Add --accept-nth option to transform the output

This option can be used to replace a sed or awk in the post-processing step.

  ps -ef | fzf --multi --header-lines 1 | awk '{print $2}'
  ps -ef | fzf --multi --header-lines 1 --accept-nth 2

This may not be a very "Unix-y" thing to do, so I've always felt that fzf
shouldn't have such an option, but I've finally changed my mind because:

* fzf can be configured with a custom delimiter that is a fixed string
  or a regular expression.
* In such cases, you'd need to repeat the delimiter again in the
  post-processing step.
* Also, tools like awk or sed may interpret a regular expression
  differently, causing mismatches.

You can still use sed, cut, or awk if you prefer.

Close #3987
Close #1323

man/man1/fzf.1

@@ -120,6 +120,10 @@ fields.
 .BI "\-\-with\-nth=" "N[,..]"
 Transform the presentation of each line using field index expressions
 .TP
+.BI "\-\-accept\-nth=" "N[,..]"
+Define which fields to print on accept. The last delimiter is stripped from the
+output.
+.TP
 .B "+s, \-\-no\-sort"
 Do not sort the result
 .TP

src/core.go

@@ -128,7 +128,7 @@ func Run(opts *Options) (int, error) {
 				}
 			}
 			trans := Transform(tokens, opts.WithNth)
-			transformed := joinTokens(trans)
+			transformed := JoinTokens(trans)
 			if len(header) < opts.HeaderLines {
 				header = append(header, transformed)
 				eventBox.Set(EvtHeader, header)

src/options.go

@@ -41,6 +41,7 @@ Usage: fzf [options]
                              integer or a range expression ([BEGIN]..[END]).
     --with-nth=N[,..]        Transform the presentation of each line using
                              field index expressions
+    --accept-nth=N[,..]      Define which fields to print on accept
     -d, --delimiter=STR      Field delimiter regex (default: AWK-style)
     +s, --no-sort            Do not sort the result
     --literal                Do not normalize latin script letters
@@ -544,6 +545,7 @@ type Options struct {
 	Normalize         bool
 	Nth               []Range
 	WithNth           []Range
+	AcceptNth         []Range
 	Delimiter         Delimiter
 	Sort              int
 	Track             trackOption
@@ -666,6 +668,7 @@ func defaultOptions() *Options {
 		Normalize:    true,
 		Nth:          make([]Range, 0),
 		WithNth:      make([]Range, 0),
+		AcceptNth:    make([]Range, 0),
 		Delimiter:    Delimiter{},
 		Sort:         1000,
 		Track:        trackDisabled,
@@ -2383,6 +2386,14 @@ func parseOptions(index *int, opts *Options, allArgs []string) error {
 			if opts.WithNth, err = splitNth(str); err != nil {
 				return err
 			}
+		case "--accept-nth":
+			str, err := nextString("nth expression required")
+			if err != nil {
+				return err
+			}
+			if opts.AcceptNth, err = splitNth(str); err != nil {
+				return err
+			}
 		case "-s", "--sort":
 			if opts.Sort, err = optionalNumeric(1); err != nil {
 				return err

src/pattern.go

@@ -403,6 +403,8 @@ func (p *Pattern) transformInput(item *Item) []Token {
 
 	tokens := Tokenize(item.text.ToString(), p.delimiter)
 	ret := Transform(tokens, p.nth)
+	// TODO: We could apply StripLastDelimiter to exclude the last delimiter from
+	// the search allowing suffix match with a string or a regex delimiter.
 	item.transformed = &transformed{p.revision, ret}
 	return ret
 }

