说明： CGI接口库源代码，提供一些方便的CGI接口函数，有一定借鉴价值。 (CGI interface library source code, CGI interface to provide some convenience functions, there is a certain reference value.)
README.txt for cgi-util 2.2.1
by: Bill Kendrick
Mike Simons
and others (Other contributors listed in "CHANGES.txt")
http://www.newbreedsoftware.com/cgi-util/
June 16, 1999 - October 10, 2005
WHAT IS IT?
-----------
"cgi-util" is a library which provides a set of C functions you can
use to create Common Gateway Interface ("CGI") programs.
Simply call cgi-util's initialization function and send HTML (or any
other kind of data) out from your program and to the user's web browser.
cgi-util allows you to grab data sent to your CGI program by way of
forms or URL-encoded data. Both "POST" and "GET" methods are handled
transparently. It also supports file-upload. (NOTE: UNDER DEVELOPMENT!)
cgi-util is probably not the most feature-filled library, but it
is fast, simple and takes virtually no time to learn how to use.
INSTALLING IT
-------------
There are a number of ways you can install the cgi-util library. These
are discussed in detail in the file "INSTALL".
CREATING A CGI
--------------
To create a CGI in C using cgi-util, simply follow these simple steps:
1. Include the cgi-util header file:
#include "cgi-util.h"
or, if you installed the header file in a shared place:
#include
2. Call cgi-util's initialization function:
cgiinit();
Note: The "cgiinit()" function returns a status result. To be
safe, you should make sure no errors occured:
if (cgiinit() != CGIERR_NONE)
exit(0);
3. Send a "Content-type" value to the webserver. Since your
program's "stdout" stream is connected to your webserver, you can
just use "printf()" or "write(0,...)":
Since most CGI's generate an HTML page (versus plain text, graphics
or some other format), you'll almost always specify the "text/html"
content type:
printf("Content-type: text/html\n\n");
Note: The two "\n"'s at the end produce a blank line, which ends
the HTTP headers. If you want to specify other HTTP headers,
you of course can. For example:
printf("Content-type: text/html\n");
printf("Pragma: no-cache\n\n");
Simply remember to add that one extra "\n" character at the end
to produce the blank line to end your HTTP headers.
4. Use the "cgi_getentry..." family of functions to grab data from the
web browser. This data may be coming from an HTML form, or it may
have been URL-encoded (in other words, the URL that the browser
was directed to included the path to your CGI program, followed by
data. For example: http://site.com/path/to/your.cgi?data...")
The "cgi_getentry..." family includes:
char * cgi_getentrystr()
Returns the string value of a form field.
int cgi_getentryint()
Returns the integer value of a form field.
double cgi_getentrydouble()
Returns the double (floating point) value of a form field.
int cgi_getentrybool()
Returns the boolean value of a form field, depending on whether
the form string value is "on" or "yes" (true, which returns 1),
or "off" or "no" (false, which returns 0).
5. Send HTML data out to the browser based on the input you received.
Most of the time, you'll probably be using "printf()" to send HTML
to the browser:
printf("...\n");
...
printf("\n");
You may also wish to copy entire files to the resulting page.
For example, rather than embeding your standard navigation links
at the bottom of the CGI's page output, you can create a file
and then use on of cgi-util's "cgi_dump..." functions:
cgi_dump("file.html");
If you need to change something in the HTML later, you need only
edit the HTML file, not the C file. No recompiling need take place.
Most of the time, you'll be "dumping" HTML files which you know
will be readable, since they are technically part of your CGI
software. The standard "dump()" function will abort your CGI with
an HTML error message in the rare case that the file can't be open.
Hopefully, this will never happen except during development, at
which point, you simply correct the problem and it should never
break again.
If, however, you're trying to open a file which may or may not
be available, you can use the "cgi_dump_no_abort()" function call
which does not abort the CGI or display any HTML error message, but
instead returns an error value:
if (cgi_dump_no_abort("file.html") != CGIERR_NONE)
...
