* [patch] Add ldap completer
@ 2018-06-19 12:57 Matthew Martin
2018-06-19 20:21 ` Oliver Kiddle
0 siblings, 1 reply; 3+ messages in thread
From: Matthew Martin @ 2018-06-19 12:57 UTC (permalink / raw)
To: zsh-workers
OpenBSD recently gained an ldap command. This is my first stab at
completing it.
For the protocol _describe, is -I correct? I don't grok the differences
between the prefixes and suffixes. For host is -S '' correct for not
adding a space after and should the completer avoid adding the space?
The ldapi protocol is not completed correctly because I don't know how
to make _files replace / with %2f.
(N.B. The URL pattern below is correct by RFC, I'll submit a patch for
the OpenBSD man page soon.)
- Matthew Martin
diff --git a/Completion/BSD/Command/_ldap b/Completion/BSD/Command/_ldap
new file mode 100644
index 000000000..ba8b981b4
--- /dev/null
+++ b/Completion/BSD/Command/_ldap
@@ -0,0 +1,79 @@
+#compdef ldap
+
+local -a commands scopes
+commands=(
+ search:'search a directory'
+)
+scopes=(
+ base:'base object only'
+ one:'one level'
+ sub:subtree
+)
+
+_ldap_url() {
+ local -a expl protocols tags
+ protocols=(
+ ldap:'TCP in plaintext'
+ ldaps:'TLS'
+ ldap+tls:'TCP and use StartTLS'
+ ldapi:'connect to a socket'
+ )
+
+ # [protocol://]host[:port][/basedn[?[attribute,...][?[scope][?[filter]]]]]
+ if ! compset -P '*://'; then
+ tags=(protocol)
+ fi
+
+ if ! compset -P '*/'; then
+ if compset -P '*:'; then
+ tags=(port)
+ else
+ tags+=(host)
+ fi
+ else
+ case $PREFIX in
+ *\?*\?*\?*) tags=(filter);;
+ *\?*\?*) tags=(scope);;
+ *\?*) tags=(attribute);;
+ *) tags=(basedn);;
+ esac
+ compset -P '*\?'
+ fi
+
+ _tags $tags
+ while _tags; do
+ _requested protocol && _describe -t protocol protocol protocols -I ://
+ _requested host && _hosts -S ''
+ _requested port && _guard '|<1-65535>' port
+ _requested basedn expl 'base DN'
+ _requested attribute expl attribute
+ _requested scope && _describe -t scope scope scopes
+ _requested filter expl filter
+ done
+}
+
+if (( CURRENT == 2 )); then
+ _describe command commands
+else
+ shift words; (( CURRENT-- ))
+ case $words[1] in
+ search)
+ _arguments -s -S -A '-*' \
+ '-b+[specify base DN]:base DN:' \
+ '-c+[specify CA file]:CA file:' \
+ '-D+[specify bind DN]:bind DN:' \
+ '-H+[specify URL]: :_ldap_url' \
+ '-L[output in LDIF]' \
+ '-l+[specify time limit or 0 for no limit]:time limit [0]:' \
+ '-s+[specify scope]:scope [sub]:(($scopes))' \
+ '-v[be verbose]' \
+ '-W[prompt for bind secret]' \
+ '-w+[specify bind secret]:bind secret:' \
+ '-x[use simple authentication]' \
+ '-Z[use StartTLS]' \
+ '-z+[specify maximum number of results or 0 for no limit]:size limit [0]:' \
+ ':: :_guard "*=*" "filter"' \
+ '*:attribute:'
+ ;;
+ esac
+fi
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [patch] Add ldap completer
2018-06-19 12:57 [patch] Add ldap completer Matthew Martin
@ 2018-06-19 20:21 ` Oliver Kiddle
2018-06-28 4:59 ` Matthew Martin
0 siblings, 1 reply; 3+ messages in thread
From: Oliver Kiddle @ 2018-06-19 20:21 UTC (permalink / raw)
To: zsh-workers
Matthew Martin wrote:
> For the protocol _describe, is -I correct? I don't grok the differences
> between the prefixes and suffixes. For host is -S '' correct for not
> adding a space after and should the completer avoid adding the space?
You should use -S.
The normal way would actually be to have:
compset -S ':*' || suf=( -S :// )
Then append $suf to _describe.
-S/-P are what you normally use for separators like : / , etc.
These work together with the suffix auto-removal. The completion
matching is aware of them and considers them for some purposes.
-s/-p suffixes/prefixes are included in matching control. So if you use, e.g:
compadd -M 'r:|/=* r:|=*' -p "head/" - tail
then h/t</tab> will complete to head/tail
This is behind things like _multi_parts. They are also quoted
automatically whereas for -P/-S you might need compquote so these are
sometimes used to take advantage of that.
-i/-I is akin to directly modifying IPREFIX and ISUFFIX. They tend
to be used when bypassing matching but can have other uses. Aside
from quoting, they don't behave exactly like a direct modification of
IPREFIX/ISUFFIX but that goes beyond my understanding.
