On 2 September 2016, Justin M. Keyes <[email protected]> wrote:
> On Thu, Sep 1, 2016 at 4:20 PM, LCD 47 <[email protected]> wrote:
> > On 1 September 2016, Nikolay Aleksandrovich Pavlov <[email protected]>
> > wrote:
> >> 2016-09-01 15:47 GMT+03:00 LCD 47 <[email protected]>:
> >> > What is the recommended way to get the buffer number of a file
> >> > starting from the filename?
> >> >
> > [...]
> >> > Based on the above (and taking a peek at the sources), I came up
> >> > with this attempt:
> >> >
> >> > function! Name2Buf(fname) abort
> >> > if exists('+shellslash')
> >> > let old_shellslash = &shellslash
> >> > let &shellslash = 1
> >> > let buf = bufnr(escape( fnamemodify(a:fname, ':p'),
> >> > '\*?,{}[' ))
> >> > let &shellslash = old_shellslash
> >> > else
> >> > let buf = bufnr(escape( fnamemodify(a:fname, ':p'),
> >> > '\*?,{}[' ))
> >> > endif
> >> >
> >> > return buf
> >> > endfunction
> >> >
> >> > It mostly works, until I try it on a file named a,b\{2,3\}.txt:
> >> >
> >> > :echo expand('%:p')
> >> > /home/lcd047/tmp/a,b\{3.4}.txt
> >> >
> >> > :Name2Buf(expand('%:p'))
> >> > -1
> >> >
> >> > However the naive bufnr(expand('%:p')) works, but it shouldn't,
> >> > because of the two commas ",":
> >> >
> >> > echomsg bufnr(expand('%:p'))
> >> > 1
> >> >
> >> > So, what _is_ the right way to do this?
> >>
> >> I would go with `fnameescape()`, but this does not work with your
> >> file name either. Additional problem is that `fnameescape()` does
> >> not work correctly on Windows.
> >
> > fnameescape() is definitely wrong here, it escapes characters
> > that have nothing to do with the "file-pattern" set (f.i. '%' and
> > '#'). Looking at the code, the "bad" characters are a mixture of
> > glob characters with regexp characters. I don't think it's possible
> > to disarm them all with a simple escape(). I wonder why it's
> > possible to do all those weird globbing in bufnr() (does anybody
> > actually use them?), but it isn't possible to do an exact string
> > match.
>
> I use basic regex with bufnr(), to *avoid* bufnr() useless magic:
>
> https://github.com/justinmk/vim-dirvish/blob/master/autoload/dirvish.vim#L306-L330
Sadly the "^ ... $" pattern still fails for a,b\{3.4}.txt:
:echo expand('%:p')
/home/lcd047/tmp/a,b\{3.4}.txt
:echo bufnr('^' . expand('%:p') . '$')
-1
But it succeeds without the anchors:
:echo bufnr(expand('%:p'))
1
> It seems to me that long ago the decision should have been to only
> have "magic" in _commands_, not functions. Plugin authors (and
> vimrc authors) usually do not want user-configured behavior, it is
> a headache. E.g. there is zero reason that substitute() should care
> about any global option. For typical editing one invokes :substitute,
> not substitute().
Absolutely, this is a long-time gripe of mine, too. Among other
things, all regexps in scripts have to be littered with "\m" or "\v",
because somebody, somewhere (presumably under a heavy rock), _might_
be running with "set nomagic". And it goes only half way in the other
direction too, since there is no such thing as "set verymagic", for
those of us that aren't so fond of ye olde BRE syntax.
> This sort of inconsistency was not "inherited from Vi", since
> FEAT_EVAL was not in Vi.
Sadly it's still 20+ years too late to fix now.
/lcd
--
--
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.
