aggregated options in completion lists

aggregated options in completion lists

[Borsenkow Andrej]

This looks strange. And of course all descriptions are cut off (I am
sorry, actually every option is on a single line, it is just I do not
know how to tell it Evolution). There probably should be some (settable)
limit on column width.

The listing is from rsync completion.

-andrej

--owner                                                        --
preserve owne
--partial                                                      -- keep
partiall
--password-file                                                -- get
password
--password-file      --include-from  --exclude-from  --config  --
specify: file
--perms                                                        --
preserve perm
--port                                                         --
specify alter
--progress                                                     -- show
progress

Re: aggregated options in completion lists

[Sven Wischnowsky]

Borsenkow Andrej wrote:

> This looks strange. And of course all descriptions are cut off (I am
> sorry, actually every option is on a single line, it is just I do not
> know how to tell it Evolution). There probably should be some (settable)
> limit on column width.
> 
> The listing is from rsync completion.
> 
> -andrej
> 
> --owner                                                        -- preserve owne

Aha. This is what I mentioned when I wrote the columnisation stuff.

Hmhm. I'll have a look. And there are probably more possible ways we
can go now than I can think of.


Bye
  Sven

-- 
Sven Wischnowsky                          wischnow@berkom.de

Re: aggregated options in completion lists

[Sven Wischnowsky]

Borsenkow Andrej wrote:

> This looks strange. And of course all descriptions are cut off (I am
> sorry, actually every option is on a single line, it is just I do not
> know how to tell it Evolution). There probably should be some (settable)
> limit on column width.

Below is a patch. This is the simple implementation. One could thing
about how nice it were if the code could do:

   % foo <TAB>
   description
   --long-option1  --long-option2
   --long-option3                   -- description

I.e., to not repeat the description for every sub-set of matches it
creates. That, however is much harder to implement and, if uses often,
could probably be slightly irritating, I think.

Opinions?


Can anyone help me with a better name for that style? `max-match-length'
just doesn't sound right. `max-match-columns' might be better, but
sound (to me) as if the number of matches per line were meant, which
it isn't.

I'll commit it anyway.


Bye
  Sven

Index: Completion/Base/Utility/_describe
===================================================================
RCS file: /cvsroot/zsh/zsh/Completion/Base/Utility/_describe,v
retrieving revision 1.11
diff -u -r1.11 _describe
--- Completion/Base/Utility/_describe	6 May 2002 07:59:50 -0000	1.11
+++ Completion/Base/Utility/_describe	14 May 2002 07:57:50 -0000
@@ -2,7 +2,7 @@
 
 # This can be used to add options or values with descriptions as matches.
 
-local _opt _expl _tmpm _tmpd
+local _opt _expl _tmpm _tmpd _mlen
 local _type=values _descr _ret=1 _showd _nm _hide _args _grp _sep
 local csl="$compstate[list]" csl2
 local _oargv _argv _new _strs _mats _opts _i _try=0
@@ -25,6 +25,8 @@
 zstyle -T ":completion:${curcontext}:$_type" verbose && _showd=yes
 
 zstyle -s ":completion:${curcontext}:$_type" list-separator _sep || _sep=--
+zstyle -s ":completion:${curcontext}:$_type" max-match-length _mlen ||
+    _mlen=$((COLUMNS/2))
 
 _descr="$1"
 shift
@@ -97,9 +99,9 @@
     fi
 
     if [[ -n "$_showd" ]]; then
-      compdescribe -I "$_hide" "$_sep " _expl "$_grp[@]" "$@"
+      compdescribe -I "$_hide" "$_mlen" "$_sep " _expl "$_grp[@]" "$@"
     else
-      compdescribe -i "$_hide" "$@"
+      compdescribe -i "$_hide" "$_mlen" "$@"
     fi
 
     compstate[list]="$csl"
Index: Completion/Zsh/Command/_zstyle
===================================================================
RCS file: /cvsroot/zsh/zsh/Completion/Zsh/Command/_zstyle,v
retrieving revision 1.9
diff -u -r1.9 _zstyle
--- Completion/Zsh/Command/_zstyle	16 Apr 2002 07:48:46 -0000	1.9
+++ Completion/Zsh/Command/_zstyle	14 May 2002 07:57:51 -0000
@@ -80,6 +80,7 @@
   matcher		 c:
   matcher-list		 c:
   max-errors		 c:
+  max-match-length       c:max-match-length
   menu			 c:boolauto
   muttrc                 c:_files
   numbers		 c:bool
@@ -380,6 +381,10 @@
 
     separator)
       _message -e separators 'separator string'
+      ;;
+
+    max-match-length)
+      _message -e numbers 'maximum display length for matches'
       ;;
 
     urgh) 
Index: Doc/Zsh/compsys.yo
===================================================================
RCS file: /cvsroot/zsh/zsh/Doc/Zsh/compsys.yo,v
retrieving revision 1.155
diff -u -r1.155 compsys.yo
--- Doc/Zsh/compsys.yo	14 May 2002 07:54:24 -0000	1.155
+++ Doc/Zsh/compsys.yo	14 May 2002 07:57:51 -0000
@@ -1766,6 +1766,20 @@
 
