Common Gateway Interface
Web Technologies
Piero Fraternali
Outline
•Architectures for dynamic content
publishing
–CGI
–Java Servlet
–Server-side scripting
–JSP tag libraries
Motivations
•Creating pages on the fly based on the user’s
request and from structured data (e.g.,
database content)
•Client-side scripting & components do not
suffice
–They manipulate an existing document/page, do
not create a new one from strutured content
•Solution:
–Server-side architectures for dynamic content
production
Common Gateway Interface
•An interface that allows the Web Server to launch
external applications that create pages dynamically
•A kind of «double client-server loop»
•Is is not
–A programming language
–A telecommunication protocol
•It is
–An interface between the web server and tha applications that
defines some standard communication variables
•The interface is implemented through system variables, a
universal mechanism present in all operating systems
•A CGI program can be written in any programming language
What CGI is/is not
Invocation
•The client specifies in the URI the name of
the program to invoke
•The program must be deployed in a
specified location at the web server (e.g.,
the cgi-bin directory)
–http://my.server.web/cgi-bin/xyz.exe
Execution
•The server recognizes from the URI that
the requested resource is an executable
–Permissions must be set in the web server for
allowing program execution
–E.g., the extensions of executable files must
be explicitly specified
•http://my.server.web/cgi-bin/xyz.exe
•The web server decodes the paramaters
sent by the client and initializes the CGI
variables
•request_method, query_string, content_length,
content_type
•http://my.server.web/cgi-bin/xyz.exe?par=val
Execution
Execution
•The server lauches the program in a new
process
Execution
•The program executes and «prints» the
response on the standard output
Execution
•The server builds the response from the
content emitted to the standard output and
sends it to the client
Handling request parameters
•Client paramaters can be sent in two ways
–With the HTTP GET method
•parameters are appended to the URL
(1)
•http://www.myserver.it/cgi-bin/xyz?par=val
–With the HTTP POST method
•Parameters are inserted as an HTTP entity in the
body of the request (when their size is substantial)
•Requires the use of HTML forms to allow users
input data onto the body of the request
– (1) The specification of HTTP does not specify any maximum URI length,
practical limits are imposed by web browser and server software
<HTML>
<BODY>
<FORM
action="http://www.mysrvr.it/cgi-bin/xyz.exe"
method=post>
<P> Tell me your name:<p>
<P><INPUT type="text"
NAME="whoareyou"> </p>
<INPUT type="submit"
VALUE="Send">
</FORM>
</BODY>
</HTML>
HTML Form
Read environment variable
Execute business logic
Print MIME heading
Print HTML markup
"Content-type: text/html"
Structure of a CGI program
Read variable
content_length
Read content_length
bytes from the
standard input
Read variable
Query_string
Read variable
Request_method
Parameter decoding
CGI development
•A CGI program can be written in any programming language:
–C/C++
–Fortran
–PERL
–TCL
–Unix shell
–Visual Basic
•In case a compiled programming language is used, the
source code must be compiled
–Normally source files are in cgi-src
–Executable binaries are in cgi-bin
•If instead an interpreted scripting language is used the source
files are deployed
–Normally in the cgi-bin folder
Overview of CGI variables
•Clustered per type:
–server
–request
–headers
Server variables
•These variables are always available, i.e.,
they do not depend on the request
–SERVER_SOFTWARE: name and version of
the server software
•Format: name/version
–SERVER_NAME: hostname or IP of the
server
–GATEWAY_INTERFACE: supported CGI
version
•Format: CGI/version
Request variables
•These variables depend on the request
–SERVER_PROTOCOL: transport protocol name
and version
•Format: protocol/version
–SERVER_PORT: port to which the request is
sent
–REQUEST_METHOD : HTTP request method
–PATH_INFO: extra path information
–PATH_TRANSLATED: translation of PATH_INFO
from virtual to physical
–SCRIPT_NAME: invoked script URL
–QUERY_STRING: the query string
Other request variables
•REMOTE_HOST: client hostname
•REMOTE_ADDR: client IP address
•AUTH_TYPE: authentication type used by
the protocol
•REMOTE_USER: username used during the
authentication
•CONTENT_TYPE : content type in case of
POST and PUT request methods
•CONTENT_LENGTH : content length
Environment variables: headers
•The HTTP headers contained in the request
are stored in the environment with the prefix
HTTP_
–HTTP_USER_AGENT: browser used for the
request
–HTTP_ACCEPT_ENCODING: encoding type
accepted by the client
–HTTP_ACCEPT_CHARSET: charset accepted
by the client
–HTTP_ACCEPT_LANGUAGE: language
accepted by the client
Problems with CGI
•Performance and security issues in web server to
application communication
•When the server receives a request, it creates a new
process in order to run the CGI program
•This requires time and significant server resources
•A CGI program cannot interact back with the web server
•The process of the CGI program is terminated when
the program finishes
•No sharing of resources between subsequen calls (e.g., reuse of
database connections)
•No main memory preservation of the user’s session (database
storage is necessary if session data are to be preserved)
•Exposing to the web the physical path to an
executable program can breach security
•CGI reference:
–http://www.w3.org/CGI/
•Security and CGI:
– http://www.w3.org/Security/Faq/index.html
Riferimenti
Form.html
Mult.c
Mult.cgi
Precedentemente
compilato in...