Finally, another function exists which lets you manually abort your
CGI with an HTML error message, "cgi_error()":
cgi_error("reason for error");
6. When you're done, free up any memory used by this instance of your
CGI. The "cgi_quit()" function frees any data allocated by
the "cgi_init()" function:
cgi_quit();
FUNCTION DESCRIPTIONS
---------------------
The following functions are available when using cgi-util:
int cgi_init(void)
------------------
Receives form data. It does this by checking environment variables
(which should be set by the webserver before the CGI process begins)
to determine how to get the data, and then grabbing it and creating
a list of name/value pairs.
The following global cgi-util variables are set:
int cgi_request_method
----------------------
CGIREQ_NONE if no request method was set (which should never happen),
CGIREQ_POST if POST method was used,
CGIREQ_GET if GET method was used, or
CGIREQ_UNKNOWN if some other method was used.
int cgi_num_entries
-------------------
The number of name/value pairs received, or
0 if none or an error occured.
int cgi_errno
-------------
CGIERR_NONE if no errors occured,
CGIERR_INCORRECT_TYPE if an incorrect "CONTENT_TYPE" was set by the
server or browser,
CGIERR_BAD_CONTENT_LENGTH if the "CONTENT_LENGTH" environment
variable was not set to an integer value,
CGIERR_CONTENT_LENGTH_DISCREPENCY if the "CONTENT_LENGTH" set does
not match the amount of data actually received by the CGI,
CGIERR_UNKNOWN_METHOD if the "REQUEST_METHOD" was not set to
"POST" or "GET" (the two understood by cgi-util), or
CGIERR_OUT_OF_MEMORY if space could not be allocated for the
data being received.
cgi_entry_type * cgi_entries
----------------------------
This will contain a collection of names/values/types/lengths
(the number of which is stored in "cgi_num_entries"), or
NULL if an error occured.
int cgi_content_type
--------------------
CGITYPE_UNKNOWN if an unknown "CONTENT_TYPE" was set,
CGITYPE_APPLICATION_X_WWW_FORM_URLENCODED if the "CONTENT_TYPE"
was set to "application/x-www-form-urlencoded" (the typical type),
CGITYPE_MULTIPART_FORM_DATA if the "CONTENT_TYPE" was set to
"multipart/form-data" (normal for file-upload).
The return value of "cgi_init()" is the value of "cgi_errno".
void cgi_quit(void)
-------------------
Clears memory allocated by "cgi_init()". Since any pointers which
may be freed are initially set to "NULL", and "free()" does nothing
to NULL pointers, this can be safely called even if "cgi_init()"
failed.
The following variables are set:
cgi_entry_type * cgi_entries
----------------------------
Set to NULL.
int cgi_num_entries
-------------------
Set to 0.
int cgi_errno
-------------
Set to CGIERR_NONE.
int cgi_request_method
----------------------
Set to CGIREQ_NONE.
char * query
------------
Set to NULL.
int cgi_content_type
--------------------
Set to CGITYPE_NONE.
Since the function is void, nothing is returned.
char * cgi_getentrystr(char * field_name)
-----------------------------------------
This function looks up a name/value pair named by "field_name"
(the name of some form field) and returns the value as a character
string pointer.
In other words, it goes through all "cgi_entries" (from 0 to
"cgi_num_entries - 1") and checks the "name" of each of them.
For the first one it finds (if any) that exactly matches the
string sent to this function ("field_name"), it will return the
"value".
This function returns a "NULL" if no name/value pair was found
that matched the "field_name" requested.
Example:
--------
If an HTML form has a text input box named "name":
...and someone types "john doe" into it and submits it to your
CGI, then the string you receive from the function call:
cgi_getentrystr("name");
will be the text "john doe".
char * cgi_getnentrystr(char * field_name, int n)
-------------------------------------------------
Identical to "cgi_getentrystr()", but for forms with multiple
fields with the same name. It returns the 'n'th field with
the name 'field_name'.
