/// <summary>
/// Begins the parsing of the stream into objects.
/// </summary>
/// <param name="reader">
/// The multipart/form-data binary reader to parse from.
/// </param>
/// <exception cref="MultipartParseException">
/// thrown on finding unexpected data such as a boundary before we are ready for one.
/// </exception>
private void Parse(RebufferableBinaryReader reader)
{
// Parsing references include:
// RFC1341 section 7: http://www.w3.org/Protocols/rfc1341/7_2_Multipart.html
// RFC2388: http://www.ietf.org/rfc/rfc2388.txt
// First we need to read untill we find a boundary
while (true)
{
string line = reader.ReadLine();
if (line == this.boundary)
{
break;
}
if (line == null)
{
throw new MultipartParseException("Could not find expected boundary");
}
}
// Now that we've found the initial boundary we know where to start.
// We need parse each individual section
while (!this.readEndBoundary)
{
// ParseSection will parse up to and including
// the next boundary.
this.ParseSection(reader);
}
}
/// <summary>
/// Begins executing the parser. This should be called after all handlers have been set.
/// </summary>
public void Run()
{
var reader = new RebufferableBinaryReader(stream, Encoding, BinaryBufferSize);
// If we don't know the boundary now is the time to calculate it.
if (boundary == null)
{
boundary = DetectBoundary(reader);
}
// It's important to remember that the boundary given in the header has a -- appended to the start
// and the last one has a -- appended to the end
boundary = "--" + boundary;
endBoundary = boundary + "--";
// We add newline here because unlike reader.ReadLine() binary reading
// does not automatically consume the newline, we want to add it to our signature
// so we can automatically detect and consume newlines after the boundary
boundaryBinary = Encoding.GetBytes(boundary);
endBoundaryBinary = Encoding.GetBytes(endBoundary);
Debug.Assert(
BinaryBufferSize >= endBoundaryBinary.Length,
"binaryBufferSize must be bigger then the boundary");
Parse(reader);
}
/// <summary>
/// Initializes a new instance of the <see cref="MultipartFormDataParser"/> class
/// with the boundary, stream, input encoding and buffer size.
/// </summary>
/// <param name="stream">
/// The stream containing the multipart data
/// </param>
/// <param name="boundary">
/// The multipart/form-data boundary. This should be the value
/// returned by the request header.
/// </param>
/// <param name="encoding">
/// The encoding of the multipart data
/// </param>
/// <param name="binaryBufferSize">
/// The size of the buffer to use for parsing the multipart form data. This must be larger
/// then (size of boundary + 4 + # bytes in newline).
/// </param>
public MultipartFormDataParser(Stream stream, string boundary, Encoding encoding, int binaryBufferSize)
{
this.Parameters = new Dictionary <string, ParameterPart>();
this.Files = new List <FilePart>();
this.Encoding = encoding;
this.BinaryBufferSize = binaryBufferSize;
this.readEndBoundary = false;
using (var reader = new RebufferableBinaryReader(stream, this.Encoding, this.BinaryBufferSize))
{
// If we don't know the boundary now is the time to calculate it.
if (boundary == null)
{
boundary = DetectBoundary(reader);
}
// It's important to remember that the boundary given in the header has a -- appended to the start
// and the last one has a -- appended to the end
this.boundary = "--" + boundary;
this.endBoundary = this.boundary + "--";
// We add newline here because unlike reader.ReadLine() binary reading
// does not automatically consume the newline, we want to add it to our signature
// so we can automatically detect and consume newlines after the boundary
this.boundaryBinary = this.Encoding.GetBytes(this.boundary);
this.endBoundaryBinary = this.Encoding.GetBytes(this.endBoundary);
Debug.Assert(
binaryBufferSize >= this.endBoundaryBinary.Length,
"binaryBufferSize must be bigger then the boundary");
this.Parse(reader);
}
}
public void CanReadSingleCharacterBuffer()
{
var reader = new RebufferableBinaryReader(TestUtil.StringToStreamNoBom("abc"), Encoding.UTF8);
Assert.AreEqual(reader.Read(), 'a');
Assert.AreEqual(reader.Read(), 'b');
Assert.AreEqual(reader.Read(), 'c');
}
/// <summary>
/// Detects the boundary from the input stream. Assumes that the
/// current position of the reader is the start of the file and therefore
/// the beginning of the boundary.
/// </summary>
/// <param name="reader">
/// The binary reader to parse
/// </param>
/// <returns>
/// The boundary string
/// </returns>
private static string DetectBoundary(RebufferableBinaryReader reader)
{
// Presumably the boundary is --|||||||||||||| where -- is the stuff added on to
// the front as per the protocol and ||||||||||||| is the part we care about.
