忘记密码   免费注册 查看新帖 | 论坛精华区

ChinaUnix.net

  平台 论坛 博客 认证专区 大话IT 视频 徽章 文库 沙龙 自测 下载 频道自动化运维 虚拟化 储存备份 C/C++ PHP MySQL 嵌入式 Linux系统
最近访问板块 发新帖
查看: 160157 | 回复: 62

comp.unix.shell FAQ(转载http://home.comcast.net/~j.p.h/) [复制链接]

论坛徽章:
7
荣誉版主
日期:2011-11-23 16:44:17狮子座
日期:2014-07-24 11:00:54巨蟹座
日期:2014-07-21 19:03:10卯兔
日期:2014-08-22 13:39:09双子座
日期:2014-05-22 12:00:09卯兔
日期:2014-05-08 19:43:17子鼠
日期:2014-07-24 15:38:07
发表于 2005-05-25 21:39 |显示全部楼层
From: j.p.h@comcast.net (Joe Halpin)
Newsgroups: comp.unix.shell
Subject: comp.unix.shell FAQ - Answers to Frequently Asked Questions
Summary: This posting answers questions which come up with some frequency on comp.unix.shell. It should be read by anyone with a question about shell programming before posting questions to the newsgroup.
Followup-To: comp.unix.shell

Archive-name: unix-faq/shell/sh
Posting-Frequency: monthly
Version: $Id: cus-faq.html,v 1.11 2005/09/01 17:39:36 jhalpin Exp $
Maintainer: Joe Halpin

This FAQ list contains the answers to some Frequently Asked Questions often seen in comp.unix.shell. It spells "unix" in lower case letters to avoid arguments about whether or not Linux, FreeBSD, etc are unix. That's not the point of this FAQ, and I'm ignoring the issue.

This document as a whole is Copyright (c) 2003 Joe Halpin. It may be copied freely. Exceptions are noted in individual answers.

Suggestions, complaints, et al, should be sent to the maintainer at j.p.h@comcast.net or posted to comp.unix.shell

There are two levels of questions about shells.

One is the use of the shell itself as an interface to the operating system. For example, "how do I run a program in the background, and go on with other things?". Or "how do I setup environmental variables when I log in?".

The other level is how to write shell scripts. This often involves having the shell execute unix utilities to perform part of the work the shell script needs to accomplish, and requires knowledge of these utilities, which isn't nominally in the scope of shell programming. However, unless the question involves something other than standard unix utilities, it should be included in this FAQ.

Standard unix utilities are defined by either POSIX or the Single Unix Specification. These are now joined and are normally abbreviated as "POSIX/SUS". This specification can be found at

http://www.opengroup.org/onlinepubs/007904975/toc.htm

The man pages found on that web page define standard behavior for any given utility (or the shell itself). However, you should also check the man page on your system for any utility or shell you need to use. There isn't always a perfect correspondence between the standard and a particular implementation (in fact, I'm not sure there's any case in which they perfectly correspond).

There is also an Austin Group FAQ, which describes the standardization effort in more detail ls at

http://www.opengroup.org/austin/faq.html

Other good web sites that provide information about shells and shell programming (including OS utilities) include:

http://www.shelldorado.com/
http://cfaj.freeshell.org/shell/
http://www.faqs.org/faqs/by-newsgroup/comp/comp.unix.shell.html

This FAQ is available at
http://home.comcast.net/~j.p.h/
http://www.newsville.com/cgi-bin ... tly_Asked_Questions

---------------------------
The predictable legal stuff

---------------------------
The answers given in this FAQ list are provided with the best intentions, but they may not be accurate for any particular
shell/os. They may be completely wrong for any shell/os. If you don't test the answers, that's a bug in your procedures.

There are no guarantees for the answers or recommendations given in this document. In fact, I don't even claim to have tested any or all of them myself. Many of the answers here have been contributed by one or more regular participants in the newsgroup, who I believe to be competent (certainly more competent than I am), but THERE ARE NO GUARANTEES.

Did I really need to make that all uppercase? Hopefully not, but there are a lot of lawyers around with too much time on their hands, so I want to make it clear that THERE ARE NO GUARANTEES about the accuracy of answers in this FAQ list. This is, hopefully, an aid to people trying to learn shell programming, but it is NOT a supported product. You have to figure out for yourself whether or not the answers here work for what you're trying to do.

Under no circumstances will the maintainer of this FAQ list, or any contributors to it, be held liable for any mistakes in this
document. If the answers work for you, well and good. If not, please tell me and I'll modify them appropriately so that this will be more useful.

If you don't agree to that, don't read any farther than this. Reading beyond this point indicates your agreement.

If you do test the answers and find a problem, please send email to the maintainer (see above), so it can be corrected, or (preferably) post a question to the newsgroup so it can be discussed and corrected if there's a problem.

A number of people have contributed to this FAQ, knowingly or unknowingly. Some of the answers were taken from previous postings in the group, and other people contributed questions and answers directly to the maintainer, which you are welcome to do as well.

Among the contributors is Heiner Steven, who also provided the momentum to get this FAQ list started. He maintains a web site about shell programming that has a lot of good stuff in it.

http://www.shelldorado.com/

======================================================================

CONTENTS:

0   COPYING

0a. Glossary
    Google
    POSIX/SUS ("the standard")
    UUOC
    dotfile
    portable
    race condition
    shebang
    shells
    top-posting


0b. Notes about using echo
1.  How can I send e-mails with attached files?
2.  How can I generate random numbers in shell scripts?
3.  How can I automatically transfer files using FTP with error checking?

4.  How can I remove whitespace characters within file names?
5.  How can I automate a telnet session?
6.  How do I do date arithmetic?

7.  Why did someone tell me to RTFM?
8.  How do I create a lock file?
9. How can I convert DOS text files to unix, and vice versa?
10. How can a shell prompt be set up to change the title of xterm?

11. How do I get the exit code of cmd1 in cmd1|cmd2
12. Why do I get "script.sh: not found"
13. Why doesn't echo do what I want?

14. How do I loop through files with spaces in their name?
15. how do I change my login shell?
16. When should I use a shell instead of perl/python/ruby/tcl...
17. Why shouldn't I use csh?

18. How do I reverse a file?
19. How do I remove last n lines?
20. how do I get file size, or file modification time?

[21. How do I get a process id given a process name? Or, how do I find out if a process is still running, given a process ID?
22. How do I get a script to update my current environment?
23. how do I rename *.foo to *.bar?

24. How do I use shell variables in awk scripts
25. How do I input the user with a timeout?

[26. How do I get one character input from the user?
27. why isn't my .profile read?
28. why do I get "[5" not found in "[$1 -eq 2]"?
29. How do I exactly display the content of $var (with a \n appended).
30. How do I split a pathname into the directory and file?

31. How do I make an alias take an argument?
32. How do I deal with a file whose name begins with a weird character
33. Why do I lose the value of global variables that are set in a loop
34. How do I batch an FTP download/upload?

Appendix A: Some example scripts
Appendix B: References. These
correspond with numbers in square brackets (e.g. [1]) which may appear
in the text.

======================================================================

[ 本帖最后由 r2007 于 2005-12-3 17:41 编辑 ]

论坛徽章:
7
荣誉版主
日期:2011-11-23 16:44:17狮子座
日期:2014-07-24 11:00:54巨蟹座
日期:2014-07-21 19:03:10卯兔
日期:2014-08-22 13:39:09双子座
日期:2014-05-22 12:00:09卯兔
日期:2014-05-08 19:43:17子鼠
日期:2014-07-24 15:38:07
发表于 2005-05-25 21:41 |显示全部楼层

comp.unix.shell FAQ(转载http://home.comcast.net/~j.p.h/)

ANSWERS

0  COPYING

   Some contributors may copyright their submissions and license them
   differently than this document.

