textutils 0.8.6 → 0.8.7

data/Manifest.txt

@@ -30,6 +30,9 @@ lib/textutils/version.rb
 test/helper.rb
 test/test_address_helper.rb
 test/test_hypertext_helper.rb
+test/test_taglist.rb
+test/test_title_finder.rb
 test/test_title_helper.rb
+test/test_title_mapper.rb
 test/test_unicode_helper.rb
 test/test_values_reader.rb

data/lib/textutils/helper/title_helper.rb

@@ -157,15 +157,35 @@ module TextUtils
 
   def title_esc_regex( title_unescaped )
       
-      ##  escape regex special chars e.g. . to \. and ( to \( etc.
+      ##  escape regex special chars e.g.
+      #    . to \. and
+      #    ( to \(
+      #    ) to \)
+      #    ? to \? -- zero or one
+      #    * to \* -- zero or more
+      #    + to \+ -- one or more
+      #    $ to \$ -- end of line
+      #    ^ to \^ -- start of line etc.
+      
+      ### add { and } ???
+      ### add [ and ] ???
+      ### add \ too ???
+      ### add | too ???
+
       # e.g. Benfica Lis.
       # e.g. Club Atlético Colón (Santa Fe)
+      # e.g. Bauer Anton (????)
 
       ## NB: cannot use Regexp.escape! will escape space '' to '\ '
       ## title = Regexp.escape( title_unescaped )
       title = title_unescaped.gsub( '.', '\.' )
       title = title.gsub( '(', '\(' )
       title = title.gsub( ')', '\)' )
+      title = title.gsub( '?', '\?' )
+      title = title.gsub( '*', '\*' )
+      title = title.gsub( '+', '\+' )
+      title = title.gsub( '$', '\$' )
+      title = title.gsub( '^', '\^' )
 
       ##  match accented char with or without accents
       ##  add (ü|ue) etc.

data/lib/textutils/helper/value_helper.rb

@@ -45,7 +45,12 @@ module TextUtils
 
   def is_region?( value )
     # assume region code e.g. TX or N
-    value =~ /^[A-Z]{1,2}$/
+    #
+    # fix: allow three letter regions too e.g. BRU (brussels)
+    match_result =  value =~ /^[A-Z]{1,2}$/
+    # match found if 0,1,2,3 etc or no match if nil
+    # note: return bool e.g. false|true  (not 0,1,2,3 etc. and nil)
+    match_result != nil
   end
 
   ## fix/todo: use match_region_for_country! w/ !!! why? why not?
@@ -88,6 +93,8 @@ module TextUtils
     end
   end
 
+  ######
+  ## fix: move to worlddb?? why why not??
   def match_metro_flag( value )
     if value =~ /^metro$/   # metro(politan area)
       yield( true )
@@ -97,6 +104,8 @@ module TextUtils
     end
   end
 
+  ######
+  ## fix: move to worlddb?? why why not??
   def match_metro_pop( value )
     if value =~ /^m:/   # m:
       num = value[2..-1].gsub(/[ _]/, '').to_i   # cut off m: prefix; allow space and _ in number
@@ -109,7 +118,9 @@ module TextUtils
 
 
 
-
+  #####
+  ## fix: move to beerdb ??? why? why not??
+  
   def match_brewery( value )
     if value =~ /^by:/   ## by:  -brewed by/brewery
       brewery_key = value[3..-1]  ## cut off by: prefix
@@ -124,9 +135,13 @@ module TextUtils
 
   def is_year?( value )
     # founded/established year e.g. 1776
-    value =~ /^[0-9]{4}$/
+    match_result =  value =~ /^[0-9]{4}$/
+    # match found if 0,1,2,3 etc or no match if nil
+    # note: return bool e.g. false|true  (not 0,1,2,3 etc. and nil)
+    match_result != nil
   end
 
+
   def match_year( value )
     if is_year?( value )  # founded/established year e.g. 1776
       yield( value.to_i )
@@ -206,7 +221,10 @@ module TextUtils
     #  - must end w/   .com
     #
     # fix: support more url format (e.g. w/o www. - look for .com .country code etc.)
-    value =~ /^www\.|\.com$/
+    match_result =  value =~ /^www\.|\.com$/
+    # match found if 0,1,2,3 etc or no match if nil
+    # note: return bool e.g. false|true  (not 0,1,2,3 etc. and nil)
+    match_result != nil
   end
 
   def match_website( value )
@@ -223,18 +241,45 @@ module TextUtils
 
   def is_address?( value )
     # if value includes // assume address e.g. 3970 Weitra // Sparkasseplatz 160
-    value =~ /\/{2}/
+    match_result =  value =~ /\/{2}/
+    # match found if 0,1,2,3 etc or no match if nil
+    # note: return bool e.g. false|true  (not 0,1,2,3 etc. and nil)
+    match_result != nil
   end
 