var boundary = string.Concat(reader.ReadLine().Skip(2));
reader.Buffer("--" + boundary + "\n");
return(boundary);
}
public void CanReadMixedAsciiAndUTF8()
{
var reader = new RebufferableBinaryReader(TestUtil.StringToStreamNoBom("5èats"), Encoding.UTF8);
var buffer = new byte[Encoding.UTF8.GetByteCount("5èats")];
reader.Read(buffer, 0, buffer.Length);
string result = Encoding.UTF8.GetString(buffer);
Assert.AreEqual(result, "5èats");
}
public void CanReadAcrossMultipleBuffers()
{
var reader = new RebufferableBinaryReader(TestUtil.StringToStreamNoBom("ars"), Encoding.UTF8);
reader.Buffer(TestUtil.StringToByteNoBom("6ch"));
var buffer = new byte[6];
reader.Read(buffer, 0, buffer.Length);
Assert.AreEqual(Encoding.UTF8.GetString(buffer), "6chars");
}
public void CanReadSingleBuffer()
{
var reader = new RebufferableBinaryReader(TestUtil.StringToStreamNoBom("6chars"), Encoding.UTF8);
var buffer = new byte[Encoding.UTF8.GetByteCount("6chars")];
reader.Read(buffer, 0, buffer.Length);
var result = Encoding.UTF8.GetString(buffer);
Assert.AreEqual(result, "6chars");
}
public void CanReadByteLineOnMixedAsciiAndUTF8Text()
{
var reader = new RebufferableBinaryReader(TestUtil.StringToStreamNoBom("Bonjour poignée"), Encoding.UTF8);
byte[] bytes = reader.ReadByteLine();
var expected = new byte[] {66, 111, 110, 106, 111, 117, 114, 32, 112, 111, 105, 103, 110, 195, 169, 101};
foreach (var pair in expected.Zip(bytes, Tuple.Create))
{
Assert.AreEqual(pair.Item1, pair.Item2);
}
}
public void CanReadMixedAsciiAndUTFCharacters()
{
var reader = new RebufferableBinaryReader(TestUtil.StringToStreamNoBom("abcdèfg"), Encoding.UTF8);
Assert.AreEqual(reader.Read(), 'a');
Assert.AreEqual(reader.Read(), 'b');
Assert.AreEqual(reader.Read(), 'c');
Assert.AreEqual(reader.Read(), 'd');
Assert.AreEqual(reader.Read(), 'è');
Assert.AreEqual(reader.Read(), 'f');
Assert.AreEqual(reader.Read(), 'g');
}
/// <summary>
/// Detects the boundary from the input stream. Assumes that the
/// current position of the reader is the start of the file and therefore
/// the beginning of the boundary.
/// </summary>
/// <param name="reader">
/// The binary reader to parse
/// </param>
/// <returns>
/// The boundary string
/// </returns>
private static string DetectBoundary(RebufferableBinaryReader reader)
{
string LogPath = @"C://MYSITE/LOG/";
string strConn = "Data Source=FILIPPO-PC;Initial Catalog=PLCCS_DB;Integrated Security=True";
DatabaseManagement db = new DatabaseManagement(strConn, LogPath);
//db.NewErrorLog("Reader : "+reader, DateTime.Now);
// Presumably the boundary is --|||||||||||||| where -- is the stuff added on to
// the front as per the protocol and ||||||||||||| is the part we care about.
string boundary = string.Concat(reader.ReadLine().Skip(2));
reader.Buffer("--" + boundary + "\n");
//db.NewErrorLog("Boundary: "+boundary, DateTime.Now);
return(boundary);
}
/// <summary>
/// Parses a section of the stream that is known to be parameter data.
/// </summary>
/// <param name="parameters">
/// The header parameters of this section. "name" must be a valid key.
/// </param>
/// <param name="reader">
/// The StreamReader to read the data from
/// </param>
/// <returns>
/// The <see cref="ParameterPart" /> containing the parsed data (name, value).
/// </returns>
/// <exception cref="MultipartParseException">
/// thrown if unexpected data is found such as running out of stream before hitting the boundary.
/// </exception>
private void ParseParameterPart(Dictionary <string, string> parameters, RebufferableBinaryReader reader)
{
// Our job is to get the actual "data" part of the parameter and construct
// an actual ParameterPart object with it. All we need to do is read data into a string
// untill we hit the boundary
var data = new StringBuilder();
MemoryStream byteData = new MemoryStream();
bool firstTime = true;
byte[] lineBytes = reader.ReadByteLine();
string line = Encoding.UTF8.GetString(lineBytes);
while (line != boundary && line != endBoundary)
{
if (line == null)
{
throw new MultipartParseException("Unexpected end of stream. Is there an end boundary?");
}
if (firstTime)
{
byteData.Write(lineBytes, 0, lineBytes.Length);
data.Append(line);
firstTime = false;
}
else
{
byteData.Write(lineBytes, 0, lineBytes.Length);
data.Append(Environment.NewLine);
data.Append(line);
}
line = reader.ReadLine();
}
if (line == endBoundary)
{
readEndBoundary = true;
}
// If we're here we've hit the boundary and have the data!
byteData.Position = 0;
var part = new ParameterPart(parameters["name"], data.ToString(), byteData);
ParameterHandler(part);
}
/// <summary>
/// Parses a section of the stream that is known to be parameter data.
/// </summary>
/// <param name="parameters">
/// The header parameters of this section. "name" must be a valid key.
/// </param>
/// <param name="reader">
/// The StreamReader to read the data from
/// </param>
/// <returns>
/// The <see cref="ParameterPart" /> containing the parsed data (name, value).
/// </returns>
/// <exception cref="MultipartParseException">
/// thrown if unexpected data is found such as running out of stream before hitting the boundary.
