '15 at 16:52

















          up vote
          3
          down vote













          That worked for me:



          sed -r 's/([0-9]2)/([0-9]2)/([0-9]4),([0-9:]5)/3-1-2 4:00/g'


          Match 2 digits (([0-9]2)), slash, 2 digits (([0-9]2)), slash, 4 digits (([0-9]4)), and then digits and : (([0-9:]5)). Replace it with the order you wish: 3-1-2 4:00 (year-month-day hour:minute:00).






          share|improve this answer
















          • 1




            I think I prefer the digit-and-count approach to the RE, if only because it's a LOT more intelligible.
            – Sobrique
            Jun 2 '15 at 15:11

















          up vote
          3
          down vote













          sed 'y|/|-|
          s/,*(.....)-*([^,]*)/2-1/
          s// 1:00/2
          ' <infile


          OUTPUT:



          1998-01-02 09:30:00,0.4571,0.4613,0.4529,0.4592,6042175
          1998-01-02 09:45:00,0.4592,0.4613,0.4529,0.4571,9956023
          1998-01-02 10:00:00,0.4571,0.4613,0.455,0.4613,8939555
          1998-01-02 10:15:00,0.4613,0.4697,0.4571,0.4697,12823627
          1998-01-02 10:30:00,0.4676,0.4969,0.4613,0.4906,28145145


          With sed you don't usually need to try so hard - it often doesn't pay to try to explicitly enumerate the matches you're looking for. Rather, it is far simpler usually just to specify a few landmarks - delimiters - and let a pattern gobble up the interim for you.



          Above sed first y/// translates / characters to - characters. Next it references the first not-comma (provided there are at least 5) character in pattern-space and and the next four characters as 1 while possibly ignoring a trailing -. It follows that on by referencing as many sequential ^not-comma chars in 2 as it might before the next occurring comma in pattern space. The result - for the first substitution - is that it puts mm-dd in 1 before matching - and then yyyy in 2. So we swap those, drop the - and insert a new one on the other side like:



          s/.../2-1/


          And last we do it again - reusing the same pattern for a different purpose. When I do:



          s// 1:00/2


          I'm instructing sed to reuse the last regexp (as signified by the // empty address), but this time to find the second occurrence of that pattern in pattern space - which does match a comma with ,* this time - it matches the comma separating this field and the last. It also matches HH:MM in 1 and (because that string is immediately followed by a comma) the '' null-string in 2. All that remains from there is to replace 1 with itself preceded by a <space> and followed by the :00 string. Both the intervening comma and the null-string are edited away.



          If you feel you would rather get a little more specific after all, though, consider how much easier it might be with just a little abstraction. The primary benefit offered by regular expressions is that they provide us a means of quickly and efficiently abstracting away a repetitive task given only a clear understanding of what makes it repetitive in the first place.



          If constructing your regexp becomes a repetitive task in and of itself, then, well... something's probably missing. One of the advantages of a simple regexp syntax, though, is that it too often makes a very good candidate for abstraction - and it is easily achieved.



          For example:



          d='[0-9][0-9]' T=$d:$d m=$d y=$d$d
          sed -E "s|($m/$d)/($y),($T)|2-1 3:00|;s|/|-|"





          share|improve this answer






















          • looks like OP wants a space after the date, not a comma
            – glenn jackman
            Jun 2 '15 at 13:54










          • @glennjackman - thanks. I didn't notice that.
            – mikeserv
            Jun 2 '15 at 14:19










          • Interesting - a bit like self-modifying code. +1
            – Peter.O
            Jun 2 '15 at 21:09

















          up vote
          2
          down vote













          I would suggest taking a slightly different approach - parse the timestamp, then spit out a formatted timestamp. And I'd use perl for this:



          #!/usr/bin/perl

          use strict;
          use warnings;

          use Time::Piece;

          my $input_format = '%m/%d/%Y,%H:%M';
          my $output_format = '%Y-%m-%d %H:%M:%S';

          while (<>)
          my ( $date, $time, @stuff ) = split(",");
          my $timestamp = Time::Piece->strptime( "$date,$time", $input_format );
          print join( ",", $timestamp->strftime($output_format), @stuff );



          Which you can reduce to a one liner thus:



          perl -MTime::Piece -lne '($date,$time,@stuff) = split; print join ( ',', Time::Piece->strptime( "$date,$time", "%m/%d/%Y,%H:%M" ) -> strftime("%Y-%m-%d %H:%M:%S"), @stuff);'


          Which with your sample data, spits out:



