diff options
author | Mattes D <github@xoft.cz> | 2016-03-01 16:58:43 +0100 |
---|---|---|
committer | Mattes D <github@xoft.cz> | 2016-03-01 16:58:43 +0100 |
commit | 7cc8f8dd2213cade66633d196f579895385a869a (patch) | |
tree | 83219a6eaf87a2050c10662fb8caa5d1514a6af4 /src/HTTP | |
parent | Merge pull request #3057 from tonibm19/master (diff) | |
parent | HTTP: Fixed typos and bad leftovers. (diff) | |
download | cuberite-7cc8f8dd2213cade66633d196f579895385a869a.tar cuberite-7cc8f8dd2213cade66633d196f579895385a869a.tar.gz cuberite-7cc8f8dd2213cade66633d196f579895385a869a.tar.bz2 cuberite-7cc8f8dd2213cade66633d196f579895385a869a.tar.lz cuberite-7cc8f8dd2213cade66633d196f579895385a869a.tar.xz cuberite-7cc8f8dd2213cade66633d196f579895385a869a.tar.zst cuberite-7cc8f8dd2213cade66633d196f579895385a869a.zip |
Diffstat (limited to 'src/HTTP')
-rw-r--r-- | src/HTTP/CMakeLists.txt | 43 | ||||
-rw-r--r-- | src/HTTP/EnvelopeParser.cpp | 132 | ||||
-rw-r--r-- | src/HTTP/EnvelopeParser.h | 73 | ||||
-rw-r--r-- | src/HTTP/HTTPFormParser.cpp | 293 | ||||
-rw-r--r-- | src/HTTP/HTTPFormParser.h | 114 | ||||
-rw-r--r-- | src/HTTP/HTTPMessage.cpp | 159 | ||||
-rw-r--r-- | src/HTTP/HTTPMessage.h | 158 | ||||
-rw-r--r-- | src/HTTP/HTTPMessageParser.cpp | 222 | ||||
-rw-r--r-- | src/HTTP/HTTPMessageParser.h | 125 | ||||
-rw-r--r-- | src/HTTP/HTTPServer.cpp | 212 | ||||
-rw-r--r-- | src/HTTP/HTTPServer.h | 101 | ||||
-rw-r--r-- | src/HTTP/HTTPServerConnection.cpp | 240 | ||||
-rw-r--r-- | src/HTTP/HTTPServerConnection.h | 118 | ||||
-rw-r--r-- | src/HTTP/MultipartParser.cpp | 254 | ||||
-rw-r--r-- | src/HTTP/MultipartParser.h | 79 | ||||
-rw-r--r-- | src/HTTP/NameValueParser.cpp | 413 | ||||
-rw-r--r-- | src/HTTP/NameValueParser.h | 70 | ||||
-rw-r--r-- | src/HTTP/SslHTTPServerConnection.cpp | 115 | ||||
-rw-r--r-- | src/HTTP/SslHTTPServerConnection.h | 47 | ||||
-rw-r--r-- | src/HTTP/TransferEncodingParser.cpp | 394 | ||||
-rw-r--r-- | src/HTTP/TransferEncodingParser.h | 76 | ||||
-rw-r--r-- | src/HTTP/UrlParser.cpp | 200 | ||||
-rw-r--r-- | src/HTTP/UrlParser.h | 58 |
23 files changed, 3696 insertions, 0 deletions
diff --git a/src/HTTP/CMakeLists.txt b/src/HTTP/CMakeLists.txt new file mode 100644 index 000000000..acb3ac2cd --- /dev/null +++ b/src/HTTP/CMakeLists.txt @@ -0,0 +1,43 @@ + +cmake_minimum_required (VERSION 2.6) +project (Cuberite) + +include_directories ("${PROJECT_SOURCE_DIR}/../") + +SET (SRCS + EnvelopeParser.cpp + HTTPFormParser.cpp + HTTPMessage.cpp + HTTPMessageParser.cpp + HTTPServer.cpp + HTTPServerConnection.cpp + MultipartParser.cpp + NameValueParser.cpp + SslHTTPServerConnection.cpp + TransferEncodingParser.cpp + UrlParser.cpp +) + +SET (HDRS + EnvelopeParser.h + HTTPFormParser.h + HTTPMessage.h + HTTPMessageParser.h + HTTPServer.h + HTTPServerConnection.h + MultipartParser.h + NameValueParser.h + SslHTTPServerConnection.h + TransferEncodingParser.h + UrlParser.h +) + +if("${CMAKE_CXX_COMPILER_ID}" STREQUAL "Clang") + set_source_files_properties(HTTPServer.cpp PROPERTIES COMPILE_FLAGS "-Wno-error=global-constructors ") + set_source_files_properties(HTTPServerConnection.cpp PROPERTIES COMPILE_FLAGS "-Wno-error=switch-enum") + set_source_files_properties(HTTPMessage.cpp PROPERTIES COMPILE_FLAGS "-Wno-error=tautological-compare") +endif() + +if(NOT MSVC) + add_library(HTTPServer ${SRCS} ${HDRS}) +endif() diff --git a/src/HTTP/EnvelopeParser.cpp b/src/HTTP/EnvelopeParser.cpp new file mode 100644 index 000000000..1c49b643f --- /dev/null +++ b/src/HTTP/EnvelopeParser.cpp @@ -0,0 +1,132 @@ + +// EnvelopeParser.cpp + +// Implements the cEnvelopeParser class representing a parser for RFC-822 envelope headers, used both in HTTP and in MIME + +#include "Globals.h" +#include "EnvelopeParser.h" + + + + + +cEnvelopeParser::cEnvelopeParser(cCallbacks & a_Callbacks) : + m_Callbacks(a_Callbacks), + m_IsInHeaders(true) +{ +} + + + + + +size_t cEnvelopeParser::Parse(const char * a_Data, size_t a_Size) +{ + if (!m_IsInHeaders) + { + return 0; + } + + // Start searching 1 char from the end of the already received data, if available: + auto searchStart = m_IncomingData.size(); + searchStart = (searchStart > 1) ? searchStart - 1 : 0; + + m_IncomingData.append(a_Data, a_Size); + + size_t idxCRLF = m_IncomingData.find("\r\n", searchStart); + if (idxCRLF == AString::npos) + { + // Not a complete line yet, all input consumed: + return a_Size; + } + + // Parse as many lines as found: + size_t Last = 0; + do + { + if (idxCRLF == Last) + { + // This was the last line of the data. Finish whatever value has been cached and return: + NotifyLast(); + m_IsInHeaders = false; + return a_Size - (m_IncomingData.size() - idxCRLF) + 2; + } + if (!ParseLine(m_IncomingData.c_str() + Last, idxCRLF - Last)) + { + // An error has occurred + m_IsInHeaders = false; + return AString::npos; + } + Last = idxCRLF + 2; + idxCRLF = m_IncomingData.find("\r\n", idxCRLF + 2); + } while (idxCRLF != AString::npos); + m_IncomingData.erase(0, Last); + + // Parsed all lines and still expecting more + return a_Size; +} + + + + + +void cEnvelopeParser::Reset(void) +{ + m_IsInHeaders = true; + m_IncomingData.clear(); + m_LastKey.clear(); + m_LastValue.clear(); +} + + + + + +void cEnvelopeParser::NotifyLast(void) +{ + if (!m_LastKey.empty()) + { + m_Callbacks.OnHeaderLine(m_LastKey, m_LastValue); + m_LastKey.clear(); + } + m_LastValue.clear(); +} + + + + + +bool cEnvelopeParser::ParseLine(const char * a_Data, size_t a_Size) +{ + ASSERT(a_Size > 0); + if (a_Data[0] <= ' ') + { + // This line is a continuation for the previous line + if (m_LastKey.empty()) + { + return false; + } + // Append, including the whitespace in a_Data[0] + m_LastValue.append(a_Data, a_Size); + return true; + } + + // This is a line with a new key: + NotifyLast(); + for (size_t i = 0; i < a_Size; i++) + { + if (a_Data[i] == ':') + { + m_LastKey.assign(a_Data, i); + m_LastValue.assign(a_Data + i + 2, a_Size - i - 2); + return true; + } + } // for i - a_Data[] + + // No colon was found, key-less header?? + return false; +} + + + + diff --git a/src/HTTP/EnvelopeParser.h b/src/HTTP/EnvelopeParser.h new file mode 100644 index 000000000..2fa930539 --- /dev/null +++ b/src/HTTP/EnvelopeParser.h @@ -0,0 +1,73 @@ + +// EnvelopeParser.h + +// Declares the cEnvelopeParser class representing a parser for RFC-822 envelope headers, used both in HTTP and in MIME + + + + + +#pragma once + + + + + +class cEnvelopeParser +{ +public: + class cCallbacks + { + public: + // Force a virtual destructor in descendants: + virtual ~cCallbacks() {} + + /** Called when a full header line is parsed */ + virtual void OnHeaderLine(const AString & a_Key, const AString & a_Value) = 0; + } ; + + + cEnvelopeParser(cCallbacks & a_Callbacks); + + /** Parses the incoming data. + Returns the number of bytes consumed from the input. The bytes not consumed are not part of the envelope header. + Returns AString::npos on error + */ + size_t Parse(const char * a_Data, size_t a_Size); + + /** Makes the parser forget everything parsed so far, so that it can be reused for parsing another datastream */ + void Reset(void); + + /** Returns true if more input is expected for the envelope header */ + bool IsInHeaders(void) const { return m_IsInHeaders; } + + /** Sets the IsInHeaders flag; used by cMultipartParser to simplify the parser initial conditions */ + void SetIsInHeaders(bool a_IsInHeaders) { m_IsInHeaders = a_IsInHeaders; } + +public: + /** Callbacks to call for the various events */ + cCallbacks & m_Callbacks; + + /** Set to true while the parser is still parsing the envelope headers. Once set to true, the parser will not consume any more data. */ + bool m_IsInHeaders; + + /** Buffer for the incoming data until it is parsed */ + AString m_IncomingData; + + /** Holds the last parsed key; used for line-wrapped values */ + AString m_LastKey; + + /** Holds the last parsed value; used for line-wrapped values */ + AString m_LastValue; + + + /** Notifies the callback of the key / value stored in m_LastKey / m_LastValue, then erases them */ + void NotifyLast(void); + + /** Parses one line of header data. Returns true if successful */ + bool ParseLine(const char * a_Data, size_t a_Size); +} ; + + + + diff --git a/src/HTTP/HTTPFormParser.cpp b/src/HTTP/HTTPFormParser.cpp new file mode 100644 index 000000000..ea5da3c18 --- /dev/null +++ b/src/HTTP/HTTPFormParser.cpp @@ -0,0 +1,293 @@ + +// HTTPFormParser.cpp + +// Implements the cHTTPFormParser class representing a parser for forms sent over HTTP + +#include "Globals.h" +#include "HTTPFormParser.h" +#include "HTTPMessage.h" +#include "MultipartParser.h" +#include "NameValueParser.h" + + + + + +cHTTPFormParser::cHTTPFormParser(const cHTTPIncomingRequest & a_Request, cCallbacks & a_Callbacks) : + m_Callbacks(a_Callbacks), + m_IsValid(true), + m_IsCurrentPartFile(false), + m_FileHasBeenAnnounced(false) +{ + if (a_Request.GetMethod() == "GET") + { + m_Kind = fpkURL; + + // Directly parse the URL in the request: + const AString & URL = a_Request.GetURL(); + size_t idxQM = URL.find('?'); + if (idxQM != AString::npos) + { + Parse(URL.c_str() + idxQM + 1, URL.size() - idxQM - 1); + } + return; + } + if ((a_Request.GetMethod() == "POST") || (a_Request.GetMethod() == "PUT")) + { + if (strncmp(a_Request.GetContentType().c_str(), "application/x-www-form-urlencoded", 33) == 0) + { + m_Kind = fpkFormUrlEncoded; + return; + } + if (strncmp(a_Request.GetContentType().c_str(), "multipart/form-data", 19) == 0) + { + m_Kind = fpkMultipart; + BeginMultipart(a_Request); + return; + } + } + // Invalid method / content type combination, this is not a HTTP form + m_IsValid = false; +} + + + + + +cHTTPFormParser::cHTTPFormParser(eKind a_Kind, const char * a_Data, size_t a_Size, cCallbacks & a_Callbacks) : + m_Callbacks(a_Callbacks), + m_Kind(a_Kind), + m_IsValid(true), + m_IsCurrentPartFile(false), + m_FileHasBeenAnnounced(false) +{ + Parse(a_Data, a_Size); +} + + + + + +void cHTTPFormParser::Parse(const char * a_Data, size_t a_Size) +{ + if (!m_IsValid) + { + return; + } + + switch (m_Kind) + { + case fpkURL: + case fpkFormUrlEncoded: + { + // This format is used for smaller forms (not file uploads), so we can delay parsing it until Finish() + m_IncomingData.append(a_Data, a_Size); + break; + } + case fpkMultipart: + { + ASSERT(m_MultipartParser.get() != nullptr); + m_MultipartParser->Parse(a_Data, a_Size); + break; + } + } +} + + + + + +bool cHTTPFormParser::Finish(void) +{ + switch (m_Kind) + { + case fpkURL: + case fpkFormUrlEncoded: + { + // m_IncomingData has all the form data, parse it now: + ParseFormUrlEncoded(); + break; + } + case fpkMultipart: + { + // Nothing needed for other formats + break; + } + } + return (m_IsValid && m_IncomingData.empty()); +} + + + + + +bool cHTTPFormParser::HasFormData(const cHTTPIncomingRequest & a_Request) +{ + const AString & ContentType = a_Request.GetContentType(); + return ( + (ContentType == "application/x-www-form-urlencoded") || + (strncmp(ContentType.c_str(), "multipart/form-data", 19) == 0) || + ( + (a_Request.GetMethod() == "GET") && + (a_Request.GetURL().find('?') != AString::npos) + ) + ); +} + + + + + +void cHTTPFormParser::BeginMultipart(const cHTTPIncomingRequest & a_Request) +{ + ASSERT(m_MultipartParser.get() == nullptr); + m_MultipartParser.reset(new cMultipartParser(a_Request.GetContentType(), *this)); +} + + + + + +void cHTTPFormParser::ParseFormUrlEncoded(void) +{ + // Parse m_IncomingData for all the variables; no more data is incoming, since this is called from Finish() + // This may not be the most performant version, but we don't care, the form data is small enough and we're not a full-fledged web server anyway + AStringVector