vendor/regex/tests/unicode.rs

   1 mat!(uni_literal, r"☃", "☃", Some((0, 3)));
   2 mat!(uni_literal_plus, r"☃+", "☃", Some((0, 3)));
   3 mat!(uni_literal_casei_plus, r"(?i)☃+", "☃", Some((0, 3)));
   4 mat!(uni_class_plus, r"[☃Ⅰ]+", "☃", Some((0, 3)));
   5 mat!(uni_one, r"\pN", "Ⅰ", Some((0, 3)));
   6 mat!(uni_mixed, r"\pN+", "Ⅰ1Ⅱ2", Some((0, 8)));
   7 mat!(uni_not, r"\PN+", "abⅠ", Some((0, 2)));
   8 mat!(uni_not_class, r"[\PN]+", "abⅠ", Some((0, 2)));
   9 mat!(uni_not_class_neg, r"[^\PN]+", "abⅠ", Some((2, 5)));
  10 mat!(uni_case, r"(?i)Δ", "δ", Some((0, 2)));
  11 mat!(uni_case_upper, r"\p{Lu}+", "ΛΘΓΔα", Some((0, 8)));
  12 mat!(uni_case_upper_nocase_flag, r"(?i)\p{Lu}+", "ΛΘΓΔα", Some((0, 10)));
  13 mat!(uni_case_upper_nocase, r"\p{L}+", "ΛΘΓΔα", Some((0, 10)));
  14 mat!(uni_case_lower, r"\p{Ll}+", "ΛΘΓΔα", Some((8, 10)));
  15
  16 // Test the Unicode friendliness of Perl character classes.
  17 mat!(uni_perl_w, r"\w+", "dδd", Some((0, 4)));
  18 mat!(uni_perl_w_not, r"\w+", "⥡", None);
  19 mat!(uni_perl_w_neg, r"\W+", "⥡", Some((0, 3)));
  20 mat!(uni_perl_d, r"\d+", "1२३9", Some((0, 8)));
  21 mat!(uni_perl_d_not, r"\d+", "Ⅱ", None);
  22 mat!(uni_perl_d_neg, r"\D+", "Ⅱ", Some((0, 3)));
  23 mat!(uni_perl_s, r"\s+", " ", Some((0, 3)));
  24 mat!(uni_perl_s_not, r"\s+", "☃", None);
  25 mat!(uni_perl_s_neg, r"\S+", "☃", Some((0, 3)));
  26
  27 // And do the same for word boundaries.
  28 mat!(uni_boundary_none, r"\d\b", "6δ", None);
  29 mat!(uni_boundary_ogham, r"\d\b", "6 ", Some((0, 1)));
  30 mat!(uni_not_boundary_none, r"\d\B", "6δ", Some((0, 1)));
  31 mat!(uni_not_boundary_ogham, r"\d\B", "6 ", None);
  32
  33 // Test general categories.
  34 //
  35 // We should test more, but there's a lot. Write a script to generate more of
  36 // these tests.
