RubyGems - regexp_parser - Versions diffs - 2.1.1 → 2.2.1 - Mend

regexp_parser 2.1.1 → 2.2.1

Files changed (129) hide show

checksums.yaml +4 -4
data/CHANGELOG.md +19 -1
data/LICENSE +1 -1
data/README.md +17 -23
data/Rakefile +10 -19
data/lib/regexp_parser/expression/base.rb +123 -0
data/lib/regexp_parser/expression/classes/anchor.rb +0 -2
data/lib/regexp_parser/expression/classes/{backref.rb → backreference.rb} +0 -0
data/lib/regexp_parser/expression/classes/{set → character_set}/intersection.rb +0 -0
data/lib/regexp_parser/expression/classes/{set → character_set}/range.rb +0 -0
data/lib/regexp_parser/expression/classes/{set.rb → character_set.rb} +0 -0
data/lib/regexp_parser/expression/classes/{escape.rb → escape_sequence.rb} +13 -7
data/lib/regexp_parser/expression/classes/free_space.rb +0 -2
data/lib/regexp_parser/expression/classes/literal.rb +1 -5
data/lib/regexp_parser/expression/classes/property.rb +0 -2
data/lib/regexp_parser/expression/classes/root.rb +0 -1
data/lib/regexp_parser/expression/classes/type.rb +0 -2
data/lib/regexp_parser/expression/methods/strfregexp.rb +1 -1
data/lib/regexp_parser/expression/quantifier.rb +1 -1
data/lib/regexp_parser/expression/sequence.rb +0 -1
data/lib/regexp_parser/expression/subexpression.rb +0 -1
data/lib/regexp_parser/expression.rb +6 -130
data/lib/regexp_parser/lexer.rb +7 -5
data/lib/regexp_parser/scanner/properties/long.csv +604 -0
data/lib/regexp_parser/scanner/properties/short.csv +242 -0
data/lib/regexp_parser/scanner/scanner.rl +6 -4
data/lib/regexp_parser/scanner.rb +126 -124
data/lib/regexp_parser/syntax/any.rb +1 -3
data/lib/regexp_parser/syntax/base.rb +12 -14
data/lib/regexp_parser/syntax/token/anchor.rb +15 -0
data/lib/regexp_parser/syntax/{tokens → token}/assertion.rb +2 -2
data/lib/regexp_parser/syntax/token/backreference.rb +30 -0
data/lib/regexp_parser/syntax/{tokens → token}/character_set.rb +2 -2
data/lib/regexp_parser/syntax/{tokens → token}/character_type.rb +3 -3
data/lib/regexp_parser/syntax/{tokens → token}/conditional.rb +3 -3
data/lib/regexp_parser/syntax/token/escape.rb +31 -0
data/lib/regexp_parser/syntax/{tokens → token}/group.rb +7 -7
data/lib/regexp_parser/syntax/{tokens → token}/keep.rb +1 -1
data/lib/regexp_parser/syntax/{tokens → token}/meta.rb +2 -2
data/lib/regexp_parser/syntax/{tokens → token}/posix_class.rb +3 -3
data/lib/regexp_parser/syntax/token/quantifier.rb +35 -0
data/lib/regexp_parser/syntax/token/unicode_property.rb +696 -0
data/lib/regexp_parser/syntax/token.rb +45 -0
data/lib/regexp_parser/syntax/versions/1.8.6.rb +2 -2
data/lib/regexp_parser/syntax/versions/1.9.1.rb +1 -1
data/lib/regexp_parser/syntax/versions/3.1.0.rb +10 -0
data/lib/regexp_parser/syntax.rb +1 -1
data/lib/regexp_parser/token.rb +9 -20
data/lib/regexp_parser/version.rb +1 -1
data/lib/regexp_parser.rb +0 -2
data/regexp_parser.gemspec +20 -22
metadata +32 -164
data/lib/regexp_parser/scanner/properties/long.yml +0 -594
data/lib/regexp_parser/scanner/properties/short.yml +0 -237
data/lib/regexp_parser/syntax/tokens/anchor.rb +0 -15
data/lib/regexp_parser/syntax/tokens/backref.rb +0 -24
data/lib/regexp_parser/syntax/tokens/escape.rb +0 -30
