Fix Gem executable

- Implement prefix caching of extended mode
- Improved ranking algorithm for extended mode
- Fix nfc conversion bug

Gemfile

@@ -1,4 +0,0 @@
-source 'https://rubygems.org'
-
-# Specify your gem's dependencies in fzf.gemspec
-gemspec

README.md

@@ -16,7 +16,8 @@ fzf requires Ruby (>= 1.8.5).
 Installation
 ------------
 
-Download fzf executable and put it somewhere in your search $PATH.
+Download [fzf executable](https://raw.github.com/junegunn/fzf/master/fzf) and
+put it somewhere in your search $PATH.
 
 ```sh
 mkdir -p ~/bin
@@ -38,15 +39,21 @@ Make sure that ~/bin is included in $PATH.
 export PATH=$PATH:~/bin
 ```
 
-Install as Vim plugin
+### Install as Ruby gem
----------------------
 
-fzf was not designed to be a Vim plugin, but you can use it as one. The only
+fzf can be installed as a Ruby gem
-reason one might consider using fzf in Vim is its speed. For a very large list
-of files, fzf is significantly faster than native Vim plugins.
 
-You can use any Vim plugin manager to install fzf as a Vim plugin. If you don't
+```
-use one, I recommend you try [vim-plug](https://github.com/junegunn/vim-plug).
+gem install fzf
+```
+
+It's a bit easier to install and update the script but the Ruby gem version
+takes slightly longer to start.
+
+### Install as Vim plugin
+
+You can use any Vim plugin manager to install fzf for Vim. If you don't use one,
+I recommend you try [vim-plug](https://github.com/junegunn/vim-plug).
 
 1. [Install vim-plug](https://github.com/junegunn/vim-plug#usage)
 2. Edit your .vimrc
@@ -64,9 +71,12 @@ Usage
 ```
 usage: fzf [options]
 
-  -s, --sort=MAX   Maximum number of matched items to sort. Default: 500
+  -m, --multi      Enable multi-select
-  +s, --no-sort    Keep the sequence unchanged.
+  -x, --extended   Extended-search mode
+  -s, --sort=MAX   Maximum number of matched items to sort. Default: 1000
+  +s, --no-sort    Do not sort the result. Keep the sequence unchanged.
   +i               Case-sensitive match
+  +c, --no-color   Disable colors
 ```
 
 fzf will launch curses-based finder, read the list from STDIN, and write the
@@ -77,10 +87,11 @@ find * -type f | fzf > selected
 ```
 
 Without STDIN pipe, fzf will use find command to fetch the list of
-files (excluding hidden ones).
+files excluding hidden ones. (You can override the default command with
+`FZF_DEFAULT_COMMAND`)
 
 ```sh
-vim `fzf`
+vim $(fzf)
 ```
 
 If you want to preserve the exact sequence of the input, provide `--no-sort` (or
@@ -100,6 +111,26 @@ The following readline key bindings should also work as expected.
 - CTRL-A / CTRL-E
 - CTRL-B / CTRL-F
 - CTRL-W / CTRL-U
+- ALT-B / ALT-F
+
+If you enable multi-select mode with `-m` option, you can select multiple items
+with TAB or Shift-TAB key.
+
+### Extended-search mode
+
+With `-x` or `--extended` option, fzf will start in "extended-search mode".
+
+In this mode, you can specify multiple patterns delimited by spaces,
+such as: `^music .mp3$ sbtrkt !rmx`
+
+| Token    | Description                      | Match type           |
+| -------- | -------------------------------- | -------------------- |
+| `^music` | Items that start with `music`    | prefix-exact-match   |
+| `.mp3$`  | Items that end with `.mp3`       | suffix-exact-match   |
+| `sbtrkt` | Items that match `sbtrkt`        | fuzzy-match          |
+| `!rmx`   | Items that do not match `rmx`    | inverse-fuzzy-match  |
+| `'wild`  | Items that include `wild`        | exact-match (quoted) |
+| `!'fire` | Items that do not include `fire` | inverse-exact-match  |
 
 Usage as Vim plugin
 -------------------
@@ -117,23 +148,27 @@ You can override the command which produces input to fzf.
 let g:fzf_command = 'find . -type f'
 ```
 
