aboutsummaryrefslogtreecommitdiff
path: root/src/zenhttp/clients/httpclientcommon.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'src/zenhttp/clients/httpclientcommon.cpp')
-rw-r--r--src/zenhttp/clients/httpclientcommon.cpp315
1 files changed, 315 insertions, 0 deletions
diff --git a/src/zenhttp/clients/httpclientcommon.cpp b/src/zenhttp/clients/httpclientcommon.cpp
index 47425e014..312ca16d2 100644
--- a/src/zenhttp/clients/httpclientcommon.cpp
+++ b/src/zenhttp/clients/httpclientcommon.cpp
@@ -382,6 +382,178 @@ namespace detail {
return Result;
}
+ MultipartBoundaryParser::MultipartBoundaryParser() : BoundaryEndMatcher("--"), HeaderEndMatcher("\r\n\r\n") {}
+
+ bool MultipartBoundaryParser::Init(const std::string_view ContentTypeHeaderValue)
+ {
+ std::string LowerCaseValue = ToLower(ContentTypeHeaderValue);
+ if (LowerCaseValue.starts_with("multipart/byteranges"))
+ {
+ size_t BoundaryPos = LowerCaseValue.find("boundary=");
+ if (BoundaryPos != std::string::npos)
+ {
+ // Yes, we do a substring of the non-lowercase value string as we want the exact boundary string
+ std::string_view BoundaryName = std::string_view(ContentTypeHeaderValue).substr(BoundaryPos + 9);
+ if (!BoundaryName.empty())
+ {
+ size_t BoundaryEnd = std::string::npos;
+ while (BoundaryName[0] == ' ')
+ {
+ BoundaryName = BoundaryName.substr(1);
+ }
+ if (!BoundaryName.empty())
+ {
+ if (BoundaryName.size() > 2 && BoundaryName.front() == '"' && BoundaryName.back() == '"')
+ {
+ BoundaryEnd = BoundaryName.find('"', 1);
+ if (BoundaryEnd != std::string::npos)
+ {
+ BoundaryBeginMatcher.Init(fmt::format("\r\n--{}", BoundaryName.substr(1, BoundaryEnd - 1)));
+ return true;
+ }
+ }
+ else
+ {
+ BoundaryEnd = BoundaryName.find_first_of(" \r\n");
+ BoundaryBeginMatcher.Init(fmt::format("\r\n--{}", BoundaryName.substr(0, BoundaryEnd)));
+ return true;
+ }
+ }
+ }
+ }
+ }
+ return false;
+ }
+
+ void MultipartBoundaryParser::InternalParseInput(std::string_view data)
+ {
+ size_t ScanPos = 0;
+ while (ScanPos < data.length())
+ {
+ const char ScanChar = data[ScanPos];
+ if (BoundaryBeginMatcher.MatchState == IncrementalStringMatcher::EMatchState::Complete)
+ {
+ if (PayloadOffset + ScanPos < (BoundaryBeginMatcher.GetMatchEndOffset() + BoundaryEndMatcher.GetMatchString().length()))
+ {
+ BoundaryEndMatcher.Match(PayloadOffset + ScanPos, ScanChar);
+ if (BoundaryEndMatcher.MatchState == IncrementalStringMatcher::EMatchState::Complete)
+ {
+ BoundaryBeginMatcher.Reset();
+ HeaderEndMatcher.Reset();
+ BoundaryEndMatcher.Reset();
+ BoundaryHeader.Reset();
+ break;
+ }
+ }
+
+ BoundaryHeader.Append(ScanChar);
+
+ HeaderEndMatcher.Match(PayloadOffset + ScanPos, ScanChar);
+
+ if (HeaderEndMatcher.MatchState == IncrementalStringMatcher::EMatchState::Complete)
+ {
+ const uint64_t HeaderStartOffset = BoundaryBeginMatcher.GetMatchEndOffset();
+ const uint64_t HeaderEndOffset = HeaderEndMatcher.GetMatchStartOffset();
+ const uint64_t HeaderLength = HeaderEndOffset - HeaderStartOffset;
+ std::string_view HeaderText(BoundaryHeader.ToView().substr(0, HeaderLength));
+
+ uint64_t OffsetInPayload = PayloadOffset + ScanPos + 1;
+
+ uint64_t RangeOffset = 0;
+ uint64_t RangeLength = 0;
+ HttpContentType ContentType = HttpContentType::kBinary;
+
+ ForEachStrTok(HeaderText, "\r\n", [&](std::string_view Line) {
+ const std::pair<std::string_view, std::string_view> KeyAndValue = GetHeaderKeyAndValue(Line);
+ const std::string_view Key = KeyAndValue.first;
+ const std::string_view Value = KeyAndValue.second;
+ if (Key == "Content-Range")
+ {
+ std::pair<uint64_t, uint64_t> ContentRange = ParseContentRange(Value);
+ if (ContentRange.second != 0)
+ {
