string_splitter 0.3.0 → 0.3.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 128e1b2cc29cb122f3d5040f7c9e115688532c7e68c59f0a5373291e995642f9
-  data.tar.gz: 6b8729b7fb59aa984c1940ff0f9a1a308dded8b77c7db2b3c2a2ad4cdbd8bd52
+  metadata.gz: 67fd08fc0c1d5928d849206b28130eadedbd7c38755f1c123f4c3d46cbbc5619
+  data.tar.gz: b102be89d4c59f9a2d3dd4661277a4fd3a31816f7dbae1630f2f6954bedad62a
 SHA512:
-  metadata.gz: 3aa949fb5ac46369af379e2fd28bc18f7c93746515aea76005d606a4cb9f20426dec353bef8406264078cdccf89cdaca99902d961687f12fb72be08d9f2b0072
-  data.tar.gz: acd982d39a003be78b4548992cf108141e89e51a58f918e10515fb01dd1fd562db319e5c0dc5475d3e74739726e1fd752bbee4850b823705fb9520ff6b05e99f
+  metadata.gz: 87d567793e20367c52625d5fa9dd6cea5470221b3d53bc54d0bd59f0f8835635d81a67e1fcecba5fcce5a116c6ba6c346c4b74fa563ac21bee5ff0d06d07ad8b
+  data.tar.gz: eab3f78e4c61e77c7bb283eb50e871d665fd4a323913e9fbf525ab2a6bfa05f0ebbabf490284371c00be6690f937f485823a8bc9dc3f59aafc9bff71c8cbe893

data/CHANGELOG.md

@@ -1,3 +1,7 @@
+## 0.3.1 - 2018-06-24
+
+- remove trailing empty field when the separator is empty (#1)
+
 ## 0.3.0 - 2018-06-23
 
 - **breaking change**: rename the `default_separator` option to `default_delimiter`

data/README.md

@@ -179,7 +179,7 @@ ss.split(line, at: [1..5, 8])
 
 # VERSION
 
-0.3.0
+0.3.1
 
 # SEE ALSO
 

data/lib/string_splitter/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 class StringSplitter
-  VERSION = '0.3.0'
+  VERSION = '0.3.1'
 end

data/lib/string_splitter.rb

@@ -137,68 +137,114 @@ class StringSplitter
     end
 
     ncaptures = match.captures.length
+    delimiter = increment_backrefs(delimiter, ncaptures)
+    parts = string.split(/(#{delimiter})/, -1)
+    remove_trailing_empty_field!(parts, ncaptures)
+    result, splits = splits_for(parts, ncaptures)
+    count = splits.length
+    block ||= at ? match_positions(at, count) : ACCEPT
 
-    if delimiter.is_a?(Regexp) && ncaptures > 0
-      # increment back-references so they remain valid when the outer capture
-      # is added e.g. to split on:
-      #
-      #   - <foo-comment> ... </foo-comment>
-      #   - <bar-comment> ... </bar-comment>
-      #
-      # etc.
-      #
-      # before:
-      #
-      #   %r|   <(\w+-comment)> [^<]* </\1>   |x
-      #
-      # after:
-      #
-      #   %r| ( <(\w+-comment)> [^<]* </\2> ) |x
+    [result, block, splits, count, -1]
+  end
 
+  # increment back-references so they remain valid when the outer capture
+  # is added.
+  #
+  # e.g. to split on:
+  #
+  #   - <foo-comment> ... </foo-comment>
+  #   - <bar-comment> ... </bar-comment>
+  #
+  # etc.
+  #
+  # before:
+  #
+  #   %r|   <(\w+-comment)> [^<]* </\1>   |x
+  #
+  # after:
+  #
+  #   %r| ( <(\w+-comment)> [^<]* </\2> ) |x
+
+  def increment_backrefs(delimiter, ncaptures)
+    if delimiter.is_a?(Regexp) && ncaptures > 0
       delimiter = delimiter.to_s.gsub(/\\(?:(\d+)|.)/) do
         match = Regexp.last_match
         match[1] ? '\\' + match[1].to_i.next.to_s : match[0]
       end
     end
 
