...
Run Format

Source file src/unicode/script_test.go

Documentation: unicode

     1  // Copyright 2009 The Go Authors. All rights reserved.
     2  // Use of this source code is governed by a BSD-style
     3  // license that can be found in the LICENSE file.
     4  
     5  package unicode_test
     6  
     7  import (
     8  	"testing"
     9  	. "unicode"
    10  )
    11  
    12  type T struct {
    13  	rune   rune
    14  	script string
    15  }
    16  
    17  // Hand-chosen tests from Unicode 5.1.0, 6.0.0, 6.2.0, 6.3.0, 7.0.0, 8.0.0,
    18  // 9.0.0, 10.0.0.
    19  // mostly to discover when new scripts and categories arise.
    20  // If this tests fails, add the missing scripts to the test and add entries
    21  // of the form
    22  //     pkg unicode, var <new script> *RangeTable
    23  // to api/next.txt.
    24  var inTest = []T{
    25  	{0x11711, "Ahom"},
    26  	{0x1e900, "Adlam"},
    27  	{0x14646, "Anatolian_Hieroglyphs"},
    28  	{0x06e2, "Arabic"},
    29  	{0x0567, "Armenian"},
    30  	{0x10b20, "Avestan"},
    31  	{0x11c00, "Bhaiksuki"},
    32  	{0x1b37, "Balinese"},
    33  	{0xa6af, "Bamum"},
    34  	{0x16ada, "Bassa_Vah"},
    35  	{0x1be1, "Batak"},
    36  	{0x09c2, "Bengali"},
    37  	{0x3115, "Bopomofo"},
    38  	{0x282d, "Braille"},
    39  	{0x1a1a, "Buginese"},
    40  	{0x1747, "Buhid"},
    41  	{0x11011, "Brahmi"},
    42  	{0x156d, "Canadian_Aboriginal"},
    43  	{0x102a9, "Carian"},
    44  	{0x10563, "Caucasian_Albanian"},
    45  	{0x11111, "Chakma"},
    46  	{0xaa4d, "Cham"},
    47  	{0x13c2, "Cherokee"},
    48  	{0x0020, "Common"},
    49  	{0x1d4a5, "Common"},
    50  	{0x2cfc, "Coptic"},
    51  	{0x12420, "Cuneiform"},
    52  	{0x1080c, "Cypriot"},
    53  	{0xa663, "Cyrillic"},
    54  	{0x10430, "Deseret"},
    55  	{0x094a, "Devanagari"},
    56  	{0x1BC00, "Duployan"},
    57  	{0x13001, "Egyptian_Hieroglyphs"},
    58  	{0x10500, "Elbasan"},
    59  	{0x1271, "Ethiopic"},
    60  	{0x10fc, "Georgian"},
    61  	{0x2c40, "Glagolitic"},
    62  	{0x10347, "Gothic"},
    63  	{0x11303, "Grantha"},
    64  	{0x03ae, "Greek"},
    65  	{0x0abf, "Gujarati"},
    66  	{0x0a24, "Gurmukhi"},
    67  	{0x3028, "Han"},
    68  	{0x11b8, "Hangul"},
    69  	{0x1727, "Hanunoo"},
    70  	{0x108FF, "Hatran"},
    71  	{0x05a0, "Hebrew"},
    72  	{0x3058, "Hiragana"},
    73  	{0x10841, "Imperial_Aramaic"},
    74  	{0x20e6, "Inherited"},
    75  	{0x10b70, "Inscriptional_Pahlavi"},
    76  	{0x10b5a, "Inscriptional_Parthian"},
    77  	{0xa9d0, "Javanese"},
    78  	{0x1109f, "Kaithi"},
    79  	{0x0cbd, "Kannada"},
    80  	{0x30a6, "Katakana"},
    81  	{0xa928, "Kayah_Li"},
    82  	{0x10a11, "Kharoshthi"},
    83  	{0x17c6, "Khmer"},
    84  	{0x11211, "Khojki"},
    85  	{0x112df, "Khudawadi"},
    86  	{0x0eaa, "Lao"},
    87  	{0x1d79, "Latin"},
    88  	{0x1c10, "Lepcha"},
    89  	{0x1930, "Limbu"},
    90  	{0x10755, "Linear_A"},
    91  	{0x1003c, "Linear_B"},
    92  	{0xa4e1, "Lisu"},
    93  	{0x10290, "Lycian"},
    94  	{0x10930, "Lydian"},
    95  	{0x11173, "Mahajani"},
    96  	{0x0d42, "Malayalam"},