   [1] Chris F.A. Johnson. Examples marked with COPYING[1] were
       contributed by Chris F.A. Johnson. He has copyrighted these
       examples, and licensed them under the GNU General Public
       License (GPL). Copying them directly into another script will
       cause that script to also come under the GPL. For details see

       http://www.fsf.org/licenses/licenses.html

0a.Glossary

   -------------------------------
   Google

      Google is one of the search engines on the Internet. It took
      over dejanews some years ago, and now is the standard reference
      when directing someone to a past thread one some topic. This is
      a very good place to start when researching a question about
      shell programming (and just about anything else).

        http://groups.google.com/advanced_group_search

   -------------------------------
   POSIX/SUS ("the standard")

      POSIX (Portable Operating System Interface) and SUS (Single Unix
      Specification) have been joined into one standard. This is what
      people usually mean when they refer to "the standard" in
      discussions about unix. When people in this group refer to the
      POSIX shell, they are talking about the shell prescribed by this
      specification. You can find this standard at

        http://www.opengroup.org/onlinepubs/007904975/toc.htm

   -------------------------------
   UUOC
   
      This is short for "Useless use of cat". It's used to point out
      that some example script has used cat when it could have used
      redirection instead. It's more efficient to redirect input than
      it is to spawn a process to run cat. For example

        $ cat file | tr -d 'xyz'

      runs two processes, one for cat and one for tr. This is less
      efficient than

        $ tr -d 'xyz' < file

      In general, "cat file | somecommand" can be more efficiently
      replaced by "somecommand < file"

      or (especially for multi-file input)

        $ somecommand file [file ...]

      but check the man page for "somecommand" to find out if it will
      accept this syntax.

      For more details about this, as well as other things like it, see
        http://rhols66.adsl.netsonic.fi/era/unix/award.html

   -------------------------------
   dotfile

      This refers to a file which starts with '.' (a dot). These files
      are not shown in directory listings without the -a (or -A in
      newer versions of ls - check the man page on your system) option
      to ls. Often they are configuration files, subdirectories used
      by applications to store configuration files, NFS swap files, et
      al.

   -------------------------------
   portable

      The word "portable" means different things to different people,
      in different situations, which is to say, there isn't one
      definition of "portable".

      At one extreme, a portable script is one which will work under
      any shell, on any operating system. At this end of the spectrum,
      there is no such thing as a portable shell script (some
      operating systems don't even have shells). If we confine the
      operating system to unix (which would make sense since this is
      comp.unix.shell), the only truly portable scripts are those
      which make no use of built-in shell facilities or syntax, but
      which only call external utilities. For example

        echo Hello World

      would probably qualify. However, that doesn't do anyone much
      good.

      Given that there are probably few (if any) scripts which have to
      meet such a standard, a more frequent use of the word "portable"
      indicates the degree to which a script will run under different
      shells and/or different operating environments.

      For example, if you're writing an installation script for an
      application, and the platforms on which that application runs
      are defined, then the problem is pretty well bounded. The choice
      of shell is one which is available on all required platforms,
      and the syntax to be used is the smallest subset of all the
      variants of that shell on the target platforms.

      The degree to which your shell script needs to be portable has
      to be determined by you, or the requirements you've been given
      for the script.

   -------------------------------
   race condition

      This is a situation in which two entities (processes, threads,
      etc) are trying to access a shared resource, or perform the same
      action, and the result depends on the order of execution of the
      two entities.

   -------------------------------
   shebang

      This is the first line of a shell script, which indicates to the
      operating system which interpreter (shell) it should invoke to
      interpret the script. It has the form

      #!/path/to/shell [ argument ]

      where /path/to/shell might be /bin/sh, /usr/local/bin/bash, etc.

      This line is only interpreted by the operating system. That is,
      if a shell script (test.sh) is executable and run from the
      command line by typing its name and giving the script as an
      argument, as in

      $ sh test.sh

      then sh interprets test.sh. For sh, the shebang line is simply a
      comment, and is ignored.

   -------------------------------
   shells

   A number of shells are discussed in this group, including
   sh
   csh
   pdksh
   ksh88
   ksh93
   tcsh
   zsh
   rc
   es
   bash
   ash
   dash

   These (and more) are names of shells which are referenced in the
   group. A comparison of some of these is available at

     http://www.faqs.org/faqs/unix-faq/shell/shell-differences/

   However, it does not make specific the differences between
   ksh88, ksh93 and pdksh, which are not entirely compatible.

   -------------------------------
   top-posting

   A. top posting
   Q. What's the most irritating way to respond on usenet?

   Please see the following:

      http://catb.org/~esr/jargon/html/T/top-post.html
      http://www.uwasa.fi/~ts/http/quote.html
      http://members.fortunecity.com/nnqweb/
      http://www.guckes.net/mail/editing.html
      
======================================================================

[ 本帖最后由 r2007 于 2005-12-3 16:19 编辑 ]

论坛徽章:
7
荣誉版主
日期:2011-11-23 16:44:17狮子座
日期:2014-07-24 11:00:54巨蟹座
日期:2014-07-21 19:03:10卯兔
日期:2014-08-22 13:39:09双子座
日期:2014-05-22 12:00:09卯兔
日期:2014-05-08 19:43:17子鼠
日期:2014-07-24 15:38:07
发表于 2005-05-25 21:41 |显示全部楼层

comp.unix.shell FAQ(转载http://home.comcast.net/~j.p.h/)

0b. Notes about using echo

   This isn't really a FAQ, but discussions about using echo come up
   often enough that it seems reasonable to have something about it in
   the FAQ list.

   The echo command is not consistent in the handling of its arguments
   from implementation to implementation. Sometimes a string with
   backslash quoted characters will be interpreted in one way, and
   sometimes another.

   Also, if the string being echoed wasn't built into the script
   itself, then it could have shell metacharacters in it, which could
   confuse things. In cases where external input is used to build a
   string to be echoed the string typically should be quoted.

   For example

   s="a string with\na newline and\ta tab"

   Following are some results with various shells:

   ------
   bash:
   $ echo "$s"
   a string with\na newline and\ta tab
   
   $ echo -e "$s"
   a string with
   a newline and        a tab

   -------
   pdksh:
   $ echo "$s"
   a string with
   a newline and        a tab

   $ echo -e "$s"
   $ echo "$s"
   a string with
   a newline and        a tab

   --------
   ksh88:
   $ echo "$s"
   a string with
   a newline and        a tab

   $ echo -e "$s"
   -e a string with
   a newline and        a tab

   -------
   ksh93:
   $ echo "$s"
   a string with\na newline and\ta tab

   $ echo -e "$s"
   -e a string with\na newline and\ta tab

   Note that ksh93 makes the handling of arguments system dependent
   when they contain '\', and/or the first argument begins with '-'.

   http://www.cs.princeton.edu/~jlk/kornshell/doc/man93.html

   POSIX does not allow the -e option. It also makes the result of
   using -n or any string with '\' in it implementation-defined.
   However, on XSI-conforming systems, it disallows options, and
   defines the use of backslash-quoted characters.

   In general, the behavior of echo is system and/or shell dependent
   if its arguments contain a backslash, or its first argument is -n
   or -e.

