regexp_parser 2.8.1 → 2.12.0

data/lib/regexp_parser/expression/shared.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Regexp::Expression
   module Shared
     module ClassMethods; end # filled in ./methods/*.rb
@@ -70,11 +72,12 @@ module Regexp::Expression
     # lit.to_s(:original) # => 'a +' # with quantifier AND intermittent decorations
     #
     def to_s(format = :full)
-      base = parts.each_with_object(''.dup) do |part, buff|
+      base = ''.dup
+      parts.each do |part|
         if part.instance_of?(String)
-          buff << part
+          base << part
         elsif !part.custom_to_s_handling
-          buff << part.to_s(:original)
+          base << part.to_s(:original)
         end
       end
       "#{base}#{pre_quantifier_decoration(format)}#{quantifier_affix(format)}"

data/lib/regexp_parser/expression/subexpression.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Regexp::Expression
   class Subexpression < Regexp::Expression::Base
     include Enumerable

data/lib/regexp_parser/expression.rb

@@ -1,36 +1,42 @@
-require 'regexp_parser/error'
+# frozen_string_literal: true
 
-require 'regexp_parser/expression/shared'
-require 'regexp_parser/expression/base'
-require 'regexp_parser/expression/quantifier'
-require 'regexp_parser/expression/subexpression'
-require 'regexp_parser/expression/sequence'
-require 'regexp_parser/expression/sequence_operation'
+require_relative 'error'
 
-require 'regexp_parser/expression/classes/alternation'
-require 'regexp_parser/expression/classes/anchor'
-require 'regexp_parser/expression/classes/backreference'
-require 'regexp_parser/expression/classes/character_set'
-require 'regexp_parser/expression/classes/character_set/intersection'
-require 'regexp_parser/expression/classes/character_set/range'
-require 'regexp_parser/expression/classes/character_type'
-require 'regexp_parser/expression/classes/conditional'
-require 'regexp_parser/expression/classes/escape_sequence'
-require 'regexp_parser/expression/classes/free_space'
-require 'regexp_parser/expression/classes/group'
-require 'regexp_parser/expression/classes/keep'
-require 'regexp_parser/expression/classes/literal'
-require 'regexp_parser/expression/classes/posix_class'
-require 'regexp_parser/expression/classes/root'
-require 'regexp_parser/expression/classes/unicode_property'
+require_relative 'expression/shared'
+require_relative 'expression/base'
+require_relative 'expression/quantifier'
+require_relative 'expression/subexpression'
+require_relative 'expression/sequence'
+require_relative 'expression/sequence_operation'
 
-require 'regexp_parser/expression/methods/construct'
-require 'regexp_parser/expression/methods/human_name'
-require 'regexp_parser/expression/methods/match'
-require 'regexp_parser/expression/methods/match_length'
-require 'regexp_parser/expression/methods/options'
-require 'regexp_parser/expression/methods/parts'
-require 'regexp_parser/expression/methods/printing'
-require 'regexp_parser/expression/methods/strfregexp'
-require 'regexp_parser/expression/methods/tests'
-require 'regexp_parser/expression/methods/traverse'
+require_relative 'expression/classes/alternation'
+require_relative 'expression/classes/anchor'
+require_relative 'expression/classes/backreference'
+require_relative 'expression/classes/character_set'
+require_relative 'expression/classes/character_set/intersection'
+require_relative 'expression/classes/character_set/range'
+require_relative 'expression/classes/character_type'
+require_relative 'expression/classes/conditional'
+require_relative 'expression/classes/escape_sequence'
+require_relative 'expression/classes/free_space'
+require_relative 'expression/classes/group'
+require_relative 'expression/classes/keep'
+require_relative 'expression/classes/literal'
+require_relative 'expression/classes/posix_class'
+require_relative 'expression/classes/root'
+require_relative 'expression/classes/unicode_property'
+
+require_relative 'expression/methods/construct'
+require_relative 'expression/methods/escape_sequence_char'
+require_relative 'expression/methods/escape_sequence_codepoint'
+require_relative 'expression/methods/human_name'
+require_relative 'expression/methods/match'
+require_relative 'expression/methods/match_length'
+require_relative 'expression/methods/negative'
+require_relative 'expression/methods/options'
+require_relative 'expression/methods/parts'
+require_relative 'expression/methods/printing'
+require_relative 'expression/methods/referenced_expressions'
+require_relative 'expression/methods/strfregexp'
+require_relative 'expression/methods/tests'
+require_relative 'expression/methods/traverse'

