2017年5月16日火曜日 14時36分04秒 UTC+9 Bram Moolenaar:
> > 2017年5月11日木曜日 21時52分48秒 UTC+9 Bram Moolenaar:
> > > Ken Hamada wrote:
> > >
> > > > Hi list, I noticed that E16 occurs with regexp.
> > > >
> > > > set re=1
> > > > echo '' =~# '[\uff-\uf0]'
> > > > E16: Invalid range
> > > > echo '' =~# '[\u3000-\u4000]'
> > > > E16: Invalid range
> > > >
> > > > I think this is unintuitive error message because E16 reminds us the
> > > > range of
> > > > commands (for example :0buffer). So I suggest two exclusive options:
> > > >
> > > > - Add a note that the error can occur with regexp at :h E16.
> > > > - Change the error message and number.
> > >
> > > From the old days a bit of memory was saved by limiting the number of
> > > error messages. Now that we are more interested in useful errors,
> > > splitting off specific cases is useful.
> > >
> > > > BTW the behaviour *Limit to a range of 256 chars* differs due to the
> > > > regex engine version.
> > > > I could not tell the difference until I dived into the source code. I
> > > > would like some warning to be written regarding this topic around :h
> > > > /[.
> > >
> > > I thought this was mentioned somewhere, but can't find it right now.
> > > Please suggest an improvement.
> >
> > Thank you Bram and Ken Tanaka.
> >
> > I think creating a new error number will be better.
> > Here's the patch. What do you think?
>
> Now that we are at it, we can make it more specific: One error for the
> reverse range and one for a too large range.
>
> It's also nice to have a test for these.
Thank you Bram for your advice.
I updated the patch and added some tests.
diff --git a/runtime/doc/pattern.txt b/runtime/doc/pattern.txt
index d6764096a..d487a3d65 100644
--- a/runtime/doc/pattern.txt
+++ b/runtime/doc/pattern.txt
@@ -1075,13 +1075,16 @@ x A single character, with no special meaning,
matches itself
`:substitute` command the whole command becomes the pattern. E.g.
":s/[/x/" searches for "[/x" and replaces it with nothing. It does
not search for "[" and replaces it with "x"!
-
+ *E944* *E945*
If the sequence begins with "^", it matches any single character NOT
in the collection: "[^xyz]" matches anything but 'x', 'y' and 'z'.
- If two characters in the sequence are separated by '-', this is
shorthand for the full list of ASCII characters between them. E.g.,
- "[0-9]" matches any decimal digit. Non-ASCII characters can be
- used, but the character values must not be more than 256 apart.
+ "[0-9]" matches any decimal digit. If the starting character exceeds
+ the ending character like [c-a], E944 occurs. Non-ASCII characters
+ can be used, but the character values must not be more than 256 apart
+ in the old regexp engine. For example, searching by [\u3000-\u4000]
+ after setting re=1 emits E945 error. Prepending \%#=2 will fix it.
- A character class expression is evaluated to the set of characters
belonging to that character class. The following character classes
are supported:
diff --git a/runtime/doc/todo.txt b/runtime/doc/todo.txt
index 0736ada58..3b9356073 100644
--- a/runtime/doc/todo.txt
+++ b/runtime/doc/todo.txt
@@ -930,8 +930,6 @@ Patch to handle integer overflow. (Aaron Burrow, 2013 Dec
12)
Patch to add "ntab" item in 'listchars' to repeat first character. (Nathaniel
Braun, pragm, 2013 Oct 13) A better solution 2014 Mar 5.
-/[b-a] gives error E16, should probably be E769.
-
7 Windows XP: When using "ClearType" for text smoothing, a column of yellow
pixels remains when typing spaces in front of a "D" ('guifont' set to
"lucida_console:h8").
diff --git a/src/regexp.c b/src/regexp.c
index e1f6484c0..93507a73a 100644
--- a/src/regexp.c
+++ b/src/regexp.c
@@ -358,6 +358,8 @@ static char_u *regprop(char_u *);
static int re_mult_next(char *what);
static char_u e_missingbracket[] = N_("E769: Missing ] after %s[");
+static char_u e_reversed_range[] = N_("E944: Reversed range in character
class");
+static char_u e_large_class[] = N_("E945: Too large range in character class");
static char_u e_unmatchedpp[] = N_("E53: Unmatched %s%%(");
static char_u e_unmatchedp[] = N_("E54: Unmatched %s(");
static char_u e_unmatchedpar[] = N_("E55: Unmatched %s)");
@@ -2426,14 +2428,14 @@ collection:
endc = coll_get_char();
if (startc > endc)
- EMSG_RET_NULL(_(e_invrange));
+ EMSG_RET_NULL(_(e_reversed_range));
#ifdef FEAT_MBYTE
if (has_mbyte && ((*mb_char2len)(startc) > 1
|| (*mb_char2len)(endc) > 1))
{
/* Limit to a range of 256 chars */
if (endc > startc + 256)
- EMSG_RET_NULL(_(e_invrange));
+ EMSG_RET_NULL(_(e_large_class));
while (++startc <= endc)
regmbc(startc);
}
diff --git a/src/regexp_nfa.c b/src/regexp_nfa.c
index 120861a46..6df50f061 100644
--- a/src/regexp_nfa.c
+++ b/src/regexp_nfa.c
@@ -1851,7 +1851,7 @@ collection:
endc = startc;
startc = oldstartc;
if (startc > endc)
- EMSG_RET_FAIL(_(e_invrange));
+ EMSG_RET_FAIL(_(e_reversed_range));
if (endc > startc + 2)
{
diff --git a/src/testdir/test_regexp_utf8.vim b/src/testdir/test_regexp_utf8.vim
index c54b65081..f38332071 100644
--- a/src/testdir/test_regexp_utf8.vim
+++ b/src/testdir/test_regexp_utf8.vim
@@ -137,3 +137,18 @@ func Test_classes_re2()
call s:classes_test()
set re=0
endfunc
+
+func Test_reversed_range()
+ for re in range(0, 2)
+ exe 'set re=' . re
+ call assert_fails('call match("abc def", "[c-a]")', 'E944:')
+ endfor
+endfunc
+
+func Test_large_class()
+ set re=1
+ call assert_fails('call match("abc def", "[\u3000-\u4000]")', 'E945:')
+ set re=2
+ call assert_equal(0, 'abc def' =~# '[\u3000-\u4000]')
+ call assert_equal(1, "\u3042" =~# '[\u3000-\u4000]')
+endfunc
Ken Hamada
--
--
You received this message from the "vim_dev" maillist.
Do not top-post! Type your reply below the text you are replying to.
For more information, visit http://www.vim.org/maillist.php
---
You received this message because you are subscribed to the Google Groups
"vim_dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