+Most of the time, you will prefer native Vim plugins with better integration
+with Vim. The only reason one might consider using fzf in Vim is its speed. For
+a very large list of files, fzf is significantly faster and it does not block.
+
 Useful bash examples
 --------------------
 
 ```sh
 # vimf - Open selected file in Vim
 vimf() {
-  FILE=`fzf` && vim "$FILE"
+  FILE=$(fzf) && vim "$FILE"
 }
 
 # fd - cd to selected directory
 fd() {
-  DIR=`find ${1:-*} -path '*/\.*' -prune -o -type d -print 2> /dev/null | fzf` && cd "$DIR"
+  DIR=$(find ${1:-*} -path '*/\.*' -prune -o -type d -print 2> /dev/null | fzf) && cd "$DIR"
 }
 
 # fda - including hidden directories
 fda() {
-  DIR=`find ${1:-*} -type d 2> /dev/null | fzf` && cd "$DIR"
+  DIR=$(find ${1:-*} -type d 2> /dev/null | fzf) && cd "$DIR"
 }
 
 # fh - repeat history
@@ -143,14 +178,84 @@ fh() {
 
 # fkill - kill process
 fkill() {
-  ps -ef | sed 1d | fzf | awk '{print $2}' | xargs kill -${1:-9}
+  ps -ef | sed 1d | fzf -m | awk '{print $2}' | xargs kill -${1:-9}
 }
 
-# CTRL-T - Open fuzzy finder and paste the selected item to the command line
+# (Assuming you don't use the default CTRL-T and CTRL-R)
+
+# CTRL-T - Paste the selected file path into the command line
 bind '"\er": redraw-current-line'
 bind '"\C-t": " \C-u \C-a\C-k$(fzf)\e\C-e\C-y\C-a\C-y\ey\C-h\C-e\er"'
+
+# CTRL-R - Paste the selected command from history into the command line
+bind '"\C-r": " \C-e\C-u$(history | fzf +s | sed \"s/ *[0-9]* *//\")\e\C-e\er"'
 ```
 
+zsh widgets
+-----------
+
+```sh
+# CTRL-T - Paste the selected file(s) path into the command line
+fzf-file-widget() {
+  local FILES
+  local IFS="
+"
+  FILES=($(
+    find * -path '*/\.*' -prune \
+    -o -type f -print \
+    -o -type l -print 2> /dev/null | fzf -m))
+  unset IFS
+  FILES=$FILES:q
+  LBUFFER="${LBUFFER%% #} $FILES"
+  zle redisplay
+}
+zle     -N   fzf-file-widget
+bindkey '^T' fzf-file-widget
+
+# ALT-C - cd into the selected directory
+fzf-cd-widget() {
+  cd "${$(find * -path '*/\.*' -prune \
+          -o -type d -print 2> /dev/null | fzf):-.}"
+  zle reset-prompt
+}
+zle     -N    fzf-cd-widget
+bindkey '\ec' fzf-cd-widget
+
+# CTRL-R - Paste the selected command from history into the command line
+fzf-history-widget() {
+  LBUFFER=$(history | fzf +s | sed "s/ *[0-9]* *//")
+  zle redisplay
+}
+zle     -N   fzf-history-widget
+bindkey '^R' fzf-history-widget
+```
+
+Tips
+----
+
+### Faster startup with `--disable-gems` options
+
+If you're running Ruby 1.9 or above, you can improve the startup time with
+`--disable-gems` option to Ruby.
+
+- `time ruby ~/bin/fzf -h`
+    - 0.077 sec
+- `time ruby --disable-gems ~/bin/fzf -h`
+    - 0.025 sec
+
+Define fzf alias with the option as follows:
+
+```sh
+alias fzf='ruby --disable-gems ~/bin/fzf'
+```
+
+### Incorrect display on Ruby 1.8
+
+It is reported that the output of fzf can become unreadable on some terminals
+when it's running on Ruby 1.8. If you experience the problem, upgrade your Ruby
+to 1.9 or above. Ruby 1.9 or above is also required for displaying Unicode
+characters.
+
 License
 -------
 