          1998-01-02 09:45:00,0.4592,0.4613,0.4529,0.4571,9956023
          1998-01-02 10:00:00,0.4571,0.4613,0.455,0.4613,8939555
          1998-01-02 10:15:00,0.4613,0.4697,0.4571,0.4697,12823627
          1998-01-02 10:30:00,0.4676,0.4969,0.4613,0.4906,28145145





          share|improve this answer



























            up vote
            2
            down vote













            And possible awk solution:



            awk 'BEGIN FS = OFS = ","; split($1, d, "/"); $2 = d[3] "-" d[1] "-" d[2] " " $2 ":00"; $1 = ""; for (i = 2; i < NF; i++) printf("%s", $i OFS); printf("%s", $NF ORS);' file





            share|improve this answer





























              up vote
              1
              down vote













              Use this:



              sed -n 's_^([^/]*)/([^]*)/([^,]*),([^:]*):([^,]*)_3-1-2 4:5:00_p' file.txt





              share|improve this answer






















              • To get the correct date format, it needs -1-2, not -2-1
                – Peter.O
                Jun 3 '15 at 7:56











              • @Peter.O edited..
                – heemayl
                Jun 3 '15 at 8:03










              Your Answer








              StackExchange.ready(function()
              var channelOptions =
              tags: "".split(" "),
              id: "106"
              ;
              initTagRenderer("".split(" "), "".split(" "), channelOptions);

              StackExchange.using("externalEditor", function()
              // Have to fire editor after snippets, if snippets enabled
              if (StackExchange.settings.snippets.snippetsEnabled)
              StackExchange.using("snippets", function()
              createEditor();
              );

              else
              createEditor();

              );

              function createEditor()
              StackExchange.prepareEditor(
              heartbeatType: 'answer',
              convertImagesToLinks: false,
              noModals: true,
              showLowRepImageUploadWarning: true,
              reputationToPostImages: null,
              bindNavPrevention: true,
              postfix: "",
              imageUploader:
              brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
              contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
              allowUrls: true
              ,
              onDemand: true,
              discardSelector: ".discard-answer"
              ,immediatelyShowMarkdownHelp:true
              );



              );













               

              draft saved


              draft discarded


















              StackExchange.ready(
              function ()
              StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2funix.stackexchange.com%2fquestions%2f207039%2fsed-command-to-swap-characters%23new-answer', 'question_page');

              );

              Post as a guest















              Required, but never shown

























              6 Answers
              6






              active

              oldest

              votes








              6 Answers
              6






              active

              oldest

              votes









              active

              oldest

              votes






              active

              oldest

              votes








              up vote
              6
              down vote



              accepted










              sed -e 's/(..)/(..)/(....),(.....),(.*)/3-1-2 4:00,5/'


              Edited to include the input from the comments below:



              sed -e 's#(..).(..).(....),(.....),#3-1-2 4:00,#'





              share|improve this answer






















              • It's not necessary to capture (.*) just to add it to the end (5) of the replacement string
                – glenn jackman
                Jun 2 '15 at 14:39






              • 1




                @MikeS - he doesn't need to match slashes - they're matched. He could just use another .
                – mikeserv
                Jun 2 '15 at 14:46






              • 1




                This to me looks the simplest, though I like chaos' solution in another answer as well. One minor change I would make: When you need a lot of backslashes in the search pattern, the first character after the 's' can define something else as a separator. E.g.: sed -e 's#(..)/(..)/(....),(.....),(.*)#3-2-1 4:00,5#' (thanks @mikeserv: my text was confusing. Fixed)
                – Mike S
                Jun 2 '15 at 15:46











              • On second thought, my bad, jhilmer, if you're of the feminine persuasion. Whatever persuades you, it's a good sed script - but glenn's right about your trailing capture - it's not doing you any favors. I would drop it.
                – mikeserv
                Jun 2 '15 at 16:52














              up vote
              6
              down vote



              accepted










              sed -e 's/(..)/(..)/(....),(.....),(.*)/3-1-2 4:00,5/'


              Edited to include the input from the comments below:



              sed -e 's#(..).(..).(....),(.....),#3-1-2 4:00,#'





              share|improve this answer






















              • It's not necessary to capture (.*) just to add it to the end (5) of the replacement string
                – glenn jackman
                Jun 2 '15 at 14:39






              • 1




                @MikeS - he doesn't need to match slashes - they're matched. He could just use another .
                – mikeserv
                Jun 2 '15 at 14:46






              • 1




                This to me looks the simplest, though I like chaos' solution in another answer as well. One minor change I would make: When you need a lot of backslashes in the search pattern, the first character after the 's' can define something else as a separator. E.g.: sed -e 's#(..)/(..)/(....),(.....),(.*)#3-2-1 4:00,5#' (thanks @mikeserv: my text was confusing. Fixed)
                – Mike S
                Jun 2 '15 at 15:46











              • On second thought, my bad, jhilmer, if you're of the feminine persuasion. Whatever persuades you, it's a good sed script - but glenn's right about your trailing capture - it's not doing you any favors. I would drop it.
                – mikeserv
                Jun 2 '15 at 16:52












              up vote
              6
              down vote



              accepted







              up vote
              6
              down vote



              accepted






              sed -e 's/(..)/(..)/(....),(.....),(.*)/3-1-2 4:00,5/'


              Edited to include the input from the comments below:



              sed -e 's#(..).(..).(....),(.....),#3-1-2 4:00,#'





              share|improve this answer














              sed -e 's/(..)/(..)/(....),(.....),(.*)/3-1-2 4:00,5/'


