Mailing-List: contact zsh-workers-help@zsh.org; run by ezmlm
Precedence: bulk
Received-SPF: pass (ns1.primenet.com.au: SPF record at _spf.mail.yahoo.com designates 212.82.97.6 as permitted sender)
In-reply-to: <20160727095421.5c919e22@pwslap01u.europe.root.pri>
From: Oliver Kiddle <okiddle@yahoo.co.uk>
References: <588168129.3340906.1467709726474.JavaMail.yahoo.ref@mail.yahoo.com> <588168129.3340906.1467709726474.JavaMail.yahoo@mail.yahoo.com> <20160705111900.7e1599f0@pwslap01u.europe.root.pri> <5374.1469315162@hydra.kiddle.eu> <20160726145224.7910b718@pwslap01u.europe.root.pri> <57127.1469557325@hydra.kiddle.eu> <20160727095421.5c919e22@pwslap01u.europe.root.pri>
To: zsh-workers@zsh.org
Subject: Re: bug with camel case and delete-whole-word-match function
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <83726.1469660746.1@hydra.kiddle.eu>
Date: Thu, 28 Jul 2016 01:05:46 +0200
Message-ID: <83727.1469660746@hydra.kiddle.eu>

Peter wrote:
> OK, we need to check that there actually is a word after the cursor, as
> we should also do before setting is-word-start.

Thanks. That now works nicely.

Attached patch adjusts select-word-match to take advantage of this.
I've also updated zstyle completion slightly for the word-style and
word-class styles. The documentation seems a bit confusing in the way it
mixes "normal" and "standard" and there might be an error there. I have
adjusted the word-context example for what looks like an error to me.

I've also now had a chance to experiment with select-word-match. I
had previously used select-bracketed for directory components and
lists using i/ a/ i, and a, etc. For things that are essentially
list separators, select-word-match works better because it can grab
the first and last components of a list and the a forms such as a/
will grab just one surrounding / rather than both. Unfortunately,
for the last component of a path, it'll grab the following whitespace
rather than the preceding slash but we can perhaps add a style for
patterns specifying preferred whitespace. That might have other
uses like prefering preceding whitespace over a following newline.

Oliver

diff --git a/Completion/Zsh/Command/_zstyle b/Completion/Zsh/Command/_zstyle
index 9a6d618..20ff47f 100644
--- a/Completion/Zsh/Command/_zstyle
+++ b/Completion/Zsh/Command/_zstyle
@@ -173,6 +173,7 @@ styles=(
   url-seps               e:
   whence                 e:
   word-chars             e:
+  word-class             e:
   word-style             e:word-style
   word-context           e:
 
@@ -241,11 +242,13 @@ while (( $#state )); do
   case "$state[1]" in
     (contexts)
       if [[ ! -prefix :*: ]]; then
-	_wanted contexts expl context compadd -P : -qS : completion vcs_info zftp
+	_wanted contexts expl context compadd -P : -qS : chpwd completion vcs_info zftp zle
       elif compset -P :completion:; then
         contexts=( functions _completers cmdorcont argument tag )
       elif compset -P :vcs_info:; then
         contexts=( vcs-string user-context repo-root-name )
+      elif compset -P :zle:; then
+	_wanted widgets expl widget _widgets -qS :
       fi
       if (( $#contexts )); then
         for ostate in $contexts; do
@@ -521,7 +524,7 @@ while (( $#state )); do
       ;;
 
     (word-style)
-      _wanted word-styles expl 'word style' compadd normal shell space
+      _wanted word-styles expl 'word style' compadd {normal,specified,unspecified,shell,whitespace}-subword
       ;;
 
     (vcs-string)
diff --git a/Doc/Zsh/contrib.yo b/Doc/Zsh/contrib.yo
index 8db7395..00ed080 100644
--- a/Doc/Zsh/contrib.yo
+++ b/Doc/Zsh/contrib.yo
@@ -2105,7 +2105,7 @@ Here are some examples of use of the tt(word-context) style to extend
 the context.
 
 example(zstyle ':zle:*' word-context \ 
-       "*/*" file "[[:space:]]" whitespace
+       "*/*" filename "[[:space:]]" whitespace
 zstyle ':zle:transpose-words:whitespace' word-style shell
 zstyle ':zle:transpose-words:filename' word-style normal
 zstyle ':zle:transpose-words:filename' word-chars '')
diff --git a/Functions/Zle/select-word-match b/Functions/Zle/select-word-match
index 24620c9..8440852 100644
--- a/Functions/Zle/select-word-match
+++ b/Functions/Zle/select-word-match
@@ -12,7 +12,7 @@ emulate -L zsh
 setopt extendedglob
 
