Improvements in performance and memory usage

I profiled fzf and it turned out that it was spending significant amount
of time repeatedly converting character arrays into Unicode codepoints.
This commit greatly improves search performance after the initial scan
by memoizing the converted results.

This commit also addresses the problem of unbounded memory usage of fzf.
fzf is a short-lived process that usually processes small input, so it
was implemented to cache the intermediate results very aggressively with
no notion of cache expiration/eviction. I still think a proper
implementation of caching scheme is definitely an overkill. Instead this
commit introduces limits to the maximum size (or minimum selectivity) of
the intermediate results that can be cached.

src/algo/algo.go

@@ -1,8 +1,9 @@
 package algo
 
 import (
-	"strings"
 	"unicode"
+
+	"github.com/junegunn/fzf/src/util"
 )
 
 /*
@@ -14,13 +15,11 @@ import (
  */
 
 // FuzzyMatch performs fuzzy-match
-func FuzzyMatch(caseSensitive bool, input *string, pattern []rune) (int, int) {
+func FuzzyMatch(caseSensitive bool, runes *[]rune, pattern []rune) (int, int) {
 	if len(pattern) == 0 {
 		return 0, 0
 	}
 
-	runes := []rune(*input)
-
 	// 0. (FIXME) How to find the shortest match?
 	//    a_____b__c__abc
 	//    ^^^^^^^^^^  ^^^
@@ -34,7 +33,7 @@ func FuzzyMatch(caseSensitive bool, input *string, pattern []rune) (int, int) {
 	sidx := -1
 	eidx := -1
 
-	for index, char := range runes {
+	for index, char := range *runes {
 		// This is considerably faster than blindly applying strings.ToLower to the
 		// whole string
 		if !caseSensitive {
@@ -43,10 +42,10 @@ func FuzzyMatch(caseSensitive bool, input *string, pattern []rune) (int, int) {
 			// compiler as of now does not inline non-leaf functions.)
 			if char >= 'A' && char <= 'Z' {
 				char += 32
-				runes[index] = char
+				(*runes)[index] = char
 			} else if char > unicode.MaxASCII {
 				char = unicode.To(unicode.LowerCase, char)
-				runes[index] = char
+				(*runes)[index] = char
 			}
 		}
 		if char == pattern[pidx] {
@@ -63,7 +62,7 @@ func FuzzyMatch(caseSensitive bool, input *string, pattern []rune) (int, int) {
 	if sidx >= 0 && eidx >= 0 {
 		pidx--
 		for index := eidx - 1; index >= sidx; index-- {
-			char := runes[index]
+			char := (*runes)[index]
 			if char == pattern[pidx] {
 				if pidx--; pidx < 0 {
 					sidx = index
@@ -76,27 +75,6 @@ func FuzzyMatch(caseSensitive bool, input *string, pattern []rune) (int, int) {
 	return -1, -1
 }
 
-// ExactMatchStrings performs exact-match using strings package.
-// Currently not used.
-func ExactMatchStrings(caseSensitive bool, input *string, pattern []rune) (int, int) {
-	if len(pattern) == 0 {
-		return 0, 0
-	}
-
-	var str string
-	if caseSensitive {
-		str = *input
-	} else {
-		str = strings.ToLower(*input)
-	}
-
-	if idx := strings.Index(str, string(pattern)); idx >= 0 {
-		prefixRuneLen := len([]rune((*input)[:idx]))
-		return prefixRuneLen, prefixRuneLen + len(pattern)
-	}
-	return -1, -1
-}
-
 // ExactMatchNaive is a basic string searching algorithm that handles case
 // sensitivity. Although naive, it still performs better than the combination
 // of strings.ToLower + strings.Index for typical fzf use cases where input
@@ -104,13 +82,12 @@ func ExactMatchStrings(caseSensitive bool, input *string, pattern []rune) (int,
 //
 // We might try to implement better algorithms in the future:
 // http://en.wikipedia.org/wiki/String_searching_algorithm
-func ExactMatchNaive(caseSensitive bool, input *string, pattern []rune) (int, int) {
+func ExactMatchNaive(caseSensitive bool, runes *[]rune, pattern []rune) (int, int) {
 	if len(pattern) == 0 {
 		return 0, 0
 	}
 
-	runes := []rune(*input)
-	numRunes := len(runes)
+	numRunes := len(*runes)
 	plen := len(pattern)
 	if numRunes < plen {
 		return -1, -1
@@ -118,7 +95,7 @@ func ExactMatchNaive(caseSensitive bool, input *string, pattern []rune) (int, in
 
 	pidx := 0
 	for index := 0; index < numRunes; index++ {
-		char := runes[index]
+		char := (*runes)[index]
 		if !caseSensitive {
 			if char >= 'A' && char <= 'Z' {
 				char += 32
@@ -140,14 +117,13 @@ func ExactMatchNaive(caseSensitive bool, input *string, pattern []rune) (int, in
 }
 
 // PrefixMatch performs prefix-match
-func PrefixMatch(caseSensitive bool, input *string, pattern []rune) (int, int) {
-	runes := []rune(*input)
-	if len(runes) < len(pattern) {
+func PrefixMatch(caseSensitive bool, runes *[]rune, pattern []rune) (int, int) {
+	if len(*runes) < len(pattern) {
 		return -1, -1
 	}
 
 	for index, r := range pattern {
-		char := runes[index]
+		char := (*runes)[index]
 		if !caseSensitive {
 			char = unicode.ToLower(char)
 		}
@@ -159,8 +135,8 @@ func PrefixMatch(caseSensitive bool, input *string, pattern []rune) (int, int) {
 }
 
 // SuffixMatch performs suffix-match
-func SuffixMatch(caseSensitive bool, input *string, pattern []rune) (int, int) {
-	runes := []rune(strings.TrimRight(*input, " "))
+func SuffixMatch(caseSensitive bool, input *[]rune, pattern []rune) (int, int) {
+	runes := util.TrimRight(input)
 	trimmedLen := len(runes)
 	diff := trimmedLen - len(pattern)
 	if diff < 0 {

src/algo/algo_test.go

@@ -5,11 +5,12 @@ import (
 	"testing"
 )
 
-func assertMatch(t *testing.T, fun func(bool, *string, []rune) (int, int), caseSensitive bool, input string, pattern string, sidx int, eidx int) {
+func assertMatch(t *testing.T, fun func(bool, *[]rune, []rune) (int, int), caseSensitive bool, input string, pattern string, sidx int, eidx int) {
 	if !caseSensitive {
 		pattern = strings.ToLower(pattern)
 	}
-	s, e := fun(caseSensitive, &input, []rune(pattern))
+	runes := []rune(input)
+	s, e := fun(caseSensitive, &runes, []rune(pattern))
 	if s != sidx {
 		t.Errorf("Invalid start index: %d (expected: %d, %s / %s)", s, sidx, input, pattern)
 	}
@@ -45,7 +46,6 @@ func TestSuffixMatch(t *testing.T) {
 
 func TestEmptyPattern(t *testing.T) {
 	assertMatch(t, FuzzyMatch, true, "foobar", "", 0, 0)
-	assertMatch(t, ExactMatchStrings, true, "foobar", "", 0, 0)
 	assertMatch(t, ExactMatchNaive, true, "foobar", "", 0, 0)
 	assertMatch(t, PrefixMatch, true, "foobar", "", 0, 0)
 	assertMatch(t, SuffixMatch, true, "foobar", "", 6, 6)