CGI

• Web Applications
• Common Gateway Interface
• Alternatives to CGI
• Environment Variables
• CGI Output

Web Applications (Review)

• Web applications are online applications that users can access via standard web browsers
• Distinct from web sites that only serve static content
• Web applications operate in background when you
  • fill out an online registration form (writing your registration information to a database)
  • make an online purchse (validating your credit card, logging the transaction)
  • request a stock chart for a given time range (dynamically generating the image)
• Use enabling technologies such as CGI, PHP, ASP, or Java servlets to create dynamic content

How Does a Web Application Work? (Review)

• When the web server gets a CGI request, it executes the application in a separate process
• The server passes the parameters received with the request (if any) to the application and collects its output
• This output is then returned to the client just as if it had been fetched from a static document

  

Time Server

• The following web application is implemented in Perl
• It computes the local time on the web server and returns the result to the browser

  #!/usr/local/bin/perl
  
  use strict;
  
  my $time = localtime;
  
  print "Content-Type: text/html\n\n";
  print "The current time on this server is: $time";

• See the output

Time Server (Comments)

#!/usr/local/bin/perl

• Tells the server to use the program "/usr/local/bin/perl" to interpret this script
• Run "which perl" in Linux to find out the location of the Perl interpreter

  use strict;

• Enables strict rules for variables, subroutines and references (recommended)
• Helps locate subtle errors, and typos that might not generate a syntax error
• Enforces good programming practices (e.g. declare all variables)

  print "Content-Type: text/html\n\n";

• Outputs header field declaring the MIME type of the document returned by the script
• Followed by an additional blank line to indicate that the header is complete

Invoking CGI Scripts

• Server typically configured to map a particular virtual directory such as /cgi-bin to CGI scripts
• On Linux, CGI script must be made executable

  > chmod 0755 <script>.cgi

• Fogetting this step is a common problem

Special Instructions for the Lab

• Web server for this course runs on sigma10 on port 8000
• Unfortunately, this host is not externally accessible
• HTML files and CGI scripts must be a special directory

   ~<user>/course_html/

• CGI scripts must have the extension ".cgi"
• Assuming the time server script was called "localtime.cgi", the full URL on the sigma10 server would therfore be

   http://sigma10:8000/~<user>/course_html/localtime.cgi

• You must also set the permissions for both your home directory and the "course_html" directory to "711"

   > cd
   > chmod 711 .
   > chmod 711 course_html

Reasons for Using Perl

• Perl is by far the most widely used language for CGI programming
• Easy to learn because of its resemblance to other languages
• Allows rapid development, because it is interpreted (i.e. code does not need to be compiled)
• Perl easily portable and multi-platform
• Has extremely powerful text manipulation operators
• Strict variable types not required
• Interfaces easily with external applications
• Countless open source modules for Perl available on CPAN (www.cpan.org)

Alternatives to CGI

• Several alternatives in recent years
• Build on CGI's legacy and share underlying goal
  • Repond to queries and present dynamic content
• Attempt to avoid main drawback of CGI scripts
  • A separate process is created every time the script is executed

Some of the Alternatives

• ASP
  Active Server Pages
  • Integrated into web server
  • Allows programmers to mix code and HTML on the same page instead of writing separate scripts
  • Multiple languages (VisualBasic, JavaScript)
• PHP
  PHP: Hypertext Processor (was: Personal Home Page)
  • Interpreter embedded within the web server
  • Supports code embedded in HTML pages
  • Syntax similar to Perl
• ColdFusion
  
  • HTML pages can contain tags that call ColdFusion functions
  • Developers can create their own functions
• Java Servlets
  
  • Similar to CGI scripts in that the are programs that generate documents
  • Must be compiled before they are run, and are dynamically loaded by the web server
  • JavaServer Pages (JSP) allows developers to embed Java into web pages

Environment Variables

• The web server passes information to CGI scripts through environment variables, and the standard input
• Script accesses environment variables via %ENV hash
• The following script displays information about a request

  #!/usr/local/bin/perl
  
  use strict;
  
  print <<END_OF_HTML;
  Content-Type: text/html
  
  <table border>
    <tr>
      <td>Request method
      <td>$ENV{'REQUEST_METHOD'}
    <tr>
      <td>Data from GET
      <td>$ENV{'QUERY_STRING'}
    <tr>
      <td>Data from POST
      <td>$ENV{'CONTENT_LENGTH'}
  </table>
  END_OF_HTML

• See the output

Standard Environment Variables

• Most environment variables based on HTTP header fields
• Others provide information about the web server
• Some variables are only set for certain requests
  • Content-Length header field only sent for POST
  • Corresponding environment variable is $ENV{'CONTENT_LENGTH'}

Standard Environment Variables

• REQUEST_METHOD
  Indicates the method by which the request was made
• QUERY_STRING
  Query information (a list of name-value pairs)
• CONTENT_LENGTH
  Length of the data passed to the CGI script through STDIN
• SCRIPT_NAME
  URL path of the script (e.g. /~weiss/request.cgi) being executed
• PATH_INFO
  Extra path information passed to a CGI script

More Standard Environment Variables

• REMOTE_HOST
  Remote hostname of the client making the request (could be address of a proxy between client and server)
• REMOTE_ADDR
  Remote IP address of the client making the request
• HTTP_USER_AGENT
  Name and version of client's browser
• HTTP_ACCEPT
  List of the MIME types the client can accept
• HTTP_COOKIE
  Name-value pair previously set (typically) by the server

Obtaining a List of Environment Variables

• The following short script produces a list of all environment variables specific to your web server

  #!/usr/local/bin/perl
  
  use strict;
  
  print "Content-Type: text/html\n\n";
  
  my $name;
  foreach $name (sort keys %ENV) {
    print "<p><b>$name</b><br>";
    print "$ENV{$name}";
  }

• See the output

CGI Output

• Every CGI script must write a header field, which the server expands into a full HTTP header
• CGI scripts must output one of the following headers:
  • Content-Type specifying the type of content
  • Location header specifying an URL to redirect the client to
  • Status header with a status code the server should include in the status line

Returning Content

• Most common response for CGI scripts is to return HTML
• Script must indicate to the server the MIME type of the content it is returning

   print "Content-Type: text/html\n\n"

• By specifying another MIME type you can output other types of documents
• New lines at the end of Content-Type tell the web server that this is the last header line

Forwarding to Another URL

• Reasons for forwarding:
  • If the same message is returned by multiple CGI scripts, forward them to a common document
  • Invoking another CGI script from the current script
• Print Location header with the URL of the new location

   print "Location: such_and_such_document.html\n\n";

• URL may be absolute or relative
  • Absolute URL or relative URL with relative path is sent back to the browser, which then creates another request for the new URL
  • Relative URL with full path (e.g. /index.html) produces an internal request, which is handled by the web server without talking to the browser

Status Codes

• 200 OK
  Request was processed successfully (added by web server)
• 302 Found
  URL has changed and browsers should direct all future requests to the URL provided in the Location header
• 400 Bad Request
  Browser send an invalid request (added by web server)
• 401 Unauthorized
  Requested resource is in a protected realm (handled by web server)
• 403 Forbidden
  Client is not allowed to access the requested resource (e.g. permissions not set correctly)
• 500 Internal Server Error
  Something happened on the server that cuased the transaction to fail (e.g. syntax errors)

Reading for Next Class

• Read Chapters 1 to 7 in Castro, Perl and CGI