              Edited to include the input from the comments below:



              sed -e 's#(..).(..).(....),(.....),#3-1-2 4:00,#'






              share|improve this answer














              share|improve this answer



              share|improve this answer








              edited Jun 2 '15 at 21:01

























              answered Jun 2 '15 at 12:05









              jhilmer

              37913




              37913











              • It's not necessary to capture (.*) just to add it to the end (5) of the replacement string
                – glenn jackman
                Jun 2 '15 at 14:39






              • 1




                @MikeS - he doesn't need to match slashes - they're matched. He could just use another .
                – mikeserv
                Jun 2 '15 at 14:46






              • 1




                This to me looks the simplest, though I like chaos' solution in another answer as well. One minor change I would make: When you need a lot of backslashes in the search pattern, the first character after the 's' can define something else as a separator. E.g.: sed -e 's#(..)/(..)/(....),(.....),(.*)#3-2-1 4:00,5#' (thanks @mikeserv: my text was confusing. Fixed)
                – Mike S
                Jun 2 '15 at 15:46











              • On second thought, my bad, jhilmer, if you're of the feminine persuasion. Whatever persuades you, it's a good sed script - but glenn's right about your trailing capture - it's not doing you any favors. I would drop it.
                – mikeserv
                Jun 2 '15 at 16:52
















              • It's not necessary to capture (.*) just to add it to the end (5) of the replacement string
                – glenn jackman
                Jun 2 '15 at 14:39






              • 1




                @MikeS - he doesn't need to match slashes - they're matched. He could just use another .
                – mikeserv
                Jun 2 '15 at 14:46






              • 1




                This to me looks the simplest, though I like chaos' solution in another answer as well. One minor change I would make: When you need a lot of backslashes in the search pattern, the first character after the 's' can define something else as a separator. E.g.: sed -e 's#(..)/(..)/(....),(.....),(.*)#3-2-1 4:00,5#' (thanks @mikeserv: my text was confusing. Fixed)
                – Mike S
                Jun 2 '15 at 15:46











              • On second thought, my bad, jhilmer, if you're of the feminine persuasion. Whatever persuades you, it's a good sed script - but glenn's right about your trailing capture - it's not doing you any favors. I would drop it.
                – mikeserv
                Jun 2 '15 at 16:52















              It's not necessary to capture (.*) just to add it to the end (5) of the replacement string
              – glenn jackman
              Jun 2 '15 at 14:39




              It's not necessary to capture (.*) just to add it to the end (5) of the replacement string
              – glenn jackman
              Jun 2 '15 at 14:39




              1




              1




              @MikeS - he doesn't need to match slashes - they're matched. He could just use another .
              – mikeserv
              Jun 2 '15 at 14:46




              @MikeS - he doesn't need to match slashes - they're matched. He could just use another .
              – mikeserv
              Jun 2 '15 at 14:46




              1




              1




              This to me looks the simplest, though I like chaos' solution in another answer as well. One minor change I would make: When you need a lot of backslashes in the search pattern, the first character after the 's' can define something else as a separator. E.g.: sed -e 's#(..)/(..)/(....),(.....),(.*)#3-2-1 4:00,5#' (thanks @mikeserv: my text was confusing. Fixed)
              – Mike S
              Jun 2 '15 at 15:46





              This to me looks the simplest, though I like chaos' solution in another answer as well. One minor change I would make: When you need a lot of backslashes in the search pattern, the first character after the 's' can define something else as a separator. E.g.: sed -e 's#(..)/(..)/(....),(.....),(.*)#3-2-1 4:00,5#' (thanks @mikeserv: my text was confusing. Fixed)
              – Mike S
              Jun 2 '15 at 15:46













              On second thought, my bad, jhilmer, if you're of the feminine persuasion. Whatever persuades you, it's a good sed script - but glenn's right about your trailing capture - it's not doing you any favors. I would drop it.
              – mikeserv
              Jun 2 '15 at 16:52




              On second thought, my bad, jhilmer, if you're of the feminine persuasion. Whatever persuades you, it's a good sed script - but glenn's right about your trailing capture - it's not doing you any favors. I would drop it.
              – mikeserv
              Jun 2 '15 at 16:52












              up vote
              3
              down vote













              That worked for me:



              sed -r 's/([0-9]2)/([0-9]2)/([0-9]4),([0-9:]5)/3-1-2 4:00/g'


              Match 2 digits (([0-9]2)), slash, 2 digits (([0-9]2)), slash, 4 digits (([0-9]4)), and then digits and : (([0-9:]5)). Replace it with the order you wish: 3-1-2 4:00 (year-month-day hour:minute:00).






              share|improve this answer
















              • 1




                I think I prefer the digit-and-count approach to the RE, if only because it's a LOT more intelligible.
                – Sobrique
                Jun 2 '15 at 15:11














              up vote
              3
              down vote













              That worked for me:



              sed -r 's/([0-9]2)/([0-9]2)/([0-9]4),([0-9:]5)/3-1-2 4:00/g'


