How to check whether a character (or one-character string) is a letter?

How to check whether a character (or one-character string) is a letter?

[Marcin Borkowski]

Hello,

this is a problem I have.

Assume that I have a character (taken from some string, which in turn is
copied from the buffer - so it need not be ASCII).  What is the best way
to check whether it is a letter within ASCII range?

The reason I'm asking is that I'm writing a function which converts an
arbitrary string to a valid (and nice) filename (e.g., only letters and
hyphens) - so basically I want to walk a string character by character
and convert any space to a hyphen and omit any other non-letter.  Am I
reinventing the wheel?

TIA,

-- 
Marcin Borkowski
http://octd.wmi.amu.edu.pl/en/Marcin_Borkowski
Adam Mickiewicz University

Re: How to check whether a character (or one-character string) is a letter?

[Thorsten Jolitz]

Marcin Borkowski <mbork@wmi.amu.edu.pl> writes:

Hello,

> this is a problem I have.
>
> Assume that I have a character (taken from some string, which in turn is
> copied from the buffer - so it need not be ASCII).  What is the best way
> to check whether it is a letter within ASCII range?
>
> The reason I'm asking is that I'm writing a function which converts an
> arbitrary string to a valid (and nice) filename (e.g., only letters and
> hyphens) - so basically I want to walk a string character by character
> and convert any space to a hyphen and omit any other non-letter.  Am I
> reinventing the wheel?

there is something similar in PicoLisp:

#+BEGIN_SRC picolisp :results pp
 (fold "abc-?/@ä-12 YZ  #+ü")
#+END_SRC

#+results:
: "abcä12yzü"

but not quite ...


-- 
cheers,
Thorsten

Fwd: How to check whether a character (or one-character string) is a letter?

[John Mastro]

[I first sent this directly to Marcin in error - yeah, I use the email gateway]

Hi Marcin,

Marcin Borkowski <mbork@wmi.amu.edu.pl> wrote:
> Assume that I have a character (taken from some string, which in turn is
> copied from the buffer - so it need not be ASCII).  What is the best way
> to check whether it is a letter within ASCII range?
>
> The reason I'm asking is that I'm writing a function which converts an
> arbitrary string to a valid (and nice) filename (e.g., only letters and
> hyphens) - so basically I want to walk a string character by character
> and convert any space to a hyphen and omit any other non-letter.  Am I
> reinventing the wheel?

There are a bunch of ways to do this, but one reasonable approach is to
use a regular expression. I think this will do what you want:

    (defun reasonable-filename (str)
      (let* ((str (replace-regexp-in-string "[ \t\n\r]" "-" str))
             (str (replace-regexp-in-string "[^a-zA-Z-]" "" str)))
        str))

This is a variation which will also allow the result to contain numbers:

    (defun reasonable-filename (str)
      (let* ((str (replace-regexp-in-string "[ \t\n\r]" "-" str))
             (str (replace-regexp-in-string "[^a-zA-Z0-9-]" "" str)))
        str))

To answer your question about identifying whether a character is an
ASCII letter, the key is that Emacs's characters are really "just"
integers. Wikipedia has some charts[1] that show the numbers associated
with the characters. The letters are conveniently grouped together, so
we can use something like this:

    (defun ascii-letter-p (char)
      (and (characterp char)
           (>= char 65)
           (<= char 122)))

(Of course, this only works if it's really a character, as opposed to a
string of length one. If it's a string of length one you could either
"extract" the character with `aref' or use a regular expression
instead.)

Hope that helps.

[1] https://en.wikipedia.org/wiki/ASCII#ASCII_printable_code_chart

--
john

Re: How to check whether a character (or one-character string) is a letter?

[Eric Abrahamsen]

Marcin Borkowski <mbork@wmi.amu.edu.pl> writes:

> Hello,
>
> this is a problem I have.
>
> Assume that I have a character (taken from some string, which in turn is
> copied from the buffer - so it need not be ASCII).  What is the best way
> to check whether it is a letter within ASCII range?
>
> The reason I'm asking is that I'm writing a function which converts an
> arbitrary string to a valid (and nice) filename (e.g., only letters and
> hyphens) - so basically I want to walk a string character by character
> and convert any space to a hyphen and omit any other non-letter.  Am I
> reinventing the wheel?

The safest thing is probably still replace-regexp-in-string. You can use
the [:ascii:] character class to strip out anything that isn't ascii,
with a regexp like "[^[:ascii:]]+". Messing with the string at the
character level shouldn't be necessary.

Eric

Re: How to check whether a character (or one-character string) is a letter?

[Yuri Khan]

On Sat, Oct 4, 2014 at 7:29 AM, Marcin Borkowski <mbork@wmi.amu.edu.pl> wrote:

> The reason I'm asking is that I'm writing a function which converts an
> arbitrary string to a valid (and nice) filename (e.g., only letters and
> hyphens) - so basically I want to walk a string character by character
> and convert any space to a hyphen and omit any other non-letter.  Am I
> reinventing the wheel?