Lines = StringSplit(m_IncomingData, "&"); + for (AStringVector::iterator itr = Lines.begin(), end = Lines.end(); itr != end; ++itr) + { + AStringVector Components = StringSplit(*itr, "="); + switch (Components.size()) + { + default: + { + // Neither name nor value, or too many "="s, mark this as invalid form: + m_IsValid = false; + return; + } + case 1: + { + // Only name present + (*this)[URLDecode(ReplaceAllCharOccurrences(Components[0], '+', ' '))] = ""; + break; + } + case 2: + { + // name=value format: + (*this)[URLDecode(ReplaceAllCharOccurrences(Components[0], '+', ' '))] = URLDecode(ReplaceAllCharOccurrences(Components[1], '+', ' ')); + break; + } + } + } // for itr - Lines[] + m_IncomingData.clear(); +} + + + + + +void cHTTPFormParser::OnPartStart(void) +{ + m_CurrentPartFileName.clear(); + m_CurrentPartName.clear(); + m_IsCurrentPartFile = false; + m_FileHasBeenAnnounced = false; +} + + + + + +void cHTTPFormParser::OnPartHeader(const AString & a_Key, const AString & a_Value) +{ + if (NoCaseCompare(a_Key, "Content-Disposition") == 0) + { + size_t len = a_Value.size(); + size_t ParamsStart = AString::npos; + for (size_t i = 0; i < len; ++i) + { + if (a_Value[i] > ' ') + { + if (strncmp(a_Value.c_str() + i, "form-data", 9) != 0) + { + // Content disposition is not "form-data", mark the whole form invalid + m_IsValid = false; + return; + } + ParamsStart = a_Value.find(';', i + 9); + break; + } + } + if (ParamsStart == AString::npos) + { + // There is data missing in the Content-Disposition field, mark the whole form invalid: + m_IsValid = false; + return; + } + + // Parse the field name and optional filename from this header: + cNameValueParser Parser(a_Value.data() + ParamsStart, a_Value.size() - ParamsStart); + Parser.Finish(); + m_CurrentPartName = Parser["name"]; + if (!Parser.IsValid() || m_CurrentPartName.empty()) + { + // The required parameter "name" is missing, mark the whole form invalid: + m_IsValid = false; + return; + } + m_CurrentPartFileName = Parser["filename"]; + } +} + + + + + +void cHTTPFormParser::OnPartData(const char * a_Data, size_t a_Size) +{ + if (m_CurrentPartName.empty()) + { + // Prologue, epilogue or invalid part + return; + } + if (m_CurrentPartFileName.empty()) + { + // This is a variable, store it in the map + iterator itr = find(m_CurrentPartName); + if (itr == end()) + { + (*this)[m_CurrentPartName] = AString(a_Data, a_Size); + } + else + { + itr->second.append(a_Data, a_Size); + } + } + else + { + // This is a file, pass it on through the callbacks + if (!m_FileHasBeenAnnounced) + { + m_Callbacks.OnFileStart(*this, m_CurrentPartFileName); + m_FileHasBeenAnnounced = true; + } + m_Callbacks.OnFileData(*this, a_Data, a_Size); + } +} + + + + + +void cHTTPFormParser::OnPartEnd(void) +{ + if (m_FileHasBeenAnnounced) + { + m_Callbacks.OnFileEnd(*this); + } + m_CurrentPartName.clear(); + m_CurrentPartFileName.clear(); +} + + + + diff --git a/src/HTTP/HTTPFormParser.h b/src/HTTP/HTTPFormParser.h new file mode 100644 index 000000000..6bf3e7d78 --- /dev/null +++ b/src/HTTP/HTTPFormParser.h @@ -0,0 +1,114 @@ + +// HTTPFormParser.h + +// Declares the cHTTPFormParser class representing a parser for forms sent over HTTP + + + + +#pragma once + +#include "MultipartParser.h" + + + + + +// fwd: +class cHTTPIncomingRequest; + + + + + +class cHTTPFormParser : + public std::map<AString, AString>, + public cMultipartParser::cCallbacks +{ +public: + enum eKind + { + fpkURL, ///< The form has been transmitted as parameters to a GET request + fpkFormUrlEncoded, ///< The form has been POSTed or PUT, with Content-Type of "application/x-www-form-urlencoded" + fpkMultipart, ///< The form has been POSTed or PUT, with Content-Type of "multipart/form-data" + } ; + + class cCallbacks + { + public: + // Force a virtual destructor in descendants: + virtual ~cCallbacks() {} + + /** Called when a new file part is encountered in the form data */ + virtual void OnFileStart(cHTTPFormParser & a_Parser, const AString & a_FileName) = 0; + + /** Called when more file data has come for the current file in the form data */ + virtual void OnFileData(cHTTPFormParser & a_Parser, const char * a_Data, size_t a_Size) = 0; + + /** Called when the current file part has ended in the form data */ + virtual void OnFileEnd(cHTTPFormParser & a_Parser) = 0; + } ; + + + /** Creates a parser that is tied to a request and notifies of various events using a callback mechanism */ + cHTTPFormParser(const cHTTPIncomingRequest & a_Request, cCallbacks & a_Callbacks); + + /** Creates a parser with the specified content type that reads data from a string */ + cHTTPFormParser(eKind a_Kind, const char * a_Data, size_t a_Size, cCallbacks & a_Callbacks); + + /** Adds more data into the parser, as the request body is received */ + void Parse(const char * a_Data, size_t a_Size); + + /** Notifies that there's no more data incoming and the parser should finish its parsing. + Returns true if parsing successful. */ + bool Finish(void); + + /** Returns true if the headers suggest the request has form data parseable by this class */ + static bool HasFormData(const cHTTPIncomingRequest & a_Request); + +protected: + + /** The callbacks to call for incoming file data */ + cCallbacks & m_Callbacks; + + /** The kind of the parser (decided in the constructor, used in Parse() */ + eKind m_Kind; + + /** Buffer for the incoming data until it's parsed */ + AString m_IncomingData; + + /** True if the information received so far is a valid form; set to false on first problem. Further parsing is skipped when false. */ + bool m_IsValid; + + /** The parser for the multipart data, if used */ + std::unique_ptr<cMultipartParser> m_MultipartParser; + + /** Name of the currently parsed part in multipart data */ + AString m_CurrentPartName; + + /** True if the currently parsed part in multipart data is a file */ + bool m_IsCurrentPartFile; + + /** Filename of the current parsed part in multipart data (for file uploads) */ + AString m_CurrentPartFileName; + + /** Set to true after m_Callbacks.OnFileStart() has been called, reset to false on PartEnd */ + bool m_FileHasBeenAnnounced; + + + /** Sets up the object for parsing a fpkMultipart request */ + void BeginMultipart(const cHTTPIncomingRequest & a_Request); + + /** Parses m_IncomingData as form-urlencoded data (fpkURL or fpkFormUrlEncoded kinds) */ + void ParseFormUrlEncoded(void); + + // cMultipartParser::cCallbacks overrides: + virtual void OnPartStart (void) override; + virtual void OnPartHeader(const AString & a_Key, const AString & a_Value) override; + virtual void OnPartData (const char * a_Data, size_t a_Size) override; + virtual void OnPartEnd (void) override; +} ; + + + + diff --git a/src/HTTP/HTTPMessage.cpp b/src/HTTP/HTTPMessage.cpp new file mode 100644 index 000000000..cf7525b6c --- /dev/null +++ b/src/HTTP/HTTPMessage.cpp @@ -0,0 +1,159 @@ + +// HTTPMessage.cpp + +// Declares the cHTTPMessage class representing the common ancestor for HTTP request and response classes + +#include "Globals.h" +#include "HTTPMessage.h" + + + + + +// Disable MSVC warnings: +#if defined(_MSC_VER) + #pragma warning(push) + #pragma warning(disable:4355) // 'this' : used in base member initializer list +#endif + + + + + +//////////////////////////////////////////////////////////////////////////////// +// cHTTPMessage: + +cHTTPMessage::cHTTPMessage(eKind a_Kind) : + m_Kind(a_Kind), + m_ContentLength(AString::npos) +{ +} + + + + + +void cHTTPMessage::AddHeader(const AString & a_Key, const AString & a_Value) +{ + auto Key = StrToLower(a_Key); + auto itr = m_Headers.find(Key); + if (itr == m_Headers.end()) + { + m_Headers[Key] = a_Value; + } + else + { + // The header-field key is specified multiple times, combine into comma-separated list (RFC 2616 @ 4.2) + itr->second.append(", "); + itr->second.append(a_Value); + } + + // Special processing for well-known headers: + if (Key == "content-type") + { + m_ContentType = m_Headers[Key]; + } + else if (Key == "content-length") + { + if (!StringToInteger(m_Headers[Key], m_ContentLength)) + { + m_ContentLength = 0; + } + } +} + + + + + +//////////////////////////////////////////////////////////////////////////////// +// cHTTPOutgoingResponse: + +cHTTPOutgoingResponse::cHTTPOutgoingResponse(void) : + super(mkResponse) +{ +} + + + + + +void cHTTPOutgoingResponse::AppendToData(AString & a_DataStream) const +{ + a_DataStream.append("HTTP/1.1 200 OK\r\nTransfer-Encoding: chunked\r\nContent-Type: "); + a_DataStream.append(m_ContentType); + a_DataStream.append("\r\n"); + for (auto itr = m_Headers.cbegin(), end = m_Headers.cend(); itr != end; ++itr) + { + if ((itr->first == "Content-Type") || (itr->first == "Content-Length")) + { + continue; + } + a_DataStream.append(itr->first); + a_DataStream.append(": "); + a_DataStream.append(itr->second); + a_DataStream.append("\r\n"); + } // for itr - m_Headers[] + a_DataStream.append("\r\n"); +} + + + + + +//////////////////////////////////////////////////////////////////////////////// +// cHTTPIncomingRequest: + +cHTTPIncomingRequest::cHTTPIncomingRequest(const AString & a_Method, const AString & a_URL): + Super(mkRequest), + m_Method(a_Method), + m_URL(a_URL) +{ +} + + + + + +AString cHTTPIncomingRequest::GetURLPath(void) const +{ + auto idxQuestionMark = m_URL.find('?'); + if (idxQuestionMark == AString::npos) + { + return m_URL; + } + else + { + return m_URL.substr(0, idxQuestionMark); + } +} + + + + + +void cHTTPIncomingRequest::AddHeader(const AString & a_Key, const AString & a_Value) +{ + if ( + (NoCaseCompare(a_Key, "Authorization") == 0) && + (strncmp(a_Value.c_str(), "Basic ", 6) == 0) + ) + { + AString UserPass = Base64Decode(a_Value.substr(6)); + size_t idxCol = UserPass.find(':'); + if (idxCol != AString::npos) + { + m_AuthUsername = UserPass.substr(0, idxCol); + m_AuthPassword = UserPass.substr(idxCol + 1); + m_HasAuth = true; + } + } + if ((a_Key == "Connection") && (NoCaseCompare(a_Value, "keep-alive") == 0)) + { + m_AllowKeepAlive = true; + } +} + + + + diff --git a/src/HTTP/HTTPMessage.h b/src/HTTP/HTTPMessage.h new file mode 100644 index 000000000..9afcd5b97 --- /dev/null +++ b/src/HTTP/HTTPMessage.h @@ -0,0 +1,158 @@ + +// HTTPMessage.h + +// Declares the cHTTPMessage class representing the common ancestor for HTTP request and response classes + + + + + +#pragma once + +#include "EnvelopeParser.h" + + + + + +class cHTTPMessage +{ +public: + enum eStatus + { + HTTP_OK = 200, + HTTP_BAD_REQUEST = 400, + } ; + + enum eKind + { + mkRequest, + mkResponse, + } ; + + cHTTPMessage(eKind a_Kind); + + // Force a virtual destructor in all descendants + virtual ~cHTTPMessage() {} + + /** Adds a header into the internal map of headers. Recognizes special headers: Content-Type and Content-Length */ + virtual void AddHeader(const AString & a_Key, const AString & a_Value); + + void SetContentType (const AString & a_ContentType) { m_ContentType = a_ContentType; } + void SetContentLength(size_t a_ContentLength) { m_ContentLength = a_ContentLength; } + + const AString & GetContentType (void) const { return m_ContentType; } + size_t GetContentLength(void) const { return m_ContentLength; } + +protected: + typedef std::map<AString, AString> cNameValueMap; + + eKind m_Kind; + + /** Map of headers, with their keys lowercased. */ + AStringMap m_Headers; + + /** Type of the content; parsed by AddHeader(), set directly by SetContentLength() */ + AString m_ContentType; + + /** Length of the content that is to be received. + AString::npos when the object is created. + Parsed by AddHeader() or set directly by SetContentLength() */ + size_t m_ContentLength; +} ; + + + + + +/** Stores outgoing response headers and serializes them