data/lib/regexp_parser/syntax/tokens/quantifier.rb +0 -35
data/lib/regexp_parser/syntax/tokens/unicode_property.rb +0 -675
data/lib/regexp_parser/syntax/tokens.rb +0 -45
data/spec/expression/base_spec.rb +0 -104
data/spec/expression/clone_spec.rb +0 -152
data/spec/expression/conditional_spec.rb +0 -89
data/spec/expression/free_space_spec.rb +0 -27
data/spec/expression/methods/match_length_spec.rb +0 -161
data/spec/expression/methods/match_spec.rb +0 -25
data/spec/expression/methods/strfregexp_spec.rb +0 -224
data/spec/expression/methods/tests_spec.rb +0 -99
data/spec/expression/methods/traverse_spec.rb +0 -161
data/spec/expression/options_spec.rb +0 -128
data/spec/expression/subexpression_spec.rb +0 -50
data/spec/expression/to_h_spec.rb +0 -26
data/spec/expression/to_s_spec.rb +0 -108
data/spec/lexer/all_spec.rb +0 -22
data/spec/lexer/conditionals_spec.rb +0 -53
data/spec/lexer/delimiters_spec.rb +0 -68
data/spec/lexer/escapes_spec.rb +0 -14
data/spec/lexer/keep_spec.rb +0 -10
data/spec/lexer/literals_spec.rb +0 -64
data/spec/lexer/nesting_spec.rb +0 -99
data/spec/lexer/refcalls_spec.rb +0 -60
data/spec/parser/all_spec.rb +0 -43
data/spec/parser/alternation_spec.rb +0 -88
data/spec/parser/anchors_spec.rb +0 -17
data/spec/parser/conditionals_spec.rb +0 -179
data/spec/parser/errors_spec.rb +0 -30
data/spec/parser/escapes_spec.rb +0 -121
data/spec/parser/free_space_spec.rb +0 -130
data/spec/parser/groups_spec.rb +0 -108
data/spec/parser/keep_spec.rb +0 -6
data/spec/parser/options_spec.rb +0 -28
data/spec/parser/posix_classes_spec.rb +0 -8
data/spec/parser/properties_spec.rb +0 -115
data/spec/parser/quantifiers_spec.rb +0 -68
data/spec/parser/refcalls_spec.rb +0 -117
data/spec/parser/set/intersections_spec.rb +0 -127
data/spec/parser/set/ranges_spec.rb +0 -111
data/spec/parser/sets_spec.rb +0 -178
data/spec/parser/types_spec.rb +0 -18
data/spec/scanner/all_spec.rb +0 -18
data/spec/scanner/anchors_spec.rb +0 -21
data/spec/scanner/conditionals_spec.rb +0 -128
data/spec/scanner/delimiters_spec.rb +0 -52
data/spec/scanner/errors_spec.rb +0 -67
data/spec/scanner/escapes_spec.rb +0 -64
data/spec/scanner/free_space_spec.rb +0 -165
data/spec/scanner/groups_spec.rb +0 -61
data/spec/scanner/keep_spec.rb +0 -10
data/spec/scanner/literals_spec.rb +0 -39
data/spec/scanner/meta_spec.rb +0 -18
data/spec/scanner/options_spec.rb +0 -36
data/spec/scanner/properties_spec.rb +0 -64
data/spec/scanner/quantifiers_spec.rb +0 -25
data/spec/scanner/refcalls_spec.rb +0 -55
data/spec/scanner/sets_spec.rb +0 -151
data/spec/scanner/types_spec.rb +0 -14
data/spec/spec_helper.rb +0 -16
data/spec/support/runner.rb +0 -42
data/spec/support/shared_examples.rb +0 -77
data/spec/support/warning_extractor.rb +0 -60
data/spec/syntax/syntax_spec.rb +0 -48
data/spec/syntax/syntax_token_map_spec.rb +0 -23
data/spec/syntax/versions/1.8.6_spec.rb +0 -17
data/spec/syntax/versions/1.9.1_spec.rb +0 -10
data/spec/syntax/versions/1.9.3_spec.rb +0 -9
data/spec/syntax/versions/2.0.0_spec.rb +0 -13
data/spec/syntax/versions/2.2.0_spec.rb +0 -9
data/spec/syntax/versions/aliases_spec.rb +0 -37
data/spec/token/token_spec.rb +0 -85