 local curcontext=:zle:$WIDGET
-local -a matched_words
+local -A matched_words
 # Start and end of range of characters
 integer pos1 pos2 num=${NUMERIC:-1}
 local style word
@@ -28,10 +28,10 @@ while (( num-- )); do
   if (( MARK > CURSOR )); then
     # if cursor is at the start of the selection, just move back a word
     match-words-by-style
-    if [[ $style = i && -n $matched_words[3] ]]; then
-      word=$matched_words[3]
+    if [[ $style = i && -n $matched_words[ws-before-cursor] ]]; then
+      word=$matched_words[ws-before-cursor]
     else
-      word=$matched_words[2]$matched_words[3]
+      word=$matched_words[word-before-cursor]$matched_words[ws-before-cursor]
     fi
     if [[ -n $word ]]; then
       (( CURSOR -= ${#word} ))
@@ -43,41 +43,40 @@ while (( num-- )); do
     (( CURSOR+1 == $#BUFFER )) && return 1
     (( CURSOR++ ))
     match-words-by-style
-    if [[ -n $matched_words[4] ]]; then
+    if [[ -n $matched_words[ws-after-cursor] ]]; then
       if [[ $style = i ]]; then
 	# just skip the whitespace
-	word=$matched_words[4]
+	word=$matched_words[ws-after-cursor]
       else
 	# skip the whitespace plus word
-	word=$matched_words[4]$matched_words[5]
+	word=$matched_words[ws-after-cursor]$matched_words[word-after-cursor]
       fi
     else
       if [[ $style = i ]]; then
 	# skip the word
-	word=$matched_words[5]
+	word=$matched_words[word-after-cursor]
       else
 	# skip word and following whitespace
-	word=$matched_words[5]$matched_words[6]
+	word=$matched_words[word-after-cursor]$matched_words[ws-after-word]
       fi
     fi
     (( CURSOR += ${#word} - 1 ))
   else
     match-words-by-style
 
-    if [[ -n "${matched_words[3]}" ]]; then
-      # There's whitespace before the cursor, so the word we are selecting
-      # starts at the cursor position.
+    if (( ${matched_words[is-word-start]} )); then
+      # The word we are selecting starts at the cursor position.
       pos1=$CURSOR
     else
       # No whitespace before us, so select any wordcharacters there.
-      pos1="${#matched_words[1]}"
+      pos1="${#matched_words[start]}"
     fi
 
-    if [[ -n "${matched_words[4]}" ]]; then
-      if [[ -n "${matched_words[3]}" ]] || (( CURSOR == 0 )); then
+    if [[ -n "${matched_words[ws-after-cursor]}" ]]; then
+      if [[ -n "${matched_words[ws-before-cursor]}" ]] || (( CURSOR == 0 )); then
         # whitespace either side, select it
-	(( pos1 = CURSOR - ${#matched_words[3]} ))
-	(( pos2 = CURSOR + ${#matched_words[4]} ))
+	(( pos1 = CURSOR - ${#matched_words[ws-before-cursor]} ))
+	(( pos2 = CURSOR + ${#matched_words[ws-after-cursor]} ))
       else
 	# There's whitespace at the cursor position, so only select
 	# up to the cursor position.
@@ -86,28 +85,28 @@ while (( num-- )); do
     else
       # No whitespace at the cursor position, so select the
       # current character and any following wordcharacters.
-      (( pos2 = CURSOR + ${#matched_words[5]} ))
+      (( pos2 = CURSOR + ${#matched_words[word-after-cursor]} ))
     fi
 
     if [[ $style = a ]]; then
-      if [[ -n "${matched_words[4]}"  && ( -n "${matched_words[3]}" || CURSOR -eq 0 ) ]]; then
+      if [[ -n "${matched_words[ws-after-cursor]}"  && ( -n "${matched_words[ws-before-cursor]}" || CURSOR -eq 0 ) ]]; then
 	# in the middle of whitespace so grab a word
-	if [[ -n "${matched_words[5]}" ]]; then
-	  (( pos2 += ${#matched_words[5]} )) # preferably the one after
+	if [[ -n "${matched_words[word-after-cursor]}" ]]; then
+	  (( pos2 += ${#matched_words[word-after-cursor]} )) # preferably the one after
 	else
-	  (( pos1 -= ${#matched_words[2]} )) # otherwise the one before
+	  (( pos1 -= ${#matched_words[word-before-cursor]} )) # otherwise the one before
 	fi
-      elif [[ -n "${matched_words[6]}" ]]; then
-	(( pos2 += ${#matched_words[6]} ))
-      elif [[ -n "${matched_words[3]}" ]]; then
+      elif [[ -n "${matched_words[ws-after-word]}" ]]; then
+	(( pos2 += ${#matched_words[ws-after-word]} ))
+      elif [[ -n "${matched_words[ws-before-cursor]}" ]]; then
 	# couldn't grab whitespace forwards so try backwards
-	(( pos1 -= ${#matched_words[3]} ))
+	(( pos1 -= ${#matched_words[ws-before-cursor]} ))
       elif (( pos1 > 0 )); then
 	# There might have been whitespace before the word
 	(( CURSOR = pos1 ))
 	match-words-by-style
-	if [[ -n "${matched_words[3]}" ]]; then
-	  (( pos1 -= ${#matched_words[3]} ))
+	if [[ -n "${matched_words[ws-before-cursor]}" ]]; then
+	  (( pos1 -= ${#matched_words[ws-before-cursor]} ))
 	fi
       fi
     fi