to an HTTP data stream. */ +class cHTTPOutgoingResponse : + public cHTTPMessage +{ + typedef cHTTPMessage super; + +public: + cHTTPOutgoingResponse(void); + + /** Appends the response to the specified datastream - response line and headers. + The body will be sent later directly through cConnection::Send() */ + void AppendToData(AString & a_DataStream) const; +} ; + + + + + +/** Provides storage for an incoming HTTP request. */ +class cHTTPIncomingRequest: + public cHTTPMessage +{ + typedef cHTTPMessage Super; +public: + /** Base class for anything that can be used as the UserData for the request. */ + class cUserData + { + public: + // Force a virtual destructor in descendants: + virtual ~cUserData() {} + }; + typedef SharedPtr<cUserData> cUserDataPtr; + + + /** Creates a new instance of the class, containing the method and URL provided by the client. */ + cHTTPIncomingRequest(const AString & a_Method, const AString & a_URL); + + /** Returns the method used in the request */ + const AString & GetMethod(void) const { return m_Method; } + + /** Returns the URL used in the request */ + const AString & GetURL(void) const { return m_URL; } + + /** Returns the path part of the URL. */ + AString GetURLPath(void) const; + + /** Returns true if the request has had the Auth header present. */ + bool HasAuth(void) const { return m_HasAuth; } + + /** Returns the username that the request presented. Only valid if HasAuth() is true */ + const AString & GetAuthUsername(void) const { return m_AuthUsername; } + + /** Returns the password that the request presented. Only valid if HasAuth() is true */ + const AString & GetAuthPassword(void) const { return m_AuthPassword; } + + bool DoesAllowKeepAlive(void) const { return m_AllowKeepAlive; } + + /** Attaches any kind of data to this request, to be later retrieved by GetUserData(). */ + void SetUserData(cUserDataPtr a_UserData) { m_UserData = a_UserData; } + + /** Returns the data attached to this request by the class client. */ + cUserDataPtr GetUserData(void) { return m_UserData; } + + /** Adds the specified header into the internal list of headers. + Overrides the parent to add recognizing additional headers: auth and keepalive. */ + virtual void AddHeader(const AString & a_Key, const AString & a_Value) override; + +protected: + + /** Method of the request (GET / PUT / POST / ...) */ + AString m_Method; + + /** Full URL of the request */ + AString m_URL; + + /** Set to true if the request contains auth data that was understood by the parser */ + bool m_HasAuth; + + /** The username used for auth */ + AString m_AuthUsername; + + /** The password used for auth */ + AString m_AuthPassword; + + /** Set to true if the request indicated that it supports keepalives. + If false, the server will close the connection once the request is finished */ + bool m_AllowKeepAlive; + + /** Any data attached to the request by the class client. */ + cUserDataPtr m_UserData; +}; diff --git a/src/HTTP/HTTPMessageParser.cpp b/src/HTTP/HTTPMessageParser.cpp new file mode 100644 index 000000000..10d3d4ad9 --- /dev/null +++ b/src/HTTP/HTTPMessageParser.cpp @@ -0,0 +1,222 @@ + +// HTTPMessageParser.cpp + +// Implements the cHTTPMessageParser class that parses HTTP messages (request or response) being pushed into the parser, +// and reports the individual parts via callbacks + +#include "Globals.h" +#include "HTTPMessageParser.h" + + + + + +cHTTPMessageParser::cHTTPMessageParser(cHTTPMessageParser::cCallbacks & a_Callbacks): + m_Callbacks(a_Callbacks), + m_EnvelopeParser(*this) +{ + Reset(); +} + + + + + +size_t cHTTPMessageParser::Parse(const char * a_Data, size_t a_Size) +{ + // If parsing already finished or errorred, let the caller keep all the data: + if (m_IsFinished || m_HasHadError) + { + return 0; + } + + // If still waiting for the status line, add to buffer and try parsing it: + auto inBufferSoFar = m_Buffer.size(); + if (m_FirstLine.empty()) + { + m_Buffer.append(a_Data, a_Size); + auto bytesConsumedFirstLine = ParseFirstLine(); + ASSERT(bytesConsumedFirstLine <= inBufferSoFar + a_Size); // Haven't consumed more data than there is in the buffer + ASSERT(bytesConsumedFirstLine > inBufferSoFar); // Have consumed at least the previous buffer contents + if (m_FirstLine.empty()) + { + // All data used, but not a complete status line yet. + return a_Size; + } + if (m_HasHadError) + { + return AString::npos; + } + // Status line completed, feed the rest of the buffer into the envelope parser: + auto bytesConsumedEnvelope = m_EnvelopeParser.Parse(m_Buffer.data(), m_Buffer.size()); + if (bytesConsumedEnvelope == AString::npos) + { + m_HasHadError = true; + m_Callbacks.OnError("Failed to parse the envelope"); + return AString::npos; + } + ASSERT(bytesConsumedEnvelope <= bytesConsumedFirstLine + a_Size); // Haven't consumed more data than there was in the buffer + m_Buffer.erase(0, bytesConsumedEnvelope); + if (!m_EnvelopeParser.IsInHeaders()) + { + HeadersFinished(); + // Process any data still left in the buffer as message body: + auto bytesConsumedBody = ParseBody(m_Buffer.data(), m_Buffer.size()); + if (bytesConsumedBody == AString::npos) + { + // Error has already been reported by ParseBody, just bail out: + return AString::npos; + } + return bytesConsumedBody + bytesConsumedEnvelope + bytesConsumedFirstLine - inBufferSoFar; + } + return a_Size; + } // if (m_FirstLine.empty()) + + // If still parsing headers, send them to the envelope parser: + if (m_EnvelopeParser.IsInHeaders()) + { + auto bytesConsumed = m_EnvelopeParser.Parse(a_Data, a_Size); + if (bytesConsumed == AString::npos) + { + m_HasHadError = true; + m_Callbacks.OnError("Failed to parse the envelope"); + return AString::npos; + } + if (!m_EnvelopeParser.IsInHeaders()) + { + HeadersFinished(); + // Process any data still left as message body: + auto bytesConsumedBody = ParseBody(a_Data + bytesConsumed, a_Size - bytesConsumed); + if (bytesConsumedBody == AString::npos) + { + // Error has already been reported by ParseBody, just bail out: + return AString::npos; + } + } + return a_Size; + } + + // Already parsing the body + return ParseBody(a_Data, a_Size); +} + + + + + +void cHTTPMessageParser::Reset(void) +{ + m_HasHadError = false; + m_IsFinished = false; + m_FirstLine.clear(); + m_Buffer.clear(); + m_EnvelopeParser.Reset(); + m_TransferEncodingParser.reset(); + m_TransferEncoding.clear(); + m_ContentLength = 0; +} + + + + +size_t cHTTPMessageParser::ParseFirstLine(void) +{ + auto idxLineEnd = m_Buffer.find("\r\n"); + if (idxLineEnd == AString::npos) + { + // Not a complete line yet + return m_Buffer.size(); + } + m_FirstLine = m_Buffer.substr(0, idxLineEnd); + m_Buffer.erase(0, idxLineEnd + 2); + m_Callbacks.OnFirstLine(m_FirstLine); + return idxLineEnd + 2; +} + + + + +size_t cHTTPMessageParser::ParseBody(const char * a_Data, size_t a_Size) +{ + if (m_TransferEncodingParser == nullptr) + { + // We have no Transfer-encoding parser assigned. This should have happened when finishing the envelope + OnError("No transfer encoding parser"); + return AString::npos; + } + + // Parse the body using the transfer encoding parser: + // (Note that TE parser returns the number of bytes left, while we return the number of bytes consumed) + return a_Size - m_TransferEncodingParser->Parse(a_Data, a_Size); +} + + + + + +void cHTTPMessageParser::HeadersFinished(void) +{ + m_Callbacks.OnHeadersFinished(); + m_TransferEncodingParser = cTransferEncodingParser::Create(*this, m_TransferEncoding, m_ContentLength); + if (m_TransferEncodingParser == nullptr) + { + OnError(Printf("Unknown transfer encoding: %s", m_TransferEncoding.c_str())); + return; + } +} + + + + + +void cHTTPMessageParser::OnHeaderLine(const AString & a_Key, const AString & a_Value) +{ + m_Callbacks.OnHeaderLine(a_Key, a_Value); + auto Key = StrToLower(a_Key); + if (Key == "content-length") + { + if (!StringToInteger(a_Value, m_ContentLength)) + { + OnError(Printf("Invalid content length header value: \"%s\"", a_Value.c_str())); + } + return; + } + if (Key == "transfer-encoding") + { + m_TransferEncoding = a_Value; + return; + } +} + + + + + +void cHTTPMessageParser::OnError(const AString & a_ErrorDescription) +{ + m_HasHadError = true; + m_Callbacks.OnError(a_ErrorDescription); +} + + + + + +void cHTTPMessageParser::OnBodyData(const void * a_Data, size_t a_Size) +{ + m_Callbacks.OnBodyData(a_Data, a_Size); +} + + + + + +void cHTTPMessageParser::OnBodyFinished(void) +{ + m_IsFinished = true; + m_Callbacks.OnBodyFinished(); +} + + + + diff --git a/src/HTTP/HTTPMessageParser.h b/src/HTTP/HTTPMessageParser.h new file mode 100644 index 000000000..f07de0492 --- /dev/null +++ b/src/HTTP/HTTPMessageParser.h @@ -0,0 +1,125 @@ + +// HTTPMessageParser.h + +// Declares the cHTTPMessageParser class that parses HTTP messages (request or response) being pushed into the parser, +// and reports the individual parts via callbacks + + + + + +#pragma once + +#include "EnvelopeParser.h" +#include "TransferEncodingParser.h" + + + + + +class cHTTPMessageParser: + protected cEnvelopeParser::cCallbacks, + protected cTransferEncodingParser::cCallbacks +{ +public: + class cCallbacks + { + public: + // Force a virtual destructor in descendants: + virtual ~cCallbacks() {} + + /** Called when an error has occured while parsing. */ + virtual void OnError(const AString & a_ErrorDescription) = 0; + + /** Called when the first line (request / status) is fully parsed. */ + virtual void OnFirstLine(const AString & a_FirstLine) = 0; + + /** Called when a single header line is parsed. */ + virtual void OnHeaderLine(const AString & a_Key, const AString & a_Value) = 0; + + /** Called when all the headers have been parsed. */ + virtual void OnHeadersFinished(void) = 0; + + /** Called for each chunk of the incoming body data. */ + virtual void OnBodyData(const void * a_Data, size_t a_Size) = 0; + + /** Called when the entire body has been reported by OnBodyData(). */ + virtual void OnBodyFinished(void) = 0; + }; + + /** Creates a new parser instance that will use the specified callbacks for reporting. */ + cHTTPMessageParser(cCallbacks & a_Callbacks); + + /** Parses the incoming data and calls the appropriate callbacks. + Returns the number of bytes consumed or AString::npos number for error. */ + size_t Parse(const char * a_Data, size_t a_Size); + + /** Called when the server indicates no more data will be sent (HTTP 1.0 socket closed). + Finishes all parsing and calls apropriate callbacks (error if incomplete response). */ + void Finish(void); + + /** Returns true if the entire response has been already parsed. */ + bool IsFinished(void) const { return m_IsFinished; } + + /** Resets the parser to the initial state, so that a new request can be parsed. */ + void Reset(void); + + +protected: + + /** The callbacks used for reporting. */ + cCallbacks & m_Callbacks; + + /** Set to true if an error has been encountered by the parser. */ + bool m_HasHadError; + + /** True if the response has been fully parsed. */ + bool m_IsFinished; + + /** The complete first line of the response. Empty if not parsed yet. */ + AString m_FirstLine; + + /** Buffer for the incoming data until the status line is parsed. */ + AString m_Buffer; + + /** Parser for the envelope data (headers) */ + cEnvelopeParser m_EnvelopeParser; + + /** The specific parser for the transfer encoding used by this response. */ + cTransferEncodingParserPtr m_TransferEncodingParser; + + /** The transfer encoding to be used by the parser. + Filled while parsing headers, used when headers are finished. */ + AString m_TransferEncoding; + + /** The content length, parsed from the headers, if available. + Unused for chunked encoding. + Filled