Re: [PATCH] _gpg: Use explicit UIDs for public / secret keys.

[Doron Behar <doron.behar@xxxxxxxxx>]

Daniel,

I would like you to tell me how can I solve better the `fpr` thing.

To tell you the truth, I have no idea what `fpr` means. I just know, by
comparing the output of `gpg --list-public-keys` with and without
`--with-colons`, that the hexadecimal number afterwards is the actual
key that needs to be used as an argument for these options. The way I
understand it, only if there is an email address / description coming
somewhere after the array element `fpr` (without another `fpr` in
between), then this field is the description.

Please tell me, how can this algorithm get better? I've had one idea in
mind: Perhaps we can perform a more strict pattern matchings test before
adding a uid for example?

On Sat, Jun 09, 2018 at 08:39:32PM +0000, Daniel Shahaf wrote:
> Good morning Doron,
> 
> Thanks for revising the patch, however, I'm afraid I do have a few more
> comments:
> 
> doron.behar@xxxxxxxxx wrote on Sat, Jun 09, 2018 at 23:09:40 +0300:
> >    secret-keys)
> > -    _wanted secret-keys expl 'secret key' compadd \
> > -	${${(Mo)$(_call_program secret-keys $words[1] $needed --list-secret-keys --list-options no-show-photos):%<*>}//(<|>)/} && return
> > +    local secret_keys=(${(@s.:.)${(f)"$(_call_program secret-keys ${(q)words[1]} ${(q)needed} --list-secret-keys --list-options no-show-photos --with-colons)"}})
> > +    local -a uids emails
> > +    local i
> > +    for i in {1..${#secret_keys[@]}}; do
> > +      if [[ ${secret_keys[$i]} == "fpr" ]]; then
> 
> I'd like to see the 'fpr' thing (which I have pointed out twice by now) fixed
> before merging.  _gpg isn't a function I'm comfortable adding
> relaxed/inaccurate parsing to, and doing the parsing correctly isn't onerous.
> 
> > +        i=$((i + 1))
> > +        local j=$i
> > +        while [[ ${secret_keys[$j]} != "fpr" ]] && [ $j -lt ${#secret_keys[@]} ]; do
> > +          if [[ ${secret_keys[$j]} =~ "@" ]]; then
> 
> This condition false negatives for me, probably because I used a test secret
> key that didn't have an email address attached.
> 
> > +            emails+="${secret_keys[$j]}"
> > +            uids+="${secret_keys[$i]}"
> > +            break
> > +          fi
> > +          j=$((j + 1))
> > +        done
> > +        i=$j
> > +      fi
> > +    done
> > +    _describe -t secret-keys 'secret key' uids_and_emails
> 
> s/uids_and_emails/emails uids/
> 
> >    ;;
> >    ciphers)
> >      _wanted ciphers expl cipher compadd \
> > -        ${${(s.,.)${(M)${(f)${"$(_call_program ciphers $words[1] $needed --version)"}//,$'\n' #/, }:#Cipher*}#*:}# } && return
> > +        ${${(s.,.)${(M)${(f)${"$(_call_program ciphers ${(q)words[1]} ${(q)needed} --version)"}//,$'\n' #/, }:#Cipher*}#*:}# } && return
> >    ;;
> >    (public-keyids)
> > +    _describe -t public-keyids 'public keyids' uids_and_emails
> 
> Ditto.
> 
> Thanks for hanging in there.
> 
> Daniel