    97  	{0x0843, "Mandaic"},
    98  	{0x10ac8, "Manichaean"},
    99  	{0x11cB6, "Marchen"},
   100  	{0x11d59, "Masaram_Gondi"},
   101  	{0xabd0, "Meetei_Mayek"},
   102  	{0x1e800, "Mende_Kikakui"},
   103  	{0x1099f, "Meroitic_Hieroglyphs"},
   104  	{0x109a0, "Meroitic_Cursive"},
   105  	{0x16f00, "Miao"},
   106  	{0x11611, "Modi"},
   107  	{0x1822, "Mongolian"},
   108  	{0x16a60, "Mro"},
   109  	{0x11293, "Multani"},
   110  	{0x104c, "Myanmar"},
   111  	{0x10880, "Nabataean"},
   112  	{0x11400, "Newa"},
   113  	{0x19c3, "New_Tai_Lue"},
   114  	{0x07f8, "Nko"},
   115  	{0x1b170, "Nushu"},
   116  	{0x169b, "Ogham"},
   117  	{0x1c6a, "Ol_Chiki"},
   118  	{0x10C80, "Old_Hungarian"},
   119  	{0x10310, "Old_Italic"},
   120  	{0x10a80, "Old_North_Arabian"},
   121  	{0x10350, "Old_Permic"},
   122  	{0x103c9, "Old_Persian"},
   123  	{0x10a6f, "Old_South_Arabian"},
   124  	{0x10c20, "Old_Turkic"},
   125  	{0x0b3e, "Oriya"},
   126  	{0x104d9, "Osage"},
   127  	{0x10491, "Osmanya"},
   128  	{0x16b2b, "Pahawh_Hmong"},
   129  	{0x10876, "Palmyrene"},
   130  	{0x11ACE, "Pau_Cin_Hau"},
   131  	{0xa860, "Phags_Pa"},
   132  	{0x10918, "Phoenician"},
   133  	{0x10baf, "Psalter_Pahlavi"},
   134  	{0xa949, "Rejang"},
   135  	{0x16c0, "Runic"},
   136  	{0x081d, "Samaritan"},
   137  	{0xa892, "Saurashtra"},
   138  	{0x111a0, "Sharada"},
   139  	{0x10463, "Shavian"},
   140  	{0x115c1, "Siddham"},
   141  	{0x1D920, "SignWriting"},
   142  	{0x0dbd, "Sinhala"},
   143  	{0x110d0, "Sora_Sompeng"},
   144  	{0x11a99, "Soyombo"},
   145  	{0x1ba3, "Sundanese"},
   146  	{0xa803, "Syloti_Nagri"},
   147  	{0x070f, "Syriac"},
   148  	{0x170f, "Tagalog"},
   149  	{0x176f, "Tagbanwa"},
   150  	{0x1972, "Tai_Le"},
   151  	{0x1a62, "Tai_Tham"},
   152  	{0xaadc, "Tai_Viet"},
   153  	{0x116c9, "Takri"},
   154  	{0x0bbf, "Tamil"},
   155  	{0x17000, "Tangut"},
   156  	{0x0c55, "Telugu"},
   157  	{0x07a7, "Thaana"},
   158  	{0x0e46, "Thai"},
   159  	{0x0f36, "Tibetan"},
   160  	{0x2d55, "Tifinagh"},
   161  	{0x114d9, "Tirhuta"},
   162  	{0x10388, "Ugaritic"},
   163  	{0xa60e, "Vai"},
   164  	{0x118ff, "Warang_Citi"},
   165  	{0xa216, "Yi"},
   166  	{0x11a0a, "Zanabazar_Square"},
   167  }
   168  
   169  var outTest = []T{ // not really worth being thorough
   170  	{0x20, "Telugu"},
   171  }
   172  
   173  var inCategoryTest = []T{
   174  	{0x0081, "Cc"},
   175  	{0x200B, "Cf"},
   176  	{0xf0000, "Co"},
   177  	{0xdb80, "Cs"},
   178  	{0x0236, "Ll"},
   179  	{0x1d9d, "Lm"},
   180  	{0x07cf, "Lo"},
   181  	{0x1f8a, "Lt"},
   182  	{0x03ff, "Lu"},
   183  	{0x0bc1, "Mc"},
   184  	{0x20df, "Me"},
   185  	{0x07f0, "Mn"},
   186  	{0x1bb2, "Nd"},
   187  	{0x10147, "Nl"},
   188  	{0x2478, "No"},
   189  	{0xfe33, "Pc"},
   190  	{0x2011, "Pd"},
   191  	{0x301e, "Pe"},
   192  	{0x2e03, "Pf"},
   193  	{0x2e02, "Pi"},
   194  	{0x0022, "Po"},
   195  	{0x2770, "Ps"},
   196  	{0x00a4, "Sc"},
   197  	{0xa711, "Sk"},
   198  	{0x25f9, "Sm"},
   199  	{0x2108, "So"},
   200  	{0x2028, "Zl"},
   201  	{0x2029, "Zp"},
   202  	{0x202f, "Zs"},
   203  	// Unifieds.