 The default value for this style is `tt(2 numeric)'.
 )
+kindex(max-match-length, completion style)
+item(tt(max-match-length))(
+This is used to define the maximum length to use for the matches when
+listing matches with descriptions. In such lists, matches with the
+same description will be grouped together, but that means that in
+cases where many matches have the same description, the matches take
+up most of the display width, leaving only little room for the
+descriptions. By setting this style one can specify whether one
+prefers to make more matches be grouped together or whether the shell
+should try keep more of the descriptions visible. 
+
+The value should give the maximum number of display columns to give to
+the matches, the default is half the screen width.
+)
 kindex(menu, completion style)
 item(tt(menu))(
 If this is set to true in a given context, using any of the tags defined
Index: Src/Zle/computil.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/Zle/computil.c,v
retrieving revision 1.77
diff -u -r1.77 computil.c
--- Src/Zle/computil.c	17 Apr 2002 07:28:19 -0000	1.77
+++ Src/Zle/computil.c	14 May 2002 07:57:52 -0000
@@ -41,6 +41,7 @@
     int showd;			/* != 0 if descriptions should be shown */
     char *sep;			/* the separator string */
     int slen;			/* its length */
+    int maxmlen;                /* maximum length to allow for the matches */
     Cdset sets;			/* the sets of matches */
     int pre;                    /* longest prefix (before description) */
     int suf;                    /* longest suffix (description) */
@@ -129,22 +130,26 @@
                 continue;
 
             num = 1;
-            len = str1->len;
+            len = str1->len + cd_state.slen;
             strp = &(str1->other);
 
-            for (set2 = set1; set2; set2 = set2->next)
+            for (set2 = set1; set2; set2 = set2->next) {
                 for (str2 = (set2 == set1 ? str1->next : set2->strs);
                      str2; str2 = str2->next)
                     if (str2->desc && !strcmp(str1->desc, str2->desc)) {
+                        len += 2 + str2->len;
+                        if (len > cd_state.maxmlen)
+                            break;
                         str1->kind = 1;
                         str2->kind = 2;
                         num++;
-                        len += str2->len;
                         *strp = str2;
                         strp = &(str2->other);
                     }
+                if (str2)
+                    break;
+            }
             *strp = NULL;
-            len += num * 2 + cd_state.slen;
 
             if (len >= columns) {
                 cd_state.groups = 0;
@@ -393,12 +398,13 @@
 /* Initialisation. Store and calculate the string and matches and so on. */
 
 static int
-cd_init(char *nam, char *hide, char *sep, char **opts, char **args, int disp)
+cd_init(char *nam, char *hide, char *mlen, char *sep,
+        char **opts, char **args, int disp)
 {
     Cdset *setp, set;
     Cdstr *strp, str;
     char **ap, *tmp;
-    int grp = 0;
+    int grp = 0, itmp;
 
     if (cd_parsed) {
 	zsfree(cd_state.sep);
@@ -411,7 +417,12 @@
     cd_state.sets = NULL;
     cd_state.showd = disp;
     cd_state.maxg = cd_state.groups = cd_state.descs = 0;
-
+    cd_state.maxmlen = atoi(mlen);
+    itmp = columns - cd_state.slen - 4;
+    if (cd_state.maxmlen > itmp)
+        cd_state.maxmlen = itmp;
+    if (cd_state.maxmlen < 4)
+        cd_state.maxmlen = 4;
     if (*args && !strcmp(*args, "-g")) {
         args++;
         grp = 1;
@@ -537,7 +548,7 @@
                 /* We are building a columnised list with dummy matches
                  * but there are also matches without descriptions.
                  * Those end up in a different group, so make sure that
-                 * groupd doesn't have an explanation. */
+                 * group doesn't have an explanation. */
 
                 for (mp = dp = opts; *mp; mp++) {
                     if (dp[0][0] == '-' && dp[0][1] == 'X') {
@@ -679,25 +690,25 @@
     }
     switch (args[0][1]) {
     case 'i':
-        if (n < 2) {
+        if (n < 3) {
             zwarnnam(nam, "not enough arguments", NULL, 0);
 
             return 1;
         }
-	return cd_init(nam, args[1], "", NULL, args + 2, 0);
+	return cd_init(nam, args[1], args[2], "", NULL, args + 3, 0);
     case 'I':
-        if (n < 5) {
+        if (n < 6) {
             zwarnnam(nam, "not enough arguments", NULL, 0);
 
             return 1;
         } else {
             char **opts;
 
