Info-search-isearch

Info-search-isearch

[Juri Linkov]

Currently only regexp isearch invoked with M-C-s can use `Info-search'
and search through multiple Info nodes.  But it would be better if
normal C-s isearch searched through multiple nodes as well.

Maybe some users will not like this.  So a new user option could be
added to info.el with the default t:

Index: lisp/info.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/info.el,v
retrieving revision 1.413
diff -u -r1.413 info.el
--- lisp/info.el	13 Dec 2004 19:31:32 -0000	1.413
+++ lisp/info.el	28 Dec 2004 02:17:19 -0000
@@ -197,6 +201,15 @@
   :type 'regexp
   :group 'info)
 
+(defcustom Info-search-isearch t
+  "*If non-nil, isearch invoked in Info mode uses `Info-search' function.
+This allows isearch to search through multiple nodes.
+When isearch fails, it wraps and restarts the search from the
+top/final node depending on search direction."
+  :version "21.4"
+  :type 'boolean
+  :group 'info)
+
 (defcustom Info-mode-hook
   ;; Try to obey obsolete Info-fontify settings.
   (unless (and (boundp 'Info-fontify) (null Info-fontify))
@@ -1632,23 +1646,24 @@
   (Info-search regexp bound noerror count 'backward))
 
 (defun Info-isearch-search ()
-  (cond
-   (isearch-word
-    (if isearch-forward 'word-search-forward 'word-search-backward))
-   (isearch-regexp
-    (lambda (regexp bound noerror)
-      (condition-case nil
-          (progn
-            (Info-search regexp bound noerror nil
-                         (unless isearch-forward 'backward))
-            (point))
-        (error nil))))
-   (t
-    (if isearch-forward 'search-forward 'search-backward))))
+  (if (and Info-search-isearch (not isearch-word))
+      (lambda (string &optional bound noerror count)
+	(condition-case nil
+	    (progn
+	      (Info-search (if isearch-regexp string (regexp-quote string))
+			   bound noerror count
+			   (unless isearch-forward 'backward))
+	      (point))
+	  (error nil)))
+    (let ((isearch-search-fun-function nil))
+      (isearch-search-fun))))
 
 (defun Info-isearch-wrap ()
-  (if isearch-regexp
-      (if isearch-forward (Info-top-node) (Info-final-node))
+  (when (and Info-search-isearch (not isearch-word))
+    (if isearch-forward (Info-top-node) (Info-final-node))
     (goto-char (if isearch-forward (point-min) (point-max)))))
 
 (defun Info-isearch-push-state ()

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: Info-search-isearch

[Dan Nicolaescu]

Juri Linkov <juri@jurta.org> writes:

  > Currently only regexp isearch invoked with M-C-s can use `Info-search'
  > and search through multiple Info nodes.  But it would be better if
  > normal C-s isearch searched through multiple nodes as well.

This would be actually great, it would match the behavior of standalone 
info (which, IMO, is much better).

Is there any reason not to check this code in? 

RE: Info-search-isearch

[Drew Adams]

      > Currently only regexp isearch invoked with M-C-s can use
        `Info-search'
      > and search through multiple Info nodes.  But it would be better if
      > normal C-s isearch searched through multiple nodes as well.

    This would be actually great, it would match the behavior of standalone
    info (which, IMO, is much better).

    Is there any reason not to check this code in?

I like being able to confine an isearch to the current node. Currently, I
can use `s' for global search and `C-s' (or `C-M-s') for local (node)
search.

So, to me, it would be desirable to have some indication that the end of the
current node was reached without finding another occurrence of the string -
e.g. "Failing isearch in node Foo: bar". Except for being able to overwrap
isearch in the node (no great loss), this would give the advantages of both
approaches (local and global search). Without such a local-failure
indication, you would have to (perhaps unwittingly) continue the search to
another node, and then use `l' to get back where you were.

IOW, why not have two levels of isearch "failure": 1) no more in the current
node, 2) no more in the current Info "manual"?