while parsing headers, used when headers are finished. */ + size_t m_ContentLength; + + + /** Parses the first line out of m_Buffer. + Removes the first line from m_Buffer, if appropriate. + Returns the number of bytes consumed out of m_Buffer, or AString::npos number for error. */ + size_t ParseFirstLine(void); + + /** Parses the message body. + Processes transfer encoding and calls the callbacks for body data. + Returns the number of bytes consumed or AString::npos number for error. */ + size_t ParseBody(const char * a_Data, size_t a_Size); + + /** Called internally when the headers-parsing has just finished. */ + void HeadersFinished(void); + + // cEnvelopeParser::cCallbacks overrides: + virtual void OnHeaderLine(const AString & a_Key, const AString & a_Value) override; + + // cTransferEncodingParser::cCallbacks overrides: + virtual void OnError(const AString & a_ErrorDescription) override; + virtual void OnBodyData(const void * a_Data, size_t a_Size) override; + virtual void OnBodyFinished(void) override; +}; + + + + diff --git a/src/HTTP/HTTPServer.cpp b/src/HTTP/HTTPServer.cpp new file mode 100644 index 000000000..5a5bee045 --- /dev/null +++ b/src/HTTP/HTTPServer.cpp @@ -0,0 +1,212 @@ + +// HTTPServer.cpp + +// Implements the cHTTPServer class representing a HTTP webserver that uses cListenThread and cSocketThreads for processing + +#include "Globals.h" +#include "HTTPServer.h" +#include "HTTPMessageParser.h" +#include "HTTPServerConnection.h" +#include "HTTPFormParser.h" +#include "SslHTTPServerConnection.h" + + + + + +// Disable MSVC warnings: +#if defined(_MSC_VER) + #pragma warning(push) + #pragma warning(disable:4355) // 'this' : used in base member initializer list +#endif + + + + + +//////////////////////////////////////////////////////////////////////////////// +// cHTTPServerListenCallbacks: + +class cHTTPServerListenCallbacks: + public cNetwork::cListenCallbacks +{ +public: + cHTTPServerListenCallbacks(cHTTPServer & a_HTTPServer, UInt16 a_Port): + m_HTTPServer(a_HTTPServer), + m_Port(a_Port) + { + } + +protected: + /** The HTTP server instance that we're attached to. */ + cHTTPServer & m_HTTPServer; + + /** The port for which this instance is responsible. */ + UInt16 m_Port; + + // cNetwork::cListenCallbacks overrides: + virtual cTCPLink::cCallbacksPtr OnIncomingConnection(const AString & a_RemoteIPAddress, UInt16 a_RemotePort) override + { + return m_HTTPServer.OnIncomingConnection(a_RemoteIPAddress, a_RemotePort); + } + virtual void OnAccepted(cTCPLink & a_Link) override {} + virtual void OnError(int a_ErrorCode, const AString & a_ErrorMsg) override + { + LOGWARNING("HTTP server error on port %d: %d (%s)", m_Port, a_ErrorCode, a_ErrorMsg.c_str()); + } +}; + + + + + +//////////////////////////////////////////////////////////////////////////////// +// cHTTPServer: + +cHTTPServer::cHTTPServer(void) : + m_Callbacks(nullptr) +{ +} + + + + + +cHTTPServer::~cHTTPServer() +{ + Stop(); +} + + + + + +bool cHTTPServer::Initialize(void) +{ + // Read the HTTPS cert + key: + AString CertFile = cFile::ReadWholeFile("webadmin/httpscert.crt"); + AString KeyFile = cFile::ReadWholeFile("webadmin/httpskey.pem"); + if (!CertFile.empty() && !KeyFile.empty()) + { + m_Cert.reset(new cX509Cert); + int res = m_Cert->Parse(CertFile.data(), CertFile.size()); + if (res == 0) + { + m_CertPrivKey.reset(new cCryptoKey); + int res2 = m_CertPrivKey->ParsePrivate(KeyFile.data(), KeyFile.size(), ""); + if (res2 != 0) + { + // Reading the private key failed, reset the cert: + LOGWARNING("WebServer: Cannot read HTTPS certificate private key: -0x%x", -res2); + m_Cert.reset(); + } + } + else + { + LOGWARNING("WebServer: Cannot read HTTPS certificate: -0x%x", -res); + } + } + + // Notify the admin about the HTTPS / HTTP status + if (m_Cert.get() == nullptr) + { + LOGWARNING("WebServer: The server is running in unsecured HTTP mode."); + LOGINFO("Put a valid HTTPS certificate in file 'webadmin/httpscert.crt' and its corresponding private key to 'webadmin/httpskey.pem' (without any password) to enable HTTPS support"); + } + else + { + LOGINFO("WebServer: The server is running in secure HTTPS mode."); + } + return true; +} + + + + + +bool cHTTPServer::Start(cCallbacks & a_Callbacks, const AStringVector & a_Ports) +{ + m_Callbacks = &a_Callbacks; + + // Open up requested ports: + for (auto port : a_Ports) + { + UInt16 PortNum; + if (!StringToInteger(port, PortNum)) + { + LOGWARNING("WebServer: Invalid port value: \"%s\". Ignoring.", port.c_str()); + continue; + } + auto Handle = cNetwork::Listen(PortNum, std::make_shared<cHTTPServerListenCallbacks>(*this, PortNum)); + if (Handle->IsListening()) + { + m_ServerHandles.push_back(Handle); + } + } // for port - a_Ports[] + + // Report success if at least one port opened successfully: + return !m_ServerHandles.empty(); +} + + + + + +void cHTTPServer::Stop(void) +{ + for (auto handle : m_ServerHandles) + { + handle->Close(); + } + m_ServerHandles.clear(); +} + + + + + +cTCPLink::cCallbacksPtr cHTTPServer::OnIncomingConnection(const AString & a_RemoteIPAddress, UInt16 a_RemotePort) +{ + UNUSED(a_RemoteIPAddress); + UNUSED(a_RemotePort); + + if (m_Cert.get() != nullptr) + { + return std::make_shared<cSslHTTPServerConnection>(*this, m_Cert, m_CertPrivKey); + } + else + { + return std::make_shared<cHTTPServerConnection>(*this); + } +} + + + + + +void cHTTPServer::NewRequest(cHTTPServerConnection & a_Connection, cHTTPIncomingRequest & a_Request) +{ + m_Callbacks->OnRequestBegun(a_Connection, a_Request); +} + + + + + +void cHTTPServer::RequestBody(cHTTPServerConnection & a_Connection, cHTTPIncomingRequest & a_Request, const void * a_Data, size_t a_Size) +{ + m_Callbacks->OnRequestBody(a_Connection, a_Request, reinterpret_cast<const char *>(a_Data), a_Size); +} + + + + + +void cHTTPServer::RequestFinished(cHTTPServerConnection & a_Connection, cHTTPIncomingRequest & a_Request) +{ + m_Callbacks->OnRequestFinished(a_Connection, a_Request); +} + + + + diff --git a/src/HTTP/HTTPServer.h b/src/HTTP/HTTPServer.h new file mode 100644 index 000000000..1de0a6ce9 --- /dev/null +++ b/src/HTTP/HTTPServer.h @@ -0,0 +1,101 @@ + +// HTTPServer.h + +// Declares the cHTTPServer class representing a HTTP webserver that uses cListenThread and cSocketThreads for processing + + + + + +#pragma once + +#include "../OSSupport/Network.h" +#include "../IniFile.h" +#include "PolarSSL++/RsaPrivateKey.h" +#include "PolarSSL++/CryptoKey.h" +#include "PolarSSL++/X509Cert.h" + + + + + +// fwd: +class cHTTPMessage; +class cHTTPRequestParser; +class cHTTPIncomingRequest; +class cHTTPServerConnection; + + + + + + +class cHTTPServer +{ +public: + class cCallbacks + { + public: + virtual ~cCallbacks() {} + + /** Called when a new request arrives over a connection and all its headers have been parsed. + The request body needn't have arrived yet. */ + virtual void OnRequestBegun(cHTTPServerConnection & a_Connection, cHTTPIncomingRequest & a_Request) = 0; + + /** Called when another part of request body has arrived. + May be called multiple times for a single request. */ + virtual void OnRequestBody(cHTTPServerConnection & a_Connection, cHTTPIncomingRequest & a_Request, const char * a_Data, size_t a_Size) = 0; + + /** Called when the request body has been fully received in previous calls to OnRequestBody() */ + virtual void OnRequestFinished(cHTTPServerConnection & a_Connection, cHTTPIncomingRequest & a_Request) = 0; + } ; + + cHTTPServer(void); + virtual ~cHTTPServer(); + + /** Initializes the server - reads the cert files etc. */ + bool Initialize(void); + + /** Starts the server and assigns the callbacks to use for incoming requests */ + bool Start(cCallbacks & a_Callbacks, const AStringVector & a_Ports); + + /** Stops the server, drops all current connections */ + void Stop(void); + +protected: + friend class cHTTPServerConnection; + friend class cSslHTTPServerConnection; + friend class cHTTPServerListenCallbacks; + + /** The cNetwork API handle for the listening socket. */ + cServerHandlePtrs m_ServerHandles; + + /** The callbacks to call for various events */ + cCallbacks * m_Callbacks; + + /** The server certificate to use for the SSL connections */ + cX509CertPtr m_Cert; + + /** The private key for m_Cert. */ + cCryptoKeyPtr m_CertPrivKey; + + + /** Called by cHTTPServerListenCallbacks when there's a new incoming connection. + Returns the connection instance to be used as the cTCPLink callbacks. */ + cTCPLink::cCallbacksPtr OnIncomingConnection(const AString & a_RemoteIPAddress, UInt16 a_RemotePort); + + /** Called by cHTTPServerConnection when it finishes parsing the request header */ + void NewRequest(cHTTPServerConnection & a_Connection, cHTTPIncomingRequest & a_Request); + + /** Called by cHTTPConenction when it receives more data for the request body. + May be called multiple times for a single request. */ + void RequestBody(cHTTPServerConnection & a_Connection, cHTTPIncomingRequest & a_Request, const void * a_Data, size_t a_Size); + + /** Called by cHTTPServerConnection when it detects that the request has finished (all of its body has been received) */ + void RequestFinished(cHTTPServerConnection & a_Connection, cHTTPIncomingRequest & a_Request); +} ; + + + + + diff --git a/src/HTTP/HTTPServerConnection.cpp b/src/HTTP/HTTPServerConnection.cpp new file mode 100644 index 000000000..9edec2886 --- /dev/null +++ b/src/HTTP/HTTPServerConnection.cpp @@ -0,0 +1,240 @@ + +// HTTPConnection.cpp + +// Implements the cHTTPServerConnection class representing a single persistent connection in the HTTP server. + +#include "Globals.h" +#include "HTTPServerConnection.h" +#include "HTTPMessage.h" +#include "HTTPMessageParser.h" +#include "HTTPServer.h" + + + + + +cHTTPServerConnection::cHTTPServerConnection(cHTTPServer & a_HTTPServer) : + m_HTTPServer(a_HTTPServer), + m_Parser(*this), + m_CurrentRequest(nullptr) +{ + // LOGD("HTTP: New connection at %p", this); +} + + + + + +cHTTPServerConnection::~cHTTPServerConnection() +{ + // LOGD("HTTP: Connection deleting: %p", this); + m_CurrentRequest.reset(); +} + + + + + +void cHTTPServerConnection::SendStatusAndReason(int a_StatusCode, const AString & a_Response) +{ + SendData(Printf("HTTP/1.1 %d %s\r\n", a_StatusCode, a_Response.c_str())); + SendData(Printf("Content-Length: %u\r\n\r\n", static_cast<unsigned>(a_Response.size()))); + SendData(a_Response.data(), a_Response.size()); + m_CurrentRequest.reset(); + m_Parser.Reset(); +} + + + + + +void cHTTPServerConnection::SendNeedAuth(const AString & a_Realm) +{ + SendData(Printf("HTTP/1.1 401 Unauthorized\r\nWWW-Authenticate: Basic realm=\"%s\"\r\nContent-Length: 0\r\n\r\n", a_Realm.c_str())); + m_CurrentRequest.reset(); + m_Parser.Reset(); +} + + + + + +void cHTTPServerConnection::Send(const cHTTPOutgoingResponse & a_Response) +{ + ASSERT(m_CurrentRequest != nullptr); + AString toSend; + a_Response.AppendToData(toSend); + SendData(toSend); +} + + + + + +void cHTTPServerConnection::Send(const void * a_Data, size_t a_Size) +{ + ASSERT(m_CurrentRequest != nullptr); + // We're sending in Chunked transfer encoding + SendData(Printf(SIZE_T_FMT_HEX "\r\n", a_Size)); + SendData(a_Data, a_Size); + SendData("\r\n"); +} + + + + + +void cHTTPServerConnection::FinishResponse(void) +{ + ASSERT(m_CurrentRequest != nullptr); + SendData("0\r\n\r\n"); + m_CurrentRequest.reset(); + m_Parser.Reset(); +} + + + + + +void cHTTPServerConnection::Terminate(void) +{ + if (m_CurrentRequest != nullptr) + { + m_HTTPServer.RequestFinished(*this, *m_CurrentRequest); + } + m_Link.reset(); +} + + + + + +void cHTTPServerConnection::OnLinkCreated(cTCPLinkPtr a_Link) +{ + ASSERT(m_Link == nullptr); + m_Link = a_Link; +} + + + + + +void