Rakefile

@@ -1 +1,8 @@
 require "bundler/gem_tasks"
+require 'rake/testtask'
+
+Rake::TestTask.new(:test) do |test|
+  test.pattern = 'test/**/test_*.rb'
+  test.verbose = true
+end
+

fzf.gemspec

@@ -1,7 +1,7 @@
 # coding: utf-8
 Gem::Specification.new do |spec|
   spec.name          = 'fzf'
-  spec.version       = '0.1.0'
+  spec.version       = '0.4.0'
   spec.authors       = ['Junegunn Choi']
   spec.email         = ['junegunn.c@gmail.com']
   spec.description   = %q{Fuzzy finder for your shell}

plugin/fzf.vim

@@ -27,9 +27,13 @@ function! s:fzf(args)
   try
     let tf = tempname()
     let prefix = exists('g:fzf_command') ? g:fzf_command.'|' : ''
-    execute "silent !".prefix."/usr/bin/env ruby ".s:exec." ".a:args." > ".tf
+    let fzf = executable(s:exec) ? s:exec : 'fzf'
+    execute "silent !".prefix.fzf." ".a:args." > ".tf
     if !v:shell_error
-      execute 'silent e '.join(readfile(tf), '')
+      let file = join(readfile(tf), '')
+      if !empty(file)
+        execute 'silent e '.file
+      endif
     endif
   finally
     silent! call delete(tf)

