The "which" utility is not guaranteed to be installed, and if it
is, its behavior is not portable.

Conversely, the "command -v" shell builtin is required to exist in all
POSIX 2008 compliant shells, and is thus guaranteed to work everywhere.

Examples of open-source shells likely to be installed as /bin/sh on
Linux, which implement the 12-year-old standard: ash, bash, busybox,
dash, ksh, mksh and zsh.

A side benefit of using the POSIX portable option is that it does not
require an external disk executable to be forked. This therefore
represents a mild speedup.

Instead of using the 'which' program to resolve a command to an on-disk
executable filename, use the 'env' program to run it. The former is
non-portable and may not be installed on all systems, while the latter
is guaranteed to be anywhere.

I believe the intention of this code was to ensure that when running
completely arbitrary third-party Exec= specifications, we do not
accidentally end up running a conflicting shell function definition.
Using 'env' fulfills this goal by entering a new environment context
with an unmodified environment (we do not make use of env's stated
purpose of creating a modified enviroment context) and then invoking the
command using exec(3) rather than a shell.

The &> redirection operator is a bash-specific shorthand for the POSIX
combined redirection operators >/dev/null 2>&1 and must not be used in
portable scripts under any circumstances.

Trailing whitespace makes my .vimrc unhappy, and serves no useful
purpose. Remove them now, so they don't intermingle with followup commit
changes.

If the desktop environment was GNOME and none of the commands for
detecting the MIME type of a file was available, xdg-mime used to return
exit code 0 even though it failed. The man page says it should return
exit code 3. This commit fixes the issue.

I have considered returning exit code 4 instead since that is what would
be returned if info_kde() or info_generic() fail to find a tool for MIME
type detection. But I have decided to implement the behavior as
specified in the man page.

The exit code of an if construct is 0 if no condition
tested true [^1]. The author of the original code probably was not aware
of this.

[^1]: https://www.gnu.org/software/bash/manual/html_node/Conditional-Constructs.html#Conditional-Constructs

Cameron Fieber authored Mar 25, 2019 and Rex Dieter committed Apr 01, 2019

As per RFC 2396[1] a valid URI scheme may contain a digit:

scheme        = alpha *( alpha | digit | + | - | . )

[1]: https://www.ietf.org/rfc/rfc2396.txt

Also bump Rex Dieter’s copyright date

Signed-off-by: Simon Lees <sflees@suse.de>

Although xdg-su is not currently shipped, fix some trivial documentation
TODOs.

Signed-off-by: Simon Lees <sflees@suse.de>

libdbus expects string arguments to be valid UTF-8. If they are not,
then it aborts, which causes our backgrounded command to terminate
abnormally.

Signed-off-by: Simon Lees <sflees@suse.de>

Signed-off-by: Simon Lees <sflees@suse.de>

xdg-terminal is not working when it needs gsettings to obtain the
default terminal. Thus, xdg-terminal cannot be used in MATE, Cinnamon
or GNOME. This issue was already reported in:
https://bugs.freedesktop.org/show_bug.cgi?id=93231

Thus, this patch provides a temporary workaround until upstream fixes
it.

Link: https://bugs.freedesktop.org/show_bug.cgi?id=93231


Signed-off-by: Simon Lees <sflees@suse.de>

It was observed that t-xdg-open.sh exits after only a fraction of the tests have
been run, e.g.

ASSERTION FAILED: expected command to be run: gio open http://www.freedesktop.org/
ASSERTION FAILED: expected command to be run: gio open http://www.freedesktop.org/
  - opens a URL with gvfs-open if gio open is missing in GNOME 3, GNOME 2, and Cinnamon
gio: http://www.freedesktop.org/: Operation not supported
make: *** [Makefile:21: t-xdg-open.sh] Error 4

Given that some tests are failing on my machine (likely because of the
peculiarities of my archlinux install), the root cause is `set -e`. The nonzero
exit code returned by xdg-open gets returned by run(), which is also returned by
e.g. test_open_url(), which causes the script to exit immediately.

All test passes/failures at present are being defined by explicit assertions,
not by exit codes, and it seems at least plausible that some xdg-open calls are
meant to fail. So rather than report the nonzero error code, j...

Presently, file://localhost/ URLs are totally unsupported: is_file_url_or_path
correctly considers them files, but they are undecoded and hence
check_input_file fails.

While the standardization surrounding file: URLs is admittedly vague [1], AFAIK,
*all* literature, and other implementations, unambiguously demonstrate that
file://localhost/ should be equivalent to file:///:

- The "File URI specification" explicitly linked to from the xdg-utils homepage [2]
- RFC 8089 section 1.1
- RFC 1738 section 3.10
- Observed implementations of Windows `start`, macOS `open`, Firefox, Chrome, IE

Fix this by adding some simple carve-outs for file://localhost specifically in
file_url_to_path.

[1] https://lists.freedesktop.org/archives/xdg/2004-November/003711.html
[2] https://edeproject.org/spec/file-uri-spec.txt



Signed-off-by: Richard Tollerton <rich.tollerton@ni.com>

Reverts a small part of prior commit
56991bc1

 Please enter the commit message for your changes. Lines starting

石博文 authored Apr 20, 2018 and Rex Dieter committed May 10, 2018

In DDE, a tools called `dde-open` is used to open preferred app.

DDE use `Deepin` as a desktop name currently,
but in some older version, it was `DEEPIN` or `deepin`

dde-open is provide by dde-api, see [1].

[1]: https://cr.deepin.io/c/dde/dde-api/+/33751

patch courtesy of Simon Lees (SUSE) <sflees@suse.de>

BUG: 106343

Luís Pereira authored May 17, 2017 and Rex Dieter committed Feb 27, 2018

Commit 6387086e introduced LXQt support.
In xdg-open it assumed that LXQt and LXDE are the same. They aren't. LXQt
does not have pcmanfm, it has pcmanfm-qt.
LXQt doesn't want to rely on pcmamfm-qt to handle it: LXQt is very modular
and can be deployed without pcmanfm-qt.

open_generic() works for LXQt but it depends on mimeopen and mimetype. In
the future we will want, for sure, to drop those dependencies. That's the
reason for the existence of open_lxqt().

Credit to Ville Skyttä <ville.skytta@iki.fi>