              Match 2 digits (([0-9]2)), slash, 2 digits (([0-9]2)), slash, 4 digits (([0-9]4)), and then digits and : (([0-9:]5)). Replace it with the order you wish: 3-1-2 4:00 (year-month-day hour:minute:00).






              share|improve this answer
















              • 1




                I think I prefer the digit-and-count approach to the RE, if only because it's a LOT more intelligible.
                – Sobrique
                Jun 2 '15 at 15:11












              up vote
              3
              down vote










              up vote
              3
              down vote









              That worked for me:



              sed -r 's/([0-9]2)/([0-9]2)/([0-9]4),([0-9:]5)/3-1-2 4:00/g'


              Match 2 digits (([0-9]2)), slash, 2 digits (([0-9]2)), slash, 4 digits (([0-9]4)), and then digits and : (([0-9:]5)). Replace it with the order you wish: 3-1-2 4:00 (year-month-day hour:minute:00).






              share|improve this answer












              That worked for me:



              sed -r 's/([0-9]2)/([0-9]2)/([0-9]4),([0-9:]5)/3-1-2 4:00/g'


              Match 2 digits (([0-9]2)), slash, 2 digits (([0-9]2)), slash, 4 digits (([0-9]4)), and then digits and : (([0-9:]5)). Replace it with the order you wish: 3-1-2 4:00 (year-month-day hour:minute:00).







              share|improve this answer












              share|improve this answer



              share|improve this answer










              answered Jun 2 '15 at 12:04









              chaos

              34.7k773115




              34.7k773115







              • 1




                I think I prefer the digit-and-count approach to the RE, if only because it's a LOT more intelligible.
                – Sobrique
                Jun 2 '15 at 15:11












              • 1




                I think I prefer the digit-and-count approach to the RE, if only because it's a LOT more intelligible.
                – Sobrique
                Jun 2 '15 at 15:11







              1




              1




              I think I prefer the digit-and-count approach to the RE, if only because it's a LOT more intelligible.
              – Sobrique
              Jun 2 '15 at 15:11




              I think I prefer the digit-and-count approach to the RE, if only because it's a LOT more intelligible.
              – Sobrique
              Jun 2 '15 at 15:11










              up vote
              3
              down vote













              sed 'y|/|-|
              s/,*(.....)-*([^,]*)/2-1/
              s// 1:00/2
              ' <infile


              OUTPUT:



              1998-01-02 09:30:00,0.4571,0.4613,0.4529,0.4592,6042175
              1998-01-02 09:45:00,0.4592,0.4613,0.4529,0.4571,9956023
              1998-01-02 10:00:00,0.4571,0.4613,0.455,0.4613,8939555
              1998-01-02 10:15:00,0.4613,0.4697,0.4571,0.4697,12823627
              1998-01-02 10:30:00,0.4676,0.4969,0.4613,0.4906,28145145


              With sed you don't usually need to try so hard - it often doesn't pay to try to explicitly enumerate the matches you're looking for. Rather, it is far simpler usually just to specify a few landmarks - delimiters - and let a pattern gobble up the interim for you.



              Above sed first y/// translates / characters to - characters. Next it references the first not-comma (provided there are at least 5) character in pattern-space and and the next four characters as 1 while possibly ignoring a trailing -. It follows that on by referencing as many sequential ^not-comma chars in 2 as it might before the next occurring comma in pattern space. The result - for the first substitution - is that it puts mm-dd in 1 before matching - and then yyyy in 2. So we swap those, drop the - and insert a new one on the other side like:



              s/.../2-1/


              And last we do it again - reusing the same pattern for a different purpose. When I do:



              s// 1:00/2


              I'm instructing sed to reuse the last regexp (as signified by the // empty address), but this time to find the second occurrence of that pattern in pattern space - which does match a comma with ,* this time - it matches the comma separating this field and the last. It also matches HH:MM in 1 and (because that string is immediately followed by a comma) the '' null-string in 2. All that remains from there is to replace 1 with itself preceded by a <space> and followed by the :00 string. Both the intervening comma and the null-string are edited away.



              If you feel you would rather get a little more specific after all, though, consider how much easier it might be with just a little abstraction. The primary benefit offered by regular expressions is that they provide us a means of quickly and efficiently abstracting away a repetitive task given only a clear understanding of what makes it repetitive in the first place.



              If constructing your regexp becomes a repetitive task in and of itself, then, well... something's probably missing. One of the advantages of a simple regexp syntax, though, is that it too often makes a very good candidate for abstraction - and it is easily achieved.