src/terminal.go

@@ -305,6 +305,7 @@ type Terminal struct {
 	nthAttr            tui.Attr
 	nth                []Range
 	nthCurrent         []Range
+	acceptNth          []Range
 	tabstop            int
 	margin             [4]sizeSpec
 	padding            [4]sizeSpec
@@ -914,6 +915,7 @@ func NewTerminal(opts *Options, eventBox *util.EventBox, executor *util.Executor
 		nthAttr:            opts.Theme.Nth.Attr,
 		nth:                opts.Nth,
 		nthCurrent:         opts.Nth,
+		acceptNth:          opts.AcceptNth,
 		tabstop:            opts.Tabstop,
 		hasStartActions:    false,
 		hasResultActions:   false,
@@ -1561,16 +1563,24 @@ func (t *Terminal) output() bool {
 	for _, s := range t.printQueue {
 		t.printer(s)
 	}
+	transform := func(item *Item) string {
+		return item.AsString(t.ansi)
+	}
+	if len(t.acceptNth) > 0 {
+		transform = func(item *Item) string {
+			return JoinTokens(StripLastDelimiter(Transform(Tokenize(item.AsString(t.ansi), t.delimiter), t.acceptNth), t.delimiter))
+		}
+	}
 	found := len(t.selected) > 0
 	if !found {
 		current := t.currentItem()
 		if current != nil {
-			t.printer(current.AsString(t.ansi))
+			t.printer(transform(current))
 			found = true
 		}
 	} else {
 		for _, sel := range t.sortSelected() {
-			t.printer(sel.item.AsString(t.ansi))
+			t.printer(transform(sel.item))
 		}
 	}
 	return found
@@ -3847,7 +3857,7 @@ func replacePlaceholder(params replacePlaceholderParams) (string, []string) {
 				elems, prefixLength := awkTokenizer(params.query)
 				tokens := withPrefixLengths(elems, prefixLength)
 				trans := Transform(tokens, nth)
-				result := joinTokens(trans)
+				result := JoinTokens(trans)
 				if !flags.preserveSpace {
 					result = strings.TrimSpace(result)
 				}
@@ -3897,7 +3907,7 @@ func replacePlaceholder(params replacePlaceholderParams) (string, []string) {
 			replace = func(item *Item) string {
 				tokens := Tokenize(item.AsString(params.stripAnsi), params.delimiter)
 				trans := Transform(tokens, ranges)
-				str := joinTokens(trans)
+				str := JoinTokens(trans)
 
 				// trim the last delimiter
 				if params.delimiter.str != nil {

src/tokenizer.go

@@ -211,7 +211,36 @@ func Tokenize(text string, delimiter Delimiter) []Token {
 	return withPrefixLengths(tokens, 0)
 }
 
-func joinTokens(tokens []Token) string {
+// StripLastDelimiter removes the trailing delimiter and whitespaces from the
+// last token.
+func StripLastDelimiter(tokens []Token, delimiter Delimiter) []Token {
+	if len(tokens) == 0 {
+		return tokens
+	}
+
+	lastToken := tokens[len(tokens)-1]
+
+	if delimiter.str == nil && delimiter.regex == nil {
+		lastToken.text.TrimTrailingWhitespaces()
+	} else {
+		if delimiter.str != nil {
+			lastToken.text.TrimSuffix([]rune(*delimiter.str))
+		} else if delimiter.regex != nil {
+			str := lastToken.text.ToString()
+			locs := delimiter.regex.FindAllStringIndex(str, -1)
+			if len(locs) > 0 {
+				lastLoc := locs[len(locs)-1]
+				lastToken.text.SliceRight(lastLoc[0])
+			}
+		}
+		lastToken.text.TrimTrailingWhitespaces()
+	}
+
+	return tokens
+}
+
+// JoinTokens concatenates the tokens into a single string
+func JoinTokens(tokens []Token) string {
 	var output bytes.Buffer
 	for _, token := range tokens {
 		output.WriteString(token.text.ToString())
@@ -229,7 +258,7 @@ func Transform(tokens []Token, withNth []Range) []Token {
 		if r.begin == r.end {
 			idx := r.begin
 			if idx == rangeEllipsis {
-				chars := util.ToChars(stringBytes(joinTokens(tokens)))
+				chars := util.ToChars(stringBytes(JoinTokens(tokens)))
 				parts = append(parts, &chars)
 			} else {
 				if idx < 0 {