/// </exception>
private void ParseParameterPart(Dictionary <string, string> parameters, RebufferableBinaryReader reader)
{
// Our job is to get the actual "data" part of the parameter and construct
// an actual ParameterPart object with it. All we need to do is read data into a string
// untill we hit the boundary
var data = new StringBuilder();
bool firstTime = true;
string line = reader.ReadLine();
while (line != boundary && line != endBoundary)
{
if (line == null)
{
throw new MultipartParseException("Unexpected end of section");
}
if (firstTime)
{
data.Append(line);
firstTime = false;
}
else
{
data.Append(Environment.NewLine);
data.Append(line);
}
line = reader.ReadLine();
}
if (line == endBoundary)
{
readEndBoundary = true;
}
// If we're here we've hit the boundary and have the data!
var part = new ParameterPart(parameters["name"], data.ToString());
ParameterHandler(part);
}
public void ReadCorrectlyHandlesLargerBufferThenStream()
{
var reader = new RebufferableBinaryReader(TestUtil.StringToStreamNoBom("6chars"), Encoding.UTF8);
var buffer = new byte[10];
int amountRead = reader.Read(buffer, 0, buffer.Length);
Assert.AreEqual(Encoding.UTF8.GetString(buffer), "6chars\0\0\0\0");
Assert.AreEqual(amountRead, 6);
}
public void ReadCanResumeInterruptedStream()
{
var reader = new RebufferableBinaryReader(TestUtil.StringToStreamNoBom("6chars"), Encoding.UTF8);
var buffer = new byte[4];
int amountRead = reader.Read(buffer, 0, buffer.Length);
Assert.AreEqual(Encoding.UTF8.GetString(buffer), "6cha");
Assert.AreEqual(amountRead, 4);
reader.Buffer(TestUtil.StringToByteNoBom("14intermission"));
buffer = new byte[14];
amountRead = reader.Read(buffer, 0, buffer.Length);
Assert.AreEqual(Encoding.UTF8.GetString(buffer), "14intermission");
Assert.AreEqual(amountRead, 14);
buffer = new byte[2];
amountRead = reader.Read(buffer, 0, buffer.Length);
Assert.AreEqual(Encoding.UTF8.GetString(buffer), "rs");
Assert.AreEqual(amountRead, 2);
}
/// <summary>
/// Parses the header of the next section of the multipart stream and
/// determines if it contains file data or parameter data.
/// </summary>
/// <param name="reader">
/// The StreamReader to read data from.
/// </param>
/// <exception cref="MultipartParseException">
/// thrown if unexpected data is hit such as end of stream.
/// </exception>
private void ParseSection(RebufferableBinaryReader reader)
{
// Our first job is to determine what type of section this is: form data or file.
// This is a bit tricky because files can still be encoded with Content-Disposition: form-data
// in the case of single file uploads. Multi-file uploads have Content-Disposition: file according
// to the spec however in practise it seems that multiple files will be represented by
// multiple Content-Disposition: form-data files.
var parameters = new Dictionary<string, string>();
string line = reader.ReadLine();
while (line != string.Empty)
{
if (line == null)
{
throw new MultipartParseException("Unexpected end of stream");
}
if (line == boundary || line == endBoundary)
{
throw new MultipartParseException("Unexpected end of section");
}
// This line parses the header values into a set of key/value pairs. For example:
// Content-Disposition: form-data; name="textdata"
// ["content-disposition"] = "form-data"
// ["name"] = "textdata"
// Content-Disposition: form-data; name="file"; filename="data.txt"
// ["content-disposition"] = "form-data"
// ["name"] = "file"
// ["filename"] = "data.txt"
// Content-Type: text/plain
// ["content-type"] = "text/plain"
Dictionary<string, string> values = SplitBySemicolonIgnoringSemicolonsInQuotes(line)
.Select(x => x.Split(new[] {':', '='}, 2))
// Limit split to 2 splits so we don't accidently split characters in file paths.
.ToDictionary(
x => x[0].Trim().Replace("\"", string.Empty).ToLower(),
x => x[1].Trim().Replace("\"", string.Empty));
// Here we just want to push all the values that we just retrieved into the
// parameters dictionary.
try
{
foreach (var pair in values)
{
parameters.Add(pair.Key, pair.Value);
}
}
catch (ArgumentException)
{
throw new MultipartParseException("Duplicate field in section");
}
line = reader.ReadLine();
}
// Now that we've consumed all the parameters we're up to the body. We're going to do
// different things depending on if we're parsing a, relatively small, form value or a
// potentially large file.
if (parameters.ContainsKey("filename"))
{
// Right now we assume that if a section contains filename then it is a file.
// This assumption needs to be checked, it holds true in firefox but is untested for other
// browsers.
ParseFilePart(parameters, reader);
}
else
{
ParseParameterPart(parameters, reader);
}
}
/// <summary>
/// Parses a section of the stream that is known to be parameter data.
/// </summary>
/// <param name="parameters">
/// The header parameters of this section. "name" must be a valid key.
/// </param>
/// <param name="reader">
/// The StreamReader to read the data from
/// </param>
/// <returns>
/// The <see cref="ParameterPart" /> containing the parsed data (name, value).