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

@@ -0,0 +1,242 @@
+# THIS FILE IS AUTO-GENERATED BY `rake props:update` - DO NOT EDIT
+adlm,adlam
+aghb,caucasian_albanian
+ahex,ascii_hex_digit
+arab,arabic
+armi,imperial_aramaic
+armn,armenian
+avst,avestan
+bali,balinese
+bamu,bamum
+bass,bassa_vah
+batk,batak
+beng,bengali
+bhks,bhaiksuki
+bidic,bidi_control
+bopo,bopomofo
+brah,brahmi
+brai,braille
+bugi,buginese
+buhd,buhid
+c,other
+cakm,chakma
+cans,canadian_aboriginal
+cari,carian
+cc,control
+cf,format
+cher,cherokee
+chrs,chorasmian
+ci,case_ignorable
+cn,unassigned
+co,private_use
+combiningmark,mark
+copt,coptic
+cprt,cypriot
+cs,surrogate
+cwcf,changes_when_casefolded
+cwcm,changes_when_casemapped
+cwl,changes_when_lowercased
+cwt,changes_when_titlecased
+cwu,changes_when_uppercased
+cyrl,cyrillic
+dep,deprecated
+deva,devanagari
+di,default_ignorable_code_point
+dia,diacritic
+diak,dives_akuru
+dogr,dogra
+dsrt,deseret
+dupl,duployan
+ebase,emoji_modifier_base
+ecomp,emoji_component
+egyp,egyptian_hieroglyphs
+elba,elbasan
+elym,elymaic
+emod,emoji_modifier
+epres,emoji_presentation
+ethi,ethiopic
+ext,extender
+geor,georgian
+glag,glagolitic
+gong,gunjala_gondi
+gonm,masaram_gondi
+goth,gothic
+gran,grantha
+grbase,grapheme_base
+grek,greek
+grext,grapheme_extend
+grlink,grapheme_link
+gujr,gujarati
+guru,gurmukhi
+hang,hangul
+hani,han
+hano,hanunoo
+hatr,hatran
+hebr,hebrew
+hex,hex_digit
+hira,hiragana
+hluw,anatolian_hieroglyphs
+hmng,pahawh_hmong
+hmnp,nyiakeng_puachue_hmong
+hung,old_hungarian
+idc,id_continue
+ideo,ideographic
+ids,id_start
+idsb,ids_binary_operator
+idst,ids_trinary_operator
+ital,old_italic
+java,javanese
+joinc,join_control
+kali,kayah_li
+kana,katakana
+khar,kharoshthi
+khmr,khmer
+khoj,khojki
+kits,khitan_small_script
+knda,kannada
+kthi,kaithi
+l,letter
+lana,tai_tham
+laoo,lao
+latn,latin
+lc,cased_letter
+lepc,lepcha
+limb,limbu
+lina,linear_a
+linb,linear_b
+ll,lowercase_letter
+lm,modifier_letter
+lo,other_letter
+loe,logical_order_exception
+lt,titlecase_letter
+lu,uppercase_letter
+lyci,lycian
+lydi,lydian
+m,mark
+mahj,mahajani
+maka,makasar
+mand,mandaic
+mani,manichaean
+marc,marchen
+mc,spacing_mark
+me,enclosing_mark
+medf,medefaidrin
+mend,mende_kikakui
+merc,meroitic_cursive
+mero,meroitic_hieroglyphs
+mlym,malayalam
+mn,nonspacing_mark
+mong,mongolian
+mroo,mro
+mtei,meetei_mayek
+mult,multani
+mymr,myanmar
+n,number
+nand,nandinagari
+narb,old_north_arabian
+nbat,nabataean
+nchar,noncharacter_code_point
+nd,decimal_number