              For example:



              d='[0-9][0-9]' T=$d:$d m=$d y=$d$d
              sed -E "s|($m/$d)/($y),($T)|2-1 3:00|;s|/|-|"





              share|improve this answer






















              • looks like OP wants a space after the date, not a comma
                – glenn jackman
                Jun 2 '15 at 13:54










              • @glennjackman - thanks. I didn't notice that.
                – mikeserv
                Jun 2 '15 at 14:19










              • Interesting - a bit like self-modifying code. +1
                – Peter.O
                Jun 2 '15 at 21:09














              up vote
              3
              down vote













              sed 'y|/|-|
              s/,*(.....)-*([^,]*)/2-1/
              s// 1:00/2
              ' <infile


              OUTPUT:



              1998-01-02 09:30:00,0.4571,0.4613,0.4529,0.4592,6042175
              1998-01-02 09:45:00,0.4592,0.4613,0.4529,0.4571,9956023
              1998-01-02 10:00:00,0.4571,0.4613,0.455,0.4613,8939555
              1998-01-02 10:15:00,0.4613,0.4697,0.4571,0.4697,12823627
              1998-01-02 10:30:00,0.4676,0.4969,0.4613,0.4906,28145145


              With sed you don't usually need to try so hard - it often doesn't pay to try to explicitly enumerate the matches you're looking for. Rather, it is far simpler usually just to specify a few landmarks - delimiters - and let a pattern gobble up the interim for you.



              Above sed first y/// translates / characters to - characters. Next it references the first not-comma (provided there are at least 5) character in pattern-space and and the next four characters as 1 while possibly ignoring a trailing -. It follows that on by referencing as many sequential ^not-comma chars in 2 as it might before the next occurring comma in pattern space. The result - for the first substitution - is that it puts mm-dd in 1 before matching - and then yyyy in 2. So we swap those, drop the - and insert a new one on the other side like:



              s/.../2-1/


              And last we do it again - reusing the same pattern for a different purpose. When I do:



              s// 1:00/2


              I'm instructing sed to reuse the last regexp (as signified by the // empty address), but this time to find the second occurrence of that pattern in pattern space - which does match a comma with ,* this time - it matches the comma separating this field and the last. It also matches HH:MM in 1 and (because that string is immediately followed by a comma) the '' null-string in 2. All that remains from there is to replace 1 with itself preceded by a <space> and followed by the :00 string. Both the intervening comma and the null-string are edited away.



              If you feel you would rather get a little more specific after all, though, consider how much easier it might be with just a little abstraction. The primary benefit offered by regular expressions is that they provide us a means of quickly and efficiently abstracting away a repetitive task given only a clear understanding of what makes it repetitive in the first place.



              If constructing your regexp becomes a repetitive task in and of itself, then, well... something's probably missing. One of the advantages of a simple regexp syntax, though, is that it too often makes a very good candidate for abstraction - and it is easily achieved.



              For example:



              d='[0-9][0-9]' T=$d:$d m=$d y=$d$d
              sed -E "s|($m/$d)/($y),($T)|2-1 3:00|;s|/|-|"





              share|improve this answer






















              • looks like OP wants a space after the date, not a comma
                – glenn jackman
                Jun 2 '15 at 13:54










              • @glennjackman - thanks. I didn't notice that.
                – mikeserv
                Jun 2 '15 at 14:19










              • Interesting - a bit like self-modifying code. +1
                – Peter.O
                Jun 2 '15 at 21:09












              up vote
              3
              down vote










              up vote
              3
              down vote









              sed 'y|/|-|
              s/,*(.....)-*([^,]*)/2-1/
              s// 1:00/2
              ' <infile


              OUTPUT:



              1998-01-02 09:30:00,0.4571,0.4613,0.4529,0.4592,6042175
              1998-01-02 09:45:00,0.4592,0.4613,0.4529,0.4571,9956023
              1998-01-02 10:00:00,0.4571,0.4613,0.455,0.4613,8939555
              1998-01-02 10:15:00,0.4613,0.4697,0.4571,0.4697,12823627
              1998-01-02 10:30:00,0.4676,0.4969,0.4613,0.4906,28145145


              With sed you don't usually need to try so hard - it often doesn't pay to try to explicitly enumerate the matches you're looking for. Rather, it is far simpler usually just to specify a few landmarks - delimiters - and let a pattern gobble up the interim for you.



              Above sed first y/// translates / characters to - characters. Next it references the first not-comma (provided there are at least 5) character in pattern-space and and the next four characters as 1 while possibly ignoring a trailing -. It follows that on by referencing as many sequential ^not-comma chars in 2 as it might before the next occurring comma in pattern space. The result - for the first substitution - is that it puts mm-dd in 1 before matching - and then yyyy in 2. So we swap those, drop the - and insert a new one on the other side like:



              s/.../2-1/


              And last we do it again - reusing the same pattern for a different purpose. When I do:



              s// 1:00/2


              I'm instructing sed to reuse the last regexp (as signified by the // empty address), but this time to find the second occurrence of that pattern in pattern space - which does match a comma with ,* this time - it matches the comma separating this field and the last. It also matches HH:MM in 1 and (because that string is immediately followed by a comma) the '' null-string in 2. All that remains from there is to replace 1 with itself preceded by a <space> and followed by the :00 string. Both the intervening comma and the null-string are edited away.



              If you feel you would rather get a little more specific after all, though, consider how much easier it might be with just a little abstraction. The primary benefit offered by regular expressions is that they provide us a means of quickly and efficiently abstracting away a repetitive task given only a clear understanding of what makes it repetitive in the first place.



