unit.rs 4.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140
  1. #![allow(deprecated)]
  2. use assert_matches::assert_matches;
  3. /// https://github.com/servo/rust-url/issues/373
  4. #[test]
  5. fn test_punycode_prefix_with_length_check() {
  6. let config = idna::Config::default()
  7. .verify_dns_length(true)
  8. .check_hyphens(true)
  9. .use_std3_ascii_rules(true);
  10. assert!(config.to_ascii("xn--").is_err());
  11. assert!(config.to_ascii("xn---").is_err());
  12. assert!(config.to_ascii("xn-----").is_err());
  13. assert!(config.to_ascii("xn--.").is_err());
  14. assert!(config.to_ascii("xn--...").is_err());
  15. assert!(config.to_ascii(".xn--").is_err());
  16. assert!(config.to_ascii("...xn--").is_err());
  17. assert!(config.to_ascii("xn--.xn--").is_err());
  18. assert!(config.to_ascii("xn--.example.org").is_err());
  19. }
  20. /// https://github.com/servo/rust-url/issues/373
  21. #[test]
  22. fn test_punycode_prefix_without_length_check() {
  23. let config = idna::Config::default()
  24. .verify_dns_length(false)
  25. .check_hyphens(true)
  26. .use_std3_ascii_rules(true);
  27. assert!(config.to_ascii("xn--").is_err());
  28. assert!(config.to_ascii("xn---").is_err());
  29. assert!(config.to_ascii("xn-----").is_err());
  30. assert!(config.to_ascii("xn--.").is_err());
  31. assert!(config.to_ascii("xn--...").is_err());
  32. assert!(config.to_ascii(".xn--").is_err());
  33. assert!(config.to_ascii("...xn--").is_err());
  34. assert!(config.to_ascii("xn--.xn--").is_err());
  35. assert!(config.to_ascii("xn--.example.org").is_err());
  36. }
  37. #[test]
  38. fn test_punycode_invalid_encoding() {
  39. let config = idna::Config::default();
  40. assert!(config.to_ascii("xn--55555577").is_err());
  41. }
  42. // http://www.unicode.org/reports/tr46/#Table_Example_Processing
  43. #[test]
  44. fn test_examples() {
  45. let mut codec = idna::Idna::default();
  46. let mut out = String::new();
  47. assert_matches!(codec.to_unicode("Bloß.de", &mut out), Ok(()));
  48. assert_eq!(out, "bloß.de");
  49. out.clear();
  50. assert_matches!(codec.to_unicode("xn--blo-7ka.de", &mut out), Ok(()));
  51. assert_eq!(out, "bloß.de");
  52. out.clear();
  53. assert_matches!(codec.to_unicode("u\u{308}.com", &mut out), Ok(()));
  54. assert_eq!(out, "ü.com");
  55. out.clear();
  56. assert_matches!(codec.to_unicode("xn--tda.com", &mut out), Ok(()));
  57. assert_eq!(out, "ü.com");
  58. out.clear();
  59. assert_matches!(codec.to_unicode("xn--u-ccb.com", &mut out), Err(_));
  60. out.clear();
  61. assert_matches!(codec.to_unicode("a⒈com", &mut out), Err(_));
  62. out.clear();
  63. assert_matches!(codec.to_unicode("xn--a-ecp.ru", &mut out), Err(_));
  64. out.clear();
  65. assert_matches!(codec.to_unicode("xn--0.pt", &mut out), Err(_));
  66. out.clear();
  67. assert_matches!(codec.to_unicode("日本語。JP", &mut out), Ok(()));
  68. assert_eq!(out, "日本語.jp");
  69. out.clear();
  70. assert_matches!(codec.to_unicode("☕.us", &mut out), Ok(()));
  71. assert_eq!(out, "☕.us");
  72. }
  73. #[test]
  74. fn test_v5() {
  75. let config = idna::Config::default()
  76. .verify_dns_length(true)
  77. .use_std3_ascii_rules(true)
  78. .check_hyphens(true);
  79. // IdnaTest:784 蔏。𑰺
  80. assert!(config.to_ascii("\u{11C3A}").is_err());
  81. assert!(config.to_ascii("\u{850f}.\u{11C3A}").is_err());
  82. assert!(config.to_ascii("\u{850f}\u{ff61}\u{11C3A}").is_err());
  83. }
  84. #[test]
  85. fn test_v8_bidi_rules() {
  86. let config = idna::Config::default()
  87. .verify_dns_length(true)
  88. .use_std3_ascii_rules(true)
  89. .check_hyphens(true);
  90. assert_eq!(config.to_ascii("abc").unwrap(), "abc");
  91. assert_eq!(config.to_ascii("123").unwrap(), "123");
  92. assert_eq!(config.to_ascii("אבּג").unwrap(), "xn--kdb3bdf");
  93. assert_eq!(config.to_ascii("ابج").unwrap(), "xn--mgbcm");
  94. assert_eq!(config.to_ascii("abc.ابج").unwrap(), "abc.xn--mgbcm");
  95. assert_eq!(config.to_ascii("אבּג.ابج").unwrap(), "xn--kdb3bdf.xn--mgbcm");
  96. // Bidi domain names cannot start with digits
  97. assert!(config.to_ascii("0a.\u{05D0}").is_err());
  98. assert!(config.to_ascii("0à.\u{05D0}").is_err());
  99. // Bidi chars may be punycode-encoded
  100. assert!(config.to_ascii("xn--0ca24w").is_err());
  101. }
  102. #[test]
  103. fn emoji_domains() {
  104. // HOT BEVERAGE is allowed here...
  105. let config = idna::Config::default()
  106. .verify_dns_length(true)
  107. .use_std3_ascii_rules(true)
  108. .check_hyphens(true);
  109. assert_eq!(config.to_ascii("☕.com").unwrap(), "xn--53h.com");
  110. }
  111. #[test]
  112. fn unicode_before_delimiter() {
  113. let config = idna::Config::default();
  114. assert!(config.to_ascii("xn--f\u{34a}-PTP").is_err());
  115. }