如何为curl命令urlencode数据?

我试图写一个bash脚本测试，需要一个参数，并通过curl发送到网站。我需要url编码的值，以确保特殊字符被正确处理。最好的方法是什么?

这是我到目前为止的基本脚本:

#!/bin/bash
host=${1:?'bad host'}
value=$2
shift
shift
curl -v -d "param=${value}" http://${host}/somepath $@

当前回答

为了完整起见，许多使用sed或awk的解决方案只翻译一组特殊的字符，因此代码大小相当大，也不翻译其他应该编码的特殊字符。

urlencode的一个安全方法是对每个字节进行编码——即使是那些允许的字节。

echo -ne 'some random\nbytes' | xxd -plain | tr -d '\n' | sed 's/\(..\)/%\1/g'

XXD在这里小心地将输入处理为字节而不是字符。

编辑:

xxd附带了Debian中的vim-common包，我只是在一个没有安装它的系统上，我不想安装它。另一种选择是使用Debian中的bsdmainutils包中的hexdump。根据下图，bsdmainutils和vim-common应该有相同的可能性被安装:

http://qa.debian.org/popcon-png.php?packages=vim-common%2Cbsdmainutils&show_installed=1&want_legend=1&want_ticks=1

但是这里有一个使用hexdump代替XXD的版本，并且允许避免tr调用:

echo -ne 'some random\nbytes' | hexdump -v -e '/1 "%02x"' | sed 's/\(..\)/%\1/g'

2011-09-21 21:10:50

其他回答

如果你想运行GET请求并使用纯curl，只需添加—得到@Jacob的解决方案。

这里有一个例子:

curl -v --get --data-urlencode "access_token=$(cat .fb_access_token)" https://graph.facebook.com/me/feed

2011-02-25 12:37:16

有什么能比javascript更好地解析url呢?

node -p "encodeURIComponent('$url')"

2017-04-12 17:42:15

awk版本的直接链接:http://www.shelldorado.com/scripts/cmds/urlencode 我用了很多年了，效果很好

:
##########################################################################
# Title      :  urlencode - encode URL data
# Author     :  Heiner Steven (heiner.steven@odn.de)
# Date       :  2000-03-15
# Requires   :  awk
# Categories :  File Conversion, WWW, CGI
# SCCS-Id.   :  @(#) urlencode  1.4 06/10/29
##########################################################################
# Description
#   Encode data according to
#       RFC 1738: "Uniform Resource Locators (URL)" and
#       RFC 1866: "Hypertext Markup Language - 2.0" (HTML)
#
#   This encoding is used i.e. for the MIME type
#   "application/x-www-form-urlencoded"
#
# Notes
#    o  The default behaviour is not to encode the line endings. This
#   may not be what was intended, because the result will be
#   multiple lines of output (which cannot be used in an URL or a
#   HTTP "POST" request). If the desired output should be one
#   line, use the "-l" option.
#
#    o  The "-l" option assumes, that the end-of-line is denoted by
#   the character LF (ASCII 10). This is not true for Windows or
#   Mac systems, where the end of a line is denoted by the two
#   characters CR LF (ASCII 13 10).
#   We use this for symmetry; data processed in the following way:
#       cat | urlencode -l | urldecode -l
#   should (and will) result in the original data
#
#    o  Large lines (or binary files) will break many AWK
#       implementations. If you get the message
#       awk: record `...' too long
#        record number xxx
#   consider using GNU AWK (gawk).
#
#    o  urlencode will always terminate it's output with an EOL
#       character
#
# Thanks to Stefan Brozinski for pointing out a bug related to non-standard
# locales.
#
# See also
#   urldecode
##########################################################################

PN=`basename "$0"`          # Program name
VER='1.4'

: ${AWK=awk}

Usage () {
    echo >&2 "$PN - encode URL data, $VER
usage: $PN [-l] [file ...]
    -l:  encode line endings (result will be one line of output)

The default is to encode each input line on its own."
    exit 1
}

Msg () {
    for MsgLine
    do echo "$PN: $MsgLine" >&2
    done
}

Fatal () { Msg "$@"; exit 1; }

set -- `getopt hl "$@" 2>/dev/null` || Usage
[ $# -lt 1 ] && Usage           # "getopt" detected an error

EncodeEOL=no
while [ $# -gt 0 ]
do
    case "$1" in
        -l) EncodeEOL=yes;;
    --) shift; break;;
    -h) Usage;;
    -*) Usage;;
    *)  break;;         # First file name
    esac
    shift
done

LANG=C  export LANG
$AWK '
    BEGIN {
    # We assume an awk implementation that is just plain dumb.
    # We will convert an character to its ASCII value with the
    # table ord[], and produce two-digit hexadecimal output
    # without the printf("%02X") feature.

    EOL = "%0A"     # "end of line" string (encoded)
    split ("1 2 3 4 5 6 7 8 9 A B C D E F", hextab, " ")
    hextab [0] = 0
    for ( i=1; i<=255; ++i ) ord [ sprintf ("%c", i) "" ] = i + 0
    if ("'"$EncodeEOL"'" == "yes") EncodeEOL = 1; else EncodeEOL = 0
    }
    {
    encoded = ""
    for ( i=1; i<=length ($0); ++i ) {
        c = substr ($0, i, 1)
        if ( c ~ /[a-zA-Z0-9.-]/ ) {
        encoded = encoded c     # safe character
        } else if ( c == " " ) {
        encoded = encoded "+"   # special handling
        } else {
        # unsafe character, encode it as a two-digit hex-number
        lo = ord [c] % 16
        hi = int (ord [c] / 16);
        encoded = encoded "%" hextab [hi] hextab [lo]
        }
    }
    if ( EncodeEOL ) {
        printf ("%s", encoded EOL)
    } else {
        print encoded
    }
    }
    END {
        #if ( EncodeEOL ) print ""
    }
' "$@"

2008-11-30 21:42:37

下面是一个使用Lua的单行转换，类似于blueyed的答案，除了所有的RFC 3986 Unreserved Characters都没有被编码(就像这个答案):

url=$(echo 'print((arg[1]:gsub("([^%w%-%.%_%~])",function(c)return("%%%02X"):format(c:byte())end)))' | lua - "$1")

此外，您可能需要确保字符串中的换行符从LF转换为CRLF，在这种情况下，您可以插入一个gsub("\r? "“\n”，“\r\n”)在百分比编码之前。

下面是一个变体，在application/x-www-form-urlencoded的非标准风格中，它执行换行规范化，并将空格编码为'+'而不是'%20'(可以使用类似的技术将其添加到Perl代码片段中)。

url=$(echo 'print((arg[1]:gsub("\r?\n", "\r\n"):gsub("([^%w%-%.%_%~ ]))",function(c)return("%%%02X"):format(c:byte())end):gsub(" ","+"))' | lua - "$1")

2011-04-29 18:32:33