Mindy Demo Programs

hello-world

This is the canonical "hello, world" demo. Type 
  mindy -f hello-world.dbc
to run it.

cat

This program demonstrates the streams library by duplicating the UNIX (tm) "cat" utility. Usage is typical for a unix program. It may be invoked either with a set of files on the command line: 
  mindy -f cat.dbc file1 file2 ....
or with no arguments, in which case it reads from "standard input". You may also specify special filename "-" which will refer to the standard input.

On most Unix systems you should be able to make it into an executable script by prepending the the line 

  #$INSTALL/bin/mindy -f
to the compiled "dbc" file (and adding "execute" to the file's protection). You could then simply type 
  cat.dbc file1 file2 ....
You must, of course, remember to specify the MINDYPATH environment variable so that it points to directories containing all of the relevant libraries. (This demo uses only "dylan" and "streams", but the "html2txt" demo also uses "collection-extensions".)

html2txt

The "html2txt" program is a filter which converts text in WWWs "HyperText Markup Language" into simple formatted text. Although it is a complete and useful application, it is included in this distribution primarily as a demonstration of a "real" (albeit small) Dylan (tm) program.

Usage is similar to "cat" above, except that it does not support the "-" argument. You may therefore set of files on the command line or use standard input by specifying no command line arguments 

  mindy -f html2txt.dbc file1.html file2.html ....
  mindy -f cat.dbc file1.html file2.html | mindy -f html2txt.dbc
At present, "html2txt" accepts no command line switches, although the behavior may be changed by changing several constant declarations towards the top of the source file.

Useful test cases for this program are the files "demos.html", which was used as the source for "README"; and "$INSTALL/doc/collection-extensions.html", which was used as the source file for the documentation file "$INSTALL/doc/collection-extensions.doc".

The basic translation strategy used by html2txt is to scan the file line by line, looking for HTML "tags" and accumulating text that lies between any two tags. For each tag type, there is a set of routines (stored in tables) which define the appropriate actions for starting and ending the "environment" defined by the tag and for dumping the collected text from within that environment as formatted text. A basic control loop in "process-HTML" is responsible for calling the appropriate tag actions. This routine may be called recusively by some of the tag actions.

The "interface" between adjacent environments is handled via the "blank" parameter which is passed around extensively. This variable states whether a blank line has just been printed. Thus environments which believe that they must be preceded or followed by a blank line can determine whetehr they must do anything about it, and we lessen the risk that multiple routines will emit blank lines when we only want a maximum of one.

The primary advantage of this organization is that it allows the specialized actions for a single tag to be grouped together, and allows new tags to be cleanly added. It benefits greatly from Dylan's ability to create anonymous methods and manipulate them as first class data objects, as well as from the rich set of available collection types.