Added completions/translations for zfs and zpool #4608

PenegalECI · 2017-12-18T15:48:03Z

Description

This PR adds completions for the OpenZFS zfs and zpool commands. It also updates the *.po files and adds corresponding translations for French, my mother tongue. There are merge conflicts, but they only belong, AFAIK, in conflicting metadata in *.po files, due to the POT regeneration I did while others did it on their side. This is only related to POT timestamp, which can be solved using either version, and to translators notes which I added in fr.po, which can be solved using my version, as there is nothing semantically conflicting here.

TODOs:

Changes to fish usage are reflected in user documentation/manpages.
Tests have been added for regressions fixed
User-visible changes noted in CHANGELOG.md

PenegalECI · 2017-12-19T14:16:56Z

I might add that I did not solve the merge conflicts, even if I could, as Github told me Don’t worry, you can still create the pull request. but, if I have to correct this myself, no problem, just tell me.

faho · 2017-12-18T19:22:34Z

share/functions/__fish_complete_zfs_pools.fish

@@ -0,0 +1,3 @@
+function __fish_complete_zfs_pools -d "Completes with available ZFS pools"
+	zpool list -o name,comment -H | sed -e 's/\t-//g'


Use string replace -a \t'-' '' instead of sed, please.

Did it, but should the \t really be outside the quotes?

Yes. string currently does not interpret these sequences itself, so you'll need to pass it a literal tab character. And the way to generate that is to use \t outside of quotes.

Ok, done then.

faho · 2017-12-18T21:40:21Z

share/functions/__fish_complete_zfs_rw_properties.fish

+		echo -e "volmode\t"(_ "How to expose volumes to OS")" (default, geom, dev, none)"
+	end		
+	# User properties; the /dev/null redirection masks the possible "no datasets available"
+	zfs list -o all ^/dev/null | head -n 1 | string replace -a -r "\s+" "\n" | grep ":" | tr '[:upper:]' '[:lower:]'


Try string match -e ":" | string lower

string lower? The current doc does not mention this function.

It's new since fish 2.7.0, and the online documentation hasn't been updated.

Then, done.

faho · 2017-12-18T21:41:16Z

share/functions/__fish_is_zfs_feature_enabled.fish

@@ -0,0 +1,18 @@
+function __fish_is_zfs_feature_enabled -a feature target -d "Returns 0 if the given ZFS feature is available or enabled for the given full-path target (zpool or dataset), or any target if none given"
+	set -l pool (echo "$target" | cut -d '/' -f1)


string replace -r '/.*' '' -- $target.

The quotes here are superfluous because function -a guarantees that it'll be one element.

OK, was unsure about that.

faho · 2017-12-18T21:42:52Z

share/functions/__fish_is_zfs_feature_enabled.fish

+function __fish_is_zfs_feature_enabled -a feature target -d "Returns 0 if the given ZFS feature is available or enabled for the given full-path target (zpool or dataset), or any target if none given"
+	set -l pool (echo "$target" | cut -d '/' -f1)
+    set -l feature_name ""
+    set -l pattern "[[:space:]]"


pcre2 (which string uses) has [[:space:]], and also \s.

You mean I can safely replace | grep "$pattern" by | string match -e '\s' in the rest of this function?

-r/--regex, not -e (which is short for --entire, which means for match to print the entire string that contained the match), but yes.

faho · 2017-12-18T21:43:21Z

share/functions/__fish_is_zfs_feature_enabled.fish

+		return 1
+	end
+	set -l state (echo $feature_name | cut -f3)
+    echo "$feature_name" | grep -q -E "(active|enabled)"


string match -qr '(active|enabled)' -- $feature_name should work.

Indeed; besides, $state is to be tested, not $feature_name.

faho · 2017-12-18T21:46:59Z

share/functions/__fish_complete_zfs_rw_properties.fish

+	echo -e "atime\t"(_ "Update access time on read")" (on, off)"
+	echo -e "canmount\t"(_ "Is the dataset mountable")" (on, off, noauto)"
+	set -l additional_algs ''
+    if test $OS = 'FreeBSD'; or test $OS = 'SunOS'


You could also use if contains -- $OS FreeBSD SunOS. Though I can see an argument for symmetry with the other checks.

Oh, your syntax is much better; I was just unsure how to make mine prettier.

Should I also use contains for other alike tests, or can I let test?

You can, sure. It's a matter of taste. Here I find it nicer because it makes working with more than one possibility easier. For just one OS, I think I personally would lean towards test, unless of course the list can change, in which case contains is easier to adapt.

Pick whichever you prefer.

Let's just say it's fine as it is.

faho · 2017-12-18T21:48:09Z

share/functions/__fish_complete_zfs_write_once_properties.fish

+	echo -e "utf8only\t"(_ "Reject non-UTF-8-compliant filenames")" (on, off)"
+	if test $OS = "Linux"
+		echo -e "overlay\t"(_ "Allow overlay mount")" (on, off)"
+		if command -s sestatus >/dev/null ^&1 # SELinux is enabled


command has a "-q" option. command -sq sestatus.

It seems that share/completions/mkdir.fish could use that too.

Yes. A bunch could - it's supported since 2.5.0, and we haven't gotten around to modifying all the scripts yet. It's just cleanup, so it's not too high priority.

faho · 2017-12-18T21:49:33Z

share/functions/__fish_is_zfs_feature_enabled.fish

+        set feature_name (zpool get all -H | grep "$pattern")
+    else
+        set pattern "$pool$pattern$feature$pattern"
+        set feature_name (zpool get all -H $pool | grep "$pattern")


Is this else needed? If $pool is empty, it'll expand to nothing, so both the $pool argument and the "$pool string should just be gone.