-            if (!(opts = getaparam(args[3]))) {
-		zwarnnam(nam, "unknown parameter: %s", args[2], 0);
+            if (!(opts = getaparam(args[4]))) {
+		zwarnnam(nam, "unknown parameter: %s", args[4], 0);
 		return 1;
             }
-            return cd_init(nam, args[1], args[2], opts, args + 4, 1);
+            return cd_init(nam, args[1], args[2], args[3], opts, args + 5, 1);
         }
     case 'g':
 	if (cd_parsed) {

-- 
Sven Wischnowsky                          wischnow@berkom.de

Re: aggregated options in completion lists

[Sven Wischnowsky]

Hi

The calculation for max-match-length (still no better name) could
fail. We have to iterate here, as for completion lists.

Bye
  Sven

Index: Src/Zle/computil.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/Zle/computil.c,v
retrieving revision 1.79
diff -u -r1.79 computil.c
--- Src/Zle/computil.c	21 May 2002 07:39:42 -0000	1.79
+++ Src/Zle/computil.c	21 May 2002 08:00:46 -0000
@@ -46,6 +46,7 @@
     int pre;                    /* longest prefix (before description) */
     int suf;                    /* longest suffix (description) */
     int maxg;                   /* size of largest group */
+    int maxglen;                /* columns for matches of largest group */
     int groups;                 /* number of groups */
     int descs;                  /* number of non-group matches with desc */
     int gpre;                   /* prefix length for group display */
@@ -118,12 +119,21 @@
 /* Find matches with same descriptions and group them. */
 
 static void
-cd_group()
+cd_group(int maxg)
 {
     Cdset set1, set2;
     Cdstr str1, str2, *strp;
     int num, len;
 
+    cd_state.groups = cd_state.descs = cd_state.maxglen = 0;
+    cd_state.maxg = 0;
+
+    for (set1 = cd_state.sets; set1; set1 = set1->next)
+        for (str1 = set1->strs; str1; str1 = str1->next) {
+            str1->kind = 0;
+            str1->other = NULL;
+        }
+
     for (set1 = cd_state.sets; set1; set1 = set1->next) {
         for (str1 = set1->strs; str1; str1 = str1->next) {
             if (!str1->desc || str1->kind != 0)
@@ -131,6 +141,8 @@
 
             num = 1;
             len = str1->len + cd_state.slen;
+            if (len > cd_state.maxglen)
+                cd_state.maxglen = len;
             strp = &(str1->other);
 
             for (set2 = set1; set2; set2 = set2->next) {
@@ -138,8 +150,10 @@
                      str2; str2 = str2->next)
                     if (str2->desc && !strcmp(str1->desc, str2->desc)) {
                         len += 2 + str2->len;
-                        if (len > cd_state.maxmlen)
+                        if (len > cd_state.maxmlen || num == maxg)
                             break;
+                        if (len > cd_state.maxglen)
+                            cd_state.maxglen = len;
                         str1->kind = 1;
                         str2->kind = 2;
                         num++;
@@ -151,11 +165,6 @@
             }
             *strp = NULL;
 
-            if (len >= columns) {
-                cd_state.groups = 0;
-
-                return;
-            }
             if (num > 1)
                 cd_state.groups++;
             else
@@ -200,7 +209,7 @@
     return strcmp((*((Cdstr *) a))->str, (*((Cdstr *) b))->str);
 }
 
-static void
+static int
 cd_prep()
 {
     Cdrun run, *runp;
@@ -253,6 +262,13 @@
 
         qsort(grps, lines, sizeof(Cdstr), cd_sort);
 
+        cd_state.gpre = 0;
+        for (i = 0; i < cd_state.maxg; i++)
+            cd_state.gpre += wids[i] + 2;
+
+        if (cd_state.gpre > cd_state.maxmlen && cd_state.maxglen > 1)
+            return 1;
+
         expl =  (Cdrun) zalloc(sizeof(*run));
         expl->type = CRT_EXPL;
         expl->strs = grps[0];
@@ -325,9 +341,6 @@
                 run->count = i;
             }
         }
-        cd_state.gpre = 0;
-        for (i = 0; i < cd_state.maxg; i++)
-            cd_state.gpre += wids[i] + 2;
     } else if (cd_state.showd) {
         for (set = cd_state.sets; set; set = set->next) {
             if (set->desc) {
@@ -370,6 +383,8 @@
             }
     }
     *runp = NULL;
+
+    return 0;
 }
 