   def is_taglist?( value )
-    value =~ /^[a-z0-9\|_ ]+$/
+    ### note: cannot start w/ number must be letter for now
+    ##  -- in the future allow free standing years (e.g. 1980 etc.?? why? why not?)
+    ##  e.g. not allowed  14 ha or 5_000 hl etc.
+    match_result =  value =~ /^([a-z][a-z0-9\|_ ]*[a-z0-9]|[a-z])$/
+    # match found if 0,1,2,3 etc or no match if nil
+    # note: return bool e.g. false|true  (not 0,1,2,3 etc. and nil)
+    match_result != nil
   end
 
 
   def find_grade( value )  # NB: returns ary [grade,value] / two values
     grade = 4  # defaults to grade 4  e.g  *** => 1, ** => 2, * => 3, -/- => 4
 
-    value = value.sub( /\s+(\*{1,3})\s*$/ ) do |_|  # NB: stars must end field/value
+    # NB: stars must end field/value or start field/value
+    #  e.g.
+    #  *** Anton Bauer   or
+    #  Anton Bauer ***
+
+    value = value.sub( /^\s*(\*{1,3})\s+/ ) do |_|
+      if $1 == '***'
+        grade = 1
+      elsif $1 == '**'
+        grade = 2
+      elsif $1 == '*'
+        grade = 3
+      else
+        # unknown grade; not possible, is'it?
+      end
+      ''  # remove * from title if found
+    end
+
+    value = value.sub( /\s+(\*{1,3})\s*$/ ) do |_|
       if $1 == '***'
         grade = 1
       elsif $1 == '**'
@@ -267,7 +312,9 @@ module TextUtils
 
     # fix/todo: add support for leading underscore _
     #   or allow keys starting w/ digits?
-    if values[0] =~ /^([a-z][a-z0-9.]*[a-z0-9]|[a-z])$/    # NB: key must start w/ a-z letter (NB: minimum one letter possible)
+    
+    # NB: key must start w/ a-z letter (NB: minimum one letter possible)
+    if values[0] =~ /^([a-z][a-z0-9.]*[a-z0-9]|[a-z])$/
       key_col         = values[0]
       title_col       = values[1]
       more_values     = values[2..-1]

data/lib/textutils/version.rb

@@ -1,6 +1,6 @@
 
 module TextUtils
 
-  VERSION = '0.8.6'
+  VERSION = '0.8.7'
 
 end   # module TextUtils

data/test/helper.rb

@@ -8,6 +8,9 @@ require 'minitest/autorun'
 
 # include MiniTest::Unit  # lets us use TestCase instead of MiniTest::Unit::TestCase
 
+## make sure activesupport gets included/required
+# note: just activesupport or active_support will NOT work
+require 'active_support/all'
 
 ## our own code
 

data/test/test_taglist.rb

@@ -0,0 +1,32 @@
+# encoding: utf-8
+
+
+require 'helper'
+
+
+class TestTaglist < MiniTest::Unit::TestCase
+
+  include TextUtils::ValueHelper   #  lets us use is_taglist?, etc.
+
+  def test_taglist_starting_w_digit
+    ## for now - taglist cannot start w/ number
+    assert is_taglist?( '20 ha' ) == false
+    assert is_taglist?( '5000 hl' ) == false
+    assert is_taglist?( '5_000 hl' ) == false
+  end
+
+  def test_taglist_upcase
+    ## taglist cannot use upcase letters
+    assert is_taglist?( 'ABC' ) == false
+  end
+
+  def test_taglist
+    assert is_taglist?( 'a' )
+    assert is_taglist?( 'a|b|c' )
+    assert is_taglist?( 'a b c' )
+    assert is_taglist?( 'a_b_c' )
+  end
+
+
+end # class TestTaglist
+

data/test/test_title_finder.rb

@@ -0,0 +1,25 @@
+# encoding: utf-8
+
+
+require 'helper'
+
+
+class TestTitleFinder < MiniTest::Unit::TestCase
+
+  include TextUtils::ValueHelper   #  lets us use find_grade, etc.
+
+  def test_grade
+
+    assert_equal [1,'Anton Bauer'], find_grade( '*** Anton Bauer' )
+    assert_equal [2,'Anton Bauer'], find_grade( '** Anton Bauer' )
+    assert_equal [3,'Anton Bauer'], find_grade( '* Anton Bauer' )
+    assert_equal [4,'Anton Bauer'], find_grade( 'Anton Bauer' )
+
+    assert_equal [1,'Anton Bauer'], find_grade( 'Anton Bauer ***' )
+
+  end
+
+
+
+end # class TestTitleFinder
+

