remembering: Refactor with awk + sort; make it faster

2 files changed, 57 insertions, 86 deletions

diff --git a/remembering b/remembering
index 2f39385..0734dae 100755
--- a/remembering
+++ b/remembering
@@ -68,96 +68,52 @@ if [ ! -e "$PROFILE" ]; then
 	touch "$PROFILE"
 fi
 
+MERGED="$(mktemp)"
+FILTERED="$(mktemp)"
 SORTED_STDIN="$(mktemp)"
-cat | sort > "$SORTED_STDIN"
-
-MERGED=''
-MERGEDF="$(mktemp)"
-FILTERED=''
-FILTEREDF="$(mktemp)"
-
-append() {
-	FILTERED="$FILTERED${FILTERED:+
-}$1"
-}
-
-accumulate() {
-	MERGED="$MERGED${MERGED:+
-}$1"
-}
-
-end() {
-	echo "$FILTERED" > "$FILTEREDF"
-	echo "$MERGED" > "$MERGEDF"
-	exit
-}
-
-get_left() {
-	LEFT="$(echo "$LEFT_L" | head -n1)"
-	LEFT_L="$(echo "$LEFT_L" | tail -n+2)"
-	if [ -z "$LEFT" ]; then
-		accumulate "$RIGHT_RANKED"
-		accumulate "$RIGHT_RANKED_L"
-		end
-	fi
-}
-
-get_right() {
-	RIGHT_RANKED="$(echo "$RIGHT_RANKED_L" | head -n1)"
-	RIGHT_RANKED_L="$(echo "$RIGHT_RANKED_L" | tail -n+2)"
-	if [ -z "$RIGHT_RANKED" ]; then
-		STDIN_LEFTOVER="$(echo "$LEFT_L" | awk '/./{print "0:" $0}')"
-		append "$STDIN_LEFTOVER"
-		accumulate "$STDIN_LEFTOVER"
-		end
-	fi
-}
-
-(
-	LEFT_L="$(cat "$SORTED_STDIN")"
-	RIGHT_RANKED_L="$(cat "$PROFILE")"
-	get_right
-	get_left
-	while true; do
-		RIGHT="$(echo "$RIGHT_RANKED" | cut -d: -f2-)"
-		if [ "$LEFT" = "$RIGHT" ]; then
-			append "$RIGHT_RANKED"
-			accumulate "$RIGHT_RANKED"
-			get_right
-			get_left
-			continue
-		fi
-
-		if [ "$(expr "$LEFT" \< "$RIGHT")" = 1 ]; then
-			append "0:$LEFT"
-			accumulate "0:$LEFT"
-			get_left
-			continue
-		else
-			get_right
-			accumulate "$RIGHT_RANKED"
-			continue
-		fi
-	done
-)
-
-CHOICE="$(sort -t: -k1nr,1 -k2,2 < "$FILTEREDF" | \
+cat | sort -u > "$SORTED_STDIN"
+
+xargs printf '0:%s\n' < "$SORTED_STDIN" | \
+	sort -t: -k2,2 -m - "$PROFILE" | \
+	tac | \
+	sort -t: -k2,2 -u > "$MERGED"
+
+xargs printf 'filter_marker:%s\n' < "$SORTED_STDIN" | \
+	cat - "$PROFILE" | \
+	sort -t: -k2,2 | \
+	awk '{
+		split($0, l, ":")
+		rank = l[1]
+		entry = substr($0, length(rank) + 2)
+		if (rank != "filter_marker") {
+			prev_rank = rank
+			prev_entry = entry
+		} else {
+			if (prev_entry == entry) {
+				print prev_rank ":" entry
+			} else {
+				print "0:" entry
+			}
+		}
+	}' > "$FILTERED"
+
+CHOICE="$(sort -t: -k1nr,1 -k2,2 < "$FILTERED" | \
 	cut -d: -f2- | \
 	sh -c "$COMMAND")"
 
-NEW_PROFILE="$(mktemp)"
 if [ -n "$CHOICE" ]; then
-	while IFS='' read -r LINE_RANKED; do
-		LINE="$(echo "$LINE_RANKED" | cut -d: -f2-)"
-		if [ "$CHOICE" = "$LINE" ]; then
-			RANK="$(echo "$LINE_RANKED" | cut -d: -f1)"
-			NEW_RANK=$((RANK + 1))
-			echo "$NEW_RANK:$LINE" >> "$NEW_PROFILE"
-		else
-			echo "$LINE_RANKED" >> "$NEW_PROFILE"
-		fi
-	done < "$MERGEDF"
+	NEW_PROFILE="$(mktemp)"
+	awk -v choice="$CHOICE" '{
+		split($0, l, ":")
+		rank = l[1]
+		entry = substr($0, length(rank) + 2)
+		if (entry == choice) {
+			# Naively increment ranking by one
+			print rank + 1 ":" entry
+		} else {
+			print rank ":" entry
+		}
+	}' "$MERGED" > "$NEW_PROFILE"
 	mv "$NEW_PROFILE" "$PROFILE"
 	echo "$CHOICE"
 fi
-
diff --git a/tests/ranking.sh b/tests/ranking.sh
index 370c61b..ea9c2c6 100755
--- a/tests/ranking.sh
+++ b/tests/ranking.sh
@@ -230,10 +230,24 @@ test_stdin_is_empty() {
 	test_ok
 }
 
+
+test_profile_does_not_exist() {
+	testing 'profile does not exist'
+	PROFILE="profile-does-not-exist-$(uuid)"
+	INPUT='a
+b
+c
+d
+e'
+	pick_x a "$INPUT"
+	assert_profile "$PROFILE" "$BASE_PROFILE_A_PICKED"
+	test_ok
+}
+
 test_profile_is_empty() {
 	testing 'profile is empty'
 	PROFILE="profile-is-empty-$(uuid)"
-	echo '' > "$XDG_DATA_HOME/$PROFILE"
+	printf '' > "$XDG_DATA_HOME/$PROFILE"
 	INPUT='a
 b
 c
@@ -253,4 +267,5 @@ test_stdin_profile_merging
 test_stdin_is_larger_than_profile
 test_stdin_is_smaller_than_profile
 test_stdin_is_empty
+test_profile_does_not_exist
 test_profile_is_empty