regexp_parser 1.3.0 → 1.6.0

data/test/scanner/test_literals.rb

@@ -1,89 +0,0 @@
-# -*- encoding: utf-8 -*-
-
-require File.expand_path("../../helpers", __FILE__)
-
-class ScannerUTF8 < Test::Unit::TestCase
-
-  tests = {
-    # ascii, single byte characters
-    'a' => {
-      0     => [:literal,     :literal,       'a',        0, 1],
-    },
-
-    'ab+' => {
-      0     => [:literal,     :literal,       'ab',       0, 2],
-      1     => [:quantifier,  :one_or_more,   '+',        2, 3],
-    },
-
-    # 2 byte wide characters, Arabic
-    'aاbبcت' => {
-      0     => [:literal,     :literal,       'aاbبcت',   0, 9],
-    },
-
-    'aاbبت?' => {
-      0     => [:literal,     :literal,       'aاbبت',    0, 8],
-      1     => [:quantifier,  :zero_or_one,   '?',        8, 9],
-    },
-
-    'aا?bبcت+' => {
-      0     => [:literal,     :literal,       'aا',       0, 3],
-      1     => [:quantifier,  :zero_or_one,   '?',        3, 4],
-      2     => [:literal,     :literal,       'bبcت',     4, 10],
-      3     => [:quantifier,  :one_or_more,   '+',        10, 11],
-    },
-
-    'a(اbب+)cت?' => {
-      0     => [:literal,     :literal,       'a',        0, 1],
-      1     => [:group,       :capture,       '(',        1, 2],
-      2     => [:literal,     :literal,       'اbب',      2, 7],
-      3     => [:quantifier,  :one_or_more,   '+',        7, 8],
-      4     => [:group,       :close,         ')',        8, 9],
-      5     => [:literal,     :literal,       'cت',       9, 12],
-      6     => [:quantifier,  :zero_or_one,   '?',        12, 13],
-    },
-
-    # 3 byte wide characters, Japanese
-    'ab?れます+cd' => {
-      0     => [:literal,     :literal,       'ab',       0, 2],
-      1     => [:quantifier,  :zero_or_one,   '?',        2, 3],
-      2     => [:literal,     :literal,       'れます',   3, 12],
-      3     => [:quantifier,  :one_or_more,   '+',        12, 13],
-      4     => [:literal,     :literal,       'cd',       13, 15],
-    },
-
-    # 4 byte wide characters, Osmanya
-    '𐒀𐒁?𐒂ab+𐒃' => {
-      0     => [:literal,     :literal,       '𐒀𐒁',       0, 8],
-      1     => [:quantifier,  :zero_or_one,   '?',        8, 9],
-      2     => [:literal,     :literal,       '𐒂ab',      9, 15],
-      3     => [:quantifier,  :one_or_more,   '+',        15, 16],
-      4     => [:literal,     :literal,       '𐒃',        16, 20],
-    },
-
-    'mu𝄞?si*𝄫c+' => {
-      0     => [:literal,     :literal,       'mu𝄞',      0, 6],
-      1     => [:quantifier,  :zero_or_one,   '?',        6, 7],
-      2     => [:literal,     :literal,       'si',       7, 9],
-      3     => [:quantifier,  :zero_or_more,  '*',        9, 10],
-      4     => [:literal,     :literal,       '𝄫c',       10, 15],
-      5     => [:quantifier,  :one_or_more,   '+',        15, 16],
-    },
-  }
-
-  tests.each_with_index do |(pattern, checks), count|
-    define_method "test_scanner_utf8_runs_#{count}" do
-      tokens = RS.scan(pattern)
-
-      checks.each do |index, (type, token, text, ts, te)|
-        result = tokens[index]
-
-        assert_equal type,  result[0]
-        assert_equal token, result[1]
-        assert_equal text,  result[2]
-        assert_equal ts,    result[3]
-        assert_equal te,    result[4]
-      end
-    end
-  end
-
-end

