bug#23325: 25.0.92; insert-char: cannot find all chars if input is unicode name

bug#23325: 25.0.92; insert-char: cannot find all chars if input is unicode name

[Tino Calancha]

Hello,

Interactive calls to `insert-char' cannot find all
characters when CHARACTER is the unicode character name.

IOW, it finds all characters when CHARACTER is the
code point.


emacs -Q:
M-x insert-char RET cjk SPC TAB
;; Buffer "*Completions*" just shows entries starting
;; with 'CJK RADICAL' or 'CJK STROKE'.

M-x insert-char RET 2eea RET
;; ok
M-x insert-char RET cjk SPC radical SPC c-simplified SPC frog RET
;; ok
M-x insert-char RET 79c1 RET
;; ok
M-x insert-char RET cjk SPC ideograph-79c1 RET
;; Signal error 'Invalid character'


In GNU Emacs 25.0.92.1 (x86_64-pc-linux-gnu, GTK+ Version 2.24.30)
  of 2016-04-21 built on calancha-pc
Repository revision: a77cf24ada2f89194c0ac64aae27bcdf7021e697

bug#23325: 25.0.92; insert-char: cannot find all chars if input is unicode name

[Eli Zaretskii]

> Date: Thu, 21 Apr 2016 13:40:01 +0900 (JST)
> From: Tino Calancha <f92capac@gmail.com>
> 
> Interactive calls to `insert-char' cannot find all
> characters when CHARACTER is the unicode character name.
> 
> IOW, it finds all characters when CHARACTER is the
> code point.
> 
> 
> emacs -Q:
> M-x insert-char RET cjk SPC TAB
> ;; Buffer "*Completions*" just shows entries starting
> ;; with 'CJK RADICAL' or 'CJK STROKE'.
> 
> M-x insert-char RET 2eea RET
> ;; ok
> M-x insert-char RET cjk SPC radical SPC c-simplified SPC frog RET
> ;; ok
> M-x insert-char RET 79c1 RET
> ;; ok
> M-x insert-char RET cjk SPC ideograph-79c1 RET
> ;; Signal error 'Invalid character'

It's a feature, see ucs-names.  We deliberately filter out
non-descriptive names like "CJK COMPATIBILITY IDEOGRAPH-2F803",
because (a) if we didn't the list of completions would be sometimes
much longer; and (b) it makes very little sense to show these names as
completions candidates, because those names include the codepoint, so
if you know which of these you need to insert, you can insert it by
the codepoint.

bug#23325: 25.0.92; insert-char: cannot find all chars if input is unicode name

[Tino Calancha]

> It's a feature, see ucs-names.  We deliberately filter out
> non-descriptive names like "CJK COMPATIBILITY IDEOGRAPH-2F803",

Yes, that was my guess, and actually i would not cache such
trivial key val relations either.

Then a minor detail is as follows:

M-x insert-char 79c1 RET
;; ok character appear
C-b
M-x describe-char RET
;; Line 8 shows:
;; to input: type "C-x 8 RET 79c1" or "C-x 8 RET CJK IDEOGRAPH-79C1"

Shouldn't `describe-char' omit the input method by unicode name in 
those cases?

bug#23325: 25.0.92; insert-char: cannot find all chars if input is unicode name

[Eli Zaretskii]

> Date: Thu, 21 Apr 2016 23:32:49 +0900 (JST)
> From: Tino Calancha <f92capac@gmail.com>
> cc: Tino Calancha <f92capac@gmail.com>, 23325@debbugs.gnu.org
> 
> M-x describe-char RET
> ;; Line 8 shows:
> ;; to input: type "C-x 8 RET 79c1" or "C-x 8 RET CJK IDEOGRAPH-79C1"
> 
> Shouldn't `describe-char' omit the input method by unicode name in 
> those cases?

Yes, it should show a method that works; patches welcome.

bug#23325: 25.0.92; insert-char: cannot find all chars if input is unicode name

[Tino Calancha]

[-- Attachment #1: Type: text/plain, Size: 289 bytes --]



> Yes, it should show a method that works; patches welcome.

With attached patch:

`describe-char' on 0x79c1 shows:
to input: type "C-x 8 RET 79c1"

IOH, for catched entries in `ucs-names' (e.g. 0x304d) it shows as usual:
to input: type "C-x 8 RET 304d" or "C-x 8 RET HIRAGANA LETTER KI"

