From f092e4038fd1344bcd0a7c11f7ab7355734f7275 Mon Sep 17 00:00:00 2001
From: Junegunn Choi <junegunn.c@gmail.com>
Date: Tue, 28 Jul 2020 13:06:57 +0900
Subject: [PATCH] Smart match of accented characters

Fix #1618
---
 CHANGELOG.md    |  4 ++++
 src/pattern.go  | 12 +++++++++---
 test/test_go.rb |  8 ++++++++
 3 files changed, 21 insertions(+), 3 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index b4d81536..5126d4cb 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -34,6 +34,10 @@ CHANGELOG
     fzf --delimiter : --preview 'nl {1}' --preview-window +{2}-5
   ```
 - Added support for ANSI colors in `--prompt` string
+- Smart match of accented characters
+    - An unaccented character in the query string will match both accented and
+      unaccented characters, while an accented character will only match
+      accented characters. This is similar to how "smart-case" match works.
 - Vim plugin
     - `tmux` layout option for using fzf-tmux
       ```vim
diff --git a/src/pattern.go b/src/pattern.go
index 4880d6e9..56dc830f 100644
--- a/src/pattern.go
+++ b/src/pattern.go
@@ -33,6 +33,7 @@ type term struct {
 	inv           bool
 	text          []rune
 	caseSensitive bool
+	normalize     bool
 }
 
 // String returns the string representation of a term.
@@ -128,6 +129,8 @@ func BuildPattern(fuzzy bool, fuzzyAlgo algo.Algo, extended bool, caseMode Case,
 		}
 	} else {
 		lowerString := strings.ToLower(asString)
+		normalize = normalize &&
+			lowerString == string(algo.NormalizeRunes([]rune(lowerString)))
 		caseSensitive = caseMode == CaseRespect ||
 			caseMode == CaseSmart && lowerString != asString
 		if !caseSensitive {
@@ -173,6 +176,8 @@ func parseTerms(fuzzy bool, caseMode Case, normalize bool, str string) []termSet
 		lowerText := strings.ToLower(text)
 		caseSensitive := caseMode == CaseRespect ||
 			caseMode == CaseSmart && text != lowerText
+		normalizeTerm := normalize &&
+			lowerText == string(algo.NormalizeRunes([]rune(lowerText)))
 		if !caseSensitive {
 			text = lowerText
 		}
@@ -222,14 +227,15 @@ func parseTerms(fuzzy bool, caseMode Case, normalize bool, str string) []termSet
 				set = termSet{}
 			}
 			textRunes := []rune(text)
-			if normalize {
+			if normalizeTerm {
 				textRunes = algo.NormalizeRunes(textRunes)
 			}
 			set = append(set, term{
 				typ:           typ,
 				inv:           inv,
 				text:          textRunes,
-				caseSensitive: caseSensitive})
+				caseSensitive: caseSensitive,
+				normalize:     normalizeTerm})
 			switchSet = true
 		}
 	}
@@ -360,7 +366,7 @@ func (p *Pattern) extendedMatch(item *Item, withPos bool, slab *util.Slab) ([]Of
 		matched := false
 		for _, term := range termSet {
 			pfun := p.procFun[term.typ]
-			off, score, pos := p.iter(pfun, input, term.caseSensitive, p.normalize, p.forward, term.text, withPos, slab)
+			off, score, pos := p.iter(pfun, input, term.caseSensitive, term.normalize, p.forward, term.text, withPos, slab)
 			if sidx := off[0]; sidx >= 0 {
 				if term.inv {
 					continue
diff --git a/test/test_go.rb b/test/test_go.rb
index 75e84ec4..2afb5b81 100755
--- a/test/test_go.rb
+++ b/test/test_go.rb
@@ -1805,6 +1805,14 @@ class TestGoFZF < TestBase
     tmux.until { |lines| lines.item_count == 1 }
     tmux.until { |lines| assert_match %r{121.*121/1000}, lines[1] }
   end
+
+  def test_normalized_match
+    echoes = '(echo a; echo á; echo A; echo Á;)'
+    assert_equal %w[a á A Á], `#{echoes} | #{FZF} -f a`.lines.map(&:chomp)
+    assert_equal %w[á Á], `#{echoes} | #{FZF} -f á`.lines.map(&:chomp)
+    assert_equal %w[A Á], `#{echoes} | #{FZF} -f A`.lines.map(&:chomp)
+    assert_equal %w[Á], `#{echoes} | #{FZF} -f Á`.lines.map(&:chomp)
+  end
 end
 
 module TestShell