idna.rs 3.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131
  1. //! International domain names
  2. //!
  3. //! https://url.spec.whatwg.org/#idna
  4. use idna_mapping::*;
  5. use punycode;
  6. use std::ascii::AsciiExt;
  7. use unicode_normalization::UnicodeNormalization;
  8. fn idna_mapped(mapping: &'static [char]) -> Result<String, Error> {
  9. let mut ret = "".to_string();
  10. for &c in mapping {
  11. ret.push(c)
  12. }
  13. return Ok(ret);
  14. }
  15. fn idna_deviation(codepoint: char, mapping: &'static [char], transitional: bool) -> Result<String, Error> {
  16. if transitional {
  17. return idna_mapped(mapping);
  18. }
  19. return Ok(codepoint.to_string());
  20. }
  21. fn idna_disallowed_std3_valid(codepoint: char, use_std3_asciirules: bool) -> Result<String, Error> {
  22. if use_std3_asciirules {
  23. return Err(Error::DissallowedByStd3AsciiRules);
  24. }
  25. return Ok(codepoint.to_string());
  26. }
  27. fn idna_disallowed_std3_mapped(mapping: &'static [char], use_std3_asciirules: bool) -> Result<String, Error> {
  28. if use_std3_asciirules {
  29. return Err(Error::DissallowedMappedInStd3);
  30. }
  31. return idna_mapped(mapping);
  32. }
  33. fn map_char(codepoint: char, flags: Uts46Flags) -> Result<String, Error> {
  34. let mut min = 0;
  35. let mut max = TABLE.len() - 1;
  36. while max > min {
  37. let mid = (min + max) >> 1;
  38. if codepoint > TABLE[mid].to {
  39. min = mid;
  40. } else if codepoint < TABLE[mid].from {
  41. max = mid;
  42. } else {
  43. min = mid;
  44. max = mid;
  45. }
  46. }
  47. let mapping = TABLE[min].mapping;
  48. match TABLE[min].status {
  49. MappingStatus::valid => Ok(codepoint.to_string()),
  50. MappingStatus::ignored => Ok("".to_string()),
  51. MappingStatus::mapped => idna_mapped(mapping),
  52. MappingStatus::deviation => {
  53. idna_deviation(codepoint, mapping, flags.transitional_processing)
  54. }
  55. MappingStatus::disallowed => Err(Error::DissallowedCharacter),
  56. MappingStatus::disallowed_STD3_valid => {
  57. idna_disallowed_std3_valid(codepoint, flags.use_std3_ascii_rules)
  58. }
  59. MappingStatus::disallowed_STD3_mapped => {
  60. idna_disallowed_std3_mapped(mapping, flags.use_std3_ascii_rules)
  61. }
  62. }
  63. }
  64. #[derive(Copy, Clone)]
  65. pub struct Uts46Flags {
  66. pub use_std3_ascii_rules: bool,
  67. pub transitional_processing: bool,
  68. }
  69. pub enum Error {
  70. PunycodeEncodingError,
  71. DissallowedByStd3AsciiRules,
  72. DissallowedMappedInStd3,
  73. DissallowedCharacter,
  74. }
  75. /// http://www.unicode.org/reports/tr46/#ToASCII
  76. pub fn uts46_to_ascii(domain: &str, flags: Uts46Flags) -> Result<String, Error> {
  77. let mut ret = String::new();
  78. for c in domain.chars() {
  79. match map_char(c, flags) {
  80. Ok(mystr) => ret.push_str(&mystr),
  81. Err(x) => return Err(x)
  82. }
  83. }
  84. // normalize NFC
  85. let ret = ret.nfc().collect::<String>();
  86. let vec: Vec<&str> = ret.split(".").collect();
  87. let mut result = String::new();
  88. for label in vec {
  89. if label.is_ascii() {
  90. if result.len() > 0 {
  91. result.push('.');
  92. }
  93. result.push_str(label);
  94. } else {
  95. match punycode::encode_str(label) {
  96. Some(x) => {
  97. if result.len() > 0 {
  98. result.push('.');
  99. }
  100. result.push_str("xn--");
  101. result.push_str(&x);
  102. },
  103. None => return Err(Error::PunycodeEncodingError)
  104. }
  105. }
  106. }
  107. return Ok(result);
  108. }
  109. /// https://url.spec.whatwg.org/#concept-domain-to-ascii
  110. pub fn domain_to_ascii(domain: &str) -> Result<String, Error> {
  111. uts46_to_ascii(domain, Uts46Flags {
  112. use_std3_ascii_rules: false,
  113. transitional_processing: true,
  114. })
  115. }