 /* Duplicate and concatenate two arrays.  Return the result. */
@@ -495,12 +510,19 @@
 	if ((*args = tmp))
 	    args++;
     }
-    if (disp && grp)
-        cd_group();
+    if (disp && grp) {
+        int mg = columns;
 
-    cd_calc();
-    cd_prep();
+        do {
+            cd_group(mg);
+            mg = cd_state.maxg - 1;
+            cd_calc();
+        } while (cd_prep());
 
+    } else {
+        cd_calc();
+        cd_prep();
+    }
     cd_parsed = 1;
     return 0;
 }

-- 
Sven Wischnowsky                          wischnow@berkom.de

Re: aggregated options in completion lists

[Oliver Kiddle]

On Tue, May 21, 2002 at 10:02:53AM +0200, Sven Wischnowsky wrote:
> 
> The calculation for max-match-length (still no better name) could

How about max-matches-width? Or maybe max-width-matches.

On Tue, May 14, 2002 at 09:59:21AM +0200, Sven Wischnowsky wrote:
> 
> Below is a patch. This is the simple implementation. One could thing
> about how nice it were if the code could do:
> 
>    % foo <TAB>
>    description
>    --long-option1  --long-option2
>    --long-option3                   -- description
> 
> I.e., to not repeat the description for every sub-set of matches it
> creates. That, however is much harder to implement and, if uses often,
> could probably be slightly irritating, I think.
> 
> Opinions?

It would be better.

Another possibility would be to vary the position of the description
from one line to the next which would be good in situations such as
where only a couple out of many options have multiple options for the
same description.

Handling of empty descriptions is perhaps not ideal. I'm getting
lines like:
% info -d ~/info zsh <tab>
menu item
Arithmetic Evaluation     --
Command Execution         --

Oliver

This e-mail and any attachment is for authorised use by the intended recipient(s) only.  It may contain proprietary material, confidential information and/or be subject to legal privilege.  It should not be copied, disclosed to, retained or used by, any other party.  If you are not an intended recipient then please promptly delete this e-mail and any attachment and all copies and inform the sender.  Thank you.

Re: aggregated options in completion lists

[Borsenkow Andrej]

On Tue, 21 May 2002, Oliver Kiddle wrote:

>
> Handling of empty descriptions is perhaps not ideal. I'm getting
> lines like:
> % info -d ~/info zsh <tab>
> menu item
> Arithmetic Evaluation     --
> Command Execution         --
>
 It is a bug in _info. I have new version that does not add empty
descriptions but I am off for this week so I commit it later.

-andrej

Re: aggregated options in completion lists

[Sven Wischnowsky]

Oliver Kiddle wrote:

> ...
>
> > Below is a patch. This is the simple implementation. One could thing
> > about how nice it were if the code could do:
> > 
> >    % foo <TAB>
> >    description
> >    --long-option1  --long-option2
> >    --long-option3                   -- description
> > 
> > I.e., to not repeat the description for every sub-set of matches it
> > creates. That, however is much harder to implement and, if uses often,
> > could probably be slightly irritating, I think.
> > 
> > Opinions?
> 
> It would be better.

Ok, here is the patch.

> Another possibility would be to vary the position of the description
> from one line to the next which would be good in situations such as
> where only a couple out of many options have multiple options for the
> same description.

I'm not too sure I'd like that and anyway we'd have to find good
thresholds. Dunno.


Bye
  Sven

Index: Src/Zle/computil.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/Zle/computil.c,v
retrieving revision 1.80
diff -u -r1.80 computil.c
--- Src/Zle/computil.c	21 May 2002 08:04:35 -0000	1.80
+++ Src/Zle/computil.c	23 May 2002 12:13:01 -0000
@@ -260,8 +260,6 @@
                 *strp++ = gs;
             }
 
-        qsort(grps, lines, sizeof(Cdstr), cd_sort);
-
         cd_state.gpre = 0;
         for (i = 0; i < cd_state.maxg; i++)
             cd_state.gpre += wids[i] + 2;
@@ -269,6 +267,23 @@
         if (cd_state.gpre > cd_state.maxmlen && cd_state.maxglen > 1)
             return 1;
 