data/test/scanner/test_meta.rb

@@ -1,40 +0,0 @@
-require File.expand_path("../../helpers", __FILE__)
-
-class ScannerMeta < Test::Unit::TestCase
-
-  tests = {
-    'abc??|def*+|ghi+' => {
-      0   => [:literal,     :literal,                 'abc',  0, 3],
-      1   => [:quantifier,  :zero_or_one_reluctant,   '??',   3, 5],
-      2   => [:meta,        :alternation,             '|',    5, 6],
-      3   => [:literal,     :literal,                 'def',  6, 9],
-      4   => [:quantifier,  :zero_or_more_possessive, '*+',   9, 11],
-      5   => [:meta,        :alternation,             '|',    11, 12],
-    },
-
-    '(a\|b)|(c|d)\|(e[|]f)' => {
-      2   => [:escape,      :alternation,             '\|',   2, 4],
-      5   => [:meta,        :alternation,             '|',    6, 7],
-      8   => [:meta,        :alternation,             '|',    9, 10],
-      11  => [:escape,      :alternation,             '\|',   12, 14],
-      15  => [:literal,     :literal,                 '|',    17, 18],
-    },
-  }
-
-  tests.each_with_index do |(pattern, checks), count|
-    define_method "test_scanner_meta_alternation_#{count}" do
-      tokens = RS.scan(pattern)
-
-      checks.each do |index, (type, token, text, ts, te)|
-        result = tokens.at(index)
-
-        assert_equal type,  result[0]
-        assert_equal token, result[1]
-        assert_equal text,  result[2]
-        assert_equal ts,    result[3]
-        assert_equal te,    result[4]
-      end
-    end
-  end
-
-end