[-- Attachment #2: Type: text/plain, Size: 1152 bytes --]

diff --git a/lisp/descr-text.el b/lisp/descr-text.el
index a352ed0..21f2214 100644
--- a/lisp/descr-text.el
+++ b/lisp/descr-text.el
@@ -619,7 +619,13 @@ describe-char
                           (let ((name
                                  (or (get-char-code-property char 'name)
                                      (get-char-code-property char 'old-name))))
-                            (if name
+                            (if (and name
+                                     (not (string-prefix-p "CJK IDEOGRAPH-" name))
+                                     (not (string-prefix-p "CJK COMPATIBILITY IDEOGRAPH-" name))
+                                     (not (string-prefix-p "LOW SURROGATE-" name))
+                                     (not (string-prefix-p "HIGH SURROGATE-" name))
+                                     (not (string-prefix-p "TANGUT IDEOGRAPH-" name))
+                                     (not (string-prefix-p "TANGUT COMPONENT-" name)))
                                 (format
                                  "type \"C-x 8 RET %x\" or \"C-x 8 RET %s\""
                                  char name)

bug#23325: 25.0.92; insert-char: cannot find all chars if input is unicode name

[Eli Zaretskii]

> Date: Fri, 22 Apr 2016 02:39:24 +0900 (JST)
> From: Tino Calancha <f92capac@gmail.com>
> cc: Tino Calancha <f92capac@gmail.com>, 23325@debbugs.gnu.org
> 
> > Yes, it should show a method that works; patches welcome.
> 
> With attached patch:
> 
> `describe-char' on 0x79c1 shows:
> to input: type "C-x 8 RET 79c1"
> 
> IOH, for catched entries in `ucs-names' (e.g. 0x304d) it shows as usual:
> to input: type "C-x 8 RET 304d" or "C-x 8 RET HIRAGANA LETTER KI"

Thanks.  But I think it would be better if the code didn't have to
define in yet another place which characters are omitted; instead, how
about checking if the character is in the ucs-names list?

bug#23325: 25.0.92; insert-char: cannot find all chars if input is unicode name

[Tino Calancha]

[-- Attachment #1: Type: text/plain, Size: 227 bytes --]


> It would be better if the code didn't have to
> define in yet another place which characters are omitted; instead, how
> about checking if the character is in the ucs-names list?

Absolutely.  Implemented in the new patch.


[-- Attachment #2: Type: text/plain, Size: 1137 bytes --]

From 592473bed4a6986e9dc96489f63a174112cc90d5 Mon Sep 17 00:00:00 2001
From: Tino Calancha <f92capac@gmail.com>
Date: Fri, 22 Apr 2016 12:00:14 +0900
Subject: [PATCH] describe-char: fix insert char documentation

* lisp/descr-text.el (describe-char):
Only 'ucs-names' entries can be inserted by unicode name (Bug#23325).
---
 lisp/descr-text.el | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/lisp/descr-text.el b/lisp/descr-text.el
index a352ed0..4dac6bb 100644
--- a/lisp/descr-text.el
+++ b/lisp/descr-text.el
@@ -619,7 +619,8 @@ describe-char
                           (let ((name
                                  (or (get-char-code-property char 'name)
                                      (get-char-code-property char 'old-name))))
-                            (if name
+                            (if (and name
+                                     (assoc-string name (or ucs-names (ucs-names))))
                                 (format
                                  "type \"C-x 8 RET %x\" or \"C-x 8 RET %s\""
                                  char name)
-- 
2.8.0.rc3

bug#23325: 25.0.92; insert-char: cannot find all chars if input is unicode name

[Andreas Schwab]

Tino Calancha <f92capac@gmail.com> writes:

> +                                     (assoc-string name (or ucs-names (ucs-names))))

There is no need to refer to the variable ucs-names as the function uses
it anyway.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

bug#23325: 25.0.92; insert-char: cannot find all chars if input is unicode name

[Tino Calancha]

[-- Attachment #1: Type: text/plain, Size: 152 bytes --]



> There is no need to refer to the variable ucs-names as the function uses
> it anyway.
Thank you.  Corrected in the attached new patch.

Cheers,
Tino

[-- Attachment #2: Type: text/plain, Size: 1081 bytes --]

From 4c9b21e7a65cfc17f679906d9ebfa6ac9a11bedb Mon Sep 17 00:00:00 2001
From: Tino Calancha <f92capac@gmail.com>
Date: Fri, 22 Apr 2016 16:50:17 +0900
Subject: [PATCH] describe-char: fix insert char documentation

* lisp/descr-text.el (describe-char):
Only 'ucs-names' entries can be inserted by unicode name (Bug#23325).
---
 lisp/descr-text.el | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lisp/descr-text.el b/lisp/descr-text.el
index a352ed0..5f1a430 100644
--- a/lisp/descr-text.el
+++ b/lisp/descr-text.el
@@ -619,7 +619,7 @@ describe-char
                           (let ((name
                                  (or (get-char-code-property char 'name)
                                      (get-char-code-property char 'old-name))))
-                            (if name
+                            (if (and name (assoc-string name (ucs-names)))
                                 (format
                                  "type \"C-x 8 RET %x\" or \"C-x 8 RET %s\""
                                  char name)
-- 
2.8.0.rc3

bug#23325: 25.0.92; insert-char: cannot find all chars if input, is unicode name

[Paul Eggert]

Thanks, I installed that patch and am closing the bug report.