gnat.spitbol.patterns.match(...)

gnat.spitbol.patterns.match(...)

[bumwad]

I am trying to use gnat's Spitbol/pattern
packages to create a string by removing a regexp
from one and replacing it with another.

for example I have a string "c:\a\b\c\d\foo"
and path "c:\a\b\" and another path "c:\x\y\".
I want to create "c:\x\y\c\d\foo". I am writing a
unifying directory trees utility.

env = Nt 4.0, gnat 3.13p

here is code ....

   Base_Dir_from_Pattern : constant
Gnat.Spitbol.Patterns.Pattern :=
Gnat.Spitbol.Patterns.Span("c:\testChamber\t99\");

...

Ada.TExt_Io.Put_Line("Subject   = " &
Ada.Strings.Unbounded.To_String
(Curr_Dir_To_Name_Vstring));
Ada.TExt_Io.Put_Line("Pat       = " &
Gnat.Spitbol.Patterns.Image
(Base_Dir_From_Pattern));
Ada.TExt_Io.Put_Line("Replace   = " &
Base_Dir_To);


Replacement_Successful :=
Gnat.Spitbol.Patterns.Match (Subject =>
Curr_Dir_To_Name_Vstring,   --
Gnat.Spitbol.Patterns.VString_Var
Pat     => Base_Dir_From_Pattern,      -- pattern
Replace => Base_Dir_To                 -- string
);


ADA.Text_IO.NEW_LINE(2);
Ada.TExt_Io.Put_Line("Subject   = " &
Ada.Strings.Unbounded.To_String
(Curr_Dir_To_Name_Vstring));
Ada.TExt_Io.Put_Line("Pat       = " &
Gnat.Spitbol.Patterns.Image
(Base_Dir_From_Pattern));
Ada.TExt_Io.Put_Line("Replace   = " &
Base_Dir_To);

ADA.Text_IO.NEW_LINE(5);




and output ( go through a loop with a recursive
call and dir tree only 2 deep). See that the
directory a_dir is missing the initial 'a' after
the call on match() but not on a_dir_in_a_dir
(!?) ....


C:\testChamber>synchem3
Subject   = c:\testChamber\t99\a_dir
Pat       = Span ("9:C\abcehmrst")
Replace   = c:\testChamber\t100\


Subject   = c:\testChamber\t100\_dir
Pat       = Span ("9:C\abcehmrst")
Replace   = c:\testChamber\t100\





Subject   = c:\testChamber\t99
\a_dir\a_dir_in_a_dir
Pat       = Span ("9:C\abcehmrst")
Replace   = c:\testChamber\t100\


Subject   = c:\testChamber\t100
\_dir\a_dir_in_a_dir
Pat       = Span ("9:C\abcehmrst")
Replace   = c:\testChamber\t100\


Thanks,
Keith


Sent via Deja.com http://www.deja.com/
Before you buy.

Re: gnat.spitbol.patterns.match(...)

[Robert Dewar]

In article <dale-2D7177.11420406102000@news.rmit.edu.au>,
  dale <dale@cs.rmit.edu.au> wrote:
> Mr bumwadwrote:
>
> [a whole lot of stuff]
>
> ...and i thought Ada was supposed to be readable!!!


The point to realize here is that this feature in GNAT
duplicates the full semantics and also somewhat the syntax
of another complex language (SNOBOL-4). You cannot expect
to begin to understand code using this feature without putting
in the effort to learn this language (which is not trivial).
The 20 page tutorial in g-spipat.ads is a bare introduction
to the language, but is only that. To fully use this feature,
if you are not familiar with SNOBOL-4, you probably need to
read some other SNOBOL-4 materials :-)


Sent via Deja.com http://www.deja.com/
Before you buy.

Re: gnat.spitbol.patterns.match(...)

[Robert Dewar]

In article <3A04F805.AB1D83DA@telepath.com>,
  Ted Dennison <dennison@telepath.com> wrote:
> Well...a lot of that was SPITBOL (a regular expression
> language).


That's a lot of misinformation in a very short sentence :-) :-)

1. Spitbol is not a language, it is an implementation. Referring
to Spitbol as a language would be like referring to GNAT as
a language. The language is SNOBOL 4, designed by Ralph Griswold
and coworkers at Bell Labs in the late 70's and early 80's.
I also partipated in the language design. The reason that we
call the package Spitbol in GNAT is fourfold:

   1. It incorporates some of the extensions of SPITBOL
   2. Most SNOBOL-4 users these days are using SPITBOL
   3. It uses the pattern matching algorithms of SPITBOL
   4. Since I wrote SPITBOL, it creates an useful connection

2. It has NOTHING at ALL to do with regular expressions. I
really don't know where you got this idea. The whole point of
SNOBOL 4 is that its pattern matching is based on context free
grammars, and actually it is better to think of them as
attribute grammars, since quite general semantic actions,
including predicates can be included in a pattern.