   The biggest problem with echo is when using it to output strings
   that the script got externally (e.g. user input, or reading from a
   file). These strings may have '\' characters in them for
   example. In this case, results may not be what you expect.

   print is available in some shells, although printf(1) is perhaps
   more portable. Additionally, a here document will give predictable
   results in that it will not expand escape sequences.

   cat <<EOF
   $s
   EOF

   produces

   a string with\na newline and\ta tab

   So consider not using echo unless you are sure what will happen,
   given the shell you're using.   

======================================================================

[ 本帖最后由 r2007 于 2005-12-3 16:27 编辑 ]

论坛徽章:
7
荣誉版主
日期:2011-11-23 16:44:17狮子座
日期:2014-07-24 11:00:54巨蟹座
日期:2014-07-21 19:03:10卯兔
日期:2014-08-22 13:39:09双子座
日期:2014-05-22 12:00:09卯兔
日期:2014-05-08 19:43:17子鼠
日期:2014-07-24 15:38:07
发表于 2005-05-25 21:42 |显示全部楼层

comp.unix.shell FAQ(转载http://home.comcast.net/~j.p.h/)

1. How can I send e-mails with attached files?

   a. Use uuencode
   
      This is the simplest way to do this. For example

      $ uuencode surfing.jpeg surfing.jpeg | mail someone@some.where

      To send regular text as well

      $ (cat mailtext; uuencode surfing.jpeg surfing.jpeg) |
        mail someone@some.where

   b. Use MIME

      $ metasend -b -t someone@some.where -s "Hear our son!" \
        -m audio/basic -f crying.au

      These examples are adapted from
      http://www.shelldorado.com/articles/mailattachments.html which
      goes into much more detail about this.

   c. Use pine (with a patch) or mutt

      

======================================================================

2. How can I generate random numbers in shell scripts?

   This depends on the shell, and the facilities available from the
   OS.

   a. Some shells have a variable called RANDOM, which evaluates to a
      different value every time you dereference it. If your shell has
      this variable,

        $ number=$RANDOM will produce a random number.

   b. Some systems have a /dev/urandom device, which generates a
      stream of bits. This can be accessed using the dd(1) utility. An
      example of this (from a more extensive discussion of different
      techniques at http://www.shelldorado.com/scripts/cmds/rand)

        n=`dd if=/dev/urandom bs=1 count=4 2>/dev/null | od -t u4 | \
        awk 'NR==1 {print $2}'`

        also:

        od -vAn -N4 -tu4 < /dev/urandom

   c. Use a utility such as awk(1), which has random number generation
      included. This approach is the most portable between shells and
      operating systems.

        awk 'BEGIN {srand();print rand()}'

      Note that this doesn't work with older versions of awk. This
      requires a version supporting the POSIX spec for srand(). For
      example, on Solaris this will not work with /usr/bin/awk, but
      will with nawk or /usr/xpg4/bin/awk.

      Also, if you call this line more than once within the same
      second, you'll get the same number you did the previous time.

======================================================================

3. How can I automatically transfer files using FTP with error
   checking?

    First, there are tools to do that: curl, wget, lftp, ncftp. But,
    they are generally not part of the base system (you need to
    install them).
   
    zsh (version 4 and above) provides a FTP facility, see "info -f
    zsh -n 'zsh/zftp Module'"
   
    #! /usr/bin/zsh
    zftp open host user passwd || exit
    zftp get /remote/file > /local/file; r=$?
    zftp close && exit r
   
    With your system "ftp" command, two ways:
   
    1- using "ftp -n". Without the -n option, ftp expects user
    interaction to enter the password, so you'd need to use
    "expect". With "-n", you provide the user and password as any
    other FTP command.
   
    #! /bin/sh
    ftp -n << EOF
    open ftp.domain.org
    user anonymous ${LOGNAME:-`who am i`}@
    binary
    get /remote/file /local/file
    bye
    EOF
   
    The error checking can't be made correctly (if "open" fails, the
    "user" command will be still sent even if it shouldn't).
   
    2- using ~/.netrc
   
    If you put:
   
    <<
    machine ftp.domain.org
    login mylogin
    password mypasswd
    macdef init
      binary
      get /remote/file /local/file
      bye
   
   
    >>
   
    (with the trailing empty line) in your ~/.netrc (ensure it's not
    world readable) and then run "ftp ftp.domain.org", ftp will find
    the matching "machine" entry in your ~/.netrc and use the
    parameters provided there to make the ftp transaction.
   
    Those work at least on Linux, FreeBSD, Solaris, HPUX

======================================================================

[ 本帖最后由 r2007 于 2005-12-3 16:30 编辑 ]

论坛徽章:
7
荣誉版主
日期:2011-11-23 16:44:17狮子座
日期:2014-07-24 11:00:54巨蟹座
日期:2014-07-21 19:03:10卯兔
日期:2014-08-22 13:39:09双子座
日期:2014-05-22 12:00:09卯兔
日期:2014-05-08 19:43:17子鼠
日期:2014-07-24 15:38:07
发表于 2005-05-25 21:43 |显示全部楼层

comp.unix.shell FAQ(转载http://home.comcast.net/~j.p.h/)

4. How can I remove whitespace characters within file names?

   File names in unix can contain all kinds of whitespace characters,
   not just spaces. The following examples only work with spaces,
   adjust accordingly.

   a. Use the substitution capabilities of awk, sed, et al.

        f=`printf '%s\n' "$filename" | sed 's/ /_/g'`

        f=`printf '%s\n' "$filename" | awk '{gsub(" ","_");print $0}'`

        f=`printf '%s\n' "$filename" | tr ' ' _`

      Add characters to the tr command line as needed (see the man
      page for tr to find out the available escape sequences).

      Additionally (although not exactly a one-liner)

        f=`tr ' ' _ <<EOF
        $filename
        EOF
        `

      See section 0a "Notes about using echo" for why echo is not used
      here.

   b. Use the substitution capabilities of the shell if it has
      them. Check the man page for your shell (probably under a
      section named something like "Parameter expansion") to see. For
      example:

      f=${filename// /_}

      With zsh:

      autoload -U zmv
      zmv '* *' '$f:gs/ /_/'

      It should be noted that the zmv solution renames the files (call
      mv internally and adress several problems that may arise) while
      the other solutions only update a variable (and then, renaming
      the files may involve a quite complicated script to do it
      reliably).

======================================================================

5. How can I automate a telnet session?

   This is outside the realm of shell programing, per se. You need
   a more special purpose scripting language such as expect. See
   http://expect.nist.gov/

   Perl scripts can also do this with the Telnet module from CPAN.

======================================================================

6. How do I do date arithmetic?

   This depends on exactly what you have in mind.

   a. Finding yesterday's date

      The GNU version of date has some nice features in this
      respect. For example

        To find yesterday's date

          $ date --date yesterday

        To find tomorrow's date

          $ date --date tomorrow

         See the man page for GNU date for other options. It can also
         provide dates more than one day in the past/future.

       The FreeBSD version of date also provides extensions that can
       do things like this.

         $ date
         Wed Oct 22 13:48:29 CDT 2003
         $ date -v-1d
         Tue Oct 21 13:45:16 CDT 2003

       Playing with the TZ variable isn't a reliable method. If you
       need to do something like this, but don't have GNU or FreeBSD
       date available, see section g. "Arbitrary date arithmetic".

   b. Determining relative ages of files

      If you want to determine whether or not one file is older than
      another, you can (with bash, pdksh, ksh93) do

        $ [[ file1 -ot file2 ]] && echo file1 is older

      or you can use find to search a directory tree for files that
      are newer/older than some file:

        $ find . -name '*.c' -newer test.c

   c. Finding elapsed time

      If you want to find elapsed time, perhaps because you want to
      know when some operation has timed out, some shells (bash, ksh
      zsh [,??])  have a SECONDS variable which tell how many seconds
      have elapsed since the invocation of the shell, or since the
      last time it was set.

      ksh93 has a floating point SECONDS which is locale dependent.