+nkoo,nko
+nl,letter_number
+no,other_number
+nshu,nushu
+oalpha,other_alphabetic
+odi,other_default_ignorable_code_point
+ogam,ogham
+ogrext,other_grapheme_extend
+oidc,other_id_continue
+oids,other_id_start
+olck,ol_chiki
+olower,other_lowercase
+omath,other_math
+orkh,old_turkic
+orya,oriya
+osge,osage
+osma,osmanya
+oupper,other_uppercase
+p,punctuation
+palm,palmyrene
+patsyn,pattern_syntax
+patws,pattern_white_space
+pauc,pau_cin_hau
+pc,connector_punctuation
+pcm,prepended_concatenation_mark
+pd,dash_punctuation
+pe,close_punctuation
+perm,old_permic
+pf,final_punctuation
+phag,phags_pa
+phli,inscriptional_pahlavi
+phlp,psalter_pahlavi
+phnx,phoenician
+pi,initial_punctuation
+plrd,miao
+po,other_punctuation
+prti,inscriptional_parthian
+ps,open_punctuation
+qaac,coptic
+qaai,inherited
+qmark,quotation_mark
+ri,regional_indicator
+rjng,rejang
+rohg,hanifi_rohingya
+runr,runic
+s,symbol
+samr,samaritan
+sarb,old_south_arabian
+saur,saurashtra
+sc,currency_symbol
+sd,soft_dotted
+sgnw,signwriting
+shaw,shavian
+shrd,sharada
+sidd,siddham
+sind,khudawadi
+sinh,sinhala
+sk,modifier_symbol
+sm,math_symbol
+so,other_symbol
+sogd,sogdian
+sogo,old_sogdian
+sora,sora_sompeng
+soyo,soyombo
+sterm,sentence_terminal
+sund,sundanese
+sylo,syloti_nagri
+syrc,syriac
+tagb,tagbanwa
+takr,takri
+tale,tai_le
+talu,new_tai_lue
+taml,tamil
+tang,tangut
+tavt,tai_viet
+telu,telugu
+term,terminal_punctuation
+tfng,tifinagh
+tglg,tagalog
+thaa,thaana
+tibt,tibetan
+tirh,tirhuta
+ugar,ugaritic
+uideo,unified_ideograph
+vaii,vai
+vs,variation_selector
+wara,warang_citi
+wcho,wancho
+wspace,white_space
+xidc,xid_continue
+xids,xid_start
+xpeo,old_persian
+xsux,cuneiform
+yezi,yezidi
+yiii,yi
+z,separator
+zanb,zanabazar_square
+zinh,inherited
+zl,line_separator
+zp,paragraph_separator
+zs,space_separator
+zyyy,common
+zzzz,unknown

data/lib/regexp_parser/scanner/scanner.rl CHANGED Viewed

@@ -759,14 +759,16 @@ class Regexp::Scanner
   end
   # lazy-load property maps when first needed
-  require 'yaml'
   def self.short_prop_map
-    @short_prop_map ||= YAML.load_file("#{__dir__}/scanner/properties/short.yml")
+    @short_prop_map ||= parse_prop_map('short')
   end
   def self.long_prop_map
-    @long_prop_map ||= YAML.load_file("#{__dir__}/scanner/properties/long.yml")
+    @long_prop_map ||= parse_prop_map('long')
+  end
+  def self.parse_prop_map(name)
+    File.read("#{__dir__}/scanner/properties/#{name}.csv").scan(/(.+),(.+)/).to_h
   end
   # Emits an array with the details of the scanned pattern