cHTTPServerConnection::OnReceivedData(const char * a_Data, size_t a_Size) +{ + ASSERT(m_Link != nullptr); + + m_Parser.Parse(a_Data, a_Size); +} + + + + + +void cHTTPServerConnection::OnRemoteClosed(void) +{ + if (m_CurrentRequest != nullptr) + { + m_HTTPServer.RequestFinished(*this, *m_CurrentRequest); + } + m_Link.reset(); +} + + + + + + +void cHTTPServerConnection::OnError(int a_ErrorCode, const AString & a_ErrorMsg) +{ + OnRemoteClosed(); +} + + + + + +void cHTTPServerConnection::OnError(const AString & a_ErrorDescription) +{ + OnRemoteClosed(); +} + + + + + +void cHTTPServerConnection::OnFirstLine(const AString & a_FirstLine) +{ + // Create a new request object for this request: + auto split = StringSplit(a_FirstLine, " "); + if (split.size() < 2) + { + // Invalid request line. We need at least the Method and URL + OnRemoteClosed(); + return; + } + m_CurrentRequest.reset(new cHTTPIncomingRequest(split[0], split[1])); +} + + + + + +void cHTTPServerConnection::OnHeaderLine(const AString & a_Key, const AString & a_Value) +{ + if (m_CurrentRequest == nullptr) + { + return; + } + m_CurrentRequest->AddHeader(a_Key, a_Value); +} + + + + + +void cHTTPServerConnection::OnHeadersFinished(void) +{ + if (m_CurrentRequest == nullptr) + { + return; + } + m_HTTPServer.NewRequest(*this, *m_CurrentRequest); +} + + + + + +void cHTTPServerConnection::OnBodyData(const void * a_Data, size_t a_Size) +{ + if (m_CurrentRequest == nullptr) + { + return; + } + m_HTTPServer.RequestBody(*this, *m_CurrentRequest, a_Data, a_Size); +} + + + + + +void cHTTPServerConnection::OnBodyFinished(void) +{ + // Process the request and reset: + m_HTTPServer.RequestFinished(*this, *m_CurrentRequest); + m_CurrentRequest.reset(); + m_Parser.Reset(); +} + + + + + +void cHTTPServerConnection::SendData(const void * a_Data, size_t a_Size) +{ + m_Link->Send(a_Data, a_Size); +} + + + + diff --git a/src/HTTP/HTTPServerConnection.h b/src/HTTP/HTTPServerConnection.h new file mode 100644 index 000000000..4390471d0 --- /dev/null +++ b/src/HTTP/HTTPServerConnection.h @@ -0,0 +1,118 @@ + +// HTTPConnection.h + +// Declares the cHTTPConnection class representing a single persistent connection in the HTTP server. + + + + + +#pragma once + +#include "../OSSupport/Network.h" +#include "HTTPMessageParser.h" + + + + + +// fwd: +class cHTTPServer; +class cHTTPOutgoingResponse; +class cHTTPIncomingRequest; + + + + +class cHTTPServerConnection : + public cTCPLink::cCallbacks, + public cHTTPMessageParser::cCallbacks +{ +public: + /** Creates a new instance, connected to the specified HTTP server instance */ + cHTTPServerConnection(cHTTPServer & a_HTTPServer); + + // Force a virtual destructor in all descendants + virtual ~cHTTPServerConnection(); + + /** Sends HTTP status code together with a_Reason (used for HTTP errors). + Sends the a_Reason as the body as well, so that browsers display it. + Clears the current request (since it's finished by this call). */ + void SendStatusAndReason(int a_StatusCode, const AString & a_Reason); + + /** Sends the "401 unauthorized" reply together with instructions on authorizing, using the specified realm. + Clears the current request (since it's finished by this call). */ + void SendNeedAuth(const AString & a_Realm); + + /** Sends the headers contained in a_Response */ + void Send(const cHTTPOutgoingResponse & a_Response); + + /** Sends the data as the response (may be called multiple times) */ + void Send(const void * a_Data, size_t a_Size); + + /** Sends the data as the response (may be called multiple times) */ + void Send(const AString & a_Data) { Send(a_Data.data(), a_Data.size()); } + + /** Indicates that the current response is finished, gets ready for receiving another request (HTTP 1.1 keepalive). + Clears the current request (since it's finished by this call). */ + void FinishResponse(void); + + /** Terminates the connection; finishes any request being currently processed */ + void Terminate(void); + +protected: + typedef std::map<AString, AString> cNameValueMap; + + /** The parent webserver that is to be notified of events on this connection */ + cHTTPServer & m_HTTPServer; + + /** The parser responsible for reading the requests. */ + cHTTPMessageParser m_Parser; + + /** The request being currently received + Valid only between having parsed the headers and finishing receiving the body. */ + std::unique_ptr<cHTTPIncomingRequest> m_CurrentRequest; + + /** The network link attached to this connection. */ + cTCPLinkPtr m_Link; + + + // cTCPLink::cCallbacks overrides: + /** The link instance has been created, remember it. */ + virtual void OnLinkCreated(cTCPLinkPtr a_Link) override; + + /** Data is received from the client. */ + virtual void OnReceivedData(const char * a_Data, size_t a_Size) override; + + /** The socket has been closed for any reason. */ + virtual void OnRemoteClosed(void) override; + + /** An error has occurred on the socket. */ + virtual void OnError(int a_ErrorCode, const AString & a_ErrorMsg) override; + + // cHTTPMessageParser::cCallbacks overrides: + virtual void OnError(const AString & a_ErrorDescription) override; + virtual void OnFirstLine(const AString & a_FirstLine) override; + virtual void OnHeaderLine(const AString & a_Key, const AString & a_Value) override; + virtual void OnHeadersFinished(void) override; + virtual void OnBodyData(const void * a_Data, size_t a_Size) override; + virtual void OnBodyFinished(void) override; + + // Overridable: + /** Called to send raw data over the link. Descendants may provide data transformations (SSL etc.) */ + virtual void SendData(const void * a_Data, size_t a_Size); + + /** Sends the raw data over the link. + Descendants may provide data transformations (SSL etc.) via the overridable SendData() function. */ + void SendData(const AString & a_Data) + { + SendData(a_Data.data(), a_Data.size()); + } +} ; + +typedef std::vector<cHTTPServerConnection *> cHTTPServerConnections; + + + + + diff --git a/src/HTTP/MultipartParser.cpp b/src/HTTP/MultipartParser.cpp new file mode 100644 index 000000000..09f4fd02a --- /dev/null +++ b/src/HTTP/MultipartParser.cpp @@ -0,0 +1,254 @@ + +// MultipartParser.cpp + +// Implements the cMultipartParser class that parses messages in "multipart/*" encoding into the separate parts + +#include "Globals.h" +#include "MultipartParser.h" +#include "NameValueParser.h" + + + + + +// Disable MSVC warnings: +#if defined(_MSC_VER) + #pragma warning(push) + #pragma warning(disable:4355) // 'this' : used in base member initializer list +#endif + + + + + +//////////////////////////////////////////////////////////////////////////////// +// self-test: + +#if 0 + +class cMultipartParserTest : + public cMultipartParser::cCallbacks +{ +public: + cMultipartParserTest(void) + { + cMultipartParser Parser("multipart/mixed; boundary=\"MyBoundaryString\"; foo=bar", *this); + const char Data[] = +"ThisIsIgnoredPrologue\r\n\ +--MyBoundaryString\r\n\ +\r\n\ +Body with confusing strings\r\n\ +--NotABoundary\r\n\ +--MyBoundaryStringWithPostfix\r\n\ +--\r\n\ +--MyBoundaryString\r\n\ +content-disposition: inline\r\n\ +\r\n\ +This is body\r\n\ +--MyBoundaryString\r\n\ +\r\n\ +Headerless body with trailing CRLF\r\n\ +\r\n\ +--MyBoundaryString--\r\n\ +ThisIsIgnoredEpilogue"; + printf("Multipart parsing test commencing.\n"); + Parser.Parse(Data, sizeof(Data) - 1); + // DEBUG: Check if the onscreen output corresponds with the data above + printf("Multipart parsing test finished\n"); + } + + virtual void OnPartStart(void) override + { + printf("Starting a new part\n"); + } + + + virtual void OnPartHeader(const AString & a_Key, const AString & a_Value) override + { + printf(" Hdr: \"%s\"=\"%s\"\n", a_Key.c_str(), a_Value.c_str()); + } + + + virtual void OnPartData(const char * a_Data, int a_Size) override + { + printf(" Data: %d bytes, \"%.*s\"\n", a_Size, a_Size, a_Data); + } + + + virtual void OnPartEnd(void) override + { + printf("Part end\n"); + } +} g_Test; + +#endif + + + + + +//////////////////////////////////////////////////////////////////////////////// +// cMultipartParser: + + +cMultipartParser::cMultipartParser(const AString & a_ContentType, cCallbacks & a_Callbacks) : + m_Callbacks(a_Callbacks), + m_IsValid(true), + m_EnvelopeParser(*this), + m_HasHadData(false) +{ + // Check that the content type is multipart: + AString ContentType(a_ContentType); + if (strncmp(ContentType.c_str(), "multipart/", 10) != 0) + { + m_IsValid = false; + return; + } + size_t idxSC = ContentType.find(';', 10); + if (idxSC == AString::npos) + { + m_IsValid = false; + return; + } + + // Find the multipart boundary: + ContentType.erase(0, idxSC + 1); + cNameValueParser CTParser(ContentType.c_str(), ContentType.size()); + CTParser.Finish(); + if (!CTParser.IsValid()) + { + m_IsValid = false; + return; + } + m_Boundary = CTParser["boundary"]; + m_IsValid = !m_Boundary.empty(); + if (!m_IsValid) + { + return; + } + + // Set the envelope parser for parsing the body, so that our Parse() function parses the ignored prefix data as a body + m_EnvelopeParser.SetIsInHeaders(false); + + // Append an initial CRLF to the incoming data, so that a body starting with the boundary line will get caught + m_IncomingData.assign("\r\n"); + + /* + m_Boundary = AString("\r\n--") + m_Boundary + m_BoundaryEnd = m_Boundary + "--\r\n"; + m_Boundary = m_Boundary + "\r\n"; + */ +} + + + + + +void cMultipartParser::Parse(const char * a_Data, size_t a_Size) +{ + // Skip parsing if invalid + if (!m_IsValid) + { + return; + } + + // Append to buffer, then parse it: + m_IncomingData.append(a_Data, a_Size); + for (;;) + { + if (m_EnvelopeParser.IsInHeaders()) + { + size_t BytesConsumed = m_EnvelopeParser.Parse(m_IncomingData.data(), m_IncomingData.size()); + if (BytesConsumed == AString::npos) + { + m_IsValid = false; + return; + } + if ((BytesConsumed == a_Size) && m_EnvelopeParser.IsInHeaders()) + { + // All the incoming data has been consumed and still waiting for more + return; + } + m_IncomingData.erase(0, BytesConsumed); + } + + // Search for boundary / boundary end: + size_t idxBoundary = m_IncomingData.find("\r\n--"); + if (idxBoundary == AString::npos) + { + // Boundary string start not present, present as much data to the part callback as possible + if (m_IncomingData.size() > m_Boundary.size() + 8) + { + size_t BytesToReport = m_IncomingData.size() - m_Boundary.size() - 8; + m_Callbacks.OnPartData(m_IncomingData.data(), BytesToReport); + m_IncomingData.erase(0, BytesToReport); + } + return; + } + if (idxBoundary > 0) + { + m_Callbacks.OnPartData(m_IncomingData.data(), idxBoundary); + m_IncomingData.erase(0, idxBoundary); + } + idxBoundary = 4; + size_t LineEnd = m_IncomingData.find("\r\n", idxBoundary); + if (LineEnd == AString::npos) + { + // Not a complete line yet, present as much data to the part callback as possible + if (m_IncomingData.size() > m_Boundary.size() + 8) + { + size_t BytesToReport = m_IncomingData.size() - m_Boundary.size() - 8; + m_Callbacks.OnPartData(m_IncomingData.data(), BytesToReport); + m_IncomingData.erase(0, BytesToReport); + } + return; + } + if ( + (LineEnd - idxBoundary != m_Boundary.size()) && // Line length not equal to boundary + (LineEnd - idxBoundary != m_Boundary.size() + 2) // Line length not equal to boundary end + ) + { + // Got a line, but it's not a boundary, report it as data: + m_Callbacks.OnPartData(m_IncomingData.data(), LineEnd); + m_IncomingData.erase(0, LineEnd); + continue; + } + + if (strncmp(m_IncomingData.c_str() + idxBoundary, m_Boundary.c_str(), m_Boundary.size()) == 0) + { + // Boundary or BoundaryEnd found: + m_Callbacks.OnPartEnd(); + size_t idxSlash = idxBoundary + m_Boundary.size(); + if ((m_IncomingData[idxSlash] == '-') && (m_IncomingData[idxSlash + 1] == '-')) + { + // This was the last part + m_Callbacks.OnPartData(m_IncomingData.data() + idxSlash + 4, m_IncomingData.size() - idxSlash - 4); + m_IncomingData.clear(); + return; + } + m_Callbacks.OnPartStart(); + m_IncomingData.erase(0, LineEnd + 2); + + // Keep parsing for the headers that may have come with this data: + m_EnvelopeParser.Reset(); + continue; + } + + // It's a line, but not a boundary. It can be fully sent to the data receiver, since a boundary cannot cross lines + m_Callbacks.OnPartData(m_IncomingData.c_str(), LineEnd); + m_IncomingData.erase(0, LineEnd); + } // while (true) +} + + + + + +void cMultipartParser::OnHeaderLine(const AString & a_Key, const AString & a_Value) +{ + m_Callbacks.OnPartHeader(a_Key, a_Value); +} + + + + diff --git a/src/HTTP/MultipartParser.h b/src/HTTP/MultipartParser.h new file mode 100644 index 000000000..4f20b2bed --- /dev/null +++ b/src/HTTP/MultipartParser.h @@ -0,0 +1,79 @@ + +// MultipartParser.h + +// Declares the cMultipartParser class that parses messages in "multipart/*" encoding into the separate parts + + + + + +#pragma once + +#include "EnvelopeParser.h" + + + + + +class cMultipartParser : + protected cEnvelopeParser::cCallbacks +{ +public: + class cCallbacks + { + public: + // Force a virtual destructor in descendants: + virtual ~cCallbacks() {} + + /** Called when a new part starts */ + virtual void OnPartStart(void) = 0; + + /** Called when a complete header line is received for a part */ + virtual void OnPartHeader(const AString & a_Key, const AString & a_Value) = 0; + + /** Called when body for a part is received */ + virtual void OnPartData(const char * a_Data, size_t a_Size) = 0; + + /** Called when the current part ends */ + virtual void OnPartEnd(void) = 0; + } ; + + /** Creates the parser, expects to find the boundary in a_ContentType */ + cMultipartParser(const AString & a_ContentType, cCallbacks & a_Callbacks); + + /** Parses more incoming data */ + void Parse(const char * a_Data, size_t a_Size); + +protected: + /** The callbacks to call for various parsing events */ + cCallbacks & m_Callbacks; + + /** True if the data parsed so far is valid; if false, further parsing is skipped */ + bool m_IsValid; + + /** Parser for each part's envelope */ + cEnvelopeParser m_EnvelopeParser; + + /** Buffer for the incoming data until it is parsed */ + AString m_IncomingData; + + /** The boundary, excluding both the initial "--" and the terminating CRLF */ + AString m_Boundary; + + /** Set to true if some data for the current part has already been signalized to m_Callbacks. Used for proper CRLF inserting. */ + bool m_HasHadData; + + + /** Parse one line of incoming data. The CRLF has already been stripped from a_Data / a_Size */ + void ParseLine(const char * a_Data, size_t a_Size); + + /** Parse one line of incoming data in the headers section of a part. The CRLF has already been stripped from a_Data / a_Size */ + void ParseHeaderLine(const char * a_Data, size_t a_Size); + + // cEnvelopeParser overrides: + virtual void OnHeaderLine(const AString & a_Key, const AString & a_Value) override; +} ; + + + + diff --git a/src/HTTP/NameValueParser.cpp b/src/HTTP/NameValueParser.cpp new file mode 100644 index 000000000..f759c4d21 --- /dev/null +++ b/src/HTTP/NameValueParser.cpp @@ -0,0 +1,413 @@ + +// NameValueParser.cpp + +// Implements the cNameValueParser class that parses strings in the "name=value;name2=value2" format into a stringmap + +#include "Globals.h" +#include "NameValueParser.h" + + + + + + +// DEBUG: Self-test + +#if 0 + +class cNameValueParserTest +{ +public: + cNameValueParserTest(void) + { + const char Data[] = " Name1=Value1;Name2 = Value 2; Name3 =\"Value 3\"; Name4 =\'Value 4\'; Name5=\"Confusing; isn\'t it?\""; + + // Now try parsing char-by-char, to debug transitions across datachunk boundaries: + cNameValueParser Parser2; + for (size_t i = 0; i < sizeof(Data) - 1; i++) + { + Parser2.Parse(Data + i, 1); + } + Parser2.Finish(); + + // Parse as a single chunk of data: + cNameValueParser Parser(Data, sizeof(Data) - 1); + + // Use the debugger to inspect the Parser variable + + // Check that the two parsers have the same content: + for (cNameValueParser::const_iterator itr = Parser.begin(), end = Parser.end(); itr != end; ++itr) + { + ASSERT(Parser2[itr->first] == itr->second); + } // for itr - Parser[] + + // Try parsing in 2-char chunks: + cNameValueParser Parser3; + for (int i = 0; i < sizeof(Data) - 2; i += 2) + { + Parser3.Parse(Data + i, 2); + } + if ((sizeof(Data) % 2) == 0) // There are even number of chars, including the NUL, so the data has an odd length. Parse one more char + { + Parser3.Parse(Data + sizeof(Data) - 2, 1); + } + Parser3.Finish(); + + // Check that the third parser has the same content: + for (cNameValueParser::const_iterator itr = Parser.begin(), end = Parser.end(); itr != end; ++itr) + { + ASSERT(Parser3[itr->first] == itr->second); + } // for itr - Parser[] + + printf("cNameValueParserTest done"); + } +} g_Test; + +#endif + + + + + +//////////////////////////////////////////////////////////////////////////////// +// cNameValueParser: + +cNameValueParser::cNameValueParser(bool a_AllowsKeyOnly) : + m_State(psKeySpace), + m_AllowsKeyOnly(a_AllowsKeyOnly) +{ +} + + + + + +cNameValueParser::cNameValueParser(const char * a_Data, size_t a_Size, bool a_AllowsKeyOnly) : + m_State(psKeySpace), + m_AllowsKeyOnly(a_AllowsKeyOnly) +{ + Parse(a_Data, a_Size); +} + + + + + +void cNameValueParser::Parse(const char * a_Data, size_t a_Size) +{ + ASSERT(m_State != psFinished); // Calling Parse() after Finish() is wrong! + + size_t Last = 0; + for (size_t i = 0; i < a_Size;) + { + switch (m_State) + { + case psInvalid: + case psFinished: + { + return; + } + + case psKeySpace: + { + // Skip whitespace until a non-whitespace is found, then start the key: + while ((i < a_Size) && (a_Data[i] <= ' ')) + { + i++; + } + if ((i < a_Size) && (a_Data[i] > ' ')) + { + m_State = psKey; + Last = i; + } + break; + } + + case psKey: + { + // Read the key until whitespace or an equal sign: + while (i < a_Size) + { + if (a_Data[i] == '=') + { + m_CurrentKey.append(a_Data + Last, i - Last); + i++; + Last = i; + m_State = psEqual; + break; + } + else if (a_Data[i] <= ' ') + { + m_CurrentKey.append(a_Data + Last, i - Last); + i++; + Last = i; + m_State = psEqualSpace; + break; + } + else if (a_Data[i] == ';') + { + if (!m_AllowsKeyOnly) + { + m_State = psInvalid; + return; + } + m_CurrentKey.append(a_Data + Last, i - Last); + i++; + Last = i; + (*this)[m_CurrentKey] = ""; + m_CurrentKey.clear(); + m_State = psKeySpace; + break; + } + else if ((a_Data[i] == '\"') || (a_Data[i] == '\'')) + { + m_State = psInvalid; + return; + } + i++; + } // while (i < a_Size) + if (i == a_Size) + { + // Still the key, ran out of data to parse, store the part of the key parsed so far: + m_CurrentKey.append(a_Data + Last, a_Size - Last); + return; + } + break; + } + + case psEqualSpace: + { + // The space before the expected equal sign; the current key is already assigned + while (i < a_Size) + { + if (a_Data[i] == '=') + { + m_State = psEqual; + i++; + Last = i; + break; + } + else if (a_Data[i] == ';') + { + // Key-only + if (!m_AllowsKeyOnly) + { + m_State = psInvalid; + return; + } + i++; + Last = i; + (*this)[m_CurrentKey] = ""; + m_CurrentKey.clear(); + m_State = psKeySpace; + break; + } + else if (a_Data[i] > ' ') + { + m_State = psInvalid; + return; + } + i++; + } // while (i < a_Size) + break; + } // case psEqualSpace + + case psEqual: + { + // just parsed the equal-sign + while (i < a_Size) + { + if (a_Data[i] == ';') + { + if (!m_AllowsKeyOnly) + { + m_State = psInvalid; + return; + } + i++; + Last = i; + (*this)[m_CurrentKey] = ""; + m_CurrentKey.clear(); + m_State = psKeySpace; + break; + } + else if (a_Data[i] == '\"') + { + i++; + Last = i; + m_State = psValueInDQuotes; + break; + } + else if (a_Data[i] == '\'') + { + i++; + Last = i; + m_State = psValueInSQuotes; + break; + } + else + { + m_CurrentValue.push_back(a_Data[i]); + i++; + Last = i; + m_State = psValueRaw; + break; + } + } // while (i < a_Size) + break; + } // case psEqual + + case psValueInDQuotes: + { + while (i < a_Size) + { + if (a_Data[i] == '\"') + { + m_CurrentValue.append(a_Data + Last, i - Last); + (*this)[m_CurrentKey] = m_CurrentValue; + m_CurrentKey.clear(); + m_CurrentValue.clear(); + m_State = psAfterValue; + i++; + Last = i; + break; + } + i++; + } // while (i < a_Size) + if (i == a_Size) + { + m_CurrentValue.append(a_Data + Last, a_Size - Last); + } + break; + } // case psValueInDQuotes + + case psValueInSQuotes: + { + while (i < a_Size) + { + if (a_Data[i] == '\'') + { + m_CurrentValue.append(a_Data + Last, i - Last); + (*this)[m_CurrentKey] = m_CurrentValue; + m_CurrentKey.clear(); + m_CurrentValue.clear(); + m_State = psAfterValue; + i++; + Last = i; + break; + } + i++; + } // while (i < a_Size) + if (i == a_Size) + { + m_CurrentValue.append(a_Data + Last, a_Size - Last); + } + break; + } // case psValueInSQuotes + + case psValueRaw: + { + while (i < a_Size) + { + if (a_Data[i] == ';') + { + m_CurrentValue.append(a_Data + Last, i - Last); + (*this)[m_CurrentKey] = m_CurrentValue; + m_CurrentKey.clear(); + m_CurrentValue.clear(); + m_State = psKeySpace; + i++; + Last = i; + break; + } + i++; + } + if (i == a_Size) + { + m_CurrentValue.append(a_Data + Last, a_Size - Last); + } + break; + } // case psValueRaw + + case psAfterValue: + { + // Between the closing DQuote or SQuote and the terminating semicolon + while (i < a_Size) + { + if (a_Data[i] == ';') + { + m_State = psKeySpace; + i++; + Last = i; + break; + } + else if (a_Data[i] < ' ') + { + i++; + continue; + } + m_State = psInvalid; + return; + } // while (i < a_Size) + break; + } + } // switch (m_State) + } // for i - a_Data[] +} + + + + + +bool cNameValueParser::Finish(void) +{ + switch (m_State) + { + case psInvalid: + { + return false; + } + case psFinished: + { + return true; + } + case psKey: + case psEqualSpace: + case psEqual: + { + if ((m_AllowsKeyOnly) && !m_CurrentKey.empty()) + { + (*this)[m_CurrentKey] = ""; + m_State = psFinished; + return true; + } + m_State = psInvalid; + return false; + } + case psValueRaw: + { + (*this)[m_CurrentKey] = m_CurrentValue; + m_State = psFinished; + return true; + } + case psValueInDQuotes: + case psValueInSQuotes: + { + // Missing the terminating quotes, this is an error + m_State = psInvalid; + return false; + } + case psKeySpace: + case psAfterValue: + { + m_State = psFinished; + return true; + } + } + ASSERT(!"Unhandled parser state!"); + return false; +} + + + + diff --git a/src/HTTP/NameValueParser.h b/src/HTTP/NameValueParser.h new file mode 100644 index 000000000..e205079db --- /dev/null +++ b/src/HTTP/NameValueParser.h @@ -0,0 +1,70 @@ + +// NameValueParser.h + +// Declares the cNameValueParser class that parses strings in the "name=value;name2=value2" format into a stringmap + + + + + +#pragma once + + + + + +class cNameValueParser : + public std::map<AString, AString> +{ +public: + /** Creates an empty parser */ + cNameValueParser(bool a_AllowsKeyOnly = true); + + /** Creates an empty parser, then parses the data given. Doesn't call Finish(), so more data can be parsed later */ + cNameValueParser(const char * a_Data, size_t a_Size, bool a_AllowsKeyOnly = true); + + /** Parses the data given */ + void Parse(const char * a_Data, size_t a_Size); + + /** Notifies the parser that no more data will be coming. Returns true if the parser state is valid */ + bool Finish(void); + + /** Returns true if the data parsed so far was valid */ + bool IsValid(void) const { return (m_State != psInvalid); } + + /** Returns