      In zsh 4.1 and above one can be made floating point with: float
      SECONDS

      zsh 4.1 and above also has $EPOCHSECONDS for seconds since
      1970-1-1 0:0:0 UTC (see zsh/datetime module).

   d. Determining leap year

      A leap year in the Gregorian calendar is defined as a year which
      is evenly divisible by 4, however, if it's also evenly divisible
      by 100 then it's not a leap year unless it's also evenly
      divisible by 400.  It gets worse than that, actually, but this
      is as far as I go :-).

      In the Julain calendar which was used before in Europe, only the
      years divisible by 4 where leap years.

      The standard "cal" utility performed the switch between Julian
      and Gregorian calendar in september 1752 (see cal 9 1752) which
      corresponds to the date used in England. The Gregorian calendar
      (created by Pope Gregory III) was first used in 1582 in many
      other countries.

      One possibility for a ksh function to do this (after 1600 AD/CE)
      is

        isleap()
        {
          y=$1
          four=$(( $y % 4 ))
          hundred=$(( $y % 100 ))
          fourhundred=$(( $y % 400 ))
          if [ $four -eq 0 ];then
            if [ $hundred -eq 0 ];then
              if [ $fourhundred -eq 0 ];then
                echo leap year
              else
                echo not a leap year
              fi
            else
              echo leap year
           fi
         else
           echo not a leap year
         fi
        }

      Or, valid with any date with the same calendar switch day as
      POSIX cal's (POSIX syntax):

        is_leap_year() # args: year
        # NB: year before year 1 is year -1, not 0.
        {
          [ "$1" -lt 0 ] && set -- "$(($1 + 1))"
            [ "$(($1 % 4))" -eq 0 ] && {
                [ "$(($1 % 100))" -ne 0 ] || [ "$(($1 % 400))" -eq 0 ] \
                      || [ "$1" -le 1752 ]
            }
        }

      Or in any Bourne shell (see COPYING[1]):

        is_leap_year() { ## USAGE: is_leap_year [year]
          isl_year=${1:-`date +%Y`}
          case $isl_year in
            *0[48] |\
            *[2468][048] |\
            *[13579][26] |\
            *[13579][26]0|\
            *[2468][048]00 |\
            *[13579][26]00 ) _IS_LEAP_YEAR=1
               return 0 ;;
            *) _IS_LEAP_YEAR=0
               return 1 ;;
          esac
        }

      On FreeBSD, use the -f option to date(1) to pass in the
      (supposed) February 29 in the current year and then print it the
      day of the month again to see if there really is such a date
      (note that you need -j as well as -f, otherwise date(1) thinks
      you want to set the clock):

      if [ $(date -jf%Y%m%d $(date +%Y0229) +%d) = 29 ]; then
        echo Leap year!
      fi

   e. Determining the last day of a month.

      There are a number of possibilities for doing this which have
      been mentioned in the group. The following is a sampling:

      In any Bourne-type shell (in conjunction with is_leap_year() as
      given above, when month is February) (see COPYING[1]):

        days_in_month() { ## USAGE: days_in_month [month [year]]
          if [ -n "$1" ]
          then
            dim_m=$1
            dim_y=$2
          else
            eval `date "+dim_m=%m dim_y=%Y"`
          fi
          case $dim_m in
            9|09|4|04|6|06|11) _DAYS_IN_MONTH=30 ;;
            1|01|3|03|5|05|7|07|8|08|10|12) _DAYS_IN_MONTH=31 ;;
            2|02) is_leap_year ${dim_y:-`date +%Y`} &&
              _DAYS_IN_MONTH=29 || _DAYS_IN_MONTH=28 ;;
          esac
          [ ${SILENT_FUNCS:-0} -eq 1 ] || echo $_DAYS_IN_MONTH
        }


      With GNU date:

        year=2003
        month=9
        date -d "$year/$month/1 +1 month -1 day" +%d

      With FreeBSD date use the -v-1d option to date(1) to get the day
      before the first day of the next month:

        $ MONTH=12
        $ date -v-1d -jf%Y-%m-%d $(date +%Y-$(((MONTH+1)%12))-01) +%d
        31
        
      In the shell using cal (But beware of implementations of cal
      which print more than one month at a time):

        month=9 ; year=2003   # adjust
        ##
        for lday in `cal $month $year` ; do : ; done
        echo $lday

        ## or
        set -- `cal $month $year` ; eval lday=\${$#}
        echo $lday
        
      In ksh, bash and zsh:

        : $(cal)
        days_in_month=$_

      In zsh:

        days_in_month=${$(cal)[-1]}

   f. Determining the day of the week for a given date.

      This algorithm is known as Zeller's congruence. An explanation
      of it is available from the Dictionary of Algorithms and Data
      Structures web page at NIST:

        http://www.nist.gov/dads/

      Also, a fuller explanation is available at

        http://www.merlyn.demon.co.uk/zeller-c.htm#ZC

      An example in C, with a short explanation, is given at

        http://wwwcdf.pd.infn.it/MLO/Calendars/Notes.html#zeller

      A shell (ksh93) implementation of a homework assignment (given
      for illustration only - don't turn this in as yours - you might
      be sorry if it's wrong :-)

        dayofweek()
        {
          # Implementation of a homework assignment given at
          # http://carbon.cudenver.edu/~traup/fa02/lec/hw3.html
          #
          # call with day:   1 - 31
          #           month: March = 1, Jan & Feb are months 11 and
          #                  12 of the previous year.
          #           year:  The year of the century
          #           c:     The previous century
          #
          # For example, for July 4, 1989,
          #   m = 5,  d = 4,  y = 89, and c = 19,
          # while for January 25, 1989,
          #   m = 11, d = 25, y = 88, and c = 19.
          #
          # The output is the day of the week with Sunday = 0,
          # Monday = 1, etc.

          d=$1
          m=$2
          y=$3
          c=$4

          A=$(( ($m * 13 - 1) / 5 ))
          B=$(( $y / 4 ))
          C=$(( $c / 4 ))
          D=$(( $A + $B + $C + $d + $y - ($c * 2) ))
          echo $(( $D % 7 ))
        }

      On FreeBSD, use the -f option to date(1) to pass in the date of
      interest and +%A to print the day of the week:

        $ date -jf%Y-%m-%d 2000-01-01 +%A
        Saturday

      (Use +%u or +%w if you want the weekday as a number.  See the
      strftime(3) manpage for details.)

   g. Arbitrary date arithmetic

      To do arbitrary date calculations is more complicated. One
      possibility is to call an external utility, or a program in
      another scripting language, which has this built in. For
      example, perl has wrappers for the unix time functions built in,
      so it can provide some relief in this regard. C programs can
      also be easily written to do date arithmetic (see the examples
      section). One thing to keep in mind, however, is that unix time
      functions are, strictly speaking, limited to the range of time
      between January 1 1970 at midnight, and January 19, 2038 at
      3:14:07. C/Perl programs which calculate dates outside this
      range might work, or they might not, that would depend on the
      implementation.

      To do arbitrary date arithmetic in the shell itself is also
      possible. An article provided on the web by SysAdmin magazine
      describes one way to do this.

        http://www.samag.com/documents/s=8284/sam0307b/0307b.htm

      Another possibility is given in the examples section, from

        http://groups.google.com/groups? ... f%40ogion.it.jyu.fi

      See also
        http://groups.google.com/groups? ... las@spam.is.invalid

      On FreeBSD, the -f and -v options to date(1) cover most things
      you might want to do, with the caveat that only dates within the
      range mentioned above are defined. Dates outside that range are
      not guaranteed to work.