1. Prima
richiesta
2. Recupero
risorsa
3. Risposta
4. Seconda
richiesta
5. Set variabili
d'ambiente e
chiamata
6. Calcolo
risposta
7. Invio
risposta
Form.html
Mult.cgi
Esempio completo
La form (form.html)
<HTML>
<HEAD><TITLE>Form di
moltiplicazione</TITLE><HEAD>
<BODY>
<FORM ACTION="http://www.polimi.it/cgi-bin/run/mult.cgi">
<P>Introdurre i moltiplicandi</P>
<INPUT NAME="m" SIZE="5"><BR/>
<INPUT NAME="n" SIZE="5"><BR/>
<INPUT TYPE="SUBMIT" VALUE="Moltiplica">
</FORM>
<BODY>
</HTML>
URL
chiamata
Vista in un
browser
Lo script
#include <stdio.h>
#include <stdlib.h>
int main(void){
char *data;
long m,n;
printf("%s%c%c\n", "Content-Type:text/html;charset=iso-8859-
1",13,10);
printf("<HTML>\n<HEAD>\n<TITLE>Risultato
moltiplicazione</TITLE>\n<HEAD>\n");
printf("<BODY>\n<H3>Risultato moltiplicazione</H3>\n");
data = getenv("QUERY_STRING");
if(data == NULL)
printf("<P>Errore! Errore nel ricevere i dati dalla form.</P>\n");
else if(sscanf(data,"m=%ld&n=%ld",&m,&n)!=2)
printf("<P>Errore! Dati non validi. Devono essere numerici.</P>\n");
else
printf("<P>Risultato: %ld * %ld = %ld</P>\n",m,n,m*n);
printf("<BODY>\n");
return 0;
}
Istruzioni di
stampa della
risposta
sull'output
Recupero di
valori dalle
variabili
d'ambiente
Compilazione e test locale
•Compilazione:
$ gcc -o mult.cgi mult.c
•Test locale:
$ export QUERY_STRING="m=2&n=3"
$ ./mult.cgi
•Risultato:
Content-Type:text/html;charset=iso-8859-1
<HTML>
<HEAD>
<TITLE>Risultato moltiplicazione</TITLE>
<HEAD>
<BODY>
<H3>Risultato moltiplicazione</H3>
<P>Risultato: 2 * 3 = 6</P>
<BODY>
Set manuale della
variabile
d'ambiente
contenente la
query string
Considerazioni su CGI
•Possibili problemi di sicurezza
•Prestazioni (overhead)
–creare e terminare processi richiede tempo
–cambi di contesto richiedono tempo
•Processi CGI:
–creati a ciascuna invocazione
–non ereditano stato di processo da invocazioni
precedenti (e.g., connessioni a database)
•CGI reference:
http://hoohoo.ncsa.uiuc.edu/cgi/overview.ht
ml
•Sicurezza e CGI:
http://www.w3.org/Security/Faq/wwwsf4.ht
ml
Riferimenti