parse.cpp 7.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207
  1. #include "parse.h"
  2. #include "common.h"
  3. #include "encode.h"
  4. namespace NUri {
  5. const TParseFlags TParser::FieldFlags[] =
  6. {
  7. TParseFlags(0 // FieldScheme
  8. | TFeature::FeatureToLower,
  9. 0)
  10. ,
  11. TParseFlags(0 // FieldUsername
  12. | TFeature::FeatureDecodeANY | TFeature::FeaturesDecode | TFeature::FeatureEncodePercent,
  13. 0 | TFeature::FeatureToLower)
  14. ,
  15. TParseFlags(0 // FieldPassword
  16. | TFeature::FeatureDecodeANY | TFeature::FeaturesDecode | TFeature::FeatureEncodePercent,
  17. 0 | TFeature::FeatureToLower)
  18. ,
  19. TParseFlags(0 // FieldHost
  20. | TFeature::FeatureToLower | TFeature::FeatureUpperEncoded | (TFeature::FeaturesMaybeEncode & ~TFeature::FeatureEncodeExtendedDelim),
  21. 0 | TFeature::FeaturesMaybeDecode)
  22. ,
  23. TParseFlags(0 // FieldPort
  24. ,
  25. 0)
  26. ,
  27. TParseFlags(0 // FieldPath
  28. | TFeature::FeaturesEncodePChar | TFeature::FeaturePathOperation,
  29. 0 | TFeature::FeatureToLower | TFeature::FeatureEncodeSpaceAsPlus)
  30. ,
  31. TParseFlags(0 // FieldQuery
  32. | TFeature::FeaturesEncodePChar | TFeature::FeatureEncodeSpaceAsPlus,
  33. 0 | TFeature::FeatureToLower)
  34. ,
  35. TParseFlags(0 // FieldFragment
  36. | TFeature::FeaturesEncodePChar,
  37. 0 | TFeature::FeatureToLower | TFeature::FeatureEncodeSpaceAsPlus)};
  38. namespace NParse {
  39. void TRange::AddRange(const TRange& range, ui64 mask) {
  40. FlagsAllPlaintext |= range.FlagsAllPlaintext;
  41. // update only if flags apply here
  42. mask &= range.FlagsEncodeMasked;
  43. if (0 == mask)
  44. return;
  45. FlagsEncodeMasked |= mask;
  46. if (mask & TFeature::FeaturesMaybeEncode)
  47. Encode += range.Encode;
  48. if (mask & TFeature::FeaturesDecode)
  49. Decode += range.Decode;
  50. }
  51. }
  52. void TParser::copyRequirementsImpl(const char* ptr) {
  53. Y_ASSERT(0 != CurRange.FlagsAllPlaintext);
  54. Y_UNUSED(ptr);
  55. #ifdef DO_PRN
  56. PrintHead(ptr, __FUNCTION__)
  57. << " all=[" << IntToString<16>(CurRange.FlagsAllPlaintext)
  58. << "] enc=[" << IntToString<16>(CurRange.FlagsEncodeMasked)
  59. << " & " << IntToString<16>(Flags.Allow | Flags.Extra) << "]";
  60. PrintTail(CurRange.Beg, ptr);
  61. #endif
  62. for (int i = 0; i < TField::FieldUrlMAX; ++i) {
  63. const TField::EField fld = TField::EField(i);
  64. TSection& section = Sections[fld];
  65. // update only sections in progress
  66. if (nullptr == section.Beg)
  67. continue;
  68. // and overlapping with the range
  69. if (nullptr != section.End && section.End < CurRange.Beg)
  70. continue;
  71. #ifdef DO_PRN
  72. PrintHead(ptr, __FUNCTION__, fld)
  73. << " all=[" << IntToString<16>(CurRange.FlagsAllPlaintext)
  74. << "] enc=[" << IntToString<16>(CurRange.FlagsEncodeMasked)
  75. << " & " << IntToString<16>(GetFieldFlags(fld)) << "]";
  76. PrintTail(section.Beg, ptr);
  77. #endif
  78. section.AddRange(CurRange, GetFieldFlags(fld));
  79. }
  80. CurRange.Reset();
  81. }
  82. void TParser::PctEndImpl(const char* ptr) {
  83. #ifdef DO_PRN
  84. PrintHead(PctBegin, __FUNCTION__);
  85. PrintTail(PctBegin, ptr);
  86. #else
  87. Y_UNUSED(ptr);
  88. #endif
  89. setRequirement(PctBegin, TEncoder::GetFlags('%').FeatFlags);