      Also, zsh 4.1 and above has the zsh/datetime module that
      provides the $EPOCHSECONDS and the strftime function.

   h. Getting the number of seconds since the epoch

    - GNU date has the %s format option which returns the epoch
      time.

    - More portably, use awk.

        awk 'BEGIN {srand(); printf("%d\n", srand())}'

      This works because srand() sets its seed value with the
      current epoch time if not given an argument. It also returns
      the previous seed value, so the second call gives the epoch
      time.

      Note that this doesn't work with older versions of awk. This
      requires a version supporting the POSIX spec for srand(). For
      example, on Solaris this will not work with /usr/bin/awk, but
      will with nawk or /usr/xpg4/bin/awk.

      Depending on scheduling, when the call is actually executed,
      etc, this might be off by a second.

    - Another way is to use perl if you have it.

        perl -le 'print time'

    - Also, zsh 4.1 and above has the zsh/datetime module that
      provides the $EPOCHSECONDS and the strftime function.

======================================================================

[ 本帖最后由 r2007 于 2005-12-3 17:44 编辑 ]

论坛徽章:
7
荣誉版主
日期:2011-11-23 16:44:17狮子座
日期:2014-07-24 11:00:54巨蟹座
日期:2014-07-21 19:03:10卯兔
日期:2014-08-22 13:39:09双子座
日期:2014-05-22 12:00:09卯兔
日期:2014-05-08 19:43:17子鼠
日期:2014-07-24 15:38:07
发表于 2005-05-25 21:43 |显示全部楼层

comp.unix.shell FAQ(转载http://home.comcast.net/~j.p.h/)

7. Why did someone tell me to RTFM?

   Because you didn't :-)

   RTFM is part of Usenet lingo, and means "Read The F-ing Manual".
   Generally people say this when someone asks a question that is
   asked so often, and is answered plainly in some relevant man page,
   that they're tired of seeing it asked.

   http://catb.org/~esr/jargon/html/R/RTFM.html

   So RTFM, and the FAQs first before asking. Also, if you're new to
   the group, search Google Groups

   http://groups.google.com/advanced_group_search

   before asking questions. And please don't post your homework
   questions to the group unless you've tried to figure them out, and
   have some specific questions. People will generally be happy to
   help you with your homework if you post what you've got and ask
   specific questions.

======================================================================

8. How do I create a lock file?

   Very carefully :-)

   The scheduler can stop one process in the middle of a non-atomic
   operation, and run another one, which wants to perform the same
   operation. The second one, having a full timeslice, might finish
   the operation. When control returns to the first process, confusion
   will reign.

   The trick is to do something atomic, so that this won't
   happen. There are a couple ways to do this. One is to create a
   directory instead of a file, the other is to create a symbolic
   link. Both operations are defined to be atomic by POSIX/SUS, by
   virtue of the fact that they both require invocation of the
   corresponding system calls, which are atomic.

   Beware of trying to create ANY kind of lock file on an NFS
   partition. NFS pretty much eliminates anything like atomicity.  If
   you're going to create a lock file, make sure you're doing it on a
   local partition, such as /tmp.

   Netscape/Mozilla uses the symbolic link method for its lockfile (in
   spite of the fact that it creates it in the user's home directory,
   which may be NFS mounted). When it starts up it creates a file
   named for the IP address of the machine it's running on, and the
   pid of the creating process. Then it tries to create a symbolic
   link named "lock", which points to that file. If this symlink
   already exists, link(2) will return an error. In a script this
   would work something like

   touch /tmp/xxx
   ln -s /tmp/xxx /tmp/lockfile 2>/dev/null
   ret=$?
   rm /tmp/xxx
   if [ $ret -ne 0 ];then
     echo lockfile already exists
     exit 1
   else
     echo success
   fi

   If you have procmail installed, another possibility is the
   lockfile(1) command that comes with it.

======================================================================

9. How can I convert DOS text files to unix, and vice versa?

    Unix text files consist of lines delimited by an LF ("line-feed")
    character (ASCII 10). DOS uses the two characters CR LF ("carriage
    return", "line feed"; ASCII 13, 10) for the same purpose.

    To convert a DOS text into unix text format, the CR characters
    (control-M) at the end of a line have to be removed. To create a
    DOS text file, the CR character should be added.

    A couple ways to remove CR characters:

      sed 's/^M$//' dos.txt > unix.txt

      tr -d '\r' < dosfile > unixfile

    To add them:   

      sed 's/$/^M/' unix.txt > dos.txt

    Note that "^M" in this case is an embedded control character, (CR,
    ASCII 13). Many shells allow embedding control characters by
    entering ^V first (control-V), resulting in the sequence

        ^V^M

    for entering "^M".

    However, zsh, bash or ksh93 allow for:

      sed $'s/$/\r/'

    There is one special case to be considered: DOS text files
    sometimes contain an explicit end-of-file character ^Z (ASCII 26,
    or octal 32), which has no correspondent character for unix text
    files, where the end-of-file condition is determined
    implicitly. To remove that as well as the CR characters:

      tr -d '\r\032' < dosfile > unixfile

    Note that sed does not understand that notation, but awk does, and
    one simple way to do the opposite conversion is

      $ awk '{printf "%s\r\n" $0}END{printf "%c", 26}' unixfile > dosfile

    This assume a not-quite-ancient awk, in practice anything
    but Solaris /bin/awk (use nawk or /usr/xpg4/bin/awk in Solaris).

    Finally, your system may come with utilities named something like
    dos2unix and unix2dos, or d2u dos2unix fromdos non-standard
    utilities, and GNU recode: recode /CRLF

======================================================================

10. How can a shell prompt be set up to change the title of xterm?

   
    Gives escape sequences for xterm. For example, to change the name
    of the current window to "XXX" (in bash), do

      $ echo -en "\033]2;XXX\007"

      or, more portably:

      $ printf '%b' '\e]2;XXX\a'

    See also "Why doesn't echo do what I want?"

[ 本帖最后由 r2007 于 2005-12-3 16:59 编辑 ]

论坛徽章:
7
荣誉版主
日期:2011-11-23 16:44:17狮子座
日期:2014-07-24 11:00:54巨蟹座
日期:2014-07-21 19:03:10卯兔
日期:2014-08-22 13:39:09双子座
日期:2014-05-22 12:00:09卯兔
日期:2014-05-08 19:43:17子鼠
日期:2014-07-24 15:38:07
发表于 2005-05-25 21:44 |显示全部楼层

comp.unix.shell FAQ(转载http://home.comcast.net/~j.p.h/)

11. How do I get the exit code of cmd1 in cmd1|cmd2

    First, note that cmd1 exit code could be non-zero and still don't
    mean an error. This happens for instance in

    cmd | head -1

    you might observe a 141 (or 269 with ksh93) exit status of cmd1,
    but it's because cmd was interrupted by a SIGPIPE signal when
    "head -1" terminated after having read one line.