data/test/scanner/test_properties.rb

@@ -1,312 +0,0 @@
-require File.expand_path("../../helpers", __FILE__)
-
-class ScannerProperties < Test::Unit::TestCase
-
-  tests = {
-    'Alnum'                               => :alnum,
-
-    'XPosixPunct'                         => :xposixpunct,
-
-    'Newline'                             => :newline,
-
-    'Any'                                 => :any,
-
-    'Assigned'                            => :assigned,
-
-    'L'                                   => :letter,
-    'Letter'                              => :letter,
-
-    'Lu'                                  => :uppercase_letter,
-    'Uppercase_Letter'                    => :uppercase_letter,
-
-    'Ll'                                  => :lowercase_letter,
-    'Lowercase_Letter'                    => :lowercase_letter,
-
-    'Lt'                                  => :titlecase_letter,
-    'Titlecase_Letter'                    => :titlecase_letter,
-
-    'Lm'                                  => :modifier_letter,
-    'Modifier_Letter'                     => :modifier_letter,
-
-    'Lo'                                  => :other_letter,
-    'Other_Letter'                        => :other_letter,
-
-    'M'                                   => :mark,
-    'Mark'                                => :mark,
-
-    'Mn'                                  => :nonspacing_mark,
-    'Nonspacing_Mark'                     => :nonspacing_mark,
-
-    'Mc'                                  => :spacing_mark,
-    'Spacing_Mark'                        => :spacing_mark,
-
-    'Me'                                  => :enclosing_mark,
-    'Enclosing_Mark'                      => :enclosing_mark,
-
-    'N'                                   => :number,
-    'Number'                              => :number,
-
-    'Nd'                                  => :decimal_number,
-    'Decimal_Number'                      => :decimal_number,
-
-    'Nl'                                  => :letter_number,
-    'Letter_Number'                       => :letter_number,
-
-    'No'                                  => :other_number,
-    'Other_Number'                        => :other_number,
-
-    'P'                                   => :punctuation,
-    'Punctuation'                         => :punctuation,
-
-    'Pc'                                  => :connector_punctuation,
-    'Connector_Punctuation'               => :connector_punctuation,
-
-    'Pd'                                  => :dash_punctuation,
-    'Dash_Punctuation'                    => :dash_punctuation,
-
-    'Ps'                                  => :open_punctuation,
-    'Open_Punctuation'                    => :open_punctuation,
-
-    'Pe'                                  => :close_punctuation,
-    'Close_Punctuation'                   => :close_punctuation,
-
-    'Pi'                                  => :initial_punctuation,
-    'Initial_Punctuation'                 => :initial_punctuation,
-
-    'Pf'                                  => :final_punctuation,
-    'Final_Punctuation'                   => :final_punctuation,
-
-    'Po'                                  => :other_punctuation,
-    'Other_Punctuation'                   => :other_punctuation,
-
-    'S'                                   => :symbol,
-    'Symbol'                              => :symbol,
-
-    'Sm'                                  => :math_symbol,
-    'Math_Symbol'                         => :math_symbol,
-
-    'Sc'                                  => :currency_symbol,
-    'Currency_Symbol'                     => :currency_symbol,
-
-    'Sk'                                  => :modifier_symbol,
-    'Modifier_Symbol'                     => :modifier_symbol,
-
-    'So'                                  => :other_symbol,
-    'Other_Symbol'                        => :other_symbol,
-
-    'Z'                                   => :separator,
-    'Separator'                           => :separator,
-
-    'Zs'                                  => :space_separator,
-    'Space_Separator'                     => :space_separator,
-
-    'Zl'                                  => :line_separator,
-    'Line_Separator'                      => :line_separator,
-
-    'Zp'                                  => :paragraph_separator,
-    'Paragraph_Separator'                 => :paragraph_separator,
-
-    'C'                                   => :other,
-    'Other'                               => :other,
-
-    'Cc'                                  => :control,
-    'Control'                             => :control,
-
-    'Cf'                                  => :format,
-    'Format'                              => :format,
-
-    'Cs'                                  => :surrogate,
-    'Surrogate'                           => :surrogate,
-
-    'Co'                                  => :private_use,
-    'Private_Use'                         => :private_use,
-
-    'Cn'                                  => :unassigned,
-    'Unassigned'                          => :unassigned,
-
-    'Age=1.1'                             => :'age=1.1',
-    'Age=6.0'                             => :'age=6.0',
-    'Age=10.0'                            => :'age=10.0',
-
-    'ahex'                                => :ascii_hex_digit,
-    'ASCII_Hex_Digit'                     => :ascii_hex_digit,
-
-    'Alphabetic'                          => :alphabetic,
-
-    'Cased'                               => :cased,
-
-    'cwcf'                                => :changes_when_casefolded,
-    'Changes_When_Casefolded'             => :changes_when_casefolded,
-
-    'cwcm'                                => :changes_when_casemapped,
-    'Changes_When_Casemapped'             => :changes_when_casemapped,
-
-    'cwl'                                 => :changes_when_lowercased,
-    'Changes_When_Lowercased'             => :changes_when_lowercased,
-
-    'cwt'                                 => :changes_when_titlecased,
-    'Changes_When_Titlecased'             => :changes_when_titlecased,
-
-    'cwu'                                 => :changes_when_uppercased,
-    'Changes_When_Uppercased'             => :changes_when_uppercased,
-
-    'ci'                                  => :case_ignorable,
-    'Case_Ignorable'                      => :case_ignorable,
-
-    'bidic'                               => :bidi_control,
-    'Bidi_Control'                        => :bidi_control,
-
-    'Dash'                                => :dash,
-
-    'dep'                                 => :deprecated,
-    'Deprecated'                          => :deprecated,
-
-    'di'                                  => :default_ignorable_code_point,
-    'Default_Ignorable_Code_Point'        => :default_ignorable_code_point,
-
-    'dia'                                 => :diacritic,
-    'Diacritic'                           => :diacritic,
-
-    'ext'                                 => :extender,
-    'Extender'                            => :extender,
-
-    'grbase'                              => :grapheme_base,
-    'Grapheme_Base'                       => :grapheme_base,
-
-    'grext'                               => :grapheme_extend,
-    'Grapheme_Extend'                     => :grapheme_extend,
-
-    'grlink'                              => :grapheme_link,
-    'Grapheme_Link'                       => :grapheme_link,
-
-    'hex'                                 => :hex_digit,
-    'Hex_Digit'                           => :hex_digit,
-
-    'Hyphen'                              => :hyphen,
-
-    'idc'                                 => :id_continue,
-    'ID_Continue'                         => :id_continue,
-
-    'ideo'                                => :ideographic,
-    'Ideographic'                         => :ideographic,
-
-    'ids'                                 => :id_start,
-    'ID_Start'                            => :id_start,
-
-    'idsb'                                => :ids_binary_operator,
-    'IDS_Binary_Operator'                 => :ids_binary_operator,
-
-    'idst'                                => :ids_trinary_operator,
-    'IDS_Trinary_Operator'                => :ids_trinary_operator,
-
-    'joinc'                               => :join_control,
-    'Join_Control'                        => :join_control,
-
-    'loe'                                 => :logical_order_exception,
-    'Logical_Order_Exception'             => :logical_order_exception,
-
-    'Lowercase'                           => :lowercase,
-
-    'Math'                                => :math,
-
-    'nchar'                               => :noncharacter_code_point,
-    'Noncharacter_Code_Point'             => :noncharacter_code_point,
-
-    'oalpha'                              => :other_alphabetic,
-    'Other_Alphabetic'                    => :other_alphabetic,
-
-    'odi'                                 => :other_default_ignorable_code_point,
-    'Other_Default_Ignorable_Code_Point'  => :other_default_ignorable_code_point,
-
-    'ogrext'                              => :other_grapheme_extend,
-    'Other_Grapheme_Extend'               => :other_grapheme_extend,
-
-    'oidc'                                => :other_id_continue,
-    'Other_ID_Continue'                   => :other_id_continue,
-
-    'oids'                                => :other_id_start,
-    'Other_ID_Start'                      => :other_id_start,
-
-    'olower'                              => :other_lowercase,
-    'Other_Lowercase'                     => :other_lowercase,
-
-    'omath'                               => :other_math,
-    'Other_Math'                          => :other_math,
-
-    'oupper'                              => :other_uppercase,
-    'Other_Uppercase'                     => :other_uppercase,
-
-    'patsyn'                              => :pattern_syntax,
-    'Pattern_Syntax'                      => :pattern_syntax,
-
-    'patws'                               => :pattern_white_space,
-    'Pattern_Whitespace'                  => :pattern_white_space,
-
-    'qmark'                               => :quotation_mark,
-    'quotationmark'                       => :quotation_mark,
-
-    'radical'                             => :radical,
-
-    'ri'                                  => :regional_indicator,
-    'Regional_Indicator'                  => :regional_indicator,
-
-    'sd'                                  => :soft_dotted,
-    'Soft-Dotted'                         => :soft_dotted, # test dash spelling
-
-    'sterm'                               => :sentence_terminal,
-
-    'term'                                => :terminal_punctuation,
-    'Terminal_Punctuation'                => :terminal_punctuation,
-
-    'uideo'                               => :unified_ideograph,
-    'Unified_Ideograph'                   => :unified_ideograph,
-
-    'Uppercase'                           => :uppercase,
-
-    'vs'                                  => :variation_selector,
-    'Variation_Selector'                  => :variation_selector,
-
-    'wspace'                              => :white_space,
-    'whitespace'                          => :white_space,
-
-    'xids'                                => :xid_start,
-    'XID_Start'                           => :xid_start,
-
-    'xidc'                                => :xid_continue,
-    'XID_Continue'                        => :xid_continue,
-  }
-
-  tests.each_with_index do |(property, token), count|
-    define_method "test_scan_property_#{token}_#{count}" do
-      tokens = RS.scan("a\\p{#{property}}c")
-      result = tokens.at(1)
-
-      assert_equal :property, result[0]
-      assert_equal token,     result[1]
-    end
-
-    define_method "test_scan_nonproperty_#{token}_#{count}" do
-      tokens = RS.scan("a\\P{#{property}}c")
-      result = tokens.at(1)
-
-      assert_equal :nonproperty, result[0]
-      assert_equal token,        result[1]
-    end
-
-    define_method "test_scan_caret_nonproperty_#{token}_#{count}" do
-      tokens = RS.scan("a\\p{^#{property}}c")
-      result = tokens.at(1)
-
-      assert_equal :nonproperty, result[0]
-      assert_equal token,        result[1]
-    end
-
-    define_method "test_scan_double_negated_property_#{token}_#{count}" do
-      tokens = RS.scan("a\\P{^#{property}}c")
-      result = tokens.at(1)
-
-      assert_equal :property, result[0]
-      assert_equal token,     result[1]
-    end
-  end
-end