src/tokenizer_test.go

@@ -85,14 +85,14 @@ func TestTransform(t *testing.T) {
 		{
 			ranges, _ := splitNth("1,2,3")
 			tx := Transform(tokens, ranges)
-			if joinTokens(tx) != "abc:  def:  ghi:  " {
+			if JoinTokens(tx) != "abc:  def:  ghi:  " {
 				t.Errorf("%s", tx)
 			}
 		}
 		{
 			ranges, _ := splitNth("1..2,3,2..,1")
 			tx := Transform(tokens, ranges)
-			if string(joinTokens(tx)) != "abc:  def:  ghi:  def:  ghi:  jklabc:  " ||
+			if string(JoinTokens(tx)) != "abc:  def:  ghi:  def:  ghi:  jklabc:  " ||
 				len(tx) != 4 ||
 				tx[0].text.ToString() != "abc:  def:  " || tx[0].prefixLength != 2 ||
 				tx[1].text.ToString() != "ghi:  " || tx[1].prefixLength != 14 ||
@@ -107,7 +107,7 @@ func TestTransform(t *testing.T) {
 		{
 			ranges, _ := splitNth("1..2,3,2..,1")
 			tx := Transform(tokens, ranges)
-			if joinTokens(tx) != "  abc:  def:  ghi:  def:  ghi:  jkl  abc:" ||
+			if JoinTokens(tx) != "  abc:  def:  ghi:  def:  ghi:  jkl  abc:" ||
 				len(tx) != 4 ||
 				tx[0].text.ToString() != "  abc:  def:" || tx[0].prefixLength != 0 ||
 				tx[1].text.ToString() != "  ghi:" || tx[1].prefixLength != 12 ||

src/util/chars.go

@@ -189,6 +189,27 @@ func (chars *Chars) TrimTrailingWhitespaces() {
 	chars.slice = chars.slice[0 : len(chars.slice)-whitespaces]
 }
 
+func (chars *Chars) TrimSuffix(runes []rune) {
+	lastIdx := len(chars.slice)
+	firstIdx := lastIdx - len(runes)
+	if firstIdx < 0 {
+		return
+	}
+
+	for i := firstIdx; i < lastIdx; i++ {
+		char := chars.Get(i)
+		if char != runes[i-firstIdx] {
+			return
+		}
+	}
+
+	chars.slice = chars.slice[0:firstIdx]
+}
+
+func (chars *Chars) SliceRight(last int) {
+	chars.slice = chars.slice[:last]
+}
+
 func (chars *Chars) ToString() string {
 	if runes := chars.optionalRunes(); runes != nil {
 		return string(runes)

test/test_core.rb

@@ -1665,4 +1665,30 @@ class TestCore < TestInteractive
       assert_equal '', File.read(tempname).chomp
     end
   end
+
+  def test_accept_nth
+    tmux.send_keys %((echo "foo  bar  baz"; echo "bar baz  foo") | #{FZF} --multi --accept-nth 2,2 --sync --bind start:select-all+accept > #{tempname}), :Enter
+    wait do
+      assert_path_exists tempname
+      assert_equal ['bar  bar', 'baz  baz'], File.readlines(tempname, chomp: true)
+    end
+  end
+
+  def test_accept_nth_string_delimiter
+    tmux.send_keys %(echo "foo  ,bar,baz" | #{FZF} -d, --accept-nth 2,2,1,3,1 --sync --bind start:accept > #{tempname}), :Enter
+    wait do
+      assert_path_exists tempname
+      # Last delimiter and the whitespaces are removed
+      assert_equal ['bar,bar,foo  ,bazfoo'], File.readlines(tempname, chomp: true)
+    end
+  end
+
+  def test_accept_nth_regex_delimiter
+    tmux.send_keys %(echo "foo  :,:bar,baz" | #{FZF} --delimiter='[:,]+' --accept-nth 2,2,1,3,1 --sync --bind start:accept > #{tempname}), :Enter
+    wait do
+      assert_path_exists tempname
+      # Last delimiter and the whitespaces are removed
+      assert_equal ['bar,bar,foo  :,:bazfoo'], File.readlines(tempname, chomp: true)
+    end
+  end
 end