   204  	{0x04aa, "L"},
   205  	{0x0009, "C"},
   206  	{0x1712, "M"},
   207  	{0x0031, "N"},
   208  	{0x00bb, "P"},
   209  	{0x00a2, "S"},
   210  	{0x00a0, "Z"},
   211  }
   212  
   213  var inPropTest = []T{
   214  	{0x0046, "ASCII_Hex_Digit"},
   215  	{0x200F, "Bidi_Control"},
   216  	{0x2212, "Dash"},
   217  	{0xE0001, "Deprecated"},
   218  	{0x00B7, "Diacritic"},
   219  	{0x30FE, "Extender"},
   220  	{0xFF46, "Hex_Digit"},
   221  	{0x2E17, "Hyphen"},
   222  	{0x2FFB, "IDS_Binary_Operator"},
   223  	{0x2FF3, "IDS_Trinary_Operator"},
   224  	{0xFA6A, "Ideographic"},
   225  	{0x200D, "Join_Control"},
   226  	{0x0EC4, "Logical_Order_Exception"},
   227  	{0x2FFFF, "Noncharacter_Code_Point"},
   228  	{0x065E, "Other_Alphabetic"},
   229  	{0x2065, "Other_Default_Ignorable_Code_Point"},
   230  	{0x0BD7, "Other_Grapheme_Extend"},
   231  	{0x0387, "Other_ID_Continue"},
   232  	{0x212E, "Other_ID_Start"},
   233  	{0x2094, "Other_Lowercase"},
   234  	{0x2040, "Other_Math"},
   235  	{0x216F, "Other_Uppercase"},
   236  	{0x0027, "Pattern_Syntax"},
   237  	{0x0020, "Pattern_White_Space"},
   238  	{0x06DD, "Prepended_Concatenation_Mark"},
   239  	{0x300D, "Quotation_Mark"},
   240  	{0x2EF3, "Radical"},
   241  	{0x1f1ff, "Regional_Indicator"},
   242  	{0x061F, "STerm"}, // Deprecated alias of Sentence_Terminal
   243  	{0x061F, "Sentence_Terminal"},
   244  	{0x2071, "Soft_Dotted"},
   245  	{0x003A, "Terminal_Punctuation"},
   246  	{0x9FC3, "Unified_Ideograph"},
   247  	{0xFE0F, "Variation_Selector"},
   248  	{0x0020, "White_Space"},
   249  }
   250  
   251  func TestScripts(t *testing.T) {
   252  	notTested := make(map[string]bool)
   253  	for k := range Scripts {
   254  		notTested[k] = true
   255  	}
   256  	for _, test := range inTest {
   257  		if _, ok := Scripts[test.script]; !ok {
   258  			t.Fatal(test.script, "not a known script")
   259  		}
   260  		if !Is(Scripts[test.script], test.rune) {
   261  			t.Errorf("IsScript(%U, %s) = false, want true", test.rune, test.script)
   262  		}
   263  		delete(notTested, test.script)
   264  	}
   265  	for _, test := range outTest {
   266  		if Is(Scripts[test.script], test.rune) {
   267  			t.Errorf("IsScript(%U, %s) = true, want false", test.rune, test.script)
   268  		}
   269  	}
   270  	for k := range notTested {
   271  		t.Error("script not tested:", k)
   272  	}
   273  }
   274  
   275  func TestCategories(t *testing.T) {
   276  	notTested := make(map[string]bool)
   277  	for k := range Categories {
   278  		notTested[k] = true
   279  	}
   280  	for _, test := range inCategoryTest {
   281  		if _, ok := Categories[test.script]; !ok {
   282  			t.Fatal(test.script, "not a known category")
   283  		}
   284  		if !Is(Categories[test.script], test.rune) {
   285  			t.Errorf("IsCategory(%U, %s) = false, want true", test.rune, test.script)
   286  		}
   287  		delete(notTested, test.script)
   288  	}
   289  	for k := range notTested {
   290  		t.Error("category not tested:", k)
   291  	}
   292  }
   293  
   294  func TestProperties(t *testing.T) {
   295  	notTested := make(map[string]bool)
   296  	for k := range Properties {
   297  		notTested[k] = true
   298  	}
   299  	for _, test := range inPropTest {
   300  		if _, ok := Properties[test.script]; !ok {
   301  			t.Fatal(test.script, "not a known prop")
   302  		}
   303  		if !Is(Properties[test.script], test.rune) {
   304  			t.Errorf("IsCategory(%U, %s) = false, want true", test.rune, test.script)
   305  		}
   306  		delete(notTested, test.script)
   307  	}
   308  	for k := range notTested {
   309  		t.Error("property not tested:", k)
   310  	}
   311  }
   312  

View as plain text