Re: Info-search-isearch

[Dan Nicolaescu]

"Drew Adams" <drew.adams@oracle.com> writes:

  >       > Currently only regexp isearch invoked with M-C-s can use
  >         `Info-search'
  >       > and search through multiple Info nodes.  But it would be better if
  >       > normal C-s isearch searched through multiple nodes as well.
  > 
  >     This would be actually great, it would match the behavior of standalone
  >     info (which, IMO, is much better).
  > 
  >     Is there any reason not to check this code in?
  > 
  > I like being able to confine an isearch to the current node. Currently, I
  > can use `s' for global search and `C-s' (or `C-M-s') for local (node)
  > search.
  > 
  > So, to me, it would be desirable to have some indication that the end of the
  > current node was reached without finding another occurrence of the string -
  > e.g. "Failing isearch in node Foo: bar". Except for being able to overwrap
  > isearch in the node (no great loss), this would give the advantages of both
  > approaches (local and global search). Without such a local-failure
  > indication, you would have to (perhaps unwittingly) continue the search to
  > another node, and then use `l' to get back where you were.
  > 
  > IOW, why not have two levels of isearch "failure": 1) no more in the current
  > node, 2) no more in the current Info "manual"?

The behavior of what you propose is different from either the current
C-s behavior in an emacs info buffer or in standalone info. 

The message I was replying to was just asking to make the C-s behavior
the same in an info buffer and standalone info, and allow to the user
to choose one or the other via a customizable variable. 

RE: Info-search-isearch

[Drew Adams]

      > I like being able to confine an isearch to the current 
      > node. Currently, I can use `s' for global search and `C-s'
      > (or `C-M-s') for local (node) search.
      > 
      > So, to me, it would be desirable to have some indication 
      > that the end of the current node was reached without
      > finding another occurrence of the string -
      > e.g. "Failing isearch in node Foo: bar". Except for being 
      > able to overwrap isearch in the node (no great loss), this
      > would give the advantages of both
      > approaches (local and global search). Without such a local-failure
      > indication, you would have to (perhaps unwittingly) 
      > continue the search to
      > another node, and then use `l' to get back where you were.
      > 
      > IOW, why not have two levels of isearch "failure": 1) no 
      > more in the current node, 2) no more in the current Info "manual"?
    
    The behavior of what you propose is different from either the current
    C-s behavior in an emacs info buffer or in standalone info.

Correct. 
    
    The message I was replying to was just asking to make the C-s behavior
    the same in an info buffer and standalone info, and allow to the user
    to choose one or the other via a customizable variable.

Correct.

Re: Info-search-isearch

[Richard Stallman]

      > Currently only regexp isearch invoked with M-C-s can use `Info-search'
      > and search through multiple Info nodes.  But it would be better if
      > normal C-s isearch searched through multiple nodes as well.

    This would be actually great, it would match the behavior of standalone 
    info (which, IMO, is much better).

    Is there any reason not to check this code in? 

Does the code look clean to you?
If so, it is ok with me.

Re: Info-search-isearch

[Richard Stallman]

    IOW, why not have two levels of isearch "failure": 1) no more in the current
    node, 2) no more in the current Info "manual"?

It sounds convenient to me.
Search could first wrap in the current node, then
when it is about to say "Failing wrapped", it would
try other nodes.

Re: Info-search-isearch

[Juri Linkov]

Richard Stallman <rms@gnu.org> writes:
>       > Currently only regexp isearch invoked with M-C-s can use `Info-search'
>       > and search through multiple Info nodes.  But it would be better if
>       > normal C-s isearch searched through multiple nodes as well.
>
>     This would be actually great, it would match the behavior of standalone 
>     info (which, IMO, is much better).
>
>     Is there any reason not to check this code in? 
>
> Does the code look clean to you?
> If so, it is ok with me.

I installed this code.  Now C-s in Info works like in standalone info.
If there are good reasons to add a "Failing isearch in node Foo: bar"
message before trying other nodes, as Drew suggested, this could be
added as a separate option value of the Info-isearch-search variable.
But I think such delay in one C-s key press is inconvenient when the
user intention is to search all nodes.

When isearch goes to the next node with the user intention to search
only in the current node then it is easy to get back by typing DEL
without leaving isearch mode.