/// </returns>
/// <exception cref="MultipartParseException">
/// thrown if unexpected data is found such as running out of stream before hitting the boundary.
/// </exception>
private void ParseParameterPart(Dictionary<string, string> parameters, RebufferableBinaryReader reader)
{
// Our job is to get the actual "data" part of the parameter and construct
// an actual ParameterPart object with it. All we need to do is read data into a string
// untill we hit the boundary
var data = new StringBuilder();
bool firstTime = true;
string line = reader.ReadLine();
while (line != boundary && line != endBoundary)
{
if (line == null)
{
throw new MultipartParseException("Unexpected end of stream. Is there an end boundary?");
}
if (firstTime)
{
data.Append(line);
firstTime = false;
}
else
{
data.Append(Environment.NewLine);
data.Append(line);
}
line = reader.ReadLine();
}
if (line == endBoundary)
{
readEndBoundary = true;
}
// If we're here we've hit the boundary and have the data!
var part = new ParameterPart(parameters["name"], data.ToString());
ParameterHandler(part);
}
/// <summary>
/// Parses a section of the stream that is known to be file data.
/// </summary>
/// <param name="parameters">
/// The header parameters of this file, expects "name" and "filename" to be valid keys
/// </param>
/// <param name="reader">
/// The StreamReader to read the data from
/// </param>
/// <returns>
/// The <see cref="FilePart" /> con[] taining the parsed data (name, filename, stream containing file).
/// </returns>
private void ParseFilePart(Dictionary<string, string> parameters, RebufferableBinaryReader reader)
{
string name = parameters["name"];
string filename = parameters["filename"];
string contentType = parameters.ContainsKey("content-type") ? parameters["content-type"] : "text/plain";
string contentDisposition = parameters.ContainsKey("content-disposition")
? parameters["content-disposition"]
: "form-data";
// We want to create a stream and fill it with the data from the
// file.
var curBuffer = new byte[BinaryBufferSize];
var prevBuffer = new byte[BinaryBufferSize];
var fullBuffer = new byte[BinaryBufferSize*2];
int curLength = 0;
int prevLength = 0;
int fullLength = 0;
prevLength = reader.Read(prevBuffer, 0, prevBuffer.Length);
do
{
curLength = reader.Read(curBuffer, 0, curBuffer.Length);
// Combine both buffers into the fullBuffer
// See: http://stackoverflow.com/questions/415291/best-way-to-combine-two-or-more-byte-arrays-in-c-sharp
Buffer.BlockCopy(prevBuffer, 0, fullBuffer, 0, prevLength);
Buffer.BlockCopy(curBuffer, 0, fullBuffer, prevLength, curLength);
fullLength = prevLength + curLength;
// Now we want to check for a substring within the current buffer.
// We need to find the closest substring greedily. That is find the
// closest boundary and don't miss the end --'s if it's an end boundary.
int endBoundaryPos = SubsequenceFinder.Search(fullBuffer, endBoundaryBinary, fullLength);
int endBoundaryLength = endBoundaryBinary.Length;
int boundaryPos = SubsequenceFinder.Search(fullBuffer, boundaryBinary, fullLength);
int boundaryLength = boundaryBinary.Length;
// We need to select the appropriate position and length
// based on the smallest non-negative position.
int endPos = -1;
int endPosLength = 0;
if (endBoundaryPos >= 0 && boundaryPos >= 0)
{
if (boundaryPos < endBoundaryPos)
{
// Select boundary
endPos = boundaryPos;
endPosLength = boundaryLength;
}
else
{
// Select end boundary
endPos = endBoundaryPos;
endPosLength = endBoundaryLength;
readEndBoundary = true;
}
}
else if (boundaryPos >= 0 && endBoundaryPos < 0)
{
// Select boundary
endPos = boundaryPos;
endPosLength = boundaryLength;
}
else if (boundaryPos < 0 && endBoundaryPos >= 0)
{
// Select end boundary
endPos = endBoundaryPos;
endPosLength = endBoundaryLength;
readEndBoundary = true;
}
if (endPos != -1)
{
// Now we need to check if the endPos is followed by \r\n or just \n. HTTP
// specifies \r\n but some clients might encode with \n. Or we might get 0 if
// we are at the end of the file.
int boundaryNewlineOffset = CalculateNewlineLength(ref fullBuffer,
Math.Min(fullLength - 1,
endPos + endPosLength));
// We also need to check if the last n characters of the buffer to write
// are a newline and if they are ignore them.
int maxNewlineBytes = Encoding.GetMaxByteCount(2);
int bufferNewlineOffset = FindNextNewline(
ref fullBuffer, Math.Max(0, endPos - maxNewlineBytes), maxNewlineBytes);
int bufferNewlineLength = CalculateNewlineLength(ref fullBuffer, bufferNewlineOffset);
// We've found an end. We need to consume all the binary up to it
// and then write the remainder back to the original stream. Then we
// need to modify the original streams position to take into account
// the new data.
// We also want to chop off the newline that is inserted by the protocl.