test/test_fzf.rb

@@ -0,0 +1,306 @@
+#!/usr/bin/env ruby
+# encoding: utf-8
+
+require 'minitest/autorun'
+$LOAD_PATH.unshift File.expand_path('../..', __FILE__)
+ENV['FZF_EXECUTABLE'] = '0'
+load 'fzf'
+
+class TestFZF < MiniTest::Unit::TestCase
+  def test_default_options
+    fzf = FZF.new []
+    assert_equal 1000, fzf.sort
+    assert_equal false, fzf.multi
+    assert_equal true, fzf.color
+    assert_equal Regexp::IGNORECASE, fzf.rxflag
+
+    begin
+      ENV['FZF_DEFAULT_SORT'] = '1500'
+      fzf = FZF.new []
+      assert_equal 1500, fzf.sort
+    ensure
+      ENV.delete 'FZF_DEFAULT_SORT'
+    end
+  end
+
+  def test_option_parser
+    # Long opts
+    fzf = FZF.new %w[--sort=2000 --no-color --multi +i]
+    assert_equal 2000, fzf.sort
+    assert_equal true, fzf.multi
+    assert_equal false, fzf.color
+    assert_equal 0, fzf.rxflag
+
+    # Short opts
+    fzf = FZF.new %w[-s 2000 +c -m +i]
+    assert_equal 2000, fzf.sort
+    assert_equal true, fzf.multi
+    assert_equal false, fzf.color
+    assert_equal 0, fzf.rxflag
+  end
+
+  def test_invalid_option
+    [%w[-s 2000 +s], %w[yo dawg]].each do |argv|
+      assert_raises(SystemExit) do
+        fzf = FZF.new argv
+      end
+    end
+  end
+
+  # FIXME Only on 1.9 or above
+  def test_width
+    fzf = FZF.new []
+    assert_equal 5, fzf.width('abcde')
+    assert_equal 4, fzf.width('한글')
+    assert_equal 5, fzf.width('한글.')
+  end
+
+  def test_trim
+    fzf = FZF.new []
+    assert_equal ['사.',     6], fzf.trim('가나다라마바사.', 4, true)
+    assert_equal ['바사.',   5], fzf.trim('가나다라마바사.', 5, true)
+    assert_equal ['바사.',   5], fzf.trim('가나다라마바사.', 6, true)
+    assert_equal ['마바사.', 4], fzf.trim('가나다라마바사.', 7, true)
+    assert_equal ['가나',    6], fzf.trim('가나다라마바사.', 4, false)
+    assert_equal ['가나',    6], fzf.trim('가나다라마바사.', 5, false)
+    assert_equal ['가나a',   6], fzf.trim('가나ab라마바사.', 5, false)
+    assert_equal ['가나ab',  5], fzf.trim('가나ab라마바사.', 6, false)
+    assert_equal ['가나ab',  5], fzf.trim('가나ab라마바사.', 7, false)
+  end
+
+  def test_format
+    fzf = FZF.new []
+    assert_equal [['01234..', false]], fzf.format('0123456789', 7, [])
+    assert_equal [['012', false], ['34', true], ['..', false]],
+      fzf.format('0123456789', 7, [[3, 5]])
+    assert_equal [['..56', false], ['789', true]],
+      fzf.format('0123456789', 7, [[7, 10]])
+    assert_equal [['..56', false], ['78', true], ['9', false]],
+      fzf.format('0123456789', 7, [[7, 9]])
+
+    (3..5).each do |i|
+      assert_equal [['..', false], ['567', true], ['89', false]],
+        fzf.format('0123456789', 7, [[i, 8]])
+    end
+
+    assert_equal [['..', false], ['345', true], ['..', false]],
+      fzf.format('0123456789', 7, [[3, 6]])
+    assert_equal [['012', false], ['34', true], ['..', false]],
+      fzf.format('0123456789', 7, [[3, 5]])
+
+    # Multi-region
+    assert_equal [["0", true], ["1", false], ["2", true], ["34..", false]],
+      fzf.format('0123456789', 7, [[0, 1], [2, 3]])
+
+    assert_equal [["..", false], ["5", true], ["6", false], ["78", true], ["9", false]],
+      fzf.format('0123456789', 7, [[3, 6], [7, 9]])
+
+    assert_equal [["..", false], ["3", true], ["4", false], ["5", true], ["..", false]],
+      fzf.format('0123456789', 7, [[3, 4], [5, 6]])
+
+    # Multi-region Overlap
+    assert_equal [["..", false], ["345", true], ["..", false]],
+      fzf.format('0123456789', 7, [[4, 5], [3, 6]])
+  end
+
+  def test_fuzzy_matcher
+    matcher = FZF::FuzzyMatcher.new Regexp::IGNORECASE
+    list = %w[
+      juice
+      juiceful
+      juiceless
+      juicily
+      juiciness
+      juicy]
+    assert matcher.caches.empty?
+    assert_equal(
+      [["juice",     [[0, 1]]],
+       ["juiceful",  [[0, 1]]],
+       ["juiceless", [[0, 1]]],
+       ["juicily",   [[0, 1]]],
+       ["juiciness", [[0, 1]]],
+       ["juicy",     [[0, 1]]]], matcher.match(list, 'j', '', '').sort)
+    assert !matcher.caches.empty?
+    assert_equal [list.object_id], matcher.caches.keys
+    assert_equal 1, matcher.caches[list.object_id].length
+    assert_equal 6, matcher.caches[list.object_id]['j'].length
+
+    assert_equal(
+      [["juicily",   [[0, 5]]],
+       ["juiciness", [[0, 5]]]], matcher.match(list, 'jii', '', '').sort)
+
+    assert_equal(
+      [["juicily",   [[2, 5]]],
+       ["juiciness", [[2, 5]]]], matcher.match(list, 'ii', '', '').sort)
+
+    assert_equal 3, matcher.caches[list.object_id].length
+    assert_equal 2, matcher.caches[list.object_id]['ii'].length
+
+    # TODO : partial_cache
+  end
+
+  def test_fuzzy_matcher_case_sensitive
+    assert_equal [['Fruit', [[0, 5]]]],
+      FZF::FuzzyMatcher.new(0).match(%w[Fruit Grapefruit], 'Fruit', '', '').sort
+
+    assert_equal [["Fruit", [[0, 5]]], ["Grapefruit", [[5, 10]]]],