I'm unsure about that, because I had a hard time testing this piece of code; I remember it failed without this else, but why? That, I can't remember. The reason may have vanished in subsequent fish releases, though, as I use 2.4.0.

You might have gotten the quoting wrong.

In the set pattern, you want "$pool$pattern..." inside quotes because you want $pool to expand to the empty string. For zpool -H, you want it without quotes (i.e. zpool get all -H $pool, because you want it to expand to no argument.

Oh, wait! Now I remember: without a target specified, I check any available pool, as, in this case, the function only checks if the feature is activated or enabled on the system. But, if a pool or a full-path target is specified, then the function does the check against this specific pool and, then, I have to find the matching line only for the specified pool. So, the else serves to narrow the zfs call to the specified pool if possible, thus decreasing processing time; I don't know if it's worth it, but it doesn't hurt.

faho · 2017-12-18T21:51:43Z

share/functions/__fish_is_zfs_feature_enabled.fish

+	if test $status -ne 0 # No such feature
+		return 1
+	end
+	set -l state (echo $feature_name | cut -f3)


This can also be echo $feature_name | read -l _ _ state. Which will use space and tab as the delimiter, IIRC.

I don't understand your replacement code. At this point, $feature_name contains something like this, in tab-separated values:

pgsql feature@lz4_compress active local

How does your suggestion work against this?

Do you want the active or local part or both?

read takes variable names (the "-l" is "--local", like in set -l) and splits the line it reads from stdin into those variables, using space and tab as the delimiter if nothing else has been given (via the new "--delimiter" option in 3.0). _ here is just a throwaway variable name (we actually have $_, but its overwritten after each command - it's a bit of a hack currently, but it works and we should allow it).

So

echo "pgsql"\t"feature@lz4_compress"\t"active"\t"local" | read -l _ _ state

writes the first "field" (pgsql) into $_, the second (lz4_compress) into $_ again and the rest (because there's no variable left) into $state (keeping the tab between "active" and "local").

So if you want just the "active" part, you'll have to add one more _ after "state" (read -l _ _ state _), if you want just the "local", add one more _ before, and if you want both in separate variables add another proper variable name.

You might be familiar with something like this from tuple-unpacking in e.g. python or go - result, _ := somefunction()).

Oh, nice construction 😍. I'm more Ruby, but I understand the concept; nice one. I was only looking for the third column, so I go with read -l _ _ state _

faho · 2017-12-18T21:54:21Z

share/completions/zfs.fish

+end
+
+function __fish_zfs_append -d "Internal completion function for appending string to the ZFS commandline"
+    set str (commandline -tc| sed -ne "s/\(.*,\)[^,]*/\1/p"|sed -e "s/--.*=//")


Should translate reasonably well to string - you don't need to mask the parentheses (characters are special by default), so something like

commandline -ct | string replace -rf '(.*,)[^,]*' '$1' | string replace -r -- '--.*=' ''

?

I'll try that. Note that __fish_append could be corrected as well, as I essentially borrowed its code and stripped off what was useless to me; that's also why I used sed: I was unsure to fully understand these regexp, so I let them as is.

In fact, it would be a bit difficult to test that, as my current config would not allow me to easily test that; as you know fish way better than me, I assume you are right and blindly replaced my code with yours.

faho · 2017-12-19T20:14:43Z

if I have to correct this myself, no problem, just tell me.

Nah, we can do that. It's probably just git merge -s theirs anyway (i.e. picking your changes).

PenegalECI · 2017-12-20T13:16:23Z

Here you are.

faho · 2017-12-21T15:41:37Z

Merged as dcf9ce6 with git rebase --interactive -X ours master. (The "-X ours" means that, in case of conflicts, the checked out branch wins. Since I had this PR checked out, it meant that @PenegalECI's changes to the translations were used)

Added completions/translations for zfs and zpool

8d3ee64

faho added completions enhancement labels Dec 18, 2017

faho added this to the fish-3.0 milestone Dec 18, 2017

faho requested changes Dec 19, 2017

View reviewed changes

Corrections following @faho guidance

0c4aaa8

faho closed this Dec 21, 2017

faho added the release notes Something that is or should be mentioned in the release notes label Dec 21, 2017

github-actions bot locked as resolved and limited conversation to collaborators Apr 17, 2020

		@@ -0,0 +1,3 @@
		function __fish_complete_zfs_pools -d "Completes with available ZFS pools"
		zpool list -o name,comment -H \| sed -e 's/\t-//g'

		@@ -0,0 +1,18 @@
		function __fish_is_zfs_feature_enabled -a feature target -d "Returns 0 if the given ZFS feature is available or enabled for the given full-path target (zpool or dataset), or any target if none given"
		set -l pool (echo "$target" \| cut -d '/' -f1)

Added completions/translations for zfs and zpool #4608

Added completions/translations for zfs and zpool #4608

Conversation

PenegalECI commented Dec 18, 2017

Description

TODOs:

PenegalECI commented Dec 19, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

PenegalECI Dec 20, 2017 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

faho commented Dec 19, 2017

PenegalECI commented Dec 20, 2017

faho commented Dec 21, 2017

PenegalECI Dec 20, 2017 •

edited