Using FlagShip on Web

This document describes how to use FlagShip for html Web document purposes. It tries not to be a detailed manual for configuration of the Web Server or browser, nor a full reference to the HTML language, but gives a general overview sufficient for the programmer to use FlagShip as a database manager for HTML documents. The examples are hold very simple and are meant as a basis for your own programs even without or with a basic HTML knowledge only.

As opposite to standard applications, which are usually screen oriented, runs locally and reacts on user entries, a Web based application is client-server oriented :

The client is your Web browser (Internet Explorer, Netscape, Mozilla, Firefox, Opera etc), running on local computer. The local browser access HTML document (located on the Web server) containing "form tags". These form tags contains either pre-defined data or variable user entries and selections. One of these tags specifies how the user request (usually processed by click on the "submit" button) should be handled. For example, the tag <FORM ACTION="/cgi-bin/my_application" METHOD="get"> will invoke the executable named "my_application" located in the directory "/cgi-bin" on the web-server and will pass current form data to the application using "get" method.
   

The server application (named here e.g. my_application) is started automatically by the <FORM ACTION=...> tag in HTML document, when the user press the "submit" button. The application is either a script, or a by FlagShip created executable, processing following three steps:

  • Receive the form/user data by "get" or "post" method, e.g. by invoking FlagShip's WebGetFormData() function
  • Check the user input and react accordingly by creating a "reply" HTML document, which is then automatically displayed by the client's browser. You may e.g. access database (located on the server) and create reply by using FlagShip's function WebHtmlBegin() ... create the text body by using standard ? or Qout(...) commands/functions ... and then invoke WebHtmlEnd() function to finish the reply document
  • Quit the application

Since the Web application is server-based, and run in background, consider following rules:

  • It must be compiled for the Web-server platform, regardless the used client/browser environment. I.e. if the Web server is Linux, you need to compile your application by FlagShip for Linux, or if the server is SUN compile by FlagShip for SUN Solaris, or if the server is Windows compile by FlagShip for MS-Windows, and so on.
  • Even when FlagShip supports dynamically linked applications, best to compile statically (by using the -stat compiler switch in VFS), you don't then need anything else on the server than the executable application self (and probably used databases)
  • Do not use any screen oriented input/output or user prompt (e.g. Inkey(), Wait etc) nor GUI or Terminal environment, but simple "basic" output to create the reply document. With VFS, use the -io=b compiler switch.

Hint: a short overview about the used Common Gateway Interface (or CGI) is e.g. here, a good introduction is available in several books, e.g. in the O'Reilly Open Book "CGI Programming on the World Wide Web" by Shishir Gundavaram
 

1. Data Structure on the Linux/Unix Server

1.a. Directory for HTML documents (Document root directory)

The html documents (*.html or *.htm) accessed by the Web browser (Internet Explorer, Netscape, Mosaic, Mozilla, Firefox, Opera etc.) usually resides in a special directory on the Web server (e.g.in /usr/local/httpd/htdocs or /var/local/public/html/docs etc). The Web server (e.g. Apache, Netscape-server, AOLserver etc.) hides (maps) the true directory structure so, that the Web browser will see this directory (named "Document root directory") as "root", i.e. the / path only. Hence, the Web user (usually of the group "nobody") can access files and subdirectories located below, but not above this Document root directory (an important security aspect). The links from a html document to another one on the same server may be done relatively, e.g. "/other.html" instead of the URL (Uniform Resource Locator) naming convention "http://www.domain.org/other.html", which reduce the Internet transfer and increases the access speed significantly.

Configuration: The name of the document root directory is specified in the /etc/httpd/access.conf and /etc/httpd/srm.conf (or similarly named) configuration files used by the Web server daemon and described in detail in the server documentation. On a remote Web server, contact your provider for the setup details.

Permissions: The document root directory should have a "read" access for everybody (e.g. 755, drwxr-xr-x). The html documents itself (*.html or *.htm) have usually only read permission (644, -rw-r--r--).
 

1.b. Directory for CGI Scripts and Executables

The Common Gateway Interface, or CGI, is a means for the HTTP server to "talk" to programs (scripts, executables) on the local or server's machine. As opposite to the static model of reading predefined html documents, the CGI allows to react on the client request and creates specific text or html documents for the browser. The requests from the browser to CGI are passed via environment variables (the most important are QUERY_STRING and PATH_INFO) and/or via stdin, depending on the request type. The QUERY_STRING is filled when the preferred "GET" method is used. The request is in the form <location>?<content> (e.g. "http://www.domain.org/cgi-bin/myapplic?CUSTOMERID=1234" or simply "/cgi-bin/myapplic?CUSTOMERID=1234" )

That script directory contains scripts, executables and data accessed by the HTML documents indirectly via the CGI interface, e.g. by the <FORM ...> tag. This directory will contain the by FlagShip created executable among the databases and indices.

