for example: dmenu -m '-9001'

This reverts commit a4053bc4e4.

My bad, was working on the wrong branch and accidently pushed it...
*facepalm*

When the input text fully matches a single item, do not draw the item
and highlight the input text to show that it matches an item in
opposition to regular input text not matching anything.

Without this, we discard the item if it's longer than assigned width
instead of truncating it.

- extract drawitem function (code deduplication)
- fix bug where inputw was not correctly calculated from the widest item, but
  just from the one with the longest strlen() which is not the same. It's better
  now, but does not account for fallback fonts, since it would be too slow to
  calculate all the correct item widths on startup.
- minor code style fixes (indentation, useless line breaks)

This prevents accessing to a potentially out-of-bounds memory section.

Signed-off-by: Lucas Gabriel Vuotto <l.vuotto92@gmail.com>

Let the user configure word boundaries other than ' ', only works with
the portable character set.

This was always broken.

Reproduce:
	focus client on second monitor, spawn dmenu with -m 0.

Result:
	Old wrong behaviour: dmenu spawns on second monitor (focused client).
	Now: dmenu spawns on specified monitor (first).

... we don't allow passing text is NULL anymore either, for that behaviour
just use drw_rect() (it is used in dwm).

don't do these checks on this level. However for resource drw_*_free
we will allow it.

... remove intermediary variables

this makes it slightly easier to add custom schemes aswell

just compare the size of the input string

look for exact matches comparing the user input against the item text

from Joshua Lloret on the ML:

"attached is a small patch to enable pasting from clipboard, as well as primary.

It seems like there was already code in there to allow this, but since there was never any case to match the upper case 'Y', that inline if would always evaluate to false."

Adds clarity. Typedefs for structs are definitely a discussion matter,
but there's no reason to hide a simple data-structure behind a
meaningless typedef.

- move main to bottom, usage above main.
- dont use variable names with function prototypes.
- space before if, for, while, etc: 'if(' -> 'if ('.

this makes the code-style more consistent

this makes the code-style more consistent aswell.

True, False are X11-specific, make sure to use c99 stdbool for
program-specific things.

... also remove left-over vim mode string in config.

this fixes a crash on NetBSD because it requires -D_XOPEN_SOURCE (strdup, usleep).

thanks k0ga and stateless for reporting and fixing this issue!

... also some style improvements.

These variables make it simpler to change the paths to this for ports.
`pkg-config` is avoided because it sucks, in particular for cross-compilation.

A commented path for *BSD is added, the Xft includes are located at:
/usr/X11R6/include there.

Also already bump the version number to 4.6, a release will come
approximately in August.

This is the style how it was before the big Xft change. The colors were
inverted, this was not the case before the change.

Reported by "zvz" on #suckless IRC, thanks!

- libdraw, util: add drw.{c,h}, util.{c,h} and update code.
- libdraw: fix drw_rect(): use w and h parameter.
- libdraw: print errstr if last character in string was ":" (sbase).
- libdraw: drw_clr_free() allow valid free(NULL).
- config.def.h: set default font to monospace.
- cleanup() on exit.
- LICENSE: update license string for dmenu -v to 2015.
- LICENSE: add myself to LICENSE

reproduce: ./dmenu; send EOF; press ctrl+enter.

f6581ca966/monarg.patch becomes upstream now