> The ldapi protocol is not completed correctly because I don't know how
> to make _files replace / with %2f.
I don't think you can currently without redoing much of _files. You
could probably still get _path_files (with -S %2f) to do much of the
work.
The completion system does track layers of quoting following compset -q
but you can't give it custom forms as %2f would be.
> + # [protocol://]host[:port][/basedn[?[attribute,...][?[scope][?[filter]]]]]
> + if ! compset -P '*://'; then
> + tags=(protocol)
> + fi
> +
> + if ! compset -P '*/'; then
> + if compset -P '*:'; then
In general, where you have compset -P. compset -S is also applicable
when completing whatever falls to the left of the separator.
> + _tags $tags
> + while _tags; do
> + _requested protocol && _describe -t protocol protocol protocols -I ://
> + _requested host && _hosts -S ''
> + _requested port && _guard '|<1-65535>' port
_guard was created to solve a particular problem. A quick demonstration
of which is the following: _arguments '--long' ':number:'
--l<tab> now does not complete to --long because number is being
offered.
I tend to limit it's use to these situations. Even if I've typed some
letters, getting the "port" description can still be useful.
> + _requested basedn expl 'base DN'
> + _requested attribute expl attribute
> + _requested scope && _describe -t scope scope scopes
If the scope is followed by '?' and a filter, you might want a ? suffix
on this. Perhaps in auto-removable form. Or the default suffix could be
auto-removable.
> + _requested filter expl filter
> + done
The tags loop is not correct. You need to break out of the loop if
matches were added. _alternative would work here. $tags never contains
more than one value, however, so the easiest would just be to directly
generate matches in each place where you assign to $tags.
If you don't like adding && ret=0 everywhere, compare
compstate[nmatches] from the start and end of the function.
> + ':: :_guard "*=*" "filter"' \
Is this really doing what you want: offering filter only if there is an
equals sign in the current word?
Oliver
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [patch] Add ldap completer
2018-06-19 20:21 ` Oliver Kiddle
@ 2018-06-28 4:59 ` Matthew Martin
0 siblings, 0 replies; 3+ messages in thread
From: Matthew Martin @ 2018-06-28 4:59 UTC (permalink / raw)
To: zsh-workers
On Tue, Jun 19, 2018 at 10:21:48PM +0200, Oliver Kiddle wrote:
> Matthew Martin wrote:
> > For the protocol _describe, is -I correct? I don't grok the differences
> > between the prefixes and suffixes. For host is -S '' correct for not
> > adding a space after and should the completer avoid adding the space?
>
> You should use -S.
> The normal way would actually be to have:
> compset -S ':*' || suf=( -S :// )
> Then append $suf to _describe.
>
> -S/-P are what you normally use for separators like : / , etc.
> These work together with the suffix auto-removal. The completion
> matching is aware of them and considers them for some purposes.
>
> -s/-p suffixes/prefixes are included in matching control. So if you use, e.g:
> compadd -M 'r:|/=* r:|=*' -p "head/" - tail
> then h/t</tab> will complete to head/tail
> This is behind things like _multi_parts. They are also quoted
> automatically whereas for -P/-S you might need compquote so these are
> sometimes used to take advantage of that.
>
> -i/-I is akin to directly modifying IPREFIX and ISUFFIX. They tend
> to be used when bypassing matching but can have other uses. Aside
> from quoting, they don't behave exactly like a direct modification of
> IPREFIX/ISUFFIX but that goes beyond my understanding.
Thanks, that helped. It seems that most times using -S sould check the
SUFFIX first, so I've added that to scope as well.
> > The ldapi protocol is not completed correctly because I don't know how
> > to make _files replace / with %2f.
>
> I don't think you can currently without redoing much of _files. You
> could probably still get _path_files (with -S %2f) to do much of the
> work.
> The completion system does track layers of quoting following compset -q
> but you can't give it custom forms as %2f would be.
(Note in case this ever happens: ed needs something similar since
ed 'foo\\bar' edits the file foo\bar.)
> > + # [protocol://]host[:port][/basedn[?[attribute,...][?[scope][?[filter]]]]]
> > + if ! compset -P '*://'; then
> > + tags=(protocol)
> > + fi
> > +
> > + if ! compset -P '*/'; then
> > + if compset -P '*:'; then
>
> In general, where you have compset -P. compset -S is also applicable
> when completing whatever falls to the left of the separator.
>
> > + _tags $tags
> > + while _tags; do
> > + _requested protocol && _describe -t protocol protocol protocols -I ://
> > + _requested host && _hosts -S ''
> > + _requested port && _guard '|<1-65535>' port
>
> _guard was created to solve a particular problem. A quick demonstration
> of which is the following: _arguments '--long' ':number:'
> --l<tab> now does not complete to --long because number is being
> offered.
> I tend to limit it's use to these situations. Even if I've typed some
> letters, getting the "port" description can still be useful.