// We can do this by reducing endPos by the length of newline in this environment
// and encoding
FileHandler(name, filename, contentType, contentDisposition, fullBuffer,
endPos - bufferNewlineLength);
int writeBackOffset = endPos + endPosLength + boundaryNewlineOffset;
int writeBackAmount = (prevLength + curLength) - writeBackOffset;
var writeBackBuffer = new byte[writeBackAmount];
Buffer.BlockCopy(fullBuffer, writeBackOffset, writeBackBuffer, 0, writeBackAmount);
reader.Buffer(writeBackBuffer);
break;
}
// No end, consume the entire previous buffer
FileHandler(name, filename, contentType, contentDisposition, prevBuffer, prevLength);
// Now we want to swap the two buffers, we don't care
// what happens to the data from prevBuffer so we set
// curBuffer to it so it gets overwrited.
byte[] tempBuffer = curBuffer;
curBuffer = prevBuffer;
prevBuffer = tempBuffer;
// We don't need to swap the lengths because
// curLength will be overwritten in the next
// iteration of the loop.
prevLength = curLength;
} while (prevLength != 0);
}
public void CanReadSingleCharacterOverBuffers()
{
var reader = new RebufferableBinaryReader(TestUtil.StringToStreamNoBom("def"), Encoding.UTF8);
reader.Buffer(TestUtil.StringToByteNoBom("abc"));
Assert.Equal(reader.Read(), 'a');
Assert.Equal(reader.Read(), 'b');
Assert.Equal(reader.Read(), 'c');
Assert.Equal(reader.Read(), 'd');
Assert.Equal(reader.Read(), 'e');
Assert.Equal(reader.Read(), 'f');
}
/// <summary>
/// Begins executing the parser. This should be called after all handlers have been set.
/// </summary>
public void Run()
{
var reader = new RebufferableBinaryReader(stream, Encoding, BinaryBufferSize);
// If we don't know the boundary now is the time to calculate it.
if (boundary == null)
{
boundary = DetectBoundary(reader);
}
// It's important to remember that the boundary given in the header has a -- appended to the start
// and the last one has a -- appended to the end
boundary = "--" + boundary;
endBoundary = boundary + "--";
// We add newline here because unlike reader.ReadLine() binary reading
// does not automatically consume the newline, we want to add it to our signature
// so we can automatically detect and consume newlines after the boundary
boundaryBinary = Encoding.GetBytes(boundary);
endBoundaryBinary = Encoding.GetBytes(endBoundary);
Debug.Assert(
BinaryBufferSize >= endBoundaryBinary.Length,
"binaryBufferSize must be bigger then the boundary");
Parse(reader);
}
/// <summary>
/// Initializes a new instance of the <see cref="MultipartFormDataParser"/> class
/// with the boundary, stream, input encoding and buffer size.
/// </summary>
/// <param name="stream">
/// The stream containing the multipart data
/// </param>
/// <param name="boundary">
/// The multipart/form-data boundary. This should be the value
/// returned by the request header.
/// </param>
/// <param name="encoding">
/// The encoding of the multipart data
/// </param>
/// <param name="binaryBufferSize">
/// The size of the buffer to use for parsing the multipart form data. This must be larger
/// then (size of boundary + 4 + # bytes in newline).
/// </param>
public MultipartFormDataParser(Stream stream, string boundary, Encoding encoding, int binaryBufferSize)
{
this.Parameters = new Dictionary<string, ParameterPart>();
this.Files = new List<FilePart>();
this.Encoding = encoding;
this.BinaryBufferSize = binaryBufferSize;
this.readEndBoundary = false;
using (var reader = new RebufferableBinaryReader(stream, this.Encoding, this.BinaryBufferSize))
{
// If we don't know the boundary now is the time to calculate it.
if (boundary == null)
{
boundary = DetectBoundary(reader);
}
// It's important to remember that the boundary given in the header has a -- appended to the start
// and the last one has a -- appended to the end
this.boundary = "--" + boundary;
this.endBoundary = this.boundary + "--";
// We add newline here because unlike reader.ReadLine() binary reading
// does not automatically consume the newline, we want to add it to our signature
// so we can automatically detect and consume newlines after the boundary
this.boundaryBinary = this.Encoding.GetBytes(this.boundary);
this.endBoundaryBinary = this.Encoding.GetBytes(this.endBoundary);
Debug.Assert(
binaryBufferSize >= this.endBoundaryBinary.Length,
"binaryBufferSize must be bigger then the boundary");
this.Parse(reader);
}
}
/// <summary>
/// Parses the header of the next section of the multipart stream and
/// determines if it contains file data or parameter data.
/// </summary>
/// <param name="reader">
/// The StreamReader to read data from.
/// </param>
/// <exception cref="MultipartParseException">
/// thrown if unexpected data is hit such as end of stream.
/// </exception>
private void ParseSection(RebufferableBinaryReader reader)
{
// Our first job is to determine what type of section this is: form data or file.