  90. PctBegin = nullptr;
  91. }
  92. void TParser::HexSet(const char* ptr) {
  93. Y_ASSERT(nullptr != PctBegin);
  94. #ifdef DO_PRN
  95. PrintHead(ptr, __FUNCTION__);
  96. PrintTail(PctBegin, ptr + 1);
  97. #endif
  98. PctBegin = nullptr;
  99. const unsigned char ch = HexValue;
  100. ui64 flags = TEncoder::GetFlags('%').FeatFlags | TEncoder::GetFlags(ch).FeatFlags;
  101. setRequirementExcept(ptr, flags, TFeature::FeaturesMaybeEncode);
  102. }
  103. TState::EParsed TParser::ParseImpl() {
  104. #ifdef DO_PRN
  105. PrintHead(UriStr.data(), "[Parsing]") << "URL";
  106. PrintTail(UriStr);
  107. #endif
  108. const bool ok = doParse(UriStr.data(), UriStr.length());
  109. #ifdef DO_PRN
  110. Cdbg << (ok ? "[Parsed]" : "[Failed]");
  111. for (int idx = 0; idx < TField::FieldUrlMAX; ++idx) {
  112. const TSection& section = Sections[idx];
  113. if (section.IsSet())
  114. Cdbg << ' ' << TField::EField(idx) << "=[" << section.Get() << ']';
  115. }
  116. Cdbg << Endl;
  117. #endif
  118. if (!ok) {
  119. if (!(Flags & TFeature::FeatureTryToFix) || !Sections[TField::FieldFrag].Beg)
  120. return TState::ParsedBadFormat;
  121. //Here: error was in fragment, just ignore it
  122. ResetSection(TField::FieldFrag);
  123. }
  124. if ((Flags & TFeature::FeatureDenyNetworkPath) && IsNetPath())
  125. return TState::ParsedBadFormat;
  126. const TSection& scheme = Sections[TField::FieldScheme];
  127. Scheme = scheme.IsSet() ? TSchemeInfo::GetKind(scheme.Get()) : TScheme::SchemeEmpty;
  128. const TSchemeInfo& schemeInfo = TSchemeInfo::Get(Scheme);
  129. if (IsRootless()) {
  130. // opaque case happens
  131. if (schemeInfo.FldReq & TField::FlagHost)
  132. return TState::ParsedBadFormat;
  133. if (TScheme::SchemeEmpty == Scheme)
  134. return TState::ParsedBadScheme;
  135. if (Flags & TFeature::FeatureAllowRootless)
  136. return TState::ParsedOK;
  137. if (!(Flags & TFeature::FeatureSchemeFlexible))
  138. return TState::ParsedBadScheme;
  139. return TState::ParsedRootless;
  140. }
  141. checkSectionCollision(TField::FieldUser, TField::FieldHost);
  142. checkSectionCollision(TField::FieldPass, TField::FieldPort);
  143. if (0 == (Flags & TFeature::FeatureAuthSupported))
  144. if (Sections[TField::FieldUser].IsSet() || Sections[TField::FieldPass].IsSet())
  145. return TState::ParsedBadAuth;
  146. TSection& host = Sections[TField::FieldHost];
  147. if (host.IsSet())
  148. for (; host.End != host.Beg && '.' == host.End[-1];)
  149. --host.End;
  150. if (scheme.IsSet()) {
  151. ui64 wantCareFlags = 0;
  152. switch (Scheme) {
  153. case TScheme::SchemeHTTP:
  154. break;
  155. case TScheme::SchemeEmpty:
  156. Scheme = TScheme::SchemeUnknown;
  157. [[fallthrough]];
  158. case TScheme::SchemeUnknown:
  159. wantCareFlags =
  160. TFeature::FeatureSchemeFlexible | TFeature::FeatureNoRelPath;
  161. break;
  162. default:
  163. wantCareFlags =
  164. TFeature::FeatureSchemeFlexible | TFeature::FeatureSchemeKnown;
  165. break;
  166. }
  167. if (0 != wantCareFlags && 0 == (Flags & wantCareFlags))
  168. return TState::ParsedBadScheme;
  169. if ((schemeInfo.FldReq & TField::FlagHost) || (Flags & TFeature::FeatureRemoteOnly))
  170. if (!host.IsSet() || 0 == host.Len())
  171. return TState::ParsedBadFormat;
  172. }
  173. return TState::ParsedOK;
  174. }
  175. }