    To know the exit status of the elements of a pipeline
    cmd1 | cmd2 | cmd3

    a. with zsh:

       The exit codes are provided in the pipestatus special array.
       cmd1 exit code is in $pipestatus[1], cmd3 exit code in
       $pipestatus[3], so that $? is always the same as
       $pipestatus[-1].

    b. with bash:

       The exit codes are provided in the PIPESTATUS special array.
       cmd1 exit code is in ${PIPESTATUS[0]}, cmd3 exit code in
       ${PIPESTATUS[2]}, so that $? is always the same as
       ${PIPESTATUS: -1}.

    c. with any other Bourne like shells

       You need to use a trick to pass the exit codes to the main
       shell.  You can do it using a pipe(2). Instead of running
       "cmd1", you run "cmd1; echo $?" and make sure $? makes it way
       to the shell.

       exec 3>&1
       eval `
         # now, inside the `...`, fd4 goes to the pipe
         # whose other end is read and passed to eval;
         # fd1 is the normal standard output preserved
         # the line before with exec 3>&1
         exec 4>&1 >&3 3>&-
         {
           cmd1 4>&-; echo "ec1=$?;" >&4
         } | {
           cmd2 4>&-; echo "ec2=$?;" >&4
         } | cmd3
         echo "ec3=$?;" >&4
       `

    d. with a POSIX shell

       You can use this function to make it easier:

       run() {
         j=1
         while eval "\${pipestatus_$j+:} false"; do
           unset pipestatus_$j
           j=$(($j+1))
         done
         j=1 com= k=1 l=
         for a; do
           if [ "x$a" = 'x|' ]; then
             com="$com { $l "'3>&-
                         echo "pipestatus_'$j'=$?" >&3
                       } 4>&- |'
             j=$(($j+1)) l=
           else
             l="$l \"\$$k\""
           fi
           k=$(($k+1))
         done
         com="$com $l"' 3>&- >&4 4>&-
                    echo "pipestatus_'$j'=$?"'
         exec 4>&1
         eval "$(exec 3>&1; eval "$com")"
         exec 4>&-
         j=1
         while eval "\${pipestatus_$j+:} false"; do
           eval "[ \$pipestatus_$j -eq 0 ]" || return 1
           j=$(($j+1))
         done
         return 0
       }
      
       use it as:
      
       run cmd1 \| cmd2 \| cmd3
       exit codes are in $pipestatus_1, $pipestatus_2, $pipestatus_3

======================================================================

12. Why do I get "script.sh: not found"

    a. While script starts with "#!/bin/sh" (^M issue)

       That's the kind of error that occurs when you transfer a file
       by FTP from a MS Windows machine. On those systems, the line
       separator is the CRLF sequence, while on unix the line
       separator is LF alone, CR being just another ordinary character
       (the problem is that it is an invisible one on your terminal
       (where it actually moves the cursor to the beginning of the
       line) or in most text editors or pagers).

       So, if a MSDOS line is "#!/bin/sh", when on a Unix system, it
       becomes "#!/bin/sh<CR>" (other names for <CR> are \r, \015, ^M,
       <Ctrl-M>).

       So, if you run the file as a script, the system will look in
       /bin for an interpreter named "sh<CR>", and report it doesn't
       exist.

       $ sed 'l;d;q' < script.sh
       #!/bin/sh\r$

       shows you the problem ($ marks the end of line, \r is the CR
       character).

    b. PATH issue

       Sometimes a shell is installed someplace other than /bin or
       /usr/bin. For example, a shell which was not part of the OS
       installation might be installed into /usr/local/bin. If the
       script was written on a machine which had ksh located in
       /usr/bin, but was run on a machine where ksh was located in
       /usr/local/bin, the shebang line would not resolve correctly.

       This is unlikely to occur when using sh. However, if the shell
       is bash, zsh, et al, it might be installed in different places
       on different machines.

       One way around this is to use the env command in the shebang
       line. So instead of

       #!/bin/sh

       use

       #!/usr/bin/env sh

       Of course, env might itself live in some other directory than
       /usr/bin, but it's not likely.

======================================================================

13. Why doesn't echo do what I want?

    See also section 0a "Notes about using echo"

    The echo command is not consistent from shell to shell. For
    example, some shells (bash, pdksh [,?]) use the following
    arguments

      -n suppress newline at the end of argument list
      -e interpret backslash-escaped characters
      -E disable interpretation of backslash-escaped characters, even
         on systems where interpretation is the default.

    However, pdksh also allows using \c to disable a newline at the
    end of the argument list.

    POSIX only allows \c to be used to suppress newlines, and doesn't
    accept any of the above arguments.

    ksh88 and ksh93 leave the interpretation of backslash-escaped
    characters up to the implementation.

    [descriptions of behavior of other shells welcome]

    In short, you have to know how echo works in any environment you
    choose to use it in, and its use can therefore be problemmatic. If
    available, print(1) or printf(1) would be better.

======================================================================

[ 本帖最后由 r2007 于 2005-12-3 17:01 编辑 ]

论坛徽章:
7
荣誉版主
日期:2011-11-23 16:44:17狮子座
日期:2014-07-24 11:00:54巨蟹座
日期:2014-07-21 19:03:10卯兔
日期:2014-08-22 13:39:09双子座
日期:2014-05-22 12:00:09卯兔
日期:2014-05-08 19:43:17子鼠
日期:2014-07-24 15:38:07
发表于 2005-05-25 21:45 |显示全部楼层

comp.unix.shell FAQ(转载http://home.comcast.net/~j.p.h/)

14. How do I loop through files with spaces in their name?

    So, you're going to loop through a list of files? How is this list
    stored? If it's stored as text, there probably was already an
    assumption about the characters allowed in a filename. Every
    character except '\0' (NUL) is allowed in a file path on Unix.  So
    the only way to store a list of file names in a file is to
    separate them by a '\0' character (if you don't use a quoting
    mechanism as for xargs input).

    Unfortunately most shells (except zsh) and most standard unix text
    utilities (except GNU ones) can't cope with "\0"
    characters. Moreover, many tools, like "ls", "find", "grep -l"
    output a \n separated list of files. So, if you want to
    postprocess this output, the simpler is to assume that the
    filenames don't contain newline characters (but beware that once
    you make that assumption, you can't pretend anymore your code is
    reliable (and thus can't be exploited)).

    So, if you've got a newline separated list of files in a
    list.txt file, Here are two ways to process it:

    1-

    while IFS= read -r file <&3; do
      something with "$file" # be sure to quote "$file"
    done 3< list.txt
    (if your read doesn't have the "-r" option, either make another
    assumption that filenames don't contain backslashes, or use:

    exec 3<&0
    sed 's/\\/&&/g' < list.txt |
    while IFS= read file; do
      something with "$file" <&3 3<&-
    done
    )

    2-

    IFS="
    " # set the internal field separator to the newline character
      # instead of the default "<space><tab><NL>".
   
    set -f # disable filename generation (or make the assumption that
           # filenames don't contain *, [ or ? characters (maybe more
           # depending on your shell)).
   
    for file in $(cat < list.txt); do
      something with "$file" # it's less a problem if you forget to
                             # quote $file here.
    done
   
    Now, beware that there are things you can do before building
    this list.txt. There are other ways to store filenames. For
    instance, you have the positional parameters.
   
    with:
    set -- ./*.txt
   
    you have the list of txt files in the current directory, and no
    problem with weird characters. Looping through them is just a
    matter of:
   
    for file
    do something with "$file"
    done
   
    You can also escape the separator. For instance, with
   
    find . -exec sh -c 'printf %s\\n "$1" | sed -n '"':1
      \$!{N;b1
      }
      s/|/|p/g;s/\n/|n/g;p'" '{}' '{}' \;
      
    instead of
   
    find . -print
   
    you have the same list of files except that the \n in filenames
    are changed to "|n" and the "|" to "|p". So that you're sure
    there's one filename per line and you have to convert back "|n"
    to "\n" and "|p" to "|" before referring to the file.

======================================================================

15. how do I change my login shell?

