hstow: speed up stowing by using more streaming

Extend the dot[-.] feature to intermediate pathname components.

We no longer match $path_dir against the ignore patterns.
I can't see how that was useful.

1 files changed, 50 insertions, 56 deletions

diff --git a/bin/hstow b/bin/hstow
index 7aba04de..a2cda67a 100755
--- a/bin/hstow
+++ b/bin/hstow
@@ -36,22 +36,14 @@ export LC_ALL=C
 tab="$(printf '\t')"
 cchars="$(printf '*[\001-\037\177]*')"
 
-if ! command -v readlink >/dev/null; then
-    readlink () {
-	# Safe parse of ls(1) output given its POSIX specification.
-	#
-	# We want the ninth field onwards in awk's sense --
-	# i.e. collapsing multiple runs of blanks between the first nine
-	# fields -- but we mustn't collapse blanks from there onwards.
-	n=$((3 + $(echo "$1" | wc -m)))
-	ls -ld "$1" \
-	    | sed -E \
-		  "s/^[[:blank:]]*([^[:blank:]]+[[:blank:]]+){8}.{$n}//"
-    }
-fi
-
-# Here we unconditionally bypass readlink(1), for speed.
+# We always bypass readlink(1), even if it's installed, for speed.
+#
+# This is a safe parse of ls(1) output given its POSIX specification.
 # The long regexp is because only BREs have backreferences in POSIX.
+#
+# We want the ninth field in the output of ls(1) in awk's sense --
+# i.e. collapsing runs of multiple blanks between the first nine fields
+# -- but we mustn't collapse blanks from there onwards.
 readlinks () {
     local restc=$1 assume="${2:-false}" first= rest= ls=
     while read -r first rest; do
@@ -73,11 +65,11 @@ readlinks () {
 '\)\([^[:blank:]]\{1,\}[[:blank:]]\{1,\}\)\{8\}\1 -> /\1\t\2/'
 }
 
-disjoin_hstow_file () {
+disjoin_file () {
     if [ -e "$DIR/$1" ]; then
 	while read -r line; do
 	    [ -n "$line" ] && printf "|$2" "$line"
-	done <"$DIR/$1" | cut -c2-
+	done <"$DIR/$1" | sed 's#^.##; s#/#\\/#g'
     fi
 }
 
@@ -100,50 +92,56 @@ stow () {
     [ -f "$HOME/.STOW/.stow" ] || touch "$HOME/.STOW/.stow"
     [ -h "$HOME/.STOW/$NAME" ] \
 	|| ( cd "$HOME/.STOW"; ln -s "$DIR" "$NAME" )
-    conflicts=
-    ignores="$(disjoin_hstow_file .hstow-local-ignore "./%s")"
+
+    conflicts="$(stow1)"
+    [ -z "$conflicts" ] && return
+    echo >&2 "hstow: encountered conflicts:"
+    for conflict in $conflicts; do echo >&2 "  $conflict"; done
+    exit 127
+}
+
+stow1 () {
+    ignores="$(disjoin_file .hstow-local-ignore "./%s")"
 
     # Files that (i) always/often have their symlinks replaced with
     # regular files when applications access them; and (ii) we don't
     # ever want to edit the copy under $DIR directly, but only via the
     # link/copy under $HOME.
     $always_adopt \
-	|| adoptions="$(disjoin_hstow_file .hstow-always-adopt "./%s")"
-
-    for file in $(find . ! -name . ! -type d ! -name "$cchars" \
-		       ! -name .gitignore \
-		       ! -name .hstow-local-ignore \
-		       ! -name .hstow-always-adopt \
-		       ! -name .hstow-unstow-ignore \
-		      | grep -v '^\./\.git/'); do
-	file_dir="$(dirname $file)"
-	if [ -n "$ignores" ]; then
-	    eval case "'$file'"     in "${ignores})" continue ";;" esac
-	    eval case "'$file_dir'" in "${ignores})" continue ";;" esac
-	fi
-
-	rel="$(echo $file|sed -E 's#/dot[-.]([^/]+)#/.\1#g; s#^\./##')"
-	dotdotslashes="$(echo $rel|sed -E 's#[^/]*$##; s#[^/]+#..#g')"
-	target="${dotdotslashes}.STOW/$NAME/$rel"
-	link="$HOME/$rel"
-	link_target=
-	[ -h "$link" ] && link_target="$(readlink $link)"
-
-	[ "$target" = "$link_target" ] && continue
-
-	if [ ! -h "$link" -a ! -h "$file" -a -f "$link" ]; then
-	    if $always_adopt \
-		    || ( [ -n "$adoptions" ] \
-			     && eval case "'$file'" in \
-				     "${adoptions})" exit 0 ";;" \
-				     "*)"            exit 1 ";;" \
-				     esac ); then
+	&& adoptp=1 \
+	|| adoptp="rel ~ /^($(disjoin_file .hstow-always-adopt "%s"))/"
+
+    # We have a separate pipeline element to find the link targets
+    # because with awk's system() we'd have to start a shell process for
+    # every record, in addition to the separate ls(1) process.
+    find . ! -name . ! -type d ! -name "$cchars" \
+	 ! -name .gitignore \
+	 ! -name .hstow-local-ignore \
+	 ! -name .hstow-always-adopt \
+	 ! -name .hstow-unstow-ignore \
+	| awk -F'\n' -vOFS='\t' '! /^(\.\/\.git\/|'"$ignores"')/ \
+{ rel = $1; gsub(/\/dot[-.]/, "/.", rel); gsub(/^\.\//, "", rel)
+  dotdotslashes = rel
+  sub(/[^\/]*$/, "", dotdotslashes)
+  gsub(/[^\/]+/, "..", dotdotslashes)
+  $2 = $1
+  $3 = $1; sub(/\/[^/]+$/, "", $3)
+  $1 = ENVIRON["HOME"] "/" rel
+  $4 = '"$adoptp"' ? "true" : "false"
+  $5 = dotdotslashes ".STOW/'"$NAME"'/" rel
+  print
+}' | readlinks 4 \
+	| while read -r link file file_dir adopt target link_target; do
+	if [ "$target" = "$link_target" ]; then
+	    continue
+	elif [ -z "$link_target" -a ! -h "$file" -a -f "$link" ]; then
+	    if $adopt; then
 		mv -f "$link" "$file"
 		ln -s "$target" "$link"
 	    else
-		conflicts="$conflicts${tab}$file"
+		echo "$file"	# into $conflicts
 	    fi
-	elif [ -h "$link" ]; then
+	elif [ -n "$link_target" ]; then
 	    # With at least GNU ln(1), passing -f, but not also -T, does
 	    # not replace an existing link in some cases.
 	    # -T is not POSIX, so we remove any existing link first.
@@ -154,17 +152,13 @@ stow () {
 	    ln -s "$target" "$link"
 	fi
     done
-    [ -z "$conflicts" ] && return
-    echo >&2 "hstow: encountered conflicts:"
-    for conflict in $conflicts; do echo >&2 "  $conflict"; done
-    exit 127
 }
 
 unstow () {
     cd "$HOME"
 
     # For speed, skip directories into which we'll never stow anything.
-    ignores="$(disjoin_hstow_file .hstow-unstow-ignore "./%s/")"
+    ignores="$(disjoin_file .hstow-unstow-ignore "./%s/")"
 
     dir_pat=".$(echo $DIR | cut -c$(echo $HOME | wc -m | tr -d ' ')-)/"
     dirs_pat="$(echo "^($dir_pat|$ignores)" | sed -e 's#\.#\\.#g')"