Indeed the use of regular expressions for matching is by
comparison kludgy and very restrictive.

To give an idea of the power of SNOBOL-4 pattern matching, it
is quite feasible to write a pattern that matches only strings
that are syntactically valid Ada 95 programs, and rejects all
other strings. The syntax of this pattern would be reminiscent
of the BNF used in the Ada RM to describe the syntax of Ada.

Robert Dewar


Sent via Deja.com http://www.deja.com/
Before you buy.

Re: gnat.spitbol.patterns.match(...)

[Dale Stanbrough]

Robert Dewar  wrote:

> The point to realize here is that this feature in GNAT
> duplicates the full semantics and also somewhat the syntax
> of another complex language (SNOBOL-4). You cannot expect
> to begin to understand code using this feature without putting
> in the effort to learn this language (which is not trivial).
> The 20 page tutorial in g-spipat.ads is a bare introduction
> to the language, but is only that. To fully use this feature,
> if you are not familiar with SNOBOL-4, you probably need to
> read some other SNOBOL-4 materials :-)


I didn't have a problem with the Snobol part, (although i find
the use of 'Unrestricted_Access can make spitbol code a bit ugly);
it was the full package name syntax, and the strange formatting
that i found hard to cope with.


Dale

Re: gnat.spitbol.patterns.match(...)

[dale]

Mr bumwadwrote:

[a whole lot of stuff]


...and i thought Ada was supposed to be readable!!!


Dale

Re: gnat.spitbol.patterns.match(...)

[Ted Dennison]

dale wrote:

> Mr bumwadwrote:
>
> [a whole lot of stuff]
>
> ...and i thought Ada was supposed to be readable!!!

Well...a lot of that was SPITBOL (a regular expression language).


--
T.E.D.

Home - mailto:dennison@telepath.com  Work - mailto:dennison@ssd.fsi.com
WWW  - http://www.telepath.com/dennison/Ted/TED.html  ICQ  - 10545591

Re: gnat.spitbol.patterns.match(...)

[Georg Bauhaus]

Ted Dennison (dennison@telepath.com) wrote:

: Well...a lot of that was SPITBOL (a regular expression language).

Not beeing the one to judge SNOBOL4, but still having
intense looks at it (and using it), thus finding "a regular
expression language" rather misleading (for reasons in
addition to those Robert Dewar has already given), here is
a quote from J. Gimpel, Algorithms in SNOBOL4:

"The SNOBOL series of programming languages through SNOBOL4 had only one
datatype, the string. Even the arithmetic facilities of SNOBOL3 were
implemented as operations on strings of digits rather than on machine
integers. Because of this historical bias, and because the language is
extaordinarily rich in string handling, SNOBOL4 is still regarded by
some as exclusively a string language. Yet, all the basic facilities
which one expects in a list processing language have been incorporated
into SNOBOL4; these include the automatic allocation and freeing of
storage, recursive functions, the pointer, and the data structure.
Moreover, the notation is, for the most part, conventional, convenient
and flexible. Were SNOBOL4 suddenly stripped of all its pattern
matching capabilities, it would still be a powerful and convenient
list-processing language." (p. 80)

And easy enough to use! :-)  With the GNAT library,
and the ./examples directory, you get Ada with it's
facilities combined with some from SNOBOL4, to make a
handy toolset for, e.g. data preparation (called "data laundry",
I believe, think of CGI as one example). 
NOte that Ada has types, so SNOBOL4's dodn't
have to be "imported" :>.  IMHO

Re: gnat.spitbol.patterns.match(...)

[Georg Bauhaus]

Georg Bauhaus (sb463ba@l1-hrz.uni-duisburg.de) wrote:
: Ted Dennison (dennison@telepath.com) wrote:


: "The SNOBOL series of programming languages through SNOBOL4 had only one
correct:                                              ^^^^^^3

Sorry.

Re: gnat.spitbol.patterns.match(...)

[Ted Dennison]

Dale Stanbrough wrote:

> I didn't have a problem with the Snobol part, (although i find
> the use of 'Unrestricted_Access can make spitbol code a bit ugly);
> it was the full package name syntax, and the strange formatting
> that i found hard to cope with.

Well, the strange formatting can probably be blamed on Deja, his
newsreader. It has a nasty habit of inserting and deleting line breaks,
seemingly at random. Whenever I post source code using Deja, I've taking
to apologizing in advance for what it might end up doing to the
formatting.

I think the full package name syntax (in isolation) looks fine. But
there really is no excuse for the random identifier casing. :-)

Now that we have finished making fun of his style and typing
capabilitites, I wonder if anyone can answer the poor bumwad's
question...

--
T.E.D.

Home - mailto:dennison@telepath.com  Work - mailto:dennison@ssd.fsi.com
WWW  - http://www.telepath.com/dennison/Ted/TED.html  ICQ  - 10545591