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: Info-search-isearch

[Richard Stallman]

    When isearch goes to the next node with the user intention to search
    only in the current node then it is easy to get back by typing DEL
    without leaving isearch mode.

It is easy to "get back".  What is not easy, with the current
behavior, is to wrap around in the current node, which is what isearch
would have done before this change.

Perhaps that case is not important enough in practice to be worth any
effort.

Re: Info-search-isearch

[Juri Linkov]

Currently multi-node isearch in Info ignores word search.  It searches
for a sequence of words only in the current node.  But it is useful to
search for words through multiple Info nodes.  I think search in Info
should support nonincremental and "half-incremental" word search.

Word search what I call "half-incremental" is an undocumented
feature that allows almost incremental search by typing
`C-s M-e C-w WORDS C-s', i.e. it starts as nonincremental search,
and after typing the second C-s in the minibuffer it continues
as incremental search.  I discovered this feature accidentally.
Perhaps, it should be documented in the Emacs manual?

Below is an implementation of word search for Info isearch which
duplicated in Lisp the C function `wordify' used by word search.

Index: lisp/info.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/info.el,v
retrieving revision 1.417
diff -u -r1.417 info.el
--- lisp/info.el	23 Feb 2005 17:03:17 -0000	1.417
+++ lisp/info.el	25 Feb 2005 19:55:18 -0000
@@ -1646,10 +1646,16 @@
   (Info-search regexp bound noerror count 'backward))
 
 (defun Info-isearch-search ()
-  (if (and Info-isearch-search (not isearch-word))
+  (if Info-isearch-search
       (lambda (string &optional bound noerror count)
 	(condition-case nil
-	    (progn
+	    (if isearch-word
+		(Info-search (concat "\\b" (replace-regexp-in-string
+					    "\\W+" "\\\\W+"
+					    (replace-regexp-in-string
+					     "^\\W+\\|\\W+$" "" string)) "\\b")
+			     bound noerror count
+			     (unless isearch-forward 'backward))
 	      (Info-search (if isearch-regexp string (regexp-quote string))
 			   bound noerror count
 			   (unless isearch-forward 'backward))
@@ -1659,7 +1665,7 @@
       (isearch-search-fun))))
 
 (defun Info-isearch-wrap ()
-  (when (and Info-isearch-search (not isearch-word))
+  (when Info-isearch-search
     (if isearch-forward (Info-top-node) (Info-final-node))
     (goto-char (if isearch-forward (point-min) (point-max)))))

-- 
Juri Linkov
http://www.jurta.org/emacs/

RE: Info-search-isearch

[Drew Adams]

I was reminded of this 2005 thread by a recent thread in help-gnu-emacs.

>From what I can see, Juri's accidental discovery (`C-x M-e C-w WORDS C-s')
never did get documented as a tip in the Emacs manual. I think it should be.

Many word-processing programs and text editors make it very easy to do word
searches, sometimes incremental word searches. In Emacs, AFAIK, we haven't
yet provided a simple way to do this. Juri's tip is a good one, until we
provide a better (quicker and easier to remember) interface to incremental
word searching.

(Yes, regexp searching with \b is good too, but that is not what many users
will want, to search for words.)

WDOT? Should Juri's tip be documented now, say in node Word Search of the
Emacs manual? If so, we should also mention how to toggle back to non-word
incremental search. (We should call this incremental word search, BTW, not
"half-incremental" search.)

Should we, after the release, come up with a quicker, easier way to toggle
incremental search to/from incremental word searching?

-------
    From: Juri Linkov
    Sent: Friday, February 25, 2005 12:00 PM
    Subject: Re: Info-search-isearch
    ...
    Word search what I call "half-incremental" is an undocumented
    feature that allows almost incremental search by typing
    `C-s M-e C-w WORDS C-s', i.e. it starts as nonincremental search,
    and after typing the second C-s in the minibuffer it continues
    as incremental search.  I discovered this feature accidentally.
    Perhaps, it should be documented in the Emacs manual?

Re: Info-search-isearch

[Juri Linkov]

> Should we, after the release, come up with a quicker, easier way to toggle
> incremental search to/from incremental word searching?