Use "cgi_numentries()" (see below) to determine how many (if any)
fields have the name you're interested in.
Example:
--------
If an HTML form has a number of checkboxes named "choice":
Chocolate
Vanilla
...you can receive the values ("choc" and/or "vani") for the
checked items using this in your C code:
for (i = 0; i < cgi_numentries("choice"); i++)
{
... cgi_getnentrystr("choice", i) ...
}
If "n" is larger than or equal the number of fields named
"field_name", or is less than 0, the global variable
"cgi_errno" is set to "CGIERR_N_OUT_OF_BOUNDS".
int cgi_getentryint(char * field_name)
--------------------------------------
Like "cgi_getentrystr()", this function looks for a name/value
pair named by "field_name". (In fact, this function is simply
a wrapper around "cgi_getentrystr()".)
It returns the integer value of the string value from the form.
(In other words, it would return the integer 10 if the string
was "10".)
If the name/value pair is not found or the string is not an
integer (for example, "abc" would return the integer 0),
the global variable "cgi_errno" is set to "CGIERR_NOT_INTEGER".
int cgi_getnentryint(char * field_name, int n)
----------------------------------------------
Identical to "cgi_getentryint()", but for forms with multiple
fields with the same name. See "cgi_getnentrystr()" above
and "cgi_getnumentries()" below.
int cgi_getentrydouble(char * field_name)
-----------------------------------------
This function is more or less identical to "cgi_getentryint()",
except it returns a double (floating point).
If the name/value pair is not found or the string is not an
integer (for example, "abc" would return the double 0.0),
the global variable "cgi_errno" is set to "CGIERR_NOT_DOUBLE".
int cgi_getnentrydouble(char * field_name, int n)
-------------------------------------------------
Identical to "cgi_getentrydouble()", but for forms with multiple
fields with the same name. See "cgi_getnentrystr()" above
and "cgi_getnumentries()" below.
int cgi_getentrybool(char * field_name, int def)
------------------------------------------------
This function returns a "1" or a "0" depending on whether
the value of the name/value pair you are looking for contains
the text string "on" or "yes", or "off" or "no", respectively.
If the name/value pair is not found or the string is not an
integer (for example, "abc"), the global variable "cgi_errno"
is set to "CGIERR_NOT_BOOL", and the value returned is "def"
(a default specified when the function is called).
int cgi_getnentrybool(char * field_name, int def, int n)
--------------------------------------------------------
Identical to "cgi_getentrybool()", but for forms with multiple
fields with the same name. See "cgi_getnentrystr()" above
and "cgi_getnumentries()" below.
int cgi_getnumentries(char * field_name)
-----------------------------------------
Returns how many fields were returned with the name "field_name".
Use to determine what the maximum value for "n" should be when
calling cgi_getn...() functions.
const char * cgi_getcookie(const char * cookie_name)
----------------------------------------------------
Searches the "HTTP_COOKIE" environment variable
(sent by the browser during the HTTP request) for a particular
cookie, and returns its value as a string.
If the cookie is not found, this function returns NULL.
This function sets the following global variable:
int cgi_errno
-------------
CGIERR_NONE if no errors occured.
CGIERR_NO_COOKIES if HTTP_COOKIE is not set.
CGIERR_COOKIE_NOT_FOUND if the cookie wasn't found.
CGIERR_OUT_OF_MEMORY if it could not create temporary space.
int cgi_dump_no_abort(char * filename)
--------------------------------------
This opens a file and sends it to "stdout" (file descriptor 0),
which results in the file being sent to the browser. This is a
very good way to include standard HTML headers and footers in
your CGI. Also, by not embeding the HTML into your C source,
you no longer need to recompile the CGI after editing that HTML.
Changes take effect immediately.
This function sets the following global variable:
int cgi_errno
-------------
CGIERR_NONE if no errors occured.
CGIERR_CANT_OPEN if the file could not be opened. (You should
look at the C global "errno" to determine the exact error.)