              If constructing your regexp becomes a repetitive task in and of itself, then, well... something's probably missing. One of the advantages of a simple regexp syntax, though, is that it too often makes a very good candidate for abstraction - and it is easily achieved.



              For example:



              d='[0-9][0-9]' T=$d:$d m=$d y=$d$d
              sed -E "s|($m/$d)/($y),($T)|2-1 3:00|;s|/|-|"





              share|improve this answer














              sed 'y|/|-|
              s/,*(.....)-*([^,]*)/2-1/
              s// 1:00/2
              ' <infile


              OUTPUT:



              1998-01-02 09:30:00,0.4571,0.4613,0.4529,0.4592,6042175
              1998-01-02 09:45:00,0.4592,0.4613,0.4529,0.4571,9956023
              1998-01-02 10:00:00,0.4571,0.4613,0.455,0.4613,8939555
              1998-01-02 10:15:00,0.4613,0.4697,0.4571,0.4697,12823627
              1998-01-02 10:30:00,0.4676,0.4969,0.4613,0.4906,28145145


              With sed you don't usually need to try so hard - it often doesn't pay to try to explicitly enumerate the matches you're looking for. Rather, it is far simpler usually just to specify a few landmarks - delimiters - and let a pattern gobble up the interim for you.



              Above sed first y/// translates / characters to - characters. Next it references the first not-comma (provided there are at least 5) character in pattern-space and and the next four characters as 1 while possibly ignoring a trailing -. It follows that on by referencing as many sequential ^not-comma chars in 2 as it might before the next occurring comma in pattern space. The result - for the first substitution - is that it puts mm-dd in 1 before matching - and then yyyy in 2. So we swap those, drop the - and insert a new one on the other side like:



              s/.../2-1/


              And last we do it again - reusing the same pattern for a different purpose. When I do:



              s// 1:00/2


              I'm instructing sed to reuse the last regexp (as signified by the // empty address), but this time to find the second occurrence of that pattern in pattern space - which does match a comma with ,* this time - it matches the comma separating this field and the last. It also matches HH:MM in 1 and (because that string is immediately followed by a comma) the '' null-string in 2. All that remains from there is to replace 1 with itself preceded by a <space> and followed by the :00 string. Both the intervening comma and the null-string are edited away.



              If you feel you would rather get a little more specific after all, though, consider how much easier it might be with just a little abstraction. The primary benefit offered by regular expressions is that they provide us a means of quickly and efficiently abstracting away a repetitive task given only a clear understanding of what makes it repetitive in the first place.



              If constructing your regexp becomes a repetitive task in and of itself, then, well... something's probably missing. One of the advantages of a simple regexp syntax, though, is that it too often makes a very good candidate for abstraction - and it is easily achieved.



              For example:



              d='[0-9][0-9]' T=$d:$d m=$d y=$d$d
              sed -E "s|($m/$d)/($y),($T)|2-1 3:00|;s|/|-|"






              share|improve this answer














              share|improve this answer



              share|improve this answer








              edited Jun 3 '15 at 3:24

























              answered Jun 2 '15 at 12:42









              mikeserv

              45k565152




              45k565152











              • looks like OP wants a space after the date, not a comma
                – glenn jackman
                Jun 2 '15 at 13:54










              • @glennjackman - thanks. I didn't notice that.
                – mikeserv
                Jun 2 '15 at 14:19










              • Interesting - a bit like self-modifying code. +1
                – Peter.O
                Jun 2 '15 at 21:09
















              • looks like OP wants a space after the date, not a comma
                – glenn jackman
                Jun 2 '15 at 13:54










              • @glennjackman - thanks. I didn't notice that.
                – mikeserv
                Jun 2 '15 at 14:19










              • Interesting - a bit like self-modifying code. +1
                – Peter.O
                Jun 2 '15 at 21:09















              looks like OP wants a space after the date, not a comma
              – glenn jackman
              Jun 2 '15 at 13:54




              looks like OP wants a space after the date, not a comma
              – glenn jackman
              Jun 2 '15 at 13:54












              @glennjackman - thanks. I didn't notice that.
              – mikeserv
              Jun 2 '15 at 14:19




              @glennjackman - thanks. I didn't notice that.
              – mikeserv
              Jun 2 '15 at 14:19












              Interesting - a bit like self-modifying code. +1
              – Peter.O
              Jun 2 '15 at 21:09




              Interesting - a bit like self-modifying code. +1
              – Peter.O
              Jun 2 '15 at 21:09










              up vote
              2
              down vote













              I would suggest taking a slightly different approach - parse the timestamp, then spit out a formatted timestamp. And I'd use perl for this:



              #!/usr/bin/perl

              use strict;
              use warnings;

              use Time::Piece;

              my $input_format = '%m/%d/%Y,%H:%M';
              my $output_format = '%Y-%m-%d %H:%M:%S';

              while (<>)
              my ( $date, $time, @stuff ) = split(",");
              my $timestamp = Time::Piece->strptime( "$date,$time", $input_format );
              print join( ",", $timestamp->strftime($output_format), @stuff );