+        qsort(grps, lines, sizeof(Cdstr), cd_sort);
+
+        for (i = lines, strp = grps; i > 1; i--, strp++) {
+            strp2 = strp + 1;
+            if (!strcmp((*strp)->desc, (*strp2)->desc))
+                continue;
+            for (j = i - 2, strp2++; j > 0; j--, strp2++)
+                if (!strcmp((*strp)->desc, (*strp2)->desc)) {
+                    Cdstr tmp = *strp2;
+
+                    memmove(strp + 2, strp + 1,
+                            (strp2 - strp - 1) * sizeof(Cdstr));
+
+                    *++strp = tmp;
+                    i--;
+                }
+        }
         expl =  (Cdrun) zalloc(sizeof(*run));
         expl->type = CRT_EXPL;
         expl->strs = grps[0];
@@ -666,6 +681,10 @@
                 dpys = (char **) zalloc((i + 1) * sizeof(char *));
 
                 for (dp = dpys, str = run->strs; str; str = str->run) {
+                    if (str->run && !strcmp(str->desc, str->run->desc)) {
+                        *dp++ = ztrdup("");
+                        continue;
+                    }
                     memset(dbuf + cd_state.slen, ' ', dlen - 1);
                     dbuf[dlen + cd_state.slen - 1] = '\0';
                     strcpy(dbuf, cd_state.sep);

-- 
Sven Wischnowsky                          wischnow@berkom.de

Re: aggregated options in completion lists

[Sven Wischnowsky]

Oliver Kiddle wrote:

> On Tue, May 21, 2002 at 10:02:53AM +0200, Sven Wischnowsky wrote:
> > 
> > The calculation for max-match-length (still no better name) could
> 
> How about max-matches-width? Or maybe max-width-matches.

I've taken the former. Thanks.


Bye
  Sven

Index: Completion/Base/Utility/_describe
===================================================================
RCS file: /cvsroot/zsh/zsh/Completion/Base/Utility/_describe,v
retrieving revision 1.12
diff -u -r1.12 _describe
--- Completion/Base/Utility/_describe	14 May 2002 08:01:18 -0000	1.12
+++ Completion/Base/Utility/_describe	23 May 2002 12:15:29 -0000
@@ -25,7 +25,7 @@
 zstyle -T ":completion:${curcontext}:$_type" verbose && _showd=yes
 
 zstyle -s ":completion:${curcontext}:$_type" list-separator _sep || _sep=--
-zstyle -s ":completion:${curcontext}:$_type" max-match-length _mlen ||
+zstyle -s ":completion:${curcontext}:$_type" max-matches-width _mlen ||
     _mlen=$((COLUMNS/2))
 
 _descr="$1"
Index: Completion/Zsh/Command/_zstyle
===================================================================
RCS file: /cvsroot/zsh/zsh/Completion/Zsh/Command/_zstyle,v
retrieving revision 1.10
diff -u -r1.10 _zstyle
--- Completion/Zsh/Command/_zstyle	14 May 2002 08:01:18 -0000	1.10
+++ Completion/Zsh/Command/_zstyle	23 May 2002 12:15:29 -0000
@@ -80,7 +80,7 @@
   matcher		 c:
   matcher-list		 c:
   max-errors		 c:
-  max-match-length       c:max-match-length
+  max-matches-width      c:max-matches-width
   menu			 c:boolauto
   muttrc                 c:_files
   numbers		 c:bool
@@ -383,8 +383,8 @@
       _message -e separators 'separator string'
       ;;
 
-    max-match-length)
-      _message -e numbers 'maximum display length for matches'
+    max-matches-width)
+      _message -e numbers 'maximum display width for matches'
       ;;
 
     urgh) 
Index: Doc/Zsh/compsys.yo
===================================================================
RCS file: /cvsroot/zsh/zsh/Doc/Zsh/compsys.yo,v
retrieving revision 1.156
diff -u -r1.156 compsys.yo
--- Doc/Zsh/compsys.yo	14 May 2002 08:01:18 -0000	1.156
+++ Doc/Zsh/compsys.yo	23 May 2002 12:15:30 -0000
@@ -1766,9 +1766,9 @@
 
 The default value for this style is `tt(2 numeric)'.
 )
-kindex(max-match-length, completion style)
-item(tt(max-match-length))(
-This is used to define the maximum length to use for the matches when
+kindex(max-matches-width, completion style)
+item(tt(max-matches-width))(
+This is used to define the maximum width to use for the matches when
 listing matches with descriptions. In such lists, matches with the
 same description will be grouped together, but that means that in
 cases where many matches have the same description, the matches take

-- 
Sven Wischnowsky                          wischnow@berkom.de