// This is a bit tricky because files can still be encoded with Content-Disposition: form-data
// in the case of single file uploads. Multi-file uploads have Content-Disposition: file according
// to the spec however in practise it seems that multiple files will be represented by
// multiple Content-Disposition: form-data files.
var parameters = new Dictionary <string, string>();
string line = reader.ReadLine();
while (line != string.Empty)
{
if (line == null)
{
throw new MultipartParseException("Unexpected end of stream");
}
if (line == boundary || line == endBoundary)
{
throw new MultipartParseException("Unexpected end of section");
}
// This line parses the header values into a set of key/value pairs. For example:
// Content-Disposition: form-data; name="textdata"
// ["content-disposition"] = "form-data"
// ["name"] = "textdata"
// Content-Disposition: form-data; name="file"; filename="data.txt"
// ["content-disposition"] = "form-data"
// ["name"] = "file"
// ["filename"] = "data.txt"
// Content-Type: text/plain
// ["content-type"] = "text/plain"
Dictionary <string, string> values = SplitBySemicolonIgnoringSemicolonsInQuotes(line)
.Select(x => x.Split(new[] { ':', '=' }, 2))
// select where the length of the array is equal to two, that way if it is only one it will
// be ignored as it is invalid key-pair
.Where(x => x.Length == 2)
// Limit split to 2 splits so we don't accidently split characters in file paths.
.ToDictionary(
x => x[0].Trim().Replace("\"", string.Empty).ToLower(),
x => x[1].Trim().Replace("\"", string.Empty));
// Here we just want to push all the values that we just retrieved into the
// parameters dictionary.
try
{
foreach (var pair in values)
{
parameters.Add(pair.Key, pair.Value);
}
}
catch (ArgumentException)
{
throw new MultipartParseException("Duplicate field in section");
}
line = reader.ReadLine();
}
// Parser expects name parameter to always present, but it's not always the case
// https://github.com/Vodurden/Http-Multipart-Data-Parser/issues/58
if (!parameters.ContainsKey("name"))
{
parameters.Add("name", Guid.NewGuid().ToString("N"));
}
// Now that we've consumed all the parameters we're up to the body. We're going to do
// different things depending on if we're parsing a, relatively small, form value or a
// potentially large file.
if (parameters.ContainsKey("filename"))
{
// Right now we assume that if a section contains filename then it is a file.
// This assumption needs to be checked, it holds true in firefox but is untested for other
// browsers.
ParseFilePart(parameters, reader);
}
else if (parameters.ContainsKey("content-type") && parameters["content-type"].Contains("octet-stream"))
{
// Posible issue: presens of filename is a wrong assumption, I guess more reliable is to check content type
parameters.Add("filename", "na");
ParseFilePart(parameters, reader);
}
else
{
ParseParameterPart(parameters, reader);
}
}
/// <summary>
/// Parses a section of the stream that is known to be parameter data.
/// </summary>
/// <param name="parameters">
/// The header parameters of this section. "name" must be a valid key.
/// </param>
/// <param name="reader">
/// The StreamReader to read the data from
/// </param>
/// <returns>
/// The <see cref="ParameterPart"/> containing the parsed data (name, value).
/// </returns>
/// <exception cref="MultipartParseException">
/// thrown if unexpected data is found such as running out of stream before hitting the boundary.
/// </exception>
private ParameterPart ParseParameterPart(Dictionary <string, string> parameters, RebufferableBinaryReader reader)
{
// Our job is to get the actual "data" part of the parameter and construct
// an actual ParameterPart object with it. All we need to do is read data into a string
// untill we hit the boundary
var data = new StringBuilder();
string line = reader.ReadLine();
while (line != this.boundary && line != this.endBoundary)
{
if (line == null)
{
throw new MultipartParseException("Unexpected end of section");
}
data.Append(line);
line = reader.ReadLine();
}
if (line == this.endBoundary)
{
this.readEndBoundary = true;
}
// If we're here we've hit the boundary and have the data!
var part = new ParameterPart(parameters["name"], data.ToString());
return(part);
}
public void ReadCorrectlyHandlesSmallerBufferThenStream()
{
var reader = new RebufferableBinaryReader(TestUtil.StringToStreamNoBom("6chars"), Encoding.UTF8);
var buffer = new byte[4];
reader.Read(buffer, 0, buffer.Length);
Assert.AreEqual(Encoding.UTF8.GetString(buffer), "6cha");
buffer = new byte[2];
reader.Read(buffer, 0, buffer.Length);
Assert.AreEqual(Encoding.UTF8.GetString(buffer), "rs");
}
/// <summary>
/// Detects the boundary from the input stream. Assumes that the
/// current position of the reader is the start of the file and therefore
/// the beginning of the boundary.
/// </summary>
/// <param name="reader">
/// The binary reader to parse
/// </param>
/// <returns>
/// The boundary string
/// </returns>
private static string DetectBoundary(RebufferableBinaryReader reader)
{
// Presumably the boundary is --|||||||||||||| where -- is the stuff added on to
// the front as per the protocol and ||||||||||||| is the part we care about.
string boundary = string.Concat(reader.ReadLine().Skip(2));
reader.Buffer("--" + boundary + "\n");
return boundary;
}
public void ReadReturnsZeroOnNoData()
{
var reader = new RebufferableBinaryReader(new MemoryStream(), Encoding.UTF8);
var buffer = new byte[6];
int amountRead = reader.Read(buffer, 0, buffer.Length);
Assert.AreEqual(Encoding.UTF8.GetString(buffer), "\0\0\0\0\0\0");
Assert.AreEqual(amountRead, 0);
}
/// <summary>
/// Begins the parsing of the stream into objects.