data/test/scanner/test_quantifiers.rb

@@ -1,37 +0,0 @@
-require File.expand_path("../../helpers", __FILE__)
-
-class ScannerQuantifiers < Test::Unit::TestCase
-
-  tests = {
-   'a?'     => [:quantifier,  :zero_or_one,             '?'],
-   'a??'    => [:quantifier,  :zero_or_one_reluctant,   '??'],
-   'a?+'    => [:quantifier,  :zero_or_one_possessive,  '?+'],
-
-   'a*'     => [:quantifier,  :zero_or_more,            '*'],
-   'a*?'    => [:quantifier,  :zero_or_more_reluctant,  '*?'],
-   'a*+'    => [:quantifier,  :zero_or_more_possessive, '*+'],
-
-   'a+'     => [:quantifier,  :one_or_more,             '+'],
-   'a+?'    => [:quantifier,  :one_or_more_reluctant,   '+?'],
-   'a++'    => [:quantifier,  :one_or_more_possessive,  '++'],
-
-   'a{2}'   => [:quantifier,  :interval,                '{2}'],
-   'a{2,}'  => [:quantifier,  :interval,                '{2,}'],
-   'a{,2}'  => [:quantifier,  :interval,                '{,2}'],
-   'a{2,4}' => [:quantifier,  :interval,                '{2,4}'],
-  }
-
-  tests.each_with_index do |(pattern, (type, token, text)), count|
-    name = (token == :interval ? "interval_#{count}" : token)
-
-    define_method "test_scan_#{type}_#{name}" do
-      tokens = RS.scan(pattern)
-      result = tokens.last
-
-      assert_equal type,  result[0]
-      assert_equal token, result[1]
-      assert_equal text,  result[2]
-    end
-  end
-
-end