"Cameron Laird" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]
> In article <[EMAIL PROTECTED]>,
> Paul McGuire <[EMAIL PROTECTED]> wrote:
>
<yet another post showing a pyparsing stab at a parsing problem, this time
parsing Tcl lists>
>
> Noooooooooooooooooo!
>
> I'll be more precise: pyparsing is quite wonderful, and I'm
> all in favor of clever demonstrations of its capabilities. I
> don't think this is one, though; in fact, although I haven't
> yet paid enough attention to the original question to provide
> an example, I'm reasonably certain that the code offered above
> mishandles at least some data (Tcl syntax is different from
> what outsiders expect).
>
> Paddy's advice elsewhere in this thread is almost entirely
> correct. Anyone who has imported Tkinter has all of Tcl avail-
> able immediately, so much the easiest, most reliable, most
> maintainable, and most lucid solution is simply to iterate in
> Tcl over the list, and construct thus the corresponding Python
> list.
>
> I'm willing to write either a correct implementation, or a
> counterexample with problematic data, if these sufficiently
> interest readers. It'll be a day or two before I can make the
> time to be careful, though.
Cameron -
Er? Thanks for the nice comments re: pyparsing, sometimes I feel a little
self-conscious always posting these pyparsing snippets. So I'm glad you
clarified your intent with your "I'll be more precise" paragraph. But I'm
not sure I see the reason for an 18-O "No!"
"Outsider"? I'm no stranger to Tcl (although I *am* a bit rusty). In the
90's I worked for several years with a script-based process control system
for semiconductor manufacturing, in which we used Tcl for customer-defined
control logic. Tcl led me to appreciate the value of a scripting language,
despite its clunky assignment syntax ("set x 1"), and "$"-happiness. When I
looked into Python a few years later, I remember thinking "Man, I wish we
could have used Python instead."
I took a look at how accessible Tcl is using Tkinter. It looks to me like
one would need to:
- write a Tcl proc to recursively traverse a list, returning a stringified
list in Python-like syntax (enclosing non-list items in ''s, separating with
commas, nesting sublists with []'s, etc.)
- construct a string passing the OP's Tcl list string "sql commands ex:
..." to this method
- invoke Tkinter.Tk().eval on this string to perform the Tcl traversal
- invoke Python's eval function on the returned string to get an actual
list.
Did I get that right? Or is there another, simpler back door into Tcl from
Tkinter?
I'll confess, I wrote my sample code based on the sample text posted by the
OP, without much extra syntax (such as negative integers or floating-point
values). So here is a version with a bit broader coverage:
======================
tcl = """sql commands ex:
{ begin { select * from foo
where baz='whatever'}
{select * from $gooble } { } end
{ insert into bar values('Tom', 25) } }"""
from pyparsing import *
tcllist = Forward()
element = quotedString | Combine(Optional("$") + Word(alphas,alphanums+"_"))
| \
Combine(Optional(oneOf(list("+-")))+ Word(nums) + "." +
Optional(Word(nums)) ) | Word(nums+"+-",nums) | \
oneOf( list(r"(),[EMAIL PROTECTED]&*-|\?/><;:") ) | Group( '{' +
tcllist
+ '}' )
tcllist << ZeroOrMore( element )
import pprint
pprint.pprint( tcllist.parseString(tcl).asList() )
======================
This should handle empty lists, signed integers and reals, and variables
with leading '$' signs.
Before putting too much effort into problematic counter-example data, it
might be a good idea to check with the OP, to see what types of values are
likely to come up in practice. Given that c.l.py posting is only a
spare-time activity (probably so for you, too), I wouldn't want to waste
time on the be-all-and-end-all Tcl list parser that handles cases the OP
wont run into.
-- Paul
--
http://mail.python.org/mailman/listinfo/python-list