+ RangeOffset = ContentRange.first;
+ RangeLength = ContentRange.second;
+ }
+ }
+ else if (Key == "Content-Type")
+ {
+ ContentType = ParseContentType(Value);
+ }
+
+ return true;
+ });
+
+ if (RangeLength > 0)
+ {
+ Boundaries.push_back(HttpClient::Response::MultipartBoundary{.OffsetInPayload = OffsetInPayload,
+ .RangeOffset = RangeOffset,
+ .RangeLength = RangeLength,
+ .ContentType = ContentType});
+ }
+
+ BoundaryBeginMatcher.Reset();
+ HeaderEndMatcher.Reset();
+ BoundaryEndMatcher.Reset();
+ BoundaryHeader.Reset();
+ }
+ }
+ else
+ {
+ BoundaryBeginMatcher.Match(PayloadOffset + ScanPos, ScanChar);
+ }
+ ScanPos++;
+ }
+ PayloadOffset += data.length();
+ }
+
+ std::pair<std::string_view, std::string_view> GetHeaderKeyAndValue(std::string_view HeaderString)
+ {
+ size_t DelimiterPos = HeaderString.find(':');
+ if (DelimiterPos != std::string::npos)
+ {
+ std::string_view Key = HeaderString.substr(0, DelimiterPos);
+ constexpr AsciiSet WhitespaceCharacters(" \v\f\t\r\n");
+ Key = AsciiSet::TrimSuffixWith(Key, WhitespaceCharacters);
+ Key = AsciiSet::TrimPrefixWith(Key, WhitespaceCharacters);
+
+ std::string_view Value = HeaderString.substr(DelimiterPos + 1);
+ Value = AsciiSet::TrimSuffixWith(Value, WhitespaceCharacters);
+ Value = AsciiSet::TrimPrefixWith(Value, WhitespaceCharacters);
+ return std::make_pair(Key, Value);
+ }
+ return std::make_pair(HeaderString, std::string_view{});
+ }
+
+ std::pair<uint64_t, uint64_t> ParseContentRange(std::string_view Value)
+ {
+ if (Value.starts_with("bytes "))
+ {
+ size_t RangeSplitPos = Value.find('-', 6);
+ if (RangeSplitPos != std::string::npos)
+ {
+ size_t RangeEndLength = Value.find('/', RangeSplitPos + 1);
+ if (RangeEndLength == std::string::npos)
+ {
+ RangeEndLength = Value.length() - (RangeSplitPos + 1);
+ }
+ else
+ {
+ RangeEndLength = RangeEndLength - (RangeSplitPos + 1);
+ }
+ std::optional<size_t> RequestedRangeStart = ParseInt<size_t>(Value.substr(6, RangeSplitPos - 6));
+ std::optional<size_t> RequestedRangeEnd = ParseInt<size_t>(Value.substr(RangeSplitPos + 1, RangeEndLength));
+ if (RequestedRangeStart.has_value() && RequestedRangeEnd.has_value())
+ {
+ uint64_t RangeOffset = RequestedRangeStart.value();
+ uint64_t RangeLength = RequestedRangeEnd.value() - RangeOffset + 1;
+ return std::make_pair(RangeOffset, RangeLength);
+ }
+ }
+ }
+ return {0, 0};
+ }
+
} // namespace detail
} // namespace zen
@@ -470,5 +642,148 @@ TEST_CASE("CompositeBufferReadStream")
CHECK_EQ(IoHash::HashBuffer(Data), testutil::HashComposite(Data));
}
+TEST_CASE("MultipartBoundaryParser")
+{
+ uint64_t Range1Offset = 2638;
+ uint64_t Range1Length = (5111437 - Range1Offset) + 1;
+
+ uint64_t Range2Offset = 5118199;
+ uint64_t Range2Length = (9147741 - Range2Offset) + 1;
+
+ std::string_view ContentTypeHeaderValue1 = "multipart/byteranges; boundary=00000000000000019229";
+ std::string_view ContentTypeHeaderValue2 = "multipart/byteranges; boundary=\"00000000000000019229\"";
+
+ {
+ std::string_view Example1 =
+ "\r\n--00000000000000019229\r\n"
+ "Content-Type: application/x-ue-comp\r\n"
+ "Content-Range: bytes 2638-5111437/44369878\r\n"
+ "\r\n"
+ "datadatadatadata"
+ "\r\n--00000000000000019229\r\n"
+ "Content-Type: application/x-ue-comp\r\n"
+ "Content-Range: bytes 5118199-9147741/44369878\r\n"
+ "\r\n"
+ "ditaditadita"
+ "\r\n--00000000000000019229--";
+
+ detail::MultipartBoundaryParser ParserExample1;
+ ParserExample1.Init(ContentTypeHeaderValue1);
+
+ const size_t InputWindow = 7;
+ for (size_t Offset = 0; Offset < Example1.length(); Offset += InputWindow)
+ {
+ ParserExample1.ParseInput(Example1.substr(Offset, Min(Example1.length() - Offset, InputWindow)));
+ }
+