true if the parser expects no more data */ + bool IsFinished(void) const { return ((m_State == psInvalid) || (m_State == psFinished)); } + +protected: + enum eState + { + psKeySpace, ///< Parsing the space in front of the next key + psKey, ///< Currently adding more chars to the key in m_CurrentKey + psEqualSpace, ///< Space after m_CurrentKey + psEqual, ///< Just parsed the = sign after a name + psValueInSQuotes, ///< Just parsed a Single-quote sign after the Equal sign + psValueInDQuotes, ///< Just parsed a Double-quote sign after the Equal sign + psValueRaw, ///< Just parsed a raw value without a quote + psAfterValue, ///< Just finished parsing the value, waiting for semicolon or data end + psInvalid, ///< The parser has encountered an invalid input; further parsing is skipped + psFinished, ///< The parser has already been instructed to finish and doesn't expect any more data + } ; + + /** The current state of the parser */ + eState m_State; + + /** If true, the parser will accept keys without an equal sign and the value */ + bool m_AllowsKeyOnly; + + /** Buffer for the current Key */ + AString m_CurrentKey; + + /** Buffer for the current Value; */ + AString m_CurrentValue; + + +} ; + + + + diff --git a/src/HTTP/SslHTTPServerConnection.cpp b/src/HTTP/SslHTTPServerConnection.cpp new file mode 100644 index 000000000..547e6de3a --- /dev/null +++ b/src/HTTP/SslHTTPServerConnection.cpp @@ -0,0 +1,115 @@ + +// SslHTTPConnection.cpp + +// Implements the cSslHTTPServerConnection class representing a HTTP connection made over a SSL link + +#include "Globals.h" +#include "SslHTTPServerConnection.h" +#include "HTTPServer.h" + + + + + +cSslHTTPServerConnection::cSslHTTPServerConnection(cHTTPServer & a_HTTPServer, const cX509CertPtr & a_Cert, const cCryptoKeyPtr & a_PrivateKey) : + super(a_HTTPServer), + m_Ssl(64000), + m_Cert(a_Cert), + m_PrivateKey(a_PrivateKey) +{ + m_Ssl.Initialize(false); + m_Ssl.SetOwnCert(a_Cert, a_PrivateKey); +} + + + + + +cSslHTTPServerConnection::~cSslHTTPServerConnection() +{ + m_Ssl.NotifyClose(); +} + + + + + +void cSslHTTPServerConnection::OnReceivedData(const char * a_Data, size_t a_Size) +{ + // Process the received data: + const char * Data = a_Data; + size_t Size = a_Size; + for (;;) + { + // Try to write as many bytes into Ssl's "incoming" buffer as possible: + size_t BytesWritten = 0; + if (Size > 0) + { + BytesWritten = m_Ssl.WriteIncoming(Data, Size); + Data += BytesWritten; + Size -= BytesWritten; + } + + // Try to read as many bytes from SSL's decryption as possible: + char Buffer[32000]; + int NumRead = m_Ssl.ReadPlain(Buffer, sizeof(Buffer)); + if (NumRead > 0) + { + super::OnReceivedData(Buffer, static_cast<size_t>(NumRead)); + // The link may have closed while processing the data, bail out: + return; + } + else if (NumRead == POLARSSL_ERR_NET_WANT_READ) + { + // SSL requires us to send data to peer first, do so by "sending" empty data: + SendData(nullptr, 0); + } + + // If both failed, bail out: + if ((BytesWritten == 0) && (NumRead <= 0)) + { + return; + } + } +} + + + + + +void cSslHTTPServerConnection::SendData(const void * a_Data, size_t a_Size) +{ + const char * OutgoingData = reinterpret_cast<const char *>(a_Data); + size_t pos = 0; + for (;;) + { + // Write as many bytes from our buffer to SSL's encryption as possible: + int NumWritten = 0; + if (pos < a_Size) + { + NumWritten = m_Ssl.WritePlain(OutgoingData + pos, a_Size - pos); + if (NumWritten > 0) + { + pos += static_cast<size_t>(NumWritten); + } + } + + // Read as many bytes from SSL's "outgoing" buffer as possible: + char Buffer[32000]; + size_t NumBytes = m_Ssl.ReadOutgoing(Buffer, sizeof(Buffer)); + if (NumBytes > 0) + { + m_Link->Send(Buffer, NumBytes); + } + + // If both failed, bail out: + if ((NumWritten <= 0) && (NumBytes == 0)) + { + return; + } + } +} + + + + diff --git a/src/HTTP/SslHTTPServerConnection.h b/src/HTTP/SslHTTPServerConnection.h new file mode 100644 index 000000000..eceb80fb7 --- /dev/null +++ b/src/HTTP/SslHTTPServerConnection.h @@ -0,0 +1,47 @@ + +// SslHTTPServerConnection.h + +// Declares the cSslHTTPServerConnection class representing a HTTP connection made over an SSL link + + + + + +#pragma once + +#include "HTTPServerConnection.h" +#include "PolarSSL++/BufferedSslContext.h" + + + + + +class cSslHTTPServerConnection : + public cHTTPServerConnection +{ + typedef cHTTPServerConnection super; + +public: + /** Creates a new connection on the specified server. + Sends the specified cert as the server certificate, uses the private key for decryption. */ + cSslHTTPServerConnection(cHTTPServer & a_HTTPServer, const cX509CertPtr & a_Cert, const cCryptoKeyPtr & a_PrivateKey); + + ~cSslHTTPServerConnection(); + +protected: + cBufferedSslContext m_Ssl; + + /** The certificate to send to the client */ + cX509CertPtr m_Cert; + + /** The private key used for the certificate */ + cCryptoKeyPtr m_PrivateKey; + + // cHTTPConnection overrides: + virtual void OnReceivedData(const char * a_Data, size_t a_Size) override; // Data is received from the client + virtual void SendData(const void * a_Data, size_t a_Size) override; // Data is to be sent to client +} ; + + + + diff --git a/src/HTTP/TransferEncodingParser.cpp b/src/HTTP/TransferEncodingParser.cpp new file mode 100644 index 000000000..a14900e9c --- /dev/null +++ b/src/HTTP/TransferEncodingParser.cpp @@ -0,0 +1,394 @@ + +// TransferEncodingParser.cpp + +// Implements the cTransferEncodingParser class and its descendants representing the parsers for the various transfer encodings (chunked etc.) + +#include "Globals.h" +#include "TransferEncodingParser.h" +#include "EnvelopeParser.h" + + + + + +//////////////////////////////////////////////////////////////////////////////// +// cChunkedTEParser: + +class cChunkedTEParser: + public cTransferEncodingParser, + public cEnvelopeParser::cCallbacks +{ + typedef cTransferEncodingParser Super; + +public: + cChunkedTEParser(Super::cCallbacks & a_Callbacks): + Super(a_Callbacks), + m_State(psChunkLength), + m_ChunkDataLengthLeft(0), + m_TrailerParser(*this) + { + } + + +protected: + enum eState + { + psChunkLength, ///< Parsing the chunk length hex number + psChunkLengthTrailer, ///< Any trailer (chunk extension) specified after the chunk length + psChunkLengthLF, ///< The LF character after the CR character terminating the chunk length + psChunkData, ///< Relaying chunk data + psChunkDataCR, ///< Skipping the extra CR character after chunk data + psChunkDataLF, ///< Skipping the extra LF character after chunk data + psTrailer, ///< Received an empty chunk, parsing the trailer (through the envelope parser) + psFinished, ///< The parser has finished parsing, either successfully or with an error + }; + + /** The current state of the parser (parsing chunk length / chunk data). */ + eState m_State; + + /** Number of bytes that still belong to the chunk currently being parsed. + When in psChunkLength, the value is the currently parsed length digits. */ + size_t m_ChunkDataLengthLeft; + + /** The parser used for the last (empty) chunk's trailer data */ + cEnvelopeParser m_TrailerParser; + + + /** Calls the OnError callback and sets parser state to finished. */ + void Error(const AString & a_ErrorMsg) + { + m_State = psFinished; + m_Callbacks.OnError(a_ErrorMsg); + } + + + /** Parses the incoming data, the current state is psChunkLength. + Stops parsing when either the chunk length has been read, or there is no more data in the input. + Returns the number of bytes consumed from the input, or AString::npos on error (calls the Error handler). */ + size_t ParseChunkLength(const char * a_Data, size_t a_Size) + { + // Expected input: <hexnumber>[;<trailer>]<CR><LF> + // Only the hexnumber is parsed into m_ChunkDataLengthLeft, the rest is postponed into psChunkLengthTrailer or psChunkLengthLF + for (size_t i = 0; i < a_Size; i++) + { + switch (a_Data[i]) + { + case '0': + case '1': + case '2': + case '3': + case '4': + case '5': + case '6': + case '7': + case '8': + case '9': + { + m_ChunkDataLengthLeft = m_ChunkDataLengthLeft * 16 + static_cast<decltype(m_ChunkDataLengthLeft)>(a_Data[i] - '0'); + break; + } + case 'a': + case 'b': + case 'c': + case 'd': + case 'e': + case 'f': + { + m_ChunkDataLengthLeft = m_ChunkDataLengthLeft * 16 + static_cast<decltype(m_ChunkDataLengthLeft)>(a_Data[i] - 'a' + 10); + break; + } + case 'A': + case 'B': + case 'C': + case 'D': + case 'E': + case 'F': + { + m_ChunkDataLengthLeft = m_ChunkDataLengthLeft * 16 + static_cast<decltype(m_ChunkDataLengthLeft)>(a_Data[i] - 'A' + 10); + break; + } + case '\r': + { + m_State = psChunkLengthLF; + return i + 1; + } + case ';': + { + m_State = psChunkLengthTrailer; + return i + 1; + } + default: + { + Error(Printf("Invalid character in chunk length line: 0x%x", a_Data[i])); + return AString::npos; + } + } // switch (a_Data[i]) + } // for i - a_Data[] + return a_Size; + } + + + /** Parses the incoming data, the current state is psChunkLengthTrailer. + Stops parsing when either the chunk length trailer has been read, or there is no more data in the input. + Returns the number of bytes consumed from the input, or AString::npos on error (calls the Error handler). */ + size_t ParseChunkLengthTrailer(const char * a_Data, size_t a_Size) + { + // Expected input: <trailer><CR><LF> + // The LF itself is not parsed, it is instead postponed into psChunkLengthLF + for (size_t i = 0; i < a_Size; i++) + { + switch (a_Data[i]) + { + case '\r': + { + m_State = psChunkLengthLF; + return i; + } + default: + { + if (a_Data[i] < 32) + { + // Only printable characters are allowed in the trailer + Error(Printf("Invalid character in chunk length line: 0x%x", a_Data[i])); + return AString::npos; + } + } + } // switch (a_Data[i]) + } // for i - a_Data[] + return a_Size; + } + + + /** Parses the incoming data, the current state is psChunkLengthLF. + Only the LF character is expected, if found, moves to psChunkData, otherwise issues an error. + If the chunk length that just finished reading is equal to 0, signals the end of stream (via psTrailer). + Returns the number of bytes consumed from the input, or AString::npos on error (calls the Error handler). */ + size_t ParseChunkLengthLF(const char * a_Data, size_t a_Size) + { + // Expected input: <LF> + if (a_Size == 0) + { + return 0; + } + if (a_Data[0] == '\n') + { + if (m_ChunkDataLengthLeft == 0) + { + m_State = psTrailer; + } + else + { + m_State = psChunkData; + } + return 1; + } + Error(Printf("Invalid character past chunk length's CR: 0x%x", a_Data[0])); + return AString::npos; + } + + + /** Consumes as much chunk data from the input as possible. + Returns the number of bytes consumed from the input, or AString::npos on error (calls the Error() handler). */ + size_t ParseChunkData(const char * a_Data, size_t a_Size) + { + ASSERT(m_ChunkDataLengthLeft > 0); + auto bytes = std::min(a_Size, m_ChunkDataLengthLeft); + m_ChunkDataLengthLeft -= bytes; + m_Callbacks.OnBodyData(a_Data, bytes); + if (m_ChunkDataLengthLeft == 0) + { + m_State = psChunkDataCR; + } + return bytes; + } + + + /** Parses the incoming data, the current state is psChunkDataCR. + Only the CR character is expected, if found, moves to psChunkDataLF, otherwise issues an error. + Returns the number of bytes consumed from the input, or AString::npos on error (calls the Error handler). */ + size_t ParseChunkDataCR(const char * a_Data, size_t a_Size) + { + // Expected input: <CR> + if (a_Size == 0) + { + return 0; + } + if (a_Data[0] == '\r') + { + m_State = psChunkDataLF; + return 1; + } + Error(Printf("Invalid character past chunk data: 0x%x", a_Data[0])); + return AString::npos; + } + + + + + /** Parses the incoming