epitools 0.5.112 → 0.5.113

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d59d9daf2f46a0daf7008b75f596bcb8164801c61ec00a8a4f779e3105e97be1
-  data.tar.gz: 97b7ef67560c7377929db8aece4fc6f7f358f3ef96501c5913f99b75a655fa07
+  metadata.gz: ac614574fc21c61b88fb73f208d16c4960c5a3165a78844a56e26c8326e7bf82
+  data.tar.gz: 75914fb5972be2115ce1704c896f183300057ab6422be80f7dd50ab5088a152c
 SHA512:
-  metadata.gz: c17bdd3fada9128955712952cfa8edfa459ff24f83fd04ba4cef9116c3a931f9caba819095a4131f24c89bc4921d574110f6ccd180310a721c5d072896901e40
-  data.tar.gz: b642d62d3c9326b2b20a85ebb8bbf92ca5cff1150c6093f7f73530c44b878ef9637a36a8f2649fcac0e4bc22fca1de75469b947f99a648d77082f4d544f08161
+  metadata.gz: fd211d087deebafc7fec99003e6dfd20293cac20b9ad7db36cbd03641b1c2b4a4ddb3f1609fd236895cb596444012f5829ee27b2562271d6ca71a1696daaa205
+  data.tar.gz: 3c5c750c2767cfc600ae78538ea6339859e8a7c50d4b4b8876efebbbbb658d8a56166e29d50f8a72e72ac12c7e7a7a64a2d179c189f1b3901bc8e26fa4370a09

data/VERSION

@@ -1 +1 @@
-0.5.112
+0.5.113

data/lib/epitools/autoloads.rb

@@ -21,6 +21,7 @@ autoload :Shellwords, 'shellwords'
 autoload :PTY,        'pty'
 autoload :CSV,        'csv'
 autoload :SDBM,       'sdbm'
+autoload :StringScanner, 'strscan'
 
 module Digest
   autoload :SHA1,     'digest/sha1'

data/lib/epitools/core_ext/object.rb

@@ -74,27 +74,6 @@ class Object
     end
   end
 
-  #
-  # Cache (memoize) the result of an instance method the first time
-  # it's called, storing this value in the "@methodname" instance variable,
-  # and always return this value on subsequent calls.
-  #
-  def self.memoize(*methods)
-    methods.each do |meth|
-      old_method = instance_method(meth)
-
-      if old_method.arity == 0
-        ivarname = "@#{meth}"
-
-        define_method meth do
-          instance_variable_get(ivarname) or instance_variable_set(ivarname, old_method.bind(self).call)
-        end
-      else
-        raise "Can't memoize methods with arguments. (YET.)"
-      end
-    end
-  end
-
   #
   # Serialize this object to a binary String, using Marshal.dump.
   #
@@ -231,3 +210,26 @@ class NotWrapper < BasicObject # :nodoc:
   end
 end
 
+
+class Module
+
+  #
+  # Cache (memoize) the result of an instance method the first time
+  # it's called, storing this value in the "@__memoized_#{methodname}_cache"
+  # instance variable, and always return this value on subsequent calls
+  # (unless the returned value is nil).
+  #
+  def memoize(*methods)
+    # alias_method is faster than define_method + old.bind(self).call
+    methods.each do |meth|
+      alias_method "__memoized__#{meth}", meth
+      module_eval <<-EOF
+        def #{meth}(*a, &b)
+          # assumes the block won't change the result if the args are the same
+          (@__memoized_#{meth}_cache ||= {})[a] ||= __memoized__#{meth}(*a, &b)
+        end
+      EOF
+    end
+  end
+
+end  

data/lib/epitools/core_ext/string.rb

@@ -108,6 +108,36 @@ class String
   alias_method :chomp_lines,   :each_chomped
 
 
+  def split_at(boundary, options={})
+    include_boundary = options[:include_boundary] || false
+
+    boundary = Regexp.new(Regexp.escape(boundary)) if boundary.is_a?(String)
+    s        = StringScanner.new(self)
+
+    Enumerator.new do |yielder|
+      loop do
+        if match = s.scan_until(boundary)
+          if include_boundary
+            yielder << match
+          else
+            yielder << match[0..-(s.matched_size+1)]
+          end
+        else
+          yielder << s.rest if s.rest?
+          break
+        end
+      end
+    end
+  end
+
+  def split_after(boundary)
+    split_at(boundary, include_boundary: true)
+  end
+
+  def split_before(boundary)
+    raise "Why would you want this? Sorry, unimplemented. Send patches."
+  end
+
   #
   # Indent all the lines, if "prefix" is a string, prepend that string
   # to each lien. If it's an integer, prepend that many spaces.
@@ -199,9 +229,12 @@ class String
   end
   alias_method :wrapdent, :wrap_and_indent
 
+  def sentences
+    split_after(/[\.\!\?]+/).lazy.map {|s| s.strip.gsub(/\s+/, " ") }
+  end
+
   def words
-    scan /\w+/
-    # scan /^[a-z]+$|^\w+\-\w+|^[a-z]+[0-9]+[a-z]+$|^[0-9]+[a-z]+|^[a-z]+[0-9]+$/
+    scan /[[:alnum:]]+/
   end
 
   def words_without_stopwords

data/spec/core_ext_spec.rb

@@ -300,6 +300,8 @@ describe String do
   it "wordses" do
     s = "This, is a bunch, of words."
     s.words.should == ["This", "is", "a", "bunch", "of", "words"]
+  
+    "Weird word-like things".words.should == ["Weird", "word", "like", "things"]
   end
 
   it "wordses without stopwords" do
@@ -355,6 +357,25 @@ describe String do
     little_endian_bytes.to_i_from_bytes.should == i
   end
 
+  it "split_ats" do
+         "hello@@there@@whee".split_at("@@").to_a.should == ["hello", "there", "whee"]
+                      "hello".split_at("@@").to_a.should == ["hello"]
+    "whatAAAtheBBBheck???oic".split_at(/[AB\?]{1,3}/).to_a.should == ["what", "the", "heck", "oic"]
+                "hello@there".split_after("@").to_a.should == ["hello@", "there"]
+               "line\nline\n".split_at("\n", include_boundary: true).to_a.should == ["line\n", "line\n"]
+  end
+
+  it "sentences" do
+    string = "Sentence the first.  Sentence\nthe second. Sentence the third!!!\n\nThe end."
+
+    string.sentences.to_a.should == [
+      "Sentence the first.",
+      "Sentence the second.",
+      "Sentence the third!!!",
+      "The end.",
+    ]
+  end
+
 end
 
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: epitools
 version: !ruby/object:Gem::Version
-  version: 0.5.112
+  version: 0.5.113
 platform: ruby
 authors:
 - epitron
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2018-05-02 00:00:00.000000000 Z
+date: 2018-06-19 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rspec
@@ -131,7 +131,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.7.6
+rubygems_version: 2.7.7
 signing_key: 
 specification_version: 3
 summary: Not utils... METILS!