I have a better idea now for implementing this.  A new command like:

(defun isearch-toggle-word ()
  "Toggle word searching on or off."
  (interactive)
  (setq isearch-word (not isearch-word))
  (setq isearch-success t isearch-adjusted t)
  (isearch-update))

and a new keybinding:

(define-key isearch-mode-map "\M-sw" 'isearch-toggle-word)

will provide a easier way to toggle on/off incremental word search.

A new prefix key M-s seems like a good key to put all other
isearch-related keybindings on it too:

(define-key isearch-mode-map "\M-s\C-y" 'isearch-yank-kill)
(define-key isearch-mode-map "\M-se" 'isearch-edit-string)
(define-key isearch-mode-map "\M-sr" 'isearch-toggle-regexp)
(define-key isearch-mode-map "\M-sc" 'isearch-toggle-case-fold)
...

> WDOT? Should Juri's tip be documented now, say in node Word Search of
> the Emacs manual? If so, we should also mention how to toggle back to
> non-word incremental search.

Since it is too late for this release to make improvements, I think
we should at least document this tip in the Emacs manual.

> (We should call this incremental word search, BTW, not
> "half-incremental" search.)

I call it "half-incremental" because when you get into this search mode
(via `C-x M-e C-w WORDS C-s'), and start typing more words, incremental
search doesn't find them until you finish typing complete words,
so word search is not fully incremental.  However, I don't propose to
use the term "half-incremental" in the documentation, since this is an
inherent limitation of incremental word search.

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: Info-search-isearch

[Richard Stallman]

Using M-s as a prefix within Isearch is ok, because it isn't used much
with other meanings.  So it would be ok for many such commands.
However, it would not be ok as a replacement for M-y, because a two-char
sequence would be too inconvenient for that job.  So we should not expect
to move ALL special isearch commands onto the M-s prefix.  But we could
move most of them.

Re: Info-search-isearch

[Richard Stallman]

Would someone please document the current behavior of C-s M-e C-w?

Re: Info-search-isearch

[Juri Linkov]

> Would someone please document the current behavior of C-s M-e C-w?

Done.  Please review it.

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: Info-search-isearch

[Richard Stallman]

    > Would someone please document the current behavior of C-s M-e C-w?

Thanks.

word-search toggle for isearch [was: Info-search-isearch]

[Drew Adams]

Both Juri and Stefan have sent (essentially the same) code for toggling word
search in incremental search:

> From: Juri Linkov Sent: Sunday, October 29, 2006 12:19 PM
> To: emacs-devel@gnu.org
> Subject: Re: Info-search-isearch
> (defun isearch-toggle-word ()
>   "Toggle word searching on or off."
>   (interactive)
>   (setq isearch-word (not isearch-word))
>   (setq isearch-success t isearch-adjusted t)
>   (isearch-update))
>
> and a new keybinding:
> (define-key isearch-mode-map "\M-sw" 'isearch-toggle-word)
> will provide a easier way to toggle on/off incremental word search.

> From: Stefan Monnier Sent: Thursday, November 23, 2006 11:25 AM
> To: help-gnu-emacs@gnu.org
> Subject: Re: incremental search
> (defun isearch-toggle-word ()
>   "Toggle word searching on or off."
>   ;; The status stack is left unchanged.
>   (interactive)
>   (setq isearch-word (not isearch-word))
>   (if isearch-word (setq isearch-regexp nil))
>   (setq isearch-success t isearch-adjusted t)
>   (isearch-update))

Juri proposed binding this to M-sw. Stefan suggested M-w instead.

Users of most editors expect a simple, quick way to toggle word searching.
How to do this in Emacs is a FAQ. We have even gone to the trouble of
documenting C-s M-e M-w WORD C-s, but we have not added a simple binding for
this.

How about adding this feature (toggle command and binding) to Emacs now?

Re: word-search toggle for isearch [was: Info-search-isearch]

[Richard Stallman]

    Juri proposed binding this to M-sw. Stefan suggested M-w instead.

Between those two, I prefer M-s, because M-w is useful for
exiting a search.

Now is the wrong time to add a new feature.  Someone could implement this
in the Unicode-2 branch instead.