data/lib/regexp_parser/lexer.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 # A very thin wrapper around the scanner that breaks quantified literal runs,
 # collects emitted tokens into an array, calculates their nesting depth, and
 # normalizes tokens for the parser, and checks if they are implemented by the

data/lib/regexp_parser/parser.rb

@@ -1,5 +1,7 @@
-require 'regexp_parser/error'
-require 'regexp_parser/expression'
+# frozen_string_literal: true
+
+require_relative 'error'
+require_relative 'expression'
 
 class Regexp::Parser
   include Regexp::Expression
@@ -319,6 +321,7 @@ class Regexp::Parser
     when :codepoint_list; node << EscapeSequence::CodepointList.new(token, active_opts)
     when :hex;            node << EscapeSequence::Hex.new(token, active_opts)
     when :octal;          node << EscapeSequence::Octal.new(token, active_opts)
+    when :utf8_hex;       node << EscapeSequence::UTF8Hex.new(token, active_opts)
 
     when :control
       if token.text =~ /\A(?:\\C-\\M|\\c\\M)/
@@ -467,6 +470,7 @@ class Regexp::Parser
     when *UPTokens::Age;          node << UP::Age.new(token, active_opts)
     when *UPTokens::Derived;      node << UP::Derived.new(token, active_opts)
     when *UPTokens::Emoji;        node << UP::Emoji.new(token, active_opts)
+    when *UPTokens::Enumerated;   node << UP::Enumerated.new(token, active_opts)
     when *UPTokens::Script;       node << UP::Script.new(token, active_opts)
     when *UPTokens::UnicodeBlock; node << UP::Block.new(token, active_opts)
 
@@ -574,21 +578,24 @@ class Regexp::Parser
     options_stack.last
   end
 
-  # Assigns referenced expressions to refering expressions, e.g. if there is
+  # Assigns referenced expressions to referring expressions, e.g. if there is
   # an instance of Backreference::Number, its #referenced_expression is set to
   # the instance of Group::Capture that it refers to via its number.
   def assign_referenced_expressions
-    # find all referencable and refering expressions
-    targets = { 0 => root }
+    # find all referenceable and referring expressions
+    targets = { 0 => [root] }
     referrers = []
     root.each_expression do |exp|
-      exp.is_a?(Group::Capture) && targets[exp.identifier] = exp
-      referrers << exp if exp.referential?
+      if exp.referential?
+        referrers << exp
+      elsif exp.is_a?(Group::Capture)
+        (targets[exp.identifier] ||= []) << exp
+      end
     end
-    # assign reference expression to refering expressions
+    # assign referenced expressions to referring expressions
     # (in a second iteration because there might be forward references)
     referrers.each do |exp|
-      exp.referenced_expression = targets[exp.reference] ||
+      exp.referenced_expressions = targets[exp.reference] ||
         raise(ParserError, "Invalid reference #{exp.reference} at pos #{exp.ts}")
     end
   end

data/lib/regexp_parser/scanner/errors/premature_end_error.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 class Regexp::Scanner
   # Unexpected end of pattern
   class PrematureEndError < ScannerError

data/lib/regexp_parser/scanner/errors/scanner_error.rb

@@ -1,4 +1,6 @@
-require 'regexp_parser/error'
+# frozen_string_literal: true
+
+require_relative '../../../regexp_parser/error'
 
 class Regexp::Scanner
   # General scanner error (catch all)

data/lib/regexp_parser/scanner/errors/validation_error.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 class Regexp::Scanner
   # Base for all scanner validation errors
   class ValidationError < ScannerError