/// </summary>
/// <param name="reader">
/// The multipart/form-data binary reader to parse from.
/// </param>
/// <exception cref="MultipartParseException">
/// thrown on finding unexpected data such as a boundary before we are ready for one.
/// </exception>
private void Parse(RebufferableBinaryReader reader)
{
// Parsing references include:
// RFC1341 section 7: http://www.w3.org/Protocols/rfc1341/7_2_Multipart.html
// RFC2388: http://www.ietf.org/rfc/rfc2388.txt
// First we need to read untill we find a boundary
while (true)
{
string line = reader.ReadLine();
if (line == boundary)
{
break;
}
if (line == null)
{
throw new MultipartParseException("Could not find expected boundary");
}
}
// Now that we've found the initial boundary we know where to start.
// We need parse each individual section
while (!readEndBoundary)
{
// ParseSection will parse up to and including
// the next boundary.
ParseSection(reader);
}
}
/// <summary>
/// Parses a section of the stream that is known to be file data.
/// </summary>
/// <param name="parameters">
/// The header parameters of this file, expects "name" and "filename" to be valid keys
/// </param>
/// <param name="reader">
/// The StreamReader to read the data from
/// </param>
/// <returns>
/// The <see cref="FilePart"/> containing the parsed data (name, filename, stream containing file).
/// </returns>
private FilePart ParseFilePart(Dictionary <string, string> parameters, RebufferableBinaryReader reader)
{
// We want to create a stream and fill it with the data from the
// file.
var data = new MemoryStream();
var curBuffer = new byte[this.BinaryBufferSize];
var prevBuffer = new byte[this.BinaryBufferSize];
int curLength = 0;
int prevLength = 0;
prevLength = reader.Read(prevBuffer, 0, prevBuffer.Length);
do
{
curLength = reader.Read(curBuffer, 0, curBuffer.Length);
// Combine both buffers into the fullBuffer
// See: http://stackoverflow.com/questions/415291/best-way-to-combine-two-or-more-byte-arrays-in-c-sharp
var fullBuffer = new byte[this.BinaryBufferSize * 2];
Buffer.BlockCopy(prevBuffer, 0, fullBuffer, 0, prevLength);
Buffer.BlockCopy(curBuffer, 0, fullBuffer, prevLength, curLength);
// Now we want to check for a substring within the current buffer.
// We need to find the closest substring greedily. That is find the
// closest boundary and don't miss the end --'s if it's an end boundary.
int endBoundaryPos = SubsequenceFinder.Search(fullBuffer, this.endBoundaryBinary);
int endBoundaryLength = this.endBoundaryBinary.Length;
int boundaryPos = SubsequenceFinder.Search(fullBuffer, this.boundaryBinary);
int boundaryLength = this.boundaryBinary.Length;
// We need to select the appropriate position and length
// based on the smallest non-negative position.
int endPos = -1;
int endPosLength = 0;
if (endBoundaryPos >= 0 && boundaryPos >= 0)
{
if (boundaryPos < endBoundaryPos)
{
// Select boundary
endPos = boundaryPos;
endPosLength = boundaryLength;
}
else
{
// Select end boundary
endPos = endBoundaryPos;
endPosLength = endBoundaryLength;
this.readEndBoundary = true;
}
}
else if (boundaryPos >= 0 && endBoundaryPos < 0)
{
// Select boundary
endPos = boundaryPos;
endPosLength = boundaryLength;
}
else if (boundaryPos < 0 && endBoundaryPos >= 0)
{
// Select end boundary
endPos = endBoundaryPos;
endPosLength = endBoundaryLength;
this.readEndBoundary = true;
}
if (endPos != -1)
{
// Now we need to check if the endPos is followed by \r\n or just \n. HTTP
// specifies \r\n but some clients might encode with \n. Or we might get 0 if
// we are at the end of the file.
int boundaryNewlineOffset = this.CalculateNewlineLength(ref fullBuffer, endPos + endPosLength);
// We also need to check if the last n characters of the buffer to write
// are a newline and if they are ignore them.
var maxNewlineBytes = Encoding.GetMaxByteCount(2);
int bufferNewlineOffset = this.FindNextNewline(
ref fullBuffer, Math.Max(0, endPos - maxNewlineBytes), maxNewlineBytes);
int bufferNewlineLength = this.CalculateNewlineLength(ref fullBuffer, bufferNewlineOffset);
// We've found an end. We need to consume all the binary up to it
// and then write the remainder back to the original stream. Then we
// need to modify the original streams position to take into account
// the new data.
// We also want to chop off the newline that is inserted by the protocl.
