cgiparse

CGIPARSE(8)

DACS Web Services and CGI

CGIPARSE(8)

NAME

cgiparse — CGI parameter parsing utility

SYNOPSIS

cgiparse [mode] [-enc { none | url | mime | dacs }] [-in filename] [-checkdup] [-d] [-duperror]
[-nodups] [-nonewline] [-qs query-string] [-copy filename] [[-n name filename]...]

DESCRIPTION

This program is part of the DACS suite. It is a stand-alone program that neither accepts the usual DACS command line options (dacsoptions) nor accesses any DACS configuration files.

This utility is used by web-based scripts (shell scripts in particular) to obtain their CGI parameters, which can be obtained from a URI's query component or in an encoded entity-body read from the standard input (as with the POST method). The form content types application/x-www-form-urlencoded and multipart/form-data are both understood.

The program has several different modes of operation, one of which may be specified by the first command line argument.

cgiparse combines query parameters found in the QUERY_STRING environment variable with parameters found in the message body it reads from the standard input. RFC 3875 states (S4.1.7) that the query string value is case-sensitive.

Duplicate parameter names are allowed by default; see -nodups and -duperror, which override duplicate handling described below.

OPTIONS

The mode may be exactly one of the following:

-arg variable-name

Emit the value of the CGI parameter variable-name, then exit. If there is no such parameter, the exit status will be 1 instead of 0. If more than one instance of variable-name is present, only one will be considered.

-checkdup

Check if any parameter name occurs more than once, then terminate. If a duplicate is found, the exit status will be 1, otherwise 0.

-targ variable-name

Test if the CGI parameter variable-name exists. If there is no such parameter, the exit status will be 1, otherwise it will be 0.

-html

Emit an HTML document that lists the CGI parameter names and their values. All instances of duplicate parameter names are output.

-one

Emit a listing of the CGI parameter values (without the names). All parameter values are output, including those associated with duplicate parameter names.

-sh --shell

Emit CGI parameters as a single line in the format:

variable-name='variable-value'; [...]

It is an error if any variable-name or variable-value is syntactically unsuitable for this format. The returned string can be used as the argument to eval to set the CGI parameters as shell variables. All parameters are output, including duplicates, in which case a variable will be assigned the value from the parameter instance that happens to appear last in the list.

-text

Like -html except emit text. This is the default. In this mode, the program's stdout is usually written to a file. Each line of the file has the format:

variable-name variable-value

A space separates the name from the corresponding value. The file is typically read by a script to obtain the parameters, or cgiparse can be run with the -in flag to retrieve a parameter. All instances of duplicate parameter names are output.

--version

Print version information to stderr and exit.

Additionally, cgiparse recognizes these options and modifiers:

[-enc { url | mime | dacs | none }]

If writing the parsed CGI parameters (-text), encode the parameter value using the specified method:

url: Selects URL encoding.
mime: Selects MIME base-64 encoding.
dacs: Selects DACS base-64 encoding.
none: Indicates that no encoding is performed (use this only when you are sure this cannot cause a problem).

For details about these encodings, please see dacs.exprs(5). The default is none. If reading the parsed CGI parameters (-in), decode the parameter values using the specified method. The default is none, which means that no decoding is performed; if the parameters were encoded, they will be returned in that encoding, but other than this case the decoding method must match the encoding method previously used or an error is likely to occur.

-qs query-string

Instead of using the environment variable QUERY_STRING to get a query component, use query-string.

-nonewline

With -arg, do not emit a newline after printing a parameter value.

-nodups

If a duplicate parameter name is read, all but one (arbitrary) instance will be discarded.

-duperror

If a duplicate parameter name is read, processing terminates immediately.

-d

Enable debugging output.

-copy filename

Append the input stream to filename. This can be useful for debugging purposes.

-in filename

Instead of parsing CGI parameters, read variable name/value pairs (in the format produced by the -text flag) from filename. If filename is "-", stdin is read.

-n name filename

If parsing succeeds, and there is a MIME body part with a name exactly matching name, then:

if the content disposition is multipart/form-data, write the content as quoted-printable text to filename;
if the content disposition is base64, write the decoded content to filename;
otherwise the content is written verbatim to filename.

If the output file exists it is truncated.

EXAMPLES

The following shell script demonstrates one way of using cgiparse.

#! /bin/sh

tmpfile=/tmp/cgiparse.$$

cgiparse > ${tmpfile}
chmod 0600 ${tmpfile}

echo "Context-Type: text/plain"
echo ""

done=
while [ "${done}x" = x ]
do
  a=
  b=
  read a b
  if [ $? = 1 ]
  then
    done=1
    break
  else
    echo "Arg: ${a}"
    echo "Is: ${b}"
  fi
done < ${tmpfile}

rm -f ${tmpfile}
exit 0

The following code fragment uses cgiparse to save and then look up its CGI parameters:

#! /bin/sh

tmpfile=/tmp/cgiparse.$$
trap 'rm -f ${tmpfile}; exit 1' EXIT 1 2 3 13 15

cgiparse -enc mime > ${tmpfile}
chmod 0600 ${tmpfile}

mode=`cgiparse -in ${tmpfile} -enc mime -arg MODE`
target=`cgiparse -in ${tmpfile} -enc mime -arg TARGET`

The following script will print "1 2 3" to its standard output:

#! /bin/sh

args=`cgiparse -sh -qs "a=1&b=2&c=3"`
eval "$args"
echo "$a $b $c"

DIAGNOSTICS

The program exits 0 if everything was fine, 1 if an error occurred.

BUGS

There do not appear to be any official recommendations concerning how to handle apparently "malformed" CGI query strings that do not look like a sequence of name=value pairs. The parsing routines that cgiparse uses will flag an error if they see strings containing a component like "=foo", for example, although "foo=" is fine.

The manner in which duplicate CGI parameters is handled is not standardized and context-specific. cgiparse could do a little better in this respect.