data/test/test_title_mapper.rb

@@ -0,0 +1,50 @@
+# encoding: utf-8
+
+
+require 'helper'
+
+
+class TestTitleMapper < MiniTest::Unit::TestCase
+
+  WineryStruct =  Struct.new(:key, :title, :synonyms)
+
+  def test_title_table
+
+    ### todo/fix: auto-add year and remove (1971) or (????) etc. from title!!!!
+
+    titles_in = [
+      WineryStruct.new( 'antonbauer',     'Anton Bauer (1971)' ),
+      WineryStruct.new( 'josefbauer',      'Weingut Josef Bauer', 'Joe Bauer|Josef Bauer (????)' ),
+      WineryStruct.new( 'bernhardott',     'Weingut Ott', 'Weingut Bernhard Ott|Bernhard Ott (1972)' ),
+      WineryStruct.new( 'andreaspolsterer', 'Weingut Andreas B. Polsterer', 'Andreas B. Polsterer (1970)' )
+    ]
+
+    ## note: for regex the following must get escaped
+    #   (  => \(
+    #   )  => \)
+    #   .  => \.
+    #   ?  => \?
+
+    titles_out2 = [
+      ['antonbauer',       [ 'Anton Bauer \(1971\)', 'Anton Bauer']],
+      ['josefbauer',       [ 'Weingut Josef Bauer', 'Josef Bauer \(\?\?\?\?\)', 'Josef Bauer', 'Joe Bauer' ]],
+      ['bernhardott',      [ 'Weingut Bernhard Ott', 'Bernhard Ott \(1972\)', 'Bernhard Ott', 'Weingut Ott' ]],
+      ['andreaspolsterer', [ 'Weingut Andreas B\. Polsterer', 'Andreas B\. Polsterer \(1970\)', 'Andreas B\. Polsterer' ]]
+    ]
+
+    titles_out = TextUtils.build_title_table_for( titles_in )
+
+    puts 'titles_out:'
+    pp titles_out
+    puts titles_out.to_s
+
+    puts 'titles_out2:'
+    pp titles_out2
+    puts titles_out.to_s
+
+    assert_equal titles_out2.to_s, titles_out.to_s
+
+  end # method test_title_table
+
+
+end # class TestTitleMapper

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: textutils
 version: !ruby/object:Gem::Version
-  version: 0.8.6
+  version: 0.8.7
   prerelease: 
 platform: ruby
 authors:
@@ -9,11 +9,11 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-02-16 00:00:00.000000000 Z
+date: 2014-03-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: logutils
-  requirement: &20840748 !ruby/object:Gem::Requirement
+  requirement: &78669010 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -21,37 +21,35 @@ dependencies:
         version: '0.5'
   type: :runtime
   prerelease: false
-  version_requirements: *20840748
+  version_requirements: *78669010
 - !ruby/object:Gem::Dependency
   name: rdoc
-  requirement: &20840412 !ruby/object:Gem::Requirement
+  requirement: &78668430 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
       - !ruby/object:Gem::Version
-        version: '4.0'
+        version: '3.10'
   type: :development
   prerelease: false
-  version_requirements: *20840412
+  version_requirements: *78668430
 - !ruby/object:Gem::Dependency
   name: hoe
-  requirement: &20839944 !ruby/object:Gem::Requirement
+  requirement: &78667530 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
       - !ruby/object:Gem::Version
-        version: '3.7'
+        version: '3.3'
   type: :development
   prerelease: false
-  version_requirements: *20839944
+  version_requirements: *78667530
 description: textutils - Text Filters, Helpers, Readers and More
 email: ruby-talk@ruby-lang.org
 executables: []
 extensions: []
 extra_rdoc_files:
-- History.md
 - Manifest.txt
-- README.md
 files:
 - History.md
 - Manifest.txt
@@ -85,7 +83,10 @@ files:
 - test/helper.rb
 - test/test_address_helper.rb
 - test/test_hypertext_helper.rb
+- test/test_taglist.rb
+- test/test_title_finder.rb
 - test/test_title_helper.rb
+- test/test_title_mapper.rb
 - test/test_unicode_helper.rb
 - test/test_values_reader.rb
 - .gemtest
@@ -112,13 +113,16 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: textutils
-rubygems_version: 1.8.16
+rubygems_version: 1.8.17
 signing_key: 
 specification_version: 3
 summary: textutils - Text Filters, Helpers, Readers and More
 test_files:
-- test/test_address_helper.rb
-- test/test_hypertext_helper.rb
-- test/test_title_helper.rb
+- test/test_title_finder.rb
 - test/test_unicode_helper.rb
+- test/test_title_mapper.rb
 - test/test_values_reader.rb
+- test/test_taglist.rb
+- test/test_hypertext_helper.rb
+- test/test_title_helper.rb
+- test/test_address_helper.rb