data/lib/regexp_parser/scanner/properties/long.csv

@@ -8,6 +8,9 @@ age=12.1,age=12.1
 age=13.0,age=13.0
 age=14.0,age=14.0
 age=15.0,age=15.0
+age=15.1,age=15.1
+age=16.0,age=16.0
+age=17.0,age=17.0
 age=2.0,age=2.0
 age=2.1,age=2.1
 age=3.0,age=3.0
@@ -42,6 +45,7 @@ bamum,bamum
 bassavah,bassa_vah
 batak,batak
 bengali,bengali
+beriaerfe,beria_erfe
 bhaiksuki,bhaiksuki
 bidicontrol,bidi_control
 blank,blank
@@ -102,18 +106,33 @@ extendedpictographic,extended_pictographic
 extender,extender
 finalpunctuation,final_punctuation
 format,format
+garay,garay
 georgian,georgian
 glagolitic,glagolitic
 gothic,gothic
 grantha,grantha
 graph,graph
 graphemebase,grapheme_base
+graphemeclusterbreak=control,grapheme_cluster_break=control
+graphemeclusterbreak=cr,grapheme_cluster_break=cr
+graphemeclusterbreak=extend,grapheme_cluster_break=extend
+graphemeclusterbreak=l,grapheme_cluster_break=l
+graphemeclusterbreak=lf,grapheme_cluster_break=lf
+graphemeclusterbreak=lv,grapheme_cluster_break=lv
+graphemeclusterbreak=lvt,grapheme_cluster_break=lvt
+graphemeclusterbreak=prepend,grapheme_cluster_break=prepend
+graphemeclusterbreak=regionalindicator,grapheme_cluster_break=regional_indicator
+graphemeclusterbreak=spacingmark,grapheme_cluster_break=spacingmark
+graphemeclusterbreak=t,grapheme_cluster_break=t
+graphemeclusterbreak=v,grapheme_cluster_break=v
+graphemeclusterbreak=zwj,grapheme_cluster_break=zwj
 graphemeextend,grapheme_extend
 graphemelink,grapheme_link
 greek,greek
 gujarati,gujarati
 gunjalagondi,gunjala_gondi
 gurmukhi,gurmukhi
+gurungkhema,gurung_khema
 han,han
 hangul,hangul
 hanifirohingya,hanifi_rohingya
@@ -123,11 +142,14 @@ hebrew,hebrew
 hexdigit,hex_digit
 hiragana,hiragana
 hyphen,hyphen
+idcompatmathcontinue,id_compat_math_continue
+idcompatmathstart,id_compat_math_start
 idcontinue,id_continue
 ideographic,ideographic
 idsbinaryoperator,ids_binary_operator
 idstart,id_start
 idstrinaryoperator,ids_trinary_operator
+idsunaryoperator,ids_unary_operator
 imperialaramaic,imperial_aramaic
 inadlam,in_adlam
 inaegeannumbers,in_aegean_numbers
@@ -156,6 +178,7 @@ inbasiclatin,in_basic_latin
 inbassavah,in_bassa_vah
 inbatak,in_batak
 inbengali,in_bengali
+inberiaerfe,in_beria_erfe
 inbhaiksuki,in_bhaiksuki
 inblockelements,in_block_elements
 inbopomofo,in_bopomofo
@@ -190,6 +213,8 @@ incjkunifiedideographsextensione,in_cjk_unified_ideographs_extension_e
 incjkunifiedideographsextensionf,in_cjk_unified_ideographs_extension_f
 incjkunifiedideographsextensiong,in_cjk_unified_ideographs_extension_g
 incjkunifiedideographsextensionh,in_cjk_unified_ideographs_extension_h
+incjkunifiedideographsextensioni,in_cjk_unified_ideographs_extension_i
+incjkunifiedideographsextensionj,in_cjk_unified_ideographs_extension_j
 incombiningdiacriticalmarks,in_combining_diacritical_marks
 incombiningdiacriticalmarksextended,in_combining_diacritical_marks_extended
 incombiningdiacriticalmarksforsymbols,in_combining_diacritical_marks_for_symbols
@@ -223,6 +248,7 @@ induployan,in_duployan
 inearlydynasticcuneiform,in_early_dynastic_cuneiform
 inegyptianhieroglyphformatcontrols,in_egyptian_hieroglyph_format_controls
 inegyptianhieroglyphs,in_egyptian_hieroglyphs
+inegyptianhieroglyphsextendeda,in_egyptian_hieroglyphs_extended_a
 inelbasan,in_elbasan
 inelymaic,in_elymaic
 inemoticons,in_emoticons
@@ -235,6 +261,7 @@ inethiopicextended,in_ethiopic_extended
 inethiopicextendeda,in_ethiopic_extended_a
 inethiopicextendedb,in_ethiopic_extended_b
 inethiopicsupplement,in_ethiopic_supplement
+ingaray,in_garay
 ingeneralpunctuation,in_general_punctuation
 ingeometricshapes,in_geometric_shapes
 ingeometricshapesextended,in_geometric_shapes_extended
@@ -250,6 +277,7 @@ ingreekextended,in_greek_extended
 ingujarati,in_gujarati
 ingunjalagondi,in_gunjala_gondi
 ingurmukhi,in_gurmukhi
+ingurungkhema,in_gurung_khema
 inhalfwidthandfullwidthforms,in_halfwidth_and_fullwidth_forms
 inhangulcompatibilityjamo,in_hangul_compatibility_jamo
 inhanguljamo,in_hangul_jamo
@@ -291,6 +319,7 @@ inkhmer,in_khmer
 inkhmersymbols,in_khmer_symbols
 inkhojki,in_khojki
 inkhudawadi,in_khudawadi
+inkiratrai,in_kirat_rai
 inlao,in_lao
 inlatin1supplement,in_latin_1_supplement
 inlatinextendeda,in_latin_extended_a
@@ -335,6 +364,7 @@ inmiscellaneousmathematicalsymbolsb,in_miscellaneous_mathematical_symbols_b
 inmiscellaneoussymbols,in_miscellaneous_symbols