              Which you can reduce to a one liner thus:



              perl -MTime::Piece -lne '($date,$time,@stuff) = split; print join ( ',', Time::Piece->strptime( "$date,$time", "%m/%d/%Y,%H:%M" ) -> strftime("%Y-%m-%d %H:%M:%S"), @stuff);'


              Which with your sample data, spits out:



              1998-01-02 09:45:00,0.4592,0.4613,0.4529,0.4571,9956023
              1998-01-02 10:00:00,0.4571,0.4613,0.455,0.4613,8939555
              1998-01-02 10:15:00,0.4613,0.4697,0.4571,0.4697,12823627
              1998-01-02 10:30:00,0.4676,0.4969,0.4613,0.4906,28145145





              share|improve this answer
























                up vote
                2
                down vote













                I would suggest taking a slightly different approach - parse the timestamp, then spit out a formatted timestamp. And I'd use perl for this:



                #!/usr/bin/perl

                use strict;
                use warnings;

                use Time::Piece;

                my $input_format = '%m/%d/%Y,%H:%M';
                my $output_format = '%Y-%m-%d %H:%M:%S';

                while (<>)
                my ( $date, $time, @stuff ) = split(",");
                my $timestamp = Time::Piece->strptime( "$date,$time", $input_format );
                print join( ",", $timestamp->strftime($output_format), @stuff );



                Which you can reduce to a one liner thus:



                perl -MTime::Piece -lne '($date,$time,@stuff) = split; print join ( ',', Time::Piece->strptime( "$date,$time", "%m/%d/%Y,%H:%M" ) -> strftime("%Y-%m-%d %H:%M:%S"), @stuff);'


                Which with your sample data, spits out:



                1998-01-02 09:45:00,0.4592,0.4613,0.4529,0.4571,9956023
                1998-01-02 10:00:00,0.4571,0.4613,0.455,0.4613,8939555
                1998-01-02 10:15:00,0.4613,0.4697,0.4571,0.4697,12823627
                1998-01-02 10:30:00,0.4676,0.4969,0.4613,0.4906,28145145





                share|improve this answer






















                  up vote
                  2
                  down vote










                  up vote
                  2
                  down vote









                  I would suggest taking a slightly different approach - parse the timestamp, then spit out a formatted timestamp. And I'd use perl for this:



                  #!/usr/bin/perl

                  use strict;
                  use warnings;

                  use Time::Piece;

                  my $input_format = '%m/%d/%Y,%H:%M';
                  my $output_format = '%Y-%m-%d %H:%M:%S';

                  while (<>)
                  my ( $date, $time, @stuff ) = split(",");
                  my $timestamp = Time::Piece->strptime( "$date,$time", $input_format );
                  print join( ",", $timestamp->strftime($output_format), @stuff );



                  Which you can reduce to a one liner thus:



                  perl -MTime::Piece -lne '($date,$time,@stuff) = split; print join ( ',', Time::Piece->strptime( "$date,$time", "%m/%d/%Y,%H:%M" ) -> strftime("%Y-%m-%d %H:%M:%S"), @stuff);'


                  Which with your sample data, spits out:



                  1998-01-02 09:45:00,0.4592,0.4613,0.4529,0.4571,9956023
                  1998-01-02 10:00:00,0.4571,0.4613,0.455,0.4613,8939555
                  1998-01-02 10:15:00,0.4613,0.4697,0.4571,0.4697,12823627
                  1998-01-02 10:30:00,0.4676,0.4969,0.4613,0.4906,28145145





                  share|improve this answer












                  I would suggest taking a slightly different approach - parse the timestamp, then spit out a formatted timestamp. And I'd use perl for this:



                  #!/usr/bin/perl

                  use strict;
                  use warnings;

                  use Time::Piece;

                  my $input_format = '%m/%d/%Y,%H:%M';
                  my $output_format = '%Y-%m-%d %H:%M:%S';

                  while (<>)
                  my ( $date, $time, @stuff ) = split(",");
                  my $timestamp = Time::Piece->strptime( "$date,$time", $input_format );
                  print join( ",", $timestamp->strftime($output_format), @stuff );



                  Which you can reduce to a one liner thus:



                  perl -MTime::Piece -lne '($date,$time,@stuff) = split; print join ( ',', Time::Piece->strptime( "$date,$time", "%m/%d/%Y,%H:%M" ) -> strftime("%Y-%m-%d %H:%M:%S"), @stuff);'


                  Which with your sample data, spits out:



                  1998-01-02 09:45:00,0.4592,0.4613,0.4529,0.4571,9956023
                  1998-01-02 10:00:00,0.4571,0.4613,0.455,0.4613,8939555
                  1998-01-02 10:15:00,0.4613,0.4697,0.4571,0.4697,12823627
                  1998-01-02 10:30:00,0.4676,0.4969,0.4613,0.4906,28145145






                  share|improve this answer












                  share|improve this answer



                  share|improve this answer










                  answered Jun 2 '15 at 13:05









                  Sobrique

                  3,759517




                  3,759517




















                      up vote
                      2
                      down vote













                      And possible awk solution:



                      awk 'BEGIN FS = OFS = ","; split($1, d, "/"); $2 = d[3] "-" d[1] "-" d[2] " " $2 ":00"; $1 = ""; for (i = 2; i < NF; i++) printf("%s", $i OFS); printf("%s", $NF ORS);' file





                      share|improve this answer


























                        up vote
                        2
                        down vote













                        And possible awk solution:



                        awk 'BEGIN FS = OFS = ","; split($1, d, "/"); $2 = d[3] "-" d[1] "-" d[2] " " $2 ":00"; $1 = ""; for (i = 2; i < NF; i++) printf("%s", $i OFS); printf("%s", $NF ORS);' file





                        share|improve this answer
























                          up vote
                          2
                          down vote










                          up vote
                          2
                          down vote









                          And possible awk solution:



                          awk 'BEGIN FS = OFS = ","; split($1, d, "/"); $2 = d[3] "-" d[1] "-" d[2] " " $2 ":00"; $1 = ""; for (i = 2; i < NF; i++) printf("%s", $i OFS); printf("%s", $NF ORS);' file





                          share|improve this answer














                          And possible awk solution:



                          awk 'BEGIN FS = OFS = ","; split($1, d, "/"); $2 = d[3] "-" d[1] "-" d[2] " " $2 ":00"; $1 = ""; for (i = 2; i < NF; i++) printf("%s", $i OFS); printf("%s", $NF ORS);' file






                          share|improve this answer














                          share|improve this answer



                          share|improve this answer








                          edited Jun 2 '15 at 21:13

























                          answered Jun 2 '15 at 12:18









                          taliezin

                          6,77011527




                          6,77011527




















                              up vote
                              1
                              down vote













                              Use this:



                              sed -n 's_^([^/]*)/([^]*)/([^,]*),([^:]*):([^,]*)_3-1-2 4:5:00_p' file.txt





                              share|improve this answer






















                              • To get the correct date format, it needs -1-2, not -2-1
                                – Peter.O
                                Jun 3 '15 at 7:56











                              • @Peter.O edited..
                                – heemayl
                                Jun 3 '15 at 8:03














                              up vote
                              1
                              down vote













                              Use this:



                              sed -n 's_^([^/]*)/([^]*)/([^,]*),([^:]*):([^,]*)_3-1-2 4:5:00_p' file.txt





                              share|improve this answer






















                              • To get the correct date format, it needs -1-2, not -2-1
                                – Peter.O
                                Jun 3 '15 at 7:56











                              • @Peter.O edited..
                                – heemayl
                                Jun 3 '15 at 8:03












                              up vote
                              1
                              down vote










                              up vote
                              1
                              down vote









                              Use this:



                              sed -n 's_^([^/]*)/([^]*)/([^,]*),([^:]*):([^,]*)_3-1-2 4:5:00_p' file.txt





                              share|improve this answer














                              Use this:



                              sed -n 's_^([^/]*)/([^]*)/([^,]*),([^:]*):([^,]*)_3-1-2 4:5:00_p' file.txt






                              share|improve this answer














                              share|improve this answer



                              share|improve this answer








                              edited Jun 3 '15 at 8:02

























                              answered Jun 2 '15 at 12:09









                              heemayl

                              34k370100




                              34k370100











                              • To get the correct date format, it needs -1-2, not -2-1
                                – Peter.O
                                Jun 3 '15 at 7:56











                              • @Peter.O edited..
                                – heemayl
                                Jun 3 '15 at 8:03
















                              • To get the correct date format, it needs -1-2, not -2-1
                                – Peter.O
                                Jun 3 '15 at 7:56











                              • @Peter.O edited..
                                – heemayl
                                Jun 3 '15 at 8:03















                              To get the correct date format, it needs -1-2, not -2-1
                              – Peter.O
                              Jun 3 '15 at 7:56





                              To get the correct date format, it needs -1-2, not -2-1
                              – Peter.O
                              Jun 3 '15 at 7:56













                              @Peter.O edited..
                              – heemayl
                              Jun 3 '15 at 8:03




                              @Peter.O edited..
                              – heemayl
                              Jun 3 '15 at 8:03

















                               

                              draft saved


                              draft discarded















































                               


                              draft saved


                              draft discarded














                              StackExchange.ready(
                              function ()
                              StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2funix.stackexchange.com%2fquestions%2f207039%2fsed-command-to-swap-characters%23new-answer', 'question_page');

                              );

                              Post as a guest















                              Required, but never shown





















































                              Required, but never shown














                              Required, but never shown












                              Required, but never shown







                              Required, but never shown

































                              Required, but never shown














                              Required, but never shown












                              Required, but never shown







                              Required, but never shown






                              Popular posts from this blog

                              How to check contact read email or not when send email to Individual?

                              Displaying single band from multi-band raster using QGIS

                              How many registers does an x86_64 CPU actually have?