What are your assumptions about input string arbitrariness, your
requirements about output filename niceness, and your requirements
about the properties of the mapping?

Because these may be in conflict.

For example, if you assume any arbitrary strings, want only
[-0-9A-Za-z_] characters, and want reasonably different strings to map
into different filenames, then you will end up having to preserve
non-nice characters as ugly character encodings (in the spirit of
urlencode, XML character references, or Punycode). Otherwise, whole
words or sentences in Russian, Japanese or Greek will map into an
empty filename.

Re: How to check whether a character (or one-character string) is a letter?

[Eli Zaretskii]

> From: Marcin Borkowski <mbork@wmi.amu.edu.pl>
> Date: Sat, 04 Oct 2014 02:29:24 +0200
> 
> Assume that I have a character (taken from some string, which in turn is
> copied from the buffer - so it need not be ASCII).  What is the best way
> to check whether it is a letter within ASCII range?

One way is this:

  (= 1 (string-bytes (char-to-string ch)))

where 'ch' is the character you want to test.  Note that this will
pass any 7-bit ASCII character, including control characters, digits,
and punctuation, not just "letters".

Re: How to check whether a character (or one-character string) is a letter?

[Marcin Borkowski]

On 2014-10-04, at 06:08, Yuri Khan wrote:

> On Sat, Oct 4, 2014 at 7:29 AM, Marcin Borkowski <mbork@wmi.amu.edu.pl> wrote:
>
>> The reason I'm asking is that I'm writing a function which converts an
>> arbitrary string to a valid (and nice) filename (e.g., only letters and
>> hyphens) - so basically I want to walk a string character by character
>> and convert any space to a hyphen and omit any other non-letter.  Am I
>> reinventing the wheel?
>
> What are your assumptions about input string arbitrariness, your
> requirements about output filename niceness, and your requirements
> about the properties of the mapping?
>
> Because these may be in conflict.
>
> For example, if you assume any arbitrary strings, want only
> [-0-9A-Za-z_] characters, and want reasonably different strings to map
> into different filenames, then you will end up having to preserve
> non-nice characters as ugly character encodings (in the spirit of
> urlencode, XML character references, or Punycode). Otherwise, whole
> words or sentences in Russian, Japanese or Greek will map into an
> empty filename.

Good point.  However, I intend to keep a list of filenames, and in case
some of them is already taken, append a number to it.  (This is an
extremely primitive hashing function, but it will suffice for my needs.)

Regards,

-- 
Marcin Borkowski
http://octd.wmi.amu.edu.pl/en/Marcin_Borkowski
Adam Mickiewicz University

Re: Fwd: How to check whether a character (or one-character string) is a letter?

[Marcin Borkowski]

On 2014-10-04, at 04:47, John Mastro wrote:

> [I first sent this directly to Marcin in error - yeah, I use the email gateway]
>
> Hi Marcin,
>
> Marcin Borkowski <mbork@wmi.amu.edu.pl> wrote:
>> Assume that I have a character (taken from some string, which in turn is
>> copied from the buffer - so it need not be ASCII).  What is the best way
>> to check whether it is a letter within ASCII range?
>>
>> The reason I'm asking is that I'm writing a function which converts an
>> arbitrary string to a valid (and nice) filename (e.g., only letters and
>> hyphens) - so basically I want to walk a string character by character
>> and convert any space to a hyphen and omit any other non-letter.  Am I
>> reinventing the wheel?
>
> There are a bunch of ways to do this, but one reasonable approach is to
> use a regular expression. I think this will do what you want:
>
>     (defun reasonable-filename (str)
>       (let* ((str (replace-regexp-in-string "[ \t\n\r]" "-" str))
>              (str (replace-regexp-in-string "[^a-zA-Z-]" "" str)))
>         str))

I think this is probably better than mapcar'ing through the string...

> This is a variation which will also allow the result to contain numbers:
>
>     (defun reasonable-filename (str)
>       (let* ((str (replace-regexp-in-string "[ \t\n\r]" "-" str))
>              (str (replace-regexp-in-string "[^a-zA-Z0-9-]" "" str)))
>         str))

This I don't want, since in case of equal filenames, I want to
differentiate them by appending a number, and allowing digits might
break this.  But thanks anyway.

> To answer your question about identifying whether a character is an
> ASCII letter, the key is that Emacs's characters are really "just"
> integers. Wikipedia has some charts[1] that show the numbers associated
> with the characters. The letters are conveniently grouped together, so
> we can use something like this:
>
>     (defun ascii-letter-p (char)
>       (and (characterp char)
>            (>= char 65)
>            (<= char 122)))
>
> (Of course, this only works if it's really a character, as opposed to a
> string of length one. If it's a string of length one you could either
> "extract" the character with `aref' or use a regular expression
> instead.)
>
> Hope that helps.

Yes it does!

Best,

-- 
Marcin Borkowski
http://octd.wmi.amu.edu.pl/en/Marcin_Borkowski
Adam Mickiewicz University