This is version 5.02 of tcllex. It's a scanner generator, producing
an output file in tcl which can be used to scan lexically.

It is certainly not quick, but it works for what I wanted it to do
1. support tcl programs that require parsing of input.
2. support rapid prototyping by means of tclsh functionality alone.
i.e. it doesn't have any C code, and there's no need for a separate
wish etc.
3.  the scanners that are produced are not covered by any copyright

It supports:
1.  flex's start states.  The only difference between tcllex and flex
is that, in flex, you have to declare the start states before you use
them.  In tcllex, you can use whatever start state you want. The
starting start state is called "normal".
2.  multiple file input streams. This is activated by the yy_push_file
and yy_pop_file procedures. I haven't tested this thoroughly, and I'm
not sure you'd always like to stack input files, so if anyone is using
this, could they let me know, please?
3.  \t and \n in regular expressions in the input language lexical
description
4.  multiple line actions in the input language lexical description
5.  alternate actions in the input language lexical description
6.  flex's behaviour vis a vis ".*" i.e. a tcl "." regexp will match a
newline character, whereas flex's one doesn't. So you specify the
input description just as you would for a normal flex description.
[Actually, if you invoke tcllex with the -C switch, tcllex will
do the usual tcl trick of matching a dot with a newline. This is quite
handy in various situations, and gives a more structured regular
expression feel to things - agc].
7.  if a regular expression begins with '^' and/or ends with '$', then
a line-orientated approach will be taken to matching.

It doesn't support:
1. flex's character class definitions - no big deal, though, just
replicate them throughout your regular expressions to match.

Installation
This release was put together with tclsh resident in /usr/local/bin -
if this isn't the case, then you'll have to create a link from there,
or modify all the occurences in these sources to point to your location.
1. Extract the gzipped tar archive.
2. Run the testit program in the current directory, which will execute
all the tests. If there are *NO* differences, congratulations - everything
has gone well. If not, please mail me the output from `testit' so that
I can try to fix it.

There are various ways to specify things. The lexer will run slowly if
single characters are always recognised. For example, there are two
example language descriptions included, using start states to identify
C comments (c.l and d.l). The speed-up is noticeable between
the two descriptions. All that has changed is that

<incomment>.	{}

in c.l has been changed to

<incomment>[^*]+	{}
<incomment>\*	{}

in d.l.

All feedback to:

Alistair G. Crooks
(agc@uts.amdahl.com)
13th December 1994.

Work: Amdahl, Dogmersfield Park, Hartley Wintney, Hants, RG27 8TE, UK.
Tel: +44 125 234 6377				Fax: +44 125 234 6400

Home: 38 Swallow Way, Wokingham, Berks, RG11 2TQ, UK.
Tel: +44 1734 773688				Fax: +44 1734 773688