-    parts = string.split(/(#{delimiter})/, -1)
-    result, splits = splits_for(parts, ncaptures)
-    count = splits.length
+    delimiter
+  end
 
-    unless block
-      if at
-        at = Array(at).map do |index|
-          if index.is_a?(Integer) && index.negative?
-            # translate 1-based negative indices to 1-based positive
-            # indices e.g:
-            #
-            #   ss.split("foo:bar:baz:quux", ":", at: -1)
-            #
-            # translates to:
-            #
-            #   ss.split("foo:bar:baz:quux", ":", at: 3)
-            #
-            # XXX note: we don't use modulo, because we don't want
-            # out-of-bounds indices to silently work e.g. we don't want:
-            #
-            #   ss.split("foo:bar:baz:quux", ":", -42)
-            #
-            # to mysteriously match when the index is 2
-
-            count + 1 + index
-          else
-            index
-          end
-        end
+  # work around Ruby's (and Perl's and Groovy's) unhelpful behavior when splitting
+  # on an empty string/pattern without removing trailing empty fields e.g.:
+  #
+  #   "foobar".split("", -1)
+  #   "foobar".split(//, -1)
+  #   # => ["f", "o", "o", "b", "a", "r", ""]
+  #
+  #   "foobar".split(/()/, -1)
+  #   # => ["f", "", "o", "", "o", "", "b", "", "a", "", "r", "", ""]
+  #
+  #   "foobar".split(/(())/, -1)
+  #   # => ["f", "", "", "o", "", "", "o", "", "", "b", "", "", "a", "", "", "r", "", "", ""]
+  #
+  # *there is no such thing as an empty field whose separator is empty*, so
+  # if String#split's result ends with an empty separator, 0 or more (empty)
+  # captures and an empty field, we can safely remove them.
+
+  def remove_trailing_empty_field!(parts, ncaptures)
+    # the trailing field is at index -1. if there are 0 captures, the separator
+    # is at -2:
+    #
+    #   [empty_separator, empty_field]
+    #
+    # if there is 1 capture, the separator is at -3:
+    #
+    #   [empty_separator, capture, empty_field]
+    #
+    # etc. therefore we find the separator by walking back
+    #
+    #  1 (empty field)
+    #  + ncaptures
+    #  + 1 (separator)
+    #
+    # steps from the end of the array i.e. ncaptures + 2
+    count = ncaptures + 2
+    separator_index = count * -1
+
+    return unless parts[-1].empty? && parts[separator_index].empty?
+
+    # drop the empty separator, the (empty) captures, and the trailing empty field
+    parts.pop(count)
+  end
 
-        block = lambda do |split|
-          case split.position when *at then true else false end
-        end
+  def match_positions(positions, nsplits)
+    positions = Array(positions).map do |position|
+      if position.is_a?(Integer) && position.negative?
+        # translate negative indices to 1-based non-negative indices e.g:
+        #
+        #   ss.split("foo:bar:baz:quux", ":", at: -1)
+        #
+        # translates to:
+        #
+        #   ss.split("foo:bar:baz:quux", ":", at: 3)
+        #
+        # XXX note: we don't use modulo, because we don't want
+        # out-of-bounds indices to silently work e.g. we don't want:
+        #
+        #   ss.split("foo:bar:baz:quux", ":", -42)
+        #
+        # to mysteriously match when the position is 2
+
+        nsplits + 1 + position
       else
-        block = ACCEPT
+        position
       end
     end
 
-    [result, block, splits, count, -1]
+    lambda do |split|
+      case split.position when *positions then true else false end
+    end
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: string_splitter
 version: !ruby/object:Gem::Version
-  version: 0.3.0
+  version: 0.3.1
 platform: ruby
 authors:
 - chocolateboy
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2018-06-23 00:00:00.000000000 Z
+date: 2018-06-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: values