 inmiscellaneoussymbolsandarrows,in_miscellaneous_symbols_and_arrows
 inmiscellaneoussymbolsandpictographs,in_miscellaneous_symbols_and_pictographs
+inmiscellaneoussymbolssupplement,in_miscellaneous_symbols_supplement
 inmiscellaneoustechnical,in_miscellaneous_technical
 inmodi,in_modi
 inmodifiertoneletters,in_modifier_tone_letters
@@ -346,6 +376,7 @@ inmusicalsymbols,in_musical_symbols
 inmyanmar,in_myanmar
 inmyanmarextendeda,in_myanmar_extended_a
 inmyanmarextendedb,in_myanmar_extended_b
+inmyanmarextendedc,in_myanmar_extended_c
 innabataean,in_nabataean
 innagmundari,in_nag_mundari
 innandinagari,in_nandinagari
@@ -367,6 +398,7 @@ inoldsogdian,in_old_sogdian
 inoldsoutharabian,in_old_south_arabian
 inoldturkic,in_old_turkic
 inolduyghur,in_old_uyghur
+inolonal,in_ol_onal
 inopticalcharacterrecognition,in_optical_character_recognition
 inoriya,in_oriya
 inornamentaldingbats,in_ornamental_dingbats
@@ -392,9 +424,11 @@ insaurashtra,in_saurashtra
 inscriptionalpahlavi,inscriptional_pahlavi
 inscriptionalparthian,inscriptional_parthian
 insharada,in_sharada
+insharadasupplement,in_sharada_supplement
 inshavian,in_shavian
 inshorthandformatcontrols,in_shorthand_format_controls
 insiddham,in_siddham
+insidetic,in_sidetic
 insinhala,in_sinhala
 insinhalaarchaicnumbers,in_sinhala_archaic_numbers
 insmallformvariants,in_small_form_variants
@@ -406,6 +440,7 @@ inspacingmodifierletters,in_spacing_modifier_letters
 inspecials,in_specials
 insundanese,in_sundanese
 insundanesesupplement,in_sundanese_supplement
+insunuwar,in_sunuwar
 insuperscriptsandsubscripts,in_superscripts_and_subscripts
 insupplementalarrowsa,in_supplemental_arrows_a
 insupplementalarrowsb,in_supplemental_arrows_b
@@ -419,6 +454,7 @@ insuttonsignwriting,in_sutton_signwriting
 insylotinagri,in_syloti_nagri
 insymbolsandpictographsextendeda,in_symbols_and_pictographs_extended_a
 insymbolsforlegacycomputing,in_symbols_for_legacy_computing
+insymbolsforlegacycomputingsupplement,in_symbols_for_legacy_computing_supplement
 insyriac,in_syriac
 insyriacsupplement,in_syriac_supplement
 intagalog,in_tagalog
@@ -428,12 +464,14 @@ intaile,in_tai_le
 intaitham,in_tai_tham
 intaiviet,in_tai_viet
 intaixuanjingsymbols,in_tai_xuan_jing_symbols
+intaiyo,in_tai_yo
 intakri,in_takri
 intamil,in_tamil
 intamilsupplement,in_tamil_supplement
 intangsa,in_tangsa
 intangut,in_tangut
 intangutcomponents,in_tangut_components
+intangutcomponentssupplement,in_tangut_components_supplement
 intangutsupplement,in_tangut_supplement
 intelugu,in_telugu
 inthaana,in_thaana
@@ -441,8 +479,11 @@ inthai,in_thai
 intibetan,in_tibetan
 intifinagh,in_tifinagh
 intirhuta,in_tirhuta
+intodhri,in_todhri
+intolongsiki,in_tolong_siki
 intoto,in_toto
 intransportandmapsymbols,in_transport_and_map_symbols
+intulutigalari,in_tulu_tigalari
 inugaritic,in_ugaritic
 inunifiedcanadianaboriginalsyllabics,in_unified_canadian_aboriginal_syllabics
 inunifiedcanadianaboriginalsyllabicsextended,in_unified_canadian_aboriginal_syllabics_extended
@@ -473,6 +514,7 @@ khitansmallscript,khitan_small_script
 khmer,khmer
 khojki,khojki
 khudawadi,khudawadi
+kiratrai,kirat_rai
 lao,lao
 latin,latin
 lepcha,lepcha
@@ -506,6 +548,7 @@ meroiticcursive,meroitic_cursive
 meroitichieroglyphs,meroitic_hieroglyphs
 miao,miao
 modi,modi
+modifiercombiningmark,modifier_combining_mark
 modifierletter,modifier_letter
 modifiersymbol,modifier_symbol
 mongolian,mongolian
@@ -535,6 +578,7 @@ oldsogdian,old_sogdian
 oldsoutharabian,old_south_arabian
 oldturkic,old_turkic
 olduyghur,old_uyghur
+olonal,ol_onal
 openpunctuation,open_punctuation
 oriya,oriya
 osage,osage
@@ -578,6 +622,7 @@ separator,separator
 sharada,sharada
 shavian,shavian
 siddham,siddham
+sidetic,sidetic
 signwriting,signwriting
 sinhala,sinhala
 softdotted,soft_dotted
@@ -588,6 +633,7 @@ space,space
 spaceseparator,space_separator
 spacingmark,spacing_mark
 sundanese,sundanese
+sunuwar,sunuwar
 surrogate,surrogate
 sylotinagri,syloti_nagri
 symbol,symbol
@@ -597,6 +643,7 @@ tagbanwa,tagbanwa
 taile,tai_le
 taitham,tai_tham
 taiviet,tai_viet
+taiyo,tai_yo
 takri,takri
 tamil,tamil
 tangsa,tangsa
@@ -609,7 +656,10 @@ tibetan,tibetan
 tifinagh,tifinagh
 tirhuta,tirhuta
 titlecaseletter,titlecase_letter
+todhri,todhri
+tolongsiki,tolong_siki
 toto,toto
+tulutigalari,tulu_tigalari
 ugaritic,ugaritic
 unassigned,unassigned
 unifiedideograph,unified_ideograph

