"cl-case" with strings

"cl-case" with strings

[Emanuel Berg]

Today when I did a new interface to my interface [1]
to `google-translate' which is an interface to
Google Translate, I used `cl-case' with strings but
was frustrated with failure, because `cl-case' uses
equality function(s) that report nil for
identical strings.

So I wrote this.

I guess I like it fine - so far.

(defun string-case (string &rest clauses)
  "Search for STRING in CLAUSES and return the corresponding value.\n
CLAUSES is a list of elements of the form ((KEY-1 ... KEY-N) VALUE).
If STRING is a `member' of (KEY-1 ... KEY-N), return VALUE.\n
The default clause is (t DEFAULT-VALUE).
When found, DEFAULT-VALUE is unconditionally returned.
Subsequent clauses are discarded, because clauses are examined left-to-right.
This is also factor if STRING matches keys in several clauses.\n
If there isn't a default clause, nil is returned if STRING isn't a key."
  (if clauses
      (let*((c      (car clauses))
            (keys   (car c)) )
        (if (or (member string keys) ; on match or
                (eq keys t) )        ;    default
            (cadr c)                 ;      return value
          (apply `(string-case ,string ,@(cdr clauses))) ))))

[1] http://user.it.uu.se/~embe8573/conf/emacs-init/translate.el

-- 
underground experts united
http://user.it.uu.se/~embe8573

Re: "cl-case" with strings

[Dmitry Gutov]

On 02/04/2016 04:13 AM, Emanuel Berg wrote:
> Today when I did a new interface to my interface [1]
> to `google-translate' which is an interface to
> Google Translate, I used `cl-case' with strings but
> was frustrated with failure, because `cl-case' uses
> equality function(s) that report nil for
> identical strings.
>
> So I wrote this.

Or you could use pcase.

Re: "cl-case" with strings

[Marcin Borkowski]

On 2016-02-04, at 02:39, Dmitry Gutov <dgutov@yandex.ru> wrote:

> On 02/04/2016 04:13 AM, Emanuel Berg wrote:
>> Today when I did a new interface to my interface [1]
>> to `google-translate' which is an interface to
>> Google Translate, I used `cl-case' with strings but
>> was frustrated with failure, because `cl-case' uses
>> equality function(s) that report nil for
>> identical strings.
>>
>> So I wrote this.
>
> Or you could use pcase.

Or this idiom:
(cl-case (intern string)
  ...)

Best,

-- 
Marcin Borkowski
http://octd.wmi.amu.edu.pl/en/Marcin_Borkowski
Faculty of Mathematics and Computer Science
Adam Mickiewicz University

Re: "cl-case" with strings

[Emanuel Berg]

Marcin Borkowski <mbork@mbork.pl> writes:

>> Or you could use pcase.
>
> Or this idiom: (cl-case (intern string) ...)

Then the keys cannot be strings - not an issue in this
"case"...

The instructions for `pcase' tho isn't exactly like
reading on the milk pack - perhaps the OP or someone
else can provide an example?

-- 
underground experts united
http://user.it.uu.se/~embe8573

Re: "cl-case" with strings

[Joost Kremers]

http://newartisans.com/2016/01/pattern-matching-with-pcase/

EN:SiS(9)
Emanuel Berg wrote:
> The instructions for `pcase' tho isn't exactly like
> reading on the milk pack - perhaps the OP or someone
> else can provide an example?

-- 
Joost Kremers                                   joostkremers@fastmail.fm
Selbst in die Unterwelt dringt durch Spalten Licht

Re: "cl-case" with strings

[Stefan Monnier]

> The instructions for `pcase' tho isn't exactly like
> reading on the milk pack - perhaps the OP or someone
> else can provide an example?

    (pcase foo
     ("toto" do-the-toto-thing)
     ("titi" do-the-titi-thing))

For some older Emacsen, IIRC, it might be necessary to do

    (pcase foo
     (`"toto" do-the-toto-thing)
     (`"titi" do-the-titi-thing))


-- Stefan

Re: "cl-case" with strings

[Emanuel Berg]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> The instructions for `pcase' tho isn't exactly like
>> reading on the milk pack - perhaps the OP or
>> someone else can provide an example?
>
> (pcase foo ("toto" do-the-toto-thing) ("titi" do-the-titi-thing))

Does it work with multiple keys (keylists) as well?

Here are the line from the original code:

    (string-case from
              '(("sv" "da") "ord")
              '(("en")      "word")
              '(("es")      "palabra") )

-- 
underground experts united
http://user.it.uu.se/~embe8573

Re: "cl-case" with strings

[Marcin Borkowski]

On 2016-02-04, at 23:12, Emanuel Berg <embe8573@student.uu.se> wrote:

> Marcin Borkowski <mbork@mbork.pl> writes:
>
>>> Or you could use pcase.
>>
>> Or this idiom: (cl-case (intern string) ...)
>
> Then the keys cannot be strings - not an issue in this
> "case"...