    See  http://www.faqs.org/faqs/unix-faq/shell/shell-differences

    Unless you have a very good reason to do so, do not change root's
    default login shell. By "default login shell" is meant the shell
    recorded in /etc/passwd. Note that "I login as root but don't like
    the default shell" isn't a good reason.

    The default shell for root is one which will work in single user
    mode, when only the root partition is mounted. This is one of the
    contexts root works in, and the default shell must accommodate
    this. So if you change it to a dynamically linked shell which
    depends on libraries that are not in the root partition, you're
    asking for trouble.

    The safest way of changing root's shell is to login as root and
    then

      # SHELL=/preferred/shell; export SHELL
      # exec <your preferred shell with login flag>

      e.g.

      # SHELL=/usr/bin/ksh; export SHELL
      # exec $SHELL -l

    Another possibility is to add something to root's .profile or
    .login which checks to see if the preferred shell is runnable, and
    then execs it. This is more complicated and has more pitfalls than
    simply typing "exec <shell>" when you login though. For example,
    one of the libraries that the desired shell relies on might have
    been mangled, etc. One suggestion that has been made is

      if [ -x /usr/bin/ksh ]; then
        SHELL=/usr/bin/ksh; export SHELL
        ENV=/root/.kshrc; export ENV   
        /usr/bin/ksh -l && exit
      fi

    A safer way is to try to run a command with the preferred shell
    before you try to exec it. This will lessen the possibility that
    the shell or one of the libraries it depends on has been
    corrupted, or that one of the libraries it depends on is not in
    the available mounted partitions.

      if [ -x /usr/bin/ksh ]; then
        /usr/bin/ksh -c echo >/dev/null 2>&1
        if [ $? -eq 0 ];then
           SHELL=/usr/bin/ksh; export SHELL
           ENV=/root/.kshrc; export ENV
           /usr/bin/ksh -l && exit
        fi
      fi

    Another common approach is to create another user with UID 0. For
    example, FreeBSD systems commonly create an account named toor,
    which can be setup however you like. This bypasses the
    controversy.
   
======================================================================

16. When should I use a shell instead of perl/python/ruby/tcl...

    a. Portability

       In many cases it can't be assumed that perl/python/etc are
       installed on the target machine. Many customer sites do not
       allow installation of such things. In cases like this, writing
       a shell script is more likely to be successful. In the extreme,
       writing a pure Bourne shell script is most likely to succeed.

    b. Maintainability

       If the script is one which serves some important purpose, and
       will need to be maintained after you get promoted, it's more
       likely that a maintainer can be found for a shell script than
       for other scripting languages (especially less used ones such
       as ruby, rexx, etc).

    c. Policy

       Sometimes you're just told what to use :-)

======================================================================

17. Why shouldn't I use csh?

       http://www.grymoire.com/Unix/CshTop10.txt
       http://www.grymoire.com/Unix/Csh.html#uh-0
       http://www.faqs.org/faqs/unix-faq/shell/csh-whynot/
   
======================================================================

[ 本帖最后由 r2007 于 2005-12-3 17:02 编辑 ]

论坛徽章:
7
荣誉版主
日期:2011-11-23 16:44:17狮子座
日期:2014-07-24 11:00:54巨蟹座
日期:2014-07-21 19:03:10卯兔
日期:2014-08-22 13:39:09双子座
日期:2014-05-22 12:00:09卯兔
日期:2014-05-08 19:43:17子鼠
日期:2014-07-24 15:38:07
发表于 2005-05-25 21:45 |显示全部楼层

comp.unix.shell FAQ(转载http://home.comcast.net/~j.p.h/)

18. How do I reverse a file?

    Non-standard commands to do so are GNU tac and "tail -r".  sed
    '1!G;h;$!d' is subject to sed limitation on the size of its hold
    space and is generally slow.
   
    The awk equivalent would be:
   
    awk '{l[n++]=$0}END{while(n--)print l[n]}'
    It stores the whole file in memory.
   
    The best approach in terms of efficiency portability and resource
    cosumption seems to be:
   
    cat -n | sort -rn | cut -f2-

    "cat -n" is not POSIX but appears to be fairly
    portable. Alternatives are "grep -n '^'", "awk '{print NR,$0}'".
    Also, nl can be used as

    nl -ba -d'
    '

    i.e. NL as the delimiter.

======================================================================

19. how do I remove the last n lines?


    First we need to tell the code how many lines we want to cut
    from the bottom of a file.

      X=10

    Then We can do this:

       head -n $(( $(wc -l < file ) - $X )) file >$$ \
         && cat $$ >file && rm $$

       The break down:  
       1) $(wc -l < file)
          Find out how many lines are in the file. Need to use
          redirection so wc won't print the file name.
       2) $(( $lines_in_file - $X ))
          Take the output from step one and do some math to find out
          how many lines we want to have when all is said and done.
       3) head -$lines_when_said_and_done file
           extracts all but the unwanted lines from the file,
           and >$$ puts those lines into a temp file that has
           the name of the pid of the current shell.
       4) && cat $$ > file
          if everything has worked so far then cat the temp file into
          the original file.  This is better than mv or cp because it
          insures that the permissions of the temp file do not
          override with the perms of the original file.
       5) && rm $$
          Remove the temp file.

    AWK solutions:

       awk 'NR<=(count-12)' count="`awk 'END{print NR}' file`" file

       awk 'NR>n{print a[NR%n]} {a[NR%n]=$0}' n=12 file

       awk 'BEGIN{n=12} NR>n{print a[NR%n]} {a[NR%n]=$0}' file

      Whenever a line is read, the line that came 12 lines ago is
      printed, and then overwritten with the newly read line, using an
      rolling array indexed 0..11.

      See also question 26. for information about setting awk
      variables on the command line.

    $SHELL/sed/mv solutions:

      L=`wc -l <file`
      DL=`expr $L - 11`
      sed "$DL,\$d" file

      L=`wc -l <file`
      DL=`expr $L - 12`
      sed "${DL}q" file

      sed "`expr \`wc -l <file\` - 12`q" file

      sed -n -e :a -e '1,12{N;ba' -e '}' -e 'P;N;D' file

      The last solution is basically same algorithm as the rolling
      array awk solutions, and shares with them the advantage that
      the file is only read once - they will even work in a
      pipe. There may be limitations in sed's pattern space which
      would make this unusable however.

    PERL solution:

      perl -ne' print shift @x if @x == 12; push @x, $_ ' file

    Using GNU dd:

      ls -l file.txt | {
        IFS=" "
        read z z z z sz z
        last=`tail -10 file.txt | wc -c`
        dd bs=1 seek=`expr $sz - $last` if=/dev/null of=file.txt
      }

======================================================================

20. how do I get file size, or file modification time?

    If your system has stat(1), use it. On Linux, for example:

      filesize=$(stat -c %s -- filename)

      or use cut, awk, etc on the output.

    Probably the most portable solution is to use wc

     filesize=`wc -c < "$file"`

    ls may be able to tell you what you want to know.  From the man
    page for ls we learn about "ls -l" the file mode, the number of
    links to the file, the owner name, the group name, the size of the
    file (in bytes), the timestamp, and the filename.  For the file
    size in human readable formate use the "-h" option.

    For example:

    $ ls -l timeTravel.html
    -rw-rw-r-- 1 user user 20624 Jun 19  2002 timeTravel1.html

    so to get the file size:

    $ set -- `ls -l timeTravel1.html`
    $ echo $5
    20624

    Note that ls doesn't always give the date in the same
    format. Check the man page for ls on your system if that
    matters. If you're interested in the file modification time.