Once the file has been sent out (if it was), the function returns
with the same value it set "cgi_errno" to.
void cgi_dump(char * filename)
------------------------------
This function does more or less the same as "cgi_dump_no_abort()"
(in fact, it's actually a wrapper around that function).
However, if it cannot open the file, rather than return with an
error code, this function sends HTML text out stating that it can't
open the file, including the C "strerror()" description of the error.
It then aborts your CGI by calling "exit(0);".
Since most of the time you will be including HTML files which are
under your control (ie, they are part of the CGI package you are
creating), you will most often wish to use this function, since it
is simpler than "cgi_dump_no_abort()" and your HTML files should
always be available to your CGI.
void cgi_error(char * reason)
-----------------------------
This function displays an HTML error message, starting with the word
"Error" as a level one header ("
") and followed by the string you
provide (the actual reason an error occured).
This function then calls "exit(0);" to abort your CGI.
int cgi_goodemailaddress(char * addr)
-------------------------------------
This function does its best to make sure that a string contains
a valid-looking e-mail address. (This is useful for when you
get users who fill out a form and refuse to enter their e-mail
address, or they don't understand what their own e-mail address is.
For example, someone with the address "abc@aol.com" might think
their address is "abc" or "abc@aol" or "abc.aol.col" or even
"http://abc@aol.com". A good rule: never trust users.)
A valid e-mail address is in the form:
text@text.text[.text...]
In other words, alphanumeric characters ("a-z", "A-Z", "0-9"),
'dashes' ("-"), underscores ("_") and 'dots' ("."), with exactly one
'at' ("@") in the middle, and at least one 'dot' appearing after
the 'at.' 'Dots' and 'ats' are not allowed at the beginning or
end of the string.
"cgi_goodemailaddress" returns a "1" if the address appears to be
in the right format (this does NOT necessarily mean it is a VALID
address), or "0" if it is not.
char * cgi_strerror(int err)
----------------------------
Like C's "strerror()" function which returns the text equivalent of
an "errno" error code, this function returns the text equivalent of
one of cgi-util's "cgi_errno" error code.
ERROR CODES
-----------
The many error codes that cgi-util will set "cgi_errno" to are described
above, but here's a reference:
CGIERR_NONE
-----------
No error occured
CGIERR_NOT_INTEGER
------------------
The field looked-up by cgi_getentryint() did not exist, or its
value was not an integer.
CGIERR_NOT_DOUBLE
-----------------
The field looked-up by cgi_getentrydouble() did not exist, or its
value was not a double.
CGIERR_NOT_BOOL
---------------
The field looked-up by cgi_getentrybool() did not exist, or its
value was not "yes", "on", "no" or "off".
CGIERR_UNKNOWN_METHOD
---------------------
The "REQUEST_METHOD" environment variable was not set to
"POST" or "GET" (the two understood by cgi-util).
CGIERR_INCORRECT_TYPE
---------------------
The "CONTENT_TYPE" environment variable was not set to
"application/x-www-form-urlencoded".
CGIERR_BAD_CONTENT_LENGTH
-------------------------
The "CONTENT_LENGTH" environment variable was not set to
an integer value.
CGIERR_CONTENT_LENGTH_DISCREPANCY
---------------------------------
The "CONTENT_LENGTH" environment variable was set to a value
different from the actual size of data received by cgi_init().
CGIERR_CANT_OPEN
----------------
The cgi_dump_no_abort() function could not open the file
specified. (Check C's "errno" value for the exact reason.)
CGIERR_OUT_OF_MEMORY
--------------------
Space could not be allocated for the data being received by
cgi_init().
THE TEST PROGRAMS
-----------------
To understand the test program, first open the "test.html" or
"filetest.html" HTML file in a web browser. (You need to open it via the
HTTP protocol, not simply opening it as a file!)
With "test.html", you'll notice the following on the page:
* A type-in field labelled "Name?"
* A type-in field labelled "Age?"
... ...