+ CHECK(ParserExample1.Boundaries.size() == 2);
+
+ CHECK(ParserExample1.Boundaries[0].RangeOffset == Range1Offset);
+ CHECK(ParserExample1.Boundaries[0].RangeLength == Range1Length);
+ CHECK(ParserExample1.Boundaries[1].RangeOffset == Range2Offset);
+ CHECK(ParserExample1.Boundaries[1].RangeLength == Range2Length);
+ }
+
+ {
+ std::string_view Example2 =
+ "\r\n--00000000000000019229\r\n"
+ "Content-Type: application/x-ue-comp\r\n"
+ "Content-Range: bytes 2638-5111437/*\r\n"
+ "\r\n"
+ "datadatadatadata"
+ "\r\n--00000000000000019229\r\n"
+ "Content-Type: application/x-ue-comp\r\n"
+ "Content-Range: bytes 5118199-9147741/44369878\r\n"
+ "\r\n"
+ "ditaditadita"
+ "\r\n--00000000000000019229--";
+
+ detail::MultipartBoundaryParser ParserExample2;
+ ParserExample2.Init(ContentTypeHeaderValue1);
+
+ const size_t InputWindow = 3;
+ for (size_t Offset = 0; Offset < Example2.length(); Offset += InputWindow)
+ {
+ std::string_view Window = Example2.substr(Offset, Min(Example2.length() - Offset, InputWindow));
+ ParserExample2.ParseInput(Window);
+ }
+
+ CHECK(ParserExample2.Boundaries.size() == 2);
+
+ CHECK(ParserExample2.Boundaries[0].RangeOffset == Range1Offset);
+ CHECK(ParserExample2.Boundaries[0].RangeLength == Range1Length);
+ CHECK(ParserExample2.Boundaries[1].RangeOffset == Range2Offset);
+ CHECK(ParserExample2.Boundaries[1].RangeLength == Range2Length);
+ }
+
+ {
+ std::string_view Example3 =
+ "\r\n--00000000000000019229\r\n"
+ "Content-Type: application/x-ue-comp\r\n"
+ "Content-Range: bytes 2638-5111437/*\r\n"
+ "\r\n"
+ "datadatadatadata"
+ "\r\n--00000000000000019229\r\n"
+ "Content-Type: application/x-ue-comp\r\n"
+ "Content-Range: bytes 5118199-9147741/44369878\r\n"
+ "\r\n"
+ "ditaditadita";
+
+ detail::MultipartBoundaryParser ParserExample3;
+ ParserExample3.Init(ContentTypeHeaderValue2);
+
+ const size_t InputWindow = 31;
+ for (size_t Offset = 0; Offset < Example3.length(); Offset += InputWindow)
+ {
+ ParserExample3.ParseInput(Example3.substr(Offset, Min(Example3.length() - Offset, InputWindow)));
+ }
+
+ CHECK(ParserExample3.Boundaries.size() == 2);
+
+ CHECK(ParserExample3.Boundaries[0].RangeOffset == Range1Offset);
+ CHECK(ParserExample3.Boundaries[0].RangeLength == Range1Length);
+ CHECK(ParserExample3.Boundaries[1].RangeOffset == Range2Offset);
+ CHECK(ParserExample3.Boundaries[1].RangeLength == Range2Length);
+ }
+
+ {
+ std::string_view Example4 =
+ "\r\n--00000000000000019229\r\n"
+ "Content-Type: application/x-ue-comp\r\n"
+ "Content-Range: bytes 2638-5111437/*\r\n"
+ "Not: really\r\n"
+ "\r\n"
+ "datadatadatadata"
+ "\r\n--000000000bait0019229\r\n"
+ "\r\n--00\r\n--000000000bait001922\r\n"
+ "\r\n\r\n\r\r\n--00000000000000019229\r\n"
+ "Content-Type: application/x-ue-comp\r\n"
+ "Content-Range: bytes 5118199-9147741/44369878\r\n"
+ "\r\n"
+ "ditaditadita"
+ "Content-Type: application/x-ue-comp\r\n"
+ "ditaditadita"
+ "Content-Range: bytes 5118199-9147741/44369878\r\n"
+ "\r\n---\r\n--00000000000000019229--";
+
+ detail::MultipartBoundaryParser ParserExample4;
+ ParserExample4.Init(ContentTypeHeaderValue1);
+
+ const size_t InputWindow = 3;
+ for (size_t Offset = 0; Offset < Example4.length(); Offset += InputWindow)
+ {
+ std::string_view Window = Example4.substr(Offset, Min(Example4.length() - Offset, InputWindow));
+ ParserExample4.ParseInput(Window);
+ }
+
+ CHECK(ParserExample4.Boundaries.size() == 2);
+
+ CHECK(ParserExample4.Boundaries[0].RangeOffset == Range1Offset);
+ CHECK(ParserExample4.Boundaries[0].RangeLength == Range1Length);
+ CHECK(ParserExample4.Boundaries[1].RangeOffset == Range2Offset);
+ CHECK(ParserExample4.Boundaries[1].RangeLength == Range2Length);
+ }
+}
+
} // namespace zen
#endif