    Another possibility is to use GNU ls, which has a -T option giving
    complete time information for the file, including month, day,
    hour, minute, second and year.

    See also GNU find (-printf), GNU stat, GNU date (-r) and zsh stat
    (+mtime).

    On FreeBSD 4, you can use the -lT option to ls(1) to get the full
    modification time and the -f option to date(1) to parse it, for
    example:

    $ FILE=/etc/motd
    $ date -jf'%b %d %T %Y' +%Y-%m-%dT%T \
      $(ls -lT $FILE|tr -s ' ' \\t|cut -f6-9)
    2003-09-09T16:04:06

    Adjust syntax as needed if your shell is FreeBSD sh

======================================================================

[ 本帖最后由 r2007 于 2005-12-3 17:04 编辑 ]

论坛徽章:
7
荣誉版主
日期:2011-11-23 16:44:17狮子座
日期:2014-07-24 11:00:54巨蟹座
日期:2014-07-21 19:03:10卯兔
日期:2014-08-22 13:39:09双子座
日期:2014-05-22 12:00:09卯兔
日期:2014-05-08 19:43:17子鼠
日期:2014-07-24 15:38:07
发表于 2005-05-25 21:46 |显示全部楼层

comp.unix.shell FAQ(转载http://home.comcast.net/~j.p.h/)

  1. 21. How do I get a process id given a process name? Or, how do I find
  2.     out if a process is still running, given a process ID?

  3.     There isn't a reliable way to to this portably in the shell. Some
  4.     systems reuse process ids much like file descriptors. That is,
  5.     they use the lowest numbered pid which is not currently in use
  6.     when starting a new process. That means that the pid you're
  7.     looking for is there, but might not refer to the process you think
  8.     it does.

  9.     The usual approach is to parse the output of ps, but that involves
  10.     a race condition, since the pid you find that way may not refer to
  11.     the same process when you actually do something with that
  12.     pid. There's no good way around that in a shell script though, so
  13.     be advised that you might be stepping into a trap.

  14.     One suggestion is to use pgrep if on Solaris, and 'ps h -o pid -C
  15.     $STRING' if not, and your ps supports that syntax, but neither of
  16.     those are perfect or ubiquitous.

  17.     The normal solution when writing C programs is to create a pid
  18.     file, and then lock it with fcntl(2). Then, if another program
  19.     wants to know if that program is really running, it can attempt to
  20.     gain a lock on the file. If the lock attempt fails, then it knows
  21.     the file is still running.

  22.     We don't have options in the shell like that, unless we can supply
  23.     a C program which can try the lock for the script. Even so, the
  24.     race condition described above still exists.

  25. ======================================================================

  26. 22. How do I get a script to update my current environment?

  27.     Processes in unix cannot update the environment of the process
  28.     that spawned them. Consequently you cannot run another process
  29.     normally and expect it to do that, since it will be a child of the
  30.     running process. There are a couple ways it can be done though.

  31.     a. source the script

  32.        This means that you use whatever syntax your shell has to read
  33.        the desired script into the current environment.

  34.        In Bourne derived shells (sh/ksh/bash/POSIX/etc) the syntax
  35.        would be

  36.          $ . script

  37.        In csh type shells this would be

  38.          $ source script

  39.      b. use eval

  40.         The eval command constructs a command by evaluating and then
  41.         executing a set of arguments. If those arguments evaluate to a
  42.         shell variable assignment, the current environment will be
  43.         updated. For example

  44.           --- exportFoo
  45.           #!/bin/ksh
  46.           echo export FOO=bar

  47.         If you run this like

  48.           eval "`exportFoo`"

  49.         the value of FOO will be set to 'bar' in the calling
  50.         shell. Note that the quotes are recommended as they will
  51.         preserve any whitespace that may be present in the variables
  52.         being set.

  53.         However, be aware that eval'ing a script written in another
  54.         shell could turn out to be the wrong thing to do. For example,
  55.         eval'ing this from a ksh script

  56.         #!/bin/csh
  57.         echo setenv FOO bar

  58.         Would not do what you expect. It would produce an error,
  59.         because ksh doesn't have a setenv command.

  60. ======================================================================

  61. 23. How do I rename *.foo to *.bar?

  62.     Naive examples in ksh/bash (which may or may not work many times)

  63.       $ ls *.foo | while read f;do mv "$f" "${f%.*}".bar;done

  64.     More generically

  65.       $ ls *.foo | while read f;do mv "$f" `basename "$f" .foo`.bar;done

  66.     However, these examples contain a potentially unnecessary use of
  67.     ls (ie, if the number of files is small enough to not overflow the
  68.     command line buffer), and will fail if any file names contain a
  69.     newline, or if there are leading or trailing spaces. An
  70.     alternative is:

  71.       for file in *.foo
  72.       do
  73.         mv -- "$file" "`basename -- \"$file\" .foo`.bar"
  74.       done

  75.     Also, tests for existence of files should also be incorporated,
  76.     e.g.:

  77.       for file in ./*.foo
  78.       do
  79.         newfile=`basename "$file" .foo`.bar
  80.         [ -f "$file" ] || continue
  81.         [ -f "$newfile" -o -d "$newfile" ] && continue
  82.         mv "$file" "$newfile"
  83.       done

  84.     In some linux distributions you may be able to use the rename
  85.     command

  86.       $ rename .foo .bar *

  87.     If not (Debian, for one, comes with a perl version of rename that
  88.     won't work with that command line) try

  89.       $ rename 's/.foo/.bar/' *.foo

  90.     More options, and much more discussion about this, is available
  91.     from [url]http://www.faqs.org/faqs/unix-faq/faq/part2/section-6.html[/url]

  92.     Note that for file specifications which don't match existing
  93.     files, the shell usually responds with something like "ls: *.foo:
  94.     No such file or directory", which will mess up your processing of
  95.     file names. One possibility is

  96.     #! /bin/sh
  97.     set x [*].foo ./*.foo
  98.     case "$2$3" in
  99.       "[*].foo./*.foo") ;;
  100.       *)
  101.         shift 2
  102.         for file
  103.         do
  104.           repl=`basename "$file" .foo`.bar
  105.           mv "$file" "$repl"
  106.         done;;
  107.      esac

  108.      Except that contrary to (zsh) mmv or zmv it doesn't check for
  109.      file overwriting and fails for filenames with NLs before the "."
  110.      and doesn't handle dotfiles.

  111. ======================================================================
复制代码

[ 本帖最后由 r2007 于 2006-8-11 14:30 编辑 ]
您需要登录后才可以回帖 登录 | 注册

本版积分规则

【有奖调查】AI时代如何迎接未知挑战?

人工智能一直在医疗行业扮演着重要角色,最早的专家系统和后来的基因诊断都是人工智能技术在医疗的行业探索。近年来,随着深度学习等技术的进步,人工智能在医疗行业的应用领域不断扩展,医学影像智能诊断、语音电子病历、癌症智能诊断等均已逐渐成为热门发展方向。
而作为人工智能重要推动力的深度学习技术的快速发展却亟需认知系统的强力支撑。
----------------------------------------
活动时间:2017年5月15日-6月5日

调查入口>>
  

北京皓辰网域网络信息技术有限公司. 版权所有 京ICP证:060528号 北京市公安局海淀分局网监中心备案编号:1101082001
广播电视节目制作经营许可证(京) 字第1234号 中国互联网协会会员  联系我们:
感谢所有关心和支持过ChinaUnix的朋友们 转载本站内容请注明原作者名及出处

清除 Cookies - ChinaUnix - Archiver - WAP - TOP