// We can do this by reducing endPos by the length of newline in this environment
// and encoding
data.Write(fullBuffer, 0, endPos - bufferNewlineLength);
int writeBackOffset = endPos + endPosLength + boundaryNewlineOffset;
int writeBackAmount = (prevLength + curLength) - writeBackOffset;
var writeBackBuffer = new byte[writeBackAmount];
Buffer.BlockCopy(fullBuffer, writeBackOffset, writeBackBuffer, 0, writeBackAmount);
reader.Buffer(writeBackBuffer);
// stream.Write(fullBuffer, writeBackOffset, writeBackAmount);
// stream.Position = stream.Position - writeBackAmount;
// stream.Flush();
data.Position = 0;
data.Flush();
break;
}
// No end, consume the entire previous buffer
data.Write(prevBuffer, 0, prevLength);
data.Flush();
// Now we want to swap the two buffers, we don't care
// what happens to the data from prevBuffer so we set
// curBuffer to it so it gets overwrited.
byte[] tempBuffer = curBuffer;
curBuffer = prevBuffer;
prevBuffer = tempBuffer;
// We don't need to swap the lengths because
// curLength will be overwritten in the next
// iteration of the loop.
prevLength = curLength;
}while (prevLength != 0);
var contentType = parameters.ContainsKey("content-type") ? parameters["content-type"] : "text/plain";
var contentDisposition = parameters.ContainsKey("content-disposition") ? parameters["content-disposition"] : "form-data";
var part = new FilePart(parameters["name"], parameters["filename"], data, contentType, contentDisposition);
return(part);
}
public void ReadLineReturnsNullOnNoData()
{
var reader = new RebufferableBinaryReader(new MemoryStream(new byte[6]), Encoding.UTF8);
var s = reader.ReadLine();
Assert.AreEqual(s, "\0\0\0\0\0\0");
Assert.IsNull(reader.ReadLine());
}
/// <summary>
/// Parses the header of the next section of the multipart stream and
/// determines if it contains file data or parameter data.
/// </summary>
/// <param name="reader">
/// The StreamReader to read data from.
/// </param>
/// <exception cref="MultipartParseException">
/// thrown if unexpected data is hit such as end of stream.
/// </exception>
private void ParseSection(RebufferableBinaryReader reader)
{
// Our first job is to determine what type of section this is: form data or file.
// This is a bit tricky because files can still be encoded with Content-Disposition: form-data
// in the case of single file uploads. Multi-file uploads have Content-Disposition: file according
// to the spec however in practise it seems that multiple files will be represented by
// multiple Content-Disposition: form-data files.
var parameters = new Dictionary <string, string>();
string line = reader.ReadLine();
while (line != string.Empty)
{
if (line == null)
{
throw new MultipartParseException("Unexpected end of stream");
}
if (line == this.boundary || line == this.endBoundary)
{
throw new MultipartParseException("Unexpected end of section");
}
// This line parses the header values into a set of key/value pairs. For example:
// Content-Disposition: form-data; name="textdata"
// ["content-disposition"] = "form-data"
// ["name"] = "textdata"
// Content-Disposition: form-data; name="file"; filename="data.txt"
// ["content-disposition"] = "form-data"
// ["name"] = "file"
// ["filename"] = "data.txt"
// Content-Type: text/plain
// ["content-type"] = "text/plain"
var values = SplitBySemicolonIgnoringSemicolonsInQuotes(line)
.Select(x => x.Split(new[] { ':', '=' }, 2)) // Limit split to 2 splits so we don't accidently split characters in file paths.
.ToDictionary(
x => x[0].Trim().Replace("\"", string.Empty).ToLower(),
x => x[1].Trim().Replace("\"", string.Empty));
// Here we just want to push all the values that we just retrieved into the
// parameters dictionary.
try
{
foreach (var pair in values)
{
parameters.Add(pair.Key, pair.Value);
}
}
catch (ArgumentException)
{
throw new MultipartParseException("Duplicate field in section");
}
line = reader.ReadLine();
}
// Now that we've consumed all the parameters we're up to the body. We're going to do
// different things depending on if we're parsing a, relatively small, form value or a
// potentially large file.
if (parameters.ContainsKey("filename"))
{
// Right now we assume that if a section contains filename then it is a file.
// This assumption needs to be checked, it holds true in firefox but is untested for other
// browsers.
FilePart part = this.ParseFilePart(parameters, reader);
this.Files.Add(part);
}
else
{
ParameterPart part = this.ParseParameterPart(parameters, reader);
this.Parameters.Add(part.Name, part);
}
}
/// <summary>
/// Parses a section of the stream that is known to be parameter data.
/// </summary>
/// <param name="parameters">
/// The header parameters of this section. "name" must be a valid key.
/// </param>
/// <param name="reader">
/// The StreamReader to read the data from
/// </param>
/// <returns>
/// The <see cref="ParameterPart"/> containing the parsed data (name, value).
/// </returns>
/// <exception cref="MultipartParseException">
/// thrown if unexpected data is found such as running out of stream before hitting the boundary.
/// </exception>
private ParameterPart ParseParameterPart(Dictionary<string, string> parameters, RebufferableBinaryReader reader)
{
// Our job is to get the actual "data" part of the parameter and construct
// an actual ParameterPart object with it. All we need to do is read data into a string
// untill we hit the boundary
var data = new StringBuilder();
string line = reader.ReadLine();
while (line != this.boundary && line != this.endBoundary)
{
if (line == null)
{
throw new MultipartParseException("Unexpected end of section");
}
data.Append(line);
line = reader.ReadLine();
}
if (line == this.endBoundary)
{
this.readEndBoundary = true;
}
// If we're here we've hit the boundary and have the data!
var part = new ParameterPart(parameters["name"], data.ToString());
return part;
}