+      FZF::FuzzyMatcher.new(Regexp::IGNORECASE).
+      match(%w[Fruit Grapefruit], 'Fruit', '', '').sort
+  end
+
+  def test_extended_fuzzy_matcher
+    matcher = FZF::ExtendedFuzzyMatcher.new Regexp::IGNORECASE
+    list = %w[
+      juice
+      juiceful
+      juiceless
+      juicily
+      juiciness
+      juicy
+      _juice]
+    match = proc { |q, prefix|
+      matcher.match(list, q, prefix, '').sort.map { |p| [p.first, p.last.sort] }
+    }
+
+    assert matcher.caches.empty?
+    3.times do
+      ['y j', 'j y'].each do |pat|
+        (0..pat.length - 1).each do |prefix_length|
+          prefix = pat[0, prefix_length]
+          assert_equal(
+            [["juicily", [[0, 1], [6, 7]]],
+             ["juicy",   [[0, 1], [4, 5]]]],
+            match.call(pat, prefix))
+        end
+      end
+
+      # $
+      assert_equal [["juiceful",  [[7, 8]]]], match.call('l$', '')
+      assert_equal [["juiceful",  [[7, 8]]],
+                    ["juiceless", [[5, 6]]],
+                    ["juicily",   [[5, 6]]]], match.call('l', '')
+
+      # ^
+      assert_equal list.length,     match.call('j', '').length
+      assert_equal list.length - 1, match.call('^j', '').length
+
+      # !
+      assert_equal 0, match.call('!j', '').length
+
+      # ! + ^
+      assert_equal [["_juice", []]], match.call('!^j', '')
+
+      # ! + $
+      assert_equal list.length - 1, match.call('!l$', '').length
+
+      # ! + f
+      assert_equal [["juicy", [[4, 5]]]], match.call('y !l', '')
+
+      # '
+      assert_equal %w[juiceful juiceless juicily],
+        match.call('il', '').map { |e| e.first }
+      assert_equal %w[juicily],
+        match.call("'il", '').map { |e| e.first }
+      assert_equal (list - %w[juicily]).sort,
+        match.call("!'il", '').map { |e| e.first }.sort
+    end
+    assert !matcher.caches.empty?
+  end
+
+  def test_xfuzzy_matcher_prefix_cache
+    matcher = FZF::ExtendedFuzzyMatcher.new Regexp::IGNORECASE
+    list = %w[
+      a.java
+      b.java
+      java.jive
+      c.java$
+      d.java
+    ]
+    2.times do
+      assert_equal 5, matcher.match(list, 'java',   'java',   '').length
+      assert_equal 3, matcher.match(list, 'java$',  'java$',  '').length
+      assert_equal 1, matcher.match(list, 'java$$', 'java$$', '').length
+
+      assert_equal 0, matcher.match(list, '!java',  '!java',  '').length
+      assert_equal 4, matcher.match(list, '!^jav',  '!^jav',  '').length
+      assert_equal 4, matcher.match(list, '!^java', '!^java', '').length
+      assert_equal 2, matcher.match(list, '!^java !b !c', '!^java', '').length
+    end
+  end
+
+  def test_sort_by_rank
+    matcher  = FZF::FuzzyMatcher.new Regexp::IGNORECASE
+    xmatcher = FZF::ExtendedFuzzyMatcher.new Regexp::IGNORECASE
+    list = %w[
+      0____1
+      0_____1
+      01
+      ____0_1
+      01_
+      _01_
+      0______1
+      ___01___
+    ]
+    assert_equal(
+      [["01",       [[0, 2]]],
+       ["01_",      [[0, 2]]],
+       ["_01_",     [[1, 3]]],
+       ["___01___", [[3, 5]]],
+       ["____0_1",  [[4, 7]]],
+       ["0____1",   [[0, 6]]],
+       ["0_____1",  [[0, 7]]],
+       ["0______1", [[0, 8]]]],
+      FZF.new([]).sort_by_rank(matcher.match(list, '01', '', '')))
+
+    assert_equal(
+      [["01",       [[0, 1], [1, 2]]],
+       ["01_",      [[0, 1], [1, 2]]],
+       ["_01_",     [[1, 2], [2, 3]]],
+       ["0____1",   [[0, 1], [5, 6]]],
+       ["0_____1",  [[0, 1], [6, 7]]],
+       ["____0_1",  [[4, 5], [6, 7]]],
+       ["0______1", [[0, 1], [7, 8]]],
+       ["___01___", [[3, 4], [4, 5]]]],
+      FZF.new([]).sort_by_rank(xmatcher.match(list, '0 1', '', '')))
+
+    assert_equal(
+      [["_01_",     [[1, 3], [0, 4]]],
+       ["0____1",   [[0, 6], [1, 3]]],
+       ["0_____1",  [[0, 7], [1, 3]]],
+       ["0______1", [[0, 8], [1, 3]]],
+       ["___01___", [[3, 5], [0, 2]]],
+       ["____0_1",  [[4, 7], [0, 2]]]],
+      FZF.new([]).sort_by_rank(xmatcher.match(list, '01 __', '', '')))
+  end
+
+  if RUBY_PLATFORM =~ /darwin/
+    NFD = '한글'
+    def test_nfc
+      assert_equal 6, NFD.length
+      assert_equal ["한글", [[0, 1], [1, 2]]],
+        FZF::UConv.nfc(NFD, [[0, 3], [3, 6]])
+
+      nfd2 = 'before' + NFD + 'after'
+      assert_equal 6 + 6 + 5, nfd2.length
+
+      nfc, offsets = FZF::UConv.nfc(nfd2, [[4, 14], [9, 13]])
+      o1, o2 = offsets
+      assert_equal 'before한글after', nfc
+      assert_equal 're한글af',        nfc[(o1.first...o1.last)]
+      assert_equal '글a',             nfc[(o2.first...o2.last)]
+    end
+
+    def test_nfd
+      nfc = '한글'
+      nfd = FZF::UConv.nfd(nfc)
+      assert_equal 6, nfd.length
+      assert_equal NFD, nfd
+    end
+  end
+
+  def test_split
+    assert_equal ["a", "b", "c", "\xFF", "d", "e", "f"],
+      FZF::UConv.split("abc\xFFdef")
+  end
+end
+