Re: gensub embeds spurious carriage return in dos-format input

[michael humphrey]

false call - there is no problem.
  ----- Original Message ----- 
  From: michael humphrey 
  To: address@hidden 
  Sent: Thursday, 02 December 2004 20:04
  Subject: gensub embeds spurious carriage return in dos-format input


  the problem occurs when the input file is dos-format and disappears when the 
input file is in unix-format.

  the problem concerns regular expression component specifiers.
  namely, a spurious carriage return ("\r") is embedded at the end of the 
replaced text called out by a component specifier.
  note that the called out text is from the middle of the input line, so the 
input line itself can not be the source of the problem.

  the console session below demonstrates the problem.
  the configuration is: intel i586/Windows 2000Pro, cygwin 1.5.5







  [console session]

  address@hidden /nth/wsr
  $ gawk -W version
  GNU Awk 3.1.3
  Copyright (C) 1989, 1991-2003 Free Software Foundation.

  address@hidden /nth/wsr
  $ cygcheck -sr
  Cygwin Win95/NT Configuration Diagnostics
  Current System Time: Thu Dec 02 18:47:21 2004
  ...
      Cygwin DLL version info:
          DLL version: 1.5.5
          DLL epoch: 19
          DLL bad signal mask: 19005
          DLL old termios: 5
          DLL malloc env: 28
          API major: 0
          API minor: 94
          Shared data: 3
          DLL identifier: cygwin1
  ...

  address@hidden /nth/wsr
  $ cat gawk-problem.awk 
  /<!-- http.* -->/ {
  # gensub embeds "\r" in result if input is dos-format
    lurl = gensub(/<!-- (http.*) -->/, "\\1", 1, $0)
  printf "KKA lurl=[%s]\n", lurl 
  # strip trailing "\r" from previous result
    nlurl = gensub(/[\n\r]*$/, "", 1, lurl)
  printf "KKA nlurl=[%s]\n", nlurl 
    }

  {
    next
    }

  address@hidden /nth/wsr
  $ cat dos-input.txt
  <!-- http://mmm.vv.nu/ee/yy -->
  <p> kkkkkkkkkkkkkk  <p> lllllllllllllll  

  address@hidden /nth/wsr
  $ gawk -f gawk-problem.awk < dos-input.txt
  ]KA lurl=[http://mmm.vv.nu/ee/yy
  KKA nlurl=[http://mmm.vv.nu/ee/yy]

  address@hidden /nth/wsr
  $ cat dos-input.txt | dos2unix > unix-input.txt

  address@hidden /nth/wsr
  $ gawk -f gawk-problem.awk < unix-input.txt
  KKA lurl=[http://mmm.vv.nu/ee/yy]
  KKA nlurl=[http://mmm.vv.nu/ee/yy]