Why not?  It works fine for me when `string' is a string.

Best,

-- 
Marcin Borkowski
http://octd.wmi.amu.edu.pl/en/Marcin_Borkowski
Faculty of Mathematics and Computer Science
Adam Mickiewicz University

Re: "cl-case" with strings

[Emanuel Berg]

Marcin Borkowski <mbork@mbork.pl> writes:

>>> Or this idiom: (cl-case (intern string) ...)
>>
>> Then the keys cannot be strings - not an issue in
>> this "case"...
>
> Why not? It works fine for me when `string' is
> a string.

It works for me too but to me it is rather a hack
than idiomatic.

I'd like the keys to be strings if the (original) data
is. Then the keys/strings are also properly
highlighted as strings because that's what they are.

Now when I have my defun I don't want to change for
this, but had I known of this before, perhaps (?)
I wouldn't have written the defun.

-- 
underground experts united
http://user.it.uu.se/~embe8573

Re: "cl-case" with strings

[Emanuel Berg]

Emanuel Berg <embe8573@student.uu.se> writes:

> So I wrote this.

Non-recursive version - perhaps (?) better:

(defun string-case-non-recursive (string &rest clauses)
  (cl-some (lambda (c)
             (let ((keys (car c)))
               (and (or (eq t          keys)
                        (member string keys) )
                    (cadr c) ))) clauses) )

-- 
underground experts united
http://user.it.uu.se/~embe8573

Re: "cl-case" with strings

[Dmitry Gutov]

On 02/05/2016 01:41 AM, Emanuel Berg wrote:

>> (pcase foo ("toto" do-the-toto-thing) ("titi" do-the-titi-thing))
>
> Does it work with multiple keys (keylists) as well?

(pcase "bar"
   ("toto" do-the-toto-thing)
   ((or "foo" "bar") do-the-foobar-thing)
   ("titi" do-the-titi-thing))

Re: "cl-case" with strings

[Emanuel Berg]

Dmitry Gutov <dgutov@yandex.ru> writes:

>> Does it work with multiple keys (keylists) as well?
>
> (pcase "bar"
>    ("toto" do-the-toto-thing)
>    ((or "foo" "bar") do-the-foobar-thing)
>    ("titi" do-the-titi-thing))

Clever!

But I still like mine the best!

-- 
underground experts united
http://user.it.uu.se/~embe8573

Re: "cl-case" with strings

[Dmitry Gutov]

On 02/05/2016 07:58 AM, Emanuel Berg wrote:

> But I still like mine the best!

Of course you do. ;-)

But if you ever write a package to distribute to other users via e.g. 
ELPA, using pcase would be a smarter choice.

Re: "cl-case" with strings

[Marcin Borkowski]

On 2016-02-05, at 05:05, Emanuel Berg <embe8573@student.uu.se> wrote:

> Marcin Borkowski <mbork@mbork.pl> writes:
>
>>>> Or this idiom: (cl-case (intern string) ...)
>>>
>>> Then the keys cannot be strings - not an issue in
>>> this "case"...
>>
>> Why not? It works fine for me when `string' is
>> a string.
>
> It works for me too but to me it is rather a hack
> than idiomatic.

Well, it was not my idea - I found it somewhere (can't find it right
now).  I'm not sure whether this is a hack or the proper way to do it;
I quite like it personally.

> I'd like the keys to be strings if the (original) data
> is. Then the keys/strings are also properly
> highlighted as strings because that's what they are.

Ah, I see now what you meant.

> Now when I have my defun I don't want to change for
> this, but had I known of this before, perhaps (?)
> I wouldn't have written the defun.

:-)

-- 
Marcin Borkowski
http://octd.wmi.amu.edu.pl/en/Marcin_Borkowski
Faculty of Mathematics and Computer Science
Adam Mickiewicz University

Re: "cl-case" with strings

[Emanuel Berg]

Dmitry Gutov <dgutov@yandex.ru> writes:

> But if you ever write a package to distribute to
> other users via e.g. ELPA, using pcase would be
> a smarter choice.

Yes.

By the way, to anyone who by chance didn't know,
"yandex.ru" is one of those old 90s Internet
"portals".

It is like the Russian Passagen (Sweden) or Yahoo!,
AltaVista, etc. in the US.

But actually Yandex is older so perhaps the comparison
should be in the opposite direction...

-- 
underground experts united
http://user.it.uu.se/~embe8573

Re: "cl-case" with strings

[Dmitry Gutov]

On 02/05/2016 09:20 PM, Emanuel Berg wrote:

> By the way, to anyone who by chance didn't know,
> "yandex.ru" is one of those old 90s Internet
> "portals".

Not really. Maybe it was a "portal" in the 90s and the beginning of 00s.

These days it's a search engine not unlike Google, handling about half 
of all search traffic in Russia. It also has a decent hosted email service.