Noted and fixed.
> > + _requested basedn expl 'base DN'
> > + _requested attribute expl attribute
> > + _requested scope && _describe -t scope scope scopes
>
> If the scope is followed by '?' and a filter, you might want a ? suffix
> on this. Perhaps in auto-removable form. Or the default suffix could be
> auto-removable.
Added with -suffix check.
> > + _requested filter expl filter
> > + done
>
> The tags loop is not correct. You need to break out of the loop if
> matches were added. _alternative would work here. $tags never contains
> more than one value, however, so the easiest would just be to directly
> generate matches in each place where you assign to $tags.
>
> If you don't like adding && ret=0 everywhere, compare
> compstate[nmatches] from the start and end of the function.
It's possible for the tags array to have both protocol and host. _tags
makes much more sense now knowing it's looping over groups. Two
questions
Why does _tags not do the compstate[nmatches] check? It would ensure
that the promises made to the user in the tag-order api are kept and
prevents mistaks.
Would it be worthwile to write a function that takes a word specifier
(e.g. [protocol://]host[:port][/basedn[?[attribute,...][?[scope][?[filter]]]]] )
to make cases like this easier? Next up I was going to submit _smtp
which has a uri argument for -s. Not sure how many other commands take
completable uris.
> > + ':: :_guard "*=*" "filter"' \
>
> Is this really doing what you want: offering filter only if there is an
> equals sign in the current word?
Indeed not and removed.
Thanks for the review!
- Matthew Martin
diff --git a/Completion/BSD/Command/_ldap b/Completion/BSD/Command/_ldap
new file mode 100644
index 000000000..8fa17e2f8
--- /dev/null
+++ b/Completion/BSD/Command/_ldap
@@ -0,0 +1,87 @@
+#compdef ldap
+
+local -a commands scopes
+commands=(
+ search:'search a directory'
+)
+scopes=(
+ base:'base object only'
+ one:'one level'
+ sub:subtree
+)
+
+_ldap_url() {
+ local nm=$compstate[nmatches]
+ local -a expl protocols suf_proto suf_scope tags
+ protocols=(
+ ldap:'TCP in plaintext'
+ ldaps:'TLS'
+ ldap+tls:'TCP and use StartTLS'
+ ldapi:'connect to a socket'
+ )
+
+ # [protocol://]host[:port][/basedn[?[attribute,...][?[scope][?[filter]]]]]
+ if ! compset -P '*://'; then
+ tags=(protocol)
+ compset -S ':*' || suf_proto=( -S :// )
+ fi
+
+ if ! compset -P '*/'; then
+ if compset -P '*:'; then
+ tags=(port)
+ compset -S '/*'
+ else
+ if ! compset -S '://*'; then
+ tags+=(host)
+ compset -S '[:/]*'
+ fi
+ fi
+ else
+ case $PREFIX in
+ *\?*\?*\?*) tags=(filter);;
+ *\?*\?*) tags=(scope); [[ -suffix \?* ]] || suf_scope=( -qS \? );;
+ *\?*) tags=(attribute);;
+ *) tags=(basedn);;
+ esac
+ compset -P '*\?'
+ compset -S '\?*'
+ fi
+
+ _tags $tags
+ while _tags; do
+ _requested protocol && _describe -t protocol protocol protocols $suf_proto
+ _requested host && _hosts -S ''
+ _requested port expl port
+ _requested basedn expl 'base DN'
+ _requested attribute expl attribute
+ _requested scope && _describe -t scope scope scopes $suf_scope
+ _requested filter expl filter
+ [[ nm -ne compstate[nmatches] ]] && return 0
+ done
+}
+
+if (( CURRENT == 2 )); then
+ _describe command commands
+else
+ shift words; (( CURRENT-- ))
+ case $words[1] in
+ search)
+ _arguments -s -S -A '-*' \
+ '-b+[specify base DN]:base DN:' \
+ '-c+[specify CA file]:CA file:' \
+ '-D+[specify bind DN]:bind DN:' \
+ '-H+[specify URL]: :_ldap_url' \
+ '-L[output in LDIF]' \
+ '-l+[specify time limit or 0 for no limit]:time limit [0]:' \
+ '-s+[specify scope]:scope [sub]:(($scopes))' \
+ '-v[be verbose]' \
+ '-W[prompt for bind secret]' \
+ '-w+[specify bind secret]:bind secret:' \
+ '-x[use simple authentication]' \
+ '-Z[use StartTLS]' \
+ '-z+[specify maximum number of results or 0 for no limit]:size limit [0]:' \
+ '::filter:' \
+ '*:attribute:'
+ ;;
+ esac
+fi
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2018-06-28 4:59 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-06-19 12:57 [patch] Add ldap completer Matthew Martin
2018-06-19 20:21 ` Oliver Kiddle
2018-06-28 4:59 ` Matthew Martin
Code repositories for project(s) associated with this public inbox
https://git.vuxu.org/mirror/zsh/
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).