data/lib/regexp_parser/scanner/properties/short.csv

@@ -11,6 +11,7 @@ bamu,bamum
 bass,bassa_vah
 batk,batak
 beng,bengali
+berf,beria_erfe
 bhks,bhaiksuki
 bidic,bidi_control
 bopo,bopomofo
@@ -58,6 +59,7 @@ epres,emoji_presentation
 ethi,ethiopic
 ext,extender
 extpict,extended_pictographic
+gara,garay
 geor,georgian
 glag,glagolitic
 gong,gunjala_gondi
@@ -69,6 +71,7 @@ grek,greek
 grext,grapheme_extend
 grlink,grapheme_link
 gujr,gujarati
+gukh,gurung_khema
 guru,gurmukhi
 hang,hangul
 hani,han
@@ -86,6 +89,7 @@ ideo,ideographic
 ids,id_start
 idsb,ids_binary_operator
 idst,ids_trinary_operator
+idsu,ids_unary_operator
 ital,old_italic
 java,javanese
 joinc,join_control
@@ -96,6 +100,7 @@ khmr,khmer
 khoj,khojki
 kits,khitan_small_script
 knda,kannada
+krai,kirat_rai
 kthi,kaithi
 l,letter
 lana,tai_tham
@@ -121,6 +126,7 @@ mand,mandaic
 mani,manichaean
 marc,marchen
 mc,spacing_mark
+mcm,modifier_combining_mark
 me,enclosing_mark
 medf,medefaidrin
 mend,mende_kikakui
@@ -153,6 +159,7 @@ oids,other_id_start
 olck,ol_chiki
 olower,other_lowercase
 omath,other_math
+onao,ol_onal
 orkh,old_turkic
 orya,oriya
 osge,osage
@@ -196,6 +203,7 @@ sgnw,signwriting
 shaw,shavian
 shrd,sharada
 sidd,siddham
+sidt,sidetic
 sind,khudawadi
 sinh,sinhala
 sk,modifier_symbol
@@ -207,6 +215,7 @@ sora,sora_sompeng
 soyo,soyombo
 sterm,sentence_terminal
 sund,sundanese
+sunu,sunuwar
 sylo,syloti_nagri
 syrc,syriac
 tagb,tagbanwa
@@ -216,6 +225,7 @@ talu,new_tai_lue
 taml,tamil
 tang,tangut
 tavt,tai_viet
+tayo,tai_yo
 telu,telugu
 term,terminal_punctuation
 tfng,tifinagh
@@ -224,6 +234,9 @@ thaa,thaana
 tibt,tibetan
 tirh,tirhuta
 tnsa,tangsa
+todr,todhri
+tols,tolong_siki
+tutg,tulu_tigalari
 ugar,ugaritic
 uideo,unified_ideograph
 vaii,vai