  37 mat!(uni_class_gencat_cased_letter,
  38      r"\p{Cased_Letter}", "Ａ", Some((0, 3)));
  39 mat!(uni_class_gencat_close_punctuation,
  40      r"\p{Close_Punctuation}", "❯", Some((0, 3)));
  41 mat!(uni_class_gencat_connector_punctuation,
  42      r"\p{Connector_Punctuation}", "⁀", Some((0, 3)));
  43 mat!(uni_class_gencat_control,
  44      r"\p{Control}", "\u{9f}", Some((0, 2)));
  45 mat!(uni_class_gencat_currency_symbol,
  46      r"\p{Currency_Symbol}", "￡", Some((0, 3)));
  47 mat!(uni_class_gencat_dash_punctuation,
  48      r"\p{Dash_Punctuation}", "〰", Some((0, 3)));
  49 mat!(uni_class_gencat_decimal_numer,
  50      r"\p{Decimal_Number}", "𑓙", Some((0, 4)));
  51 mat!(uni_class_gencat_enclosing_mark,
  52      r"\p{Enclosing_Mark}", "\u{A672}", Some((0, 3)));
  53 mat!(uni_class_gencat_final_punctuation,
  54      r"\p{Final_Punctuation}", "⸡", Some((0, 3)));
  55 mat!(uni_class_gencat_format,
  56      r"\p{Format}", "\u{E007F}", Some((0, 4)));
  57 mat!(uni_class_gencat_initial_punctuation,
  58      r"\p{Initial_Punctuation}", "⸜", Some((0, 3)));
  59 mat!(uni_class_gencat_letter,
  60      r"\p{Letter}", "Έ", Some((0, 2)));
  61 mat!(uni_class_gencat_letter_number,
  62      r"\p{Letter_Number}", "ↂ", Some((0, 3)));
  63 mat!(uni_class_gencat_line_separator,
  64      r"\p{Line_Separator}", "\u{2028}", Some((0, 3)));
  65 mat!(uni_class_gencat_lowercase_letter,
  66      r"\p{Lowercase_Letter}", "ϛ", Some((0, 2)));
  67 mat!(uni_class_gencat_mark,
  68      r"\p{Mark}", "\u{E01EF}", Some((0, 4)));
  69 mat!(uni_class_gencat_math,
  70      r"\p{Math}", "⋿", Some((0, 3)));
  71 mat!(uni_class_gencat_modifier_letter,
  72      r"\p{Modifier_Letter}", "𖭃", Some((0, 4)));
  73 mat!(uni_class_gencat_modifier_symbol,
  74      r"\p{Modifier_Symbol}", "🏿", Some((0, 4)));
  75 mat!(uni_class_gencat_nonspacing_mark,
  76      r"\p{Nonspacing_Mark}", "\u{1E94A}", Some((0, 4)));
  77 mat!(uni_class_gencat_number,
  78      r"\p{Number}", "⓿", Some((0, 3)));
  79 mat!(uni_class_gencat_open_punctuation,
  80      r"\p{Open_Punctuation}", "｟", Some((0, 3)));
  81 mat!(uni_class_gencat_other,
  82      r"\p{Other}", "\u{bc9}", Some((0, 3)));
  83 mat!(uni_class_gencat_other_letter,
  84      r"\p{Other_Letter}", "ꓷ", Some((0, 3)));
  85 mat!(uni_class_gencat_other_number,
  86      r"\p{Other_Number}", "㉏", Some((0, 3)));
  87 mat!(uni_class_gencat_other_punctuation,
  88      r"\p{Other_Punctuation}", "𞥞", Some((0, 4)));
  89 mat!(uni_class_gencat_other_symbol,
  90      r"\p{Other_Symbol}", "⅌", Some((0, 3)));
  91 mat!(uni_class_gencat_paragraph_separator,
  92      r"\p{Paragraph_Separator}", "\u{2029}", Some((0, 3)));
  93 mat!(uni_class_gencat_private_use,
  94      r"\p{Private_Use}", "\u{10FFFD}", Some((0, 4)));
  95 mat!(uni_class_gencat_punctuation,
  96      r"\p{Punctuation}", "𑁍", Some((0, 4)));
  97 mat!(uni_class_gencat_separator,
  98      r"\p{Separator}", "\u{3000}", Some((0, 3)));
  99 mat!(uni_class_gencat_space_separator,
 100      r"\p{Space_Separator}", "\u{205F}", Some((0, 3)));
 101 mat!(uni_class_gencat_spacing_mark,
 102      r"\p{Spacing_Mark}", "\u{16F7E}", Some((0, 4)));
 103 mat!(uni_class_gencat_symbol,
 104      r"\p{Symbol}", "⯈", Some((0, 3)));
 105 mat!(uni_class_gencat_titlecase_letter,
 106      r"\p{Titlecase_Letter}", "ῼ", Some((0, 3)));
 107 mat!(uni_class_gencat_unassigned,
 108      r"\p{Unassigned}", "\u{10FFFF}", Some((0, 4)));
 109 mat!(uni_class_gencat_uppercase_letter,
 110      r"\p{Uppercase_Letter}", "Ꝋ", Some((0, 3)));