For security reasons, this CGI directory should not be equivalent to, or reside below the Document root directory. Usually, this directory is named cgi-bin and reside parallel to the Document root directory, e.g. /usr/local/httpd/cgi-bin  or  /var/local/public/html/cgi-bin. Keep in mind, your applicattion will usually see this cgi-bin directory as root path "/", you hence can access data and subdirectories in and below cgi-bin, but not in another directories on the server.

Configuration: The location and its mapping to the Web browser (e.g. visible as /cgi-bin) is specified in the above named Web server configuration files. You should also change the properties of the directory to Options none, AllowOveride none in the access.conf file. On a remote Web server, contact your provider for the setup details.

Permissions: The cgi-bin directory may have only execute, or execute and read permissions for everybody (711, drwx--x--x or 755, drwxr-xr-x). The scripts or executables have usually read & execute permission for everybody (755, -rwxr-xr-x), while the databases and indices are r/w (666, -rw-rw-rw-).
 

2. Access the database executable via a HTML document

2.a. The HTML document

The most common (and on some commercial servers the only) method to access a CGI script or executable is via the <FORM ...> tag.  Here an example named e.g. "example1.html" posted in the document directory (see 1.a) and invoked by the browser as URL http://www.my.org/example1.html  or locally  http://localhost/example1.html . Note: if you invoke file:/usr/local/httpd/htdocs/example1.html instead, you will see and can enter the data in the the document, but will most probably not be able to access the CGI directory properly. Note the name of the shell script "example1.sh" posted in the CGI directory (see 1.b).

2.b The interface script (optional)

Usually, you will invoke directly the executable posted in the /cgi-bin directory. In special cases, you may also use shell script, e.g. the "/cgi-bin/example1.sh" (invoked in 2.a via the commented-out statement <FORM ACTION...> as alternative 1) which sets the required environment and invokes the by FlagShip compiled executable named "example1" :
Don't forget to make the script executable, i.e. "chmod 755 example1.sh"

2.c. The Web application/executable

The following FlagShip program is very simple. It checks the incoming request and prints the html reply-data on stdout. Here, we get a customer ID (via command line for test purposes, or passed by the Web server in the environment variable QUERY_STRING which contain here the string "CUSTOMERID=<input-value>" according to the NAME parameter in the SETECT (or INPUT) tag.

Note: this example use the simple GET method. For larger documents, usually the POST method is used, which does not restrict the amount of passed data. For processing both GET and POST methods, there is a function WebGetFormData() available in FlagShip and used in this example. It determines the used request form, decodes the passed data and returns an array, which is then simply processed by your application. There are also several additional Web*() functions for handling the HTML documents and sending e-mails, see details in the section FUN of the on-line FlagShip manual named "fsman", or in the corresponding section of fthe pdf manual. You may decode the data manually without Web*() functions, see the full source in fs4web.tar. If you wish to pass HTML documents or bitmaps links in the reply from your application, use full qualified URLs like http://www.mydomain.com/mydocument.html

With Visual FlagShip, compile on the server in basic i/o mode mode by
   FlagShip -stat -io=b example1.prg -o example1
With FlagShip version 4.4x, use the example1.prg source in fs4web.tar and compile by
   FlagShip example1.prg -o example1

and post the executable into the /cgi-bin directory (see 1.b).

Note: of course, the executable posted on the Web server (e.g. at your provider site) needs to be created by a FlagShip version corresponding to this operating system. Since this may differ to the locally used FlagShip, check for details of the operating system with your provider.

You may invoke it from the command-line by
   ./example1 1234
It displays a HTML document but first reports en error "database not available...".
 

2.d Database maintenance

We have to create a small maintenance program first (or use e.g. dbu). Often the main program checks for the existence of database and indices and creates it when required. But for security purposes on the server, it is better to maintain it separately by the web administrator which has sufficient access rights, or maintain it locally and then copy the database back to Web server.

With Visual FlagShip VFS6 and VFS7, compile on the server in GUI or hybrid mode
   FlagShip -stat mainten1.prg -o mainten1
or for Terminal i/o mode when accessing it via ssh or terminal emulator
   FlagShip -stat -io=t mainten1.prg -o mainten1

With FlagShip version 4.48, use
   FlagShip mainten1.prg -o mainten1

On Linux/Unix based Web server, set the access rights for the web administrator only
   chmod 700 mainten1
and invoke it to create (or modify) the database and index
   ./mainten1
After the first invocation,  change the permissions of the database and index:
   chmod 666 example1.dbf example1.idx

Check the output of the invocation
   ./example1 1
it should display a HTML document with two lines of data from the database.

If everything is ok, you're now ready to use the document example1.html (see 2.a) by your preferred Web browser. If something fails, check the access rights.

3. Additional hints

 

Created 08-May-1998, last updated 5-Jan-2012  by webmaster@fship.com

Back to FlagShip home page