data, the current state is psChunkDataCR. + Only the CR character is expected, if found, moves to psChunkDataLF, otherwise issues an error. + Returns the number of bytes consumed from the input, or AString::npos on error (calls the Error handler). */ + size_t ParseChunkDataLF(const char * a_Data, size_t a_Size) + { + // Expected input: <LF> + if (a_Size == 0) + { + return 0; + } + if (a_Data[0] == '\n') + { + m_State = psChunkLength; + return 1; + } + Error(Printf("Invalid character past chunk data's CR: 0x%x", a_Data[0])); + return AString::npos; + } + + + /** Parses the incoming data, the current state is psChunkDataCR. + The trailer is normally a set of "Header: Value" lines, terminated by an empty line. Use the m_TrailerParser for that. + Returns the number of bytes consumed from the input, or AString::npos on error (calls the Error handler). */ + size_t ParseTrailer(const char * a_Data, size_t a_Size) + { + auto res = m_TrailerParser.Parse(a_Data, a_Size); + if (res == AString::npos) + { + Error("Error while parsing the trailer"); + } + if ((res < a_Size) || !m_TrailerParser.IsInHeaders()) + { + m_Callbacks.OnBodyFinished(); + m_State = psFinished; + } + return res; + } + + + // cTransferEncodingParser overrides: + virtual size_t Parse(const char * a_Data, size_t a_Size) override + { + while ((a_Size > 0) && (m_State != psFinished)) + { + size_t consumed = 0; + switch (m_State) + { + case psChunkLength: consumed = ParseChunkLength (a_Data, a_Size); break; + case psChunkLengthTrailer: consumed = ParseChunkLengthTrailer(a_Data, a_Size); break; + case psChunkLengthLF: consumed = ParseChunkLengthLF (a_Data, a_Size); break; + case psChunkData: consumed = ParseChunkData (a_Data, a_Size); break; + case psChunkDataCR: consumed = ParseChunkDataCR (a_Data, a_Size); break; + case psChunkDataLF: consumed = ParseChunkDataLF (a_Data, a_Size); break; + case psTrailer: consumed = ParseTrailer (a_Data, a_Size); break; + case psFinished: consumed = 0; break; // Not supposed to happen, but Clang complains without it + } + if (consumed == AString::npos) + { + return AString::npos; + } + a_Data += consumed; + a_Size -= consumed; + } + return a_Size; + } + + virtual void Finish(void) override + { + if (m_State != psFinished) + { + Error(Printf("ChunkedTransferEncoding: Finish signal received before the data stream ended (state: %d)", m_State)); + } + m_State = psFinished; + } + + + // cEnvelopeParser::cCallbacks overrides: + virtual void OnHeaderLine(const AString & a_Key, const AString & a_Value) override + { + // Ignored + } +}; + + + + + +//////////////////////////////////////////////////////////////////////////////// +// cIdentityTEParser: + +class cIdentityTEParser: + public cTransferEncodingParser +{ + typedef cTransferEncodingParser Super; + +public: + cIdentityTEParser(cCallbacks & a_Callbacks, size_t a_ContentLength): + Super(a_Callbacks), + m_BytesLeft(a_ContentLength) + { + } + + +protected: + /** How many bytes of content are left before the message ends. */ + size_t m_BytesLeft; + + // cTransferEncodingParser overrides: + virtual size_t Parse(const char * a_Data, size_t a_Size) override + { + auto size = std::min(a_Size, m_BytesLeft); + if (size > 0) + { + m_Callbacks.OnBodyData(a_Data, size); + } + m_BytesLeft -= size; + if (m_BytesLeft == 0) + { + m_Callbacks.OnBodyFinished(); + } + return a_Size - size; + } + + virtual void Finish(void) override + { + if (m_BytesLeft > 0) + { + m_Callbacks.OnError("IdentityTransferEncoding: body was truncated"); + } + else + { + // BodyFinished has already been called, just bail out + } + } +}; + + + + + +//////////////////////////////////////////////////////////////////////////////// +// cTransferEncodingParser: + +cTransferEncodingParserPtr cTransferEncodingParser::Create( + cCallbacks & a_Callbacks, + const AString & a_TransferEncoding, + size_t a_ContentLength +) +{ + if (a_TransferEncoding == "chunked") + { + return std::make_shared<cChunkedTEParser>(a_Callbacks); + } + if (a_TransferEncoding == "identity") + { + return std::make_shared<cIdentityTEParser>(a_Callbacks, a_ContentLength); + } + if (a_TransferEncoding.empty()) + { + return std::make_shared<cIdentityTEParser>(a_Callbacks, a_ContentLength); + } + return nullptr; +} + + + + diff --git a/src/HTTP/TransferEncodingParser.h b/src/HTTP/TransferEncodingParser.h new file mode 100644 index 000000000..ce3d01df7 --- /dev/null +++ b/src/HTTP/TransferEncodingParser.h @@ -0,0 +1,76 @@ + +// TransferEncodingParser.h + +// Declares the cTransferEncodingParser class representing the parser for the various transfer encodings (chunked etc.) + +#pragma once + + + + + +// fwd: +class cTransferEncodingParser; +typedef SharedPtr<cTransferEncodingParser> cTransferEncodingParserPtr; + + + + + +/** Used as both the interface that all the parsers share and the (static) factory creating such parsers. */ +class cTransferEncodingParser +{ +public: + class cCallbacks + { + public: + // Force a virtual destructor in descendants + virtual ~cCallbacks() {} + + /** Called when an error has occured while parsing. */ + virtual void OnError(const AString & a_ErrorDescription) = 0; + + /** Called for each chunk of the incoming body data. */ + virtual void OnBodyData(const void * a_Data, size_t a_Size) = 0; + + /** Called when the entire body has been reported by OnBodyData(). */ + virtual void OnBodyFinished(void) = 0; + }; + + + // Force a virtual destructor in all descendants + virtual ~cTransferEncodingParser() {} + + /** Parses the incoming data and calls the appropriate callbacks. + Returns the number of bytes from the end of a_Data that is already not part of this message (if the parser can detect it). + Returns AString::npos on an error. */ + virtual size_t Parse(const char * a_Data, size_t a_Size) = 0; + + /** To be called when the stream is terminated from the source (connection closed). + Flushes any buffers and calls appropriate callbacks. */ + virtual void Finish(void) = 0; + + /** Creates a new parser for the specified encoding. + If the encoding is not known, returns a nullptr. + a_ContentLength is the length of the content, received in a Content-Length header. + It is used for the Identity encoding, it is ignored for the Chunked encoding. */ + static cTransferEncodingParserPtr Create( + cCallbacks & a_Callbacks, + const AString & a_TransferEncoding, + size_t a_ContentLength + ); + +protected: + /** The callbacks used to report progress. */ + cCallbacks & m_Callbacks; + + + cTransferEncodingParser(cCallbacks & a_Callbacks): + m_Callbacks(a_Callbacks) + { + } +}; + + + + diff --git a/src/HTTP/UrlParser.cpp b/src/HTTP/UrlParser.cpp new file mode 100644 index 000000000..05db3e413 --- /dev/null +++ b/src/HTTP/UrlParser.cpp @@ -0,0 +1,200 @@ + +// UrlParser.cpp + +// Implements the cUrlParser class that parses string URL into individual parts + +#include "Globals.h" +#include "UrlParser.h" + + + + + +UInt16 cUrlParser::GetDefaultPort(const AString & a_Scheme) +{ + if (a_Scheme == "http") + { + return 80; + } + else if (a_Scheme == "https") + { + return 443; + } + else if (a_Scheme == "ftp") + { + return 21; + } + else if (a_Scheme == "mailto") + { + return 25; + } + return 0; +} + + + + + +std::pair<bool, AString> cUrlParser::ParseAuthorityPart( + const AString & a_AuthorityPart, + AString & a_Username, + AString & a_Password, + AString & a_Host, + UInt16 & a_Port +) +{ + /* + a_AuthorityPart format: + [user:password@]host[:port] + host can be an IPv4, hostname, or an IPv6 enclosed in brackets + Assume only the password can contain an additional at-sign + */ + + // Split the authority on the last at-sign, if present: + auto idxLastAtSign = a_AuthorityPart.find_last_of('@'); + auto credPart = (idxLastAtSign == AString::npos) ? AString() : a_AuthorityPart.substr(0, idxLastAtSign); + auto srvrPart = (idxLastAtSign == AString::npos) ? a_AuthorityPart : a_AuthorityPart.substr(idxLastAtSign + 1); + + // User credentials are completely optional: + auto idxCredColon = credPart.find(':'); + a_Username = credPart.substr(0, idxCredColon); + a_Password = (idxCredColon == AString::npos) ? AString() : credPart.substr(idxCredColon + 1); + + // Host can be a hostname, IPv4 or [IPv6]. If in brackets, search for the closing bracket first + if (srvrPart.empty()) + { + // No host information at all. Bail out with success + a_Host.clear(); + return std::make_pair(true, AString()); + } + if (srvrPart[0] == '[') + { + // [IPv6] host, search for the closing bracket + auto idxClosingBracket = srvrPart.find(']'); + if (idxClosingBracket == AString::npos) + { + return std::make_pair(false, "Invalid IPv6-like address, missing closing bracket"); + } + a_Host = srvrPart.substr(0, idxClosingBracket); + auto portPart = srvrPart.substr(idxClosingBracket + 1); + if (portPart.empty()) + { + // No port was specified, return success + return std::make_pair(true, AString()); + } + if (portPart[0] != ':') + { + return std::make_pair(false, "Invalid port format after IPv6 address, mising colon"); + } + if (!StringToInteger(portPart.substr(2), a_Port)) + { + return std::make_pair(false, "Failed to parse port number after IPv6 address"); + } + return std::make_pair(true, AString()); + } + + // Not an [IPv6] address, split on the last colon: + auto idxLastColon = srvrPart.find_last_of(':'); + a_Host = srvrPart.substr(0, idxLastColon); + if (idxLastColon == AString::npos) + { + // No port was specified, return success + return std::make_pair(true, AString()); + } + auto portPart = srvrPart.substr(idxLastColon + 1); + if (!StringToInteger(portPart, a_Port)) + { + return std::make_pair(false, "Failed to parse port number after hostname"); + } + return std::make_pair(true, AString()); +} + + + + + +std::pair<bool, AString> cUrlParser::Parse( + const AString & a_Url, + AString & a_Scheme, + AString & a_Username, + AString & a_Password, + AString & a_Host, + UInt16 & a_Port, + AString & a_Path, + AString & a_Query, + AString & a_Fragment +) +{ + // Find the scheme - the text before the first colon: + auto idxColon = a_Url.find(':'); + if (idxColon == AString::npos) + { + return std::make_pair(false, "Cannot parse the Scheme part of the URL"); + } + a_Scheme = StrToLower(a_Url.substr(0, idxColon)); + a_Port = GetDefaultPort(a_Scheme); + if (a_Port == 0) + { + return std::make_pair(false, Printf("Unknown URL scheme: \"%s\"", a_Scheme.c_str())); + } + + // If the next two chars are a double-slash, skip them: + auto authStart = idxColon + 1; + if (a_Url.substr(authStart, 2) == "//") + { + authStart += 2; + } + + // The Authority part follows the Scheme, until the first slash: + auto idxFirstSlash = a_Url.find('/', authStart + 1); + if (idxFirstSlash == AString::npos) + { + // No slash, the whole end of the Url is the authority part + idxFirstSlash = a_Url.size(); + } + + // Parse the Authority part into individual components: + auto res = ParseAuthorityPart( + a_Url.substr(authStart, idxFirstSlash - authStart), + a_Username, a_Password, + a_Host, a_Port + ); + if (!res.first) + { + return res; + } + + // Parse the rest into a path, query and fragment: + a_Path.clear(); + a_Query.clear(); + a_Fragment.clear(); + if (idxFirstSlash == a_Url.size()) + { + // No additional data, bail out with success + return std::make_pair(true, AString()); + } + auto idxPathEnd = a_Url.find_first_of("?#", idxFirstSlash + 1); + if (idxPathEnd == AString::npos) + { + a_Path = a_Url.substr(idxFirstSlash); + return std::make_pair(true, AString()); + } + a_Path = a_Url.substr(idxFirstSlash, idxPathEnd - idxFirstSlash); + auto idxHash = a_Url.find('#', idxPathEnd); + if (idxHash == AString::npos) + { + a_Query = a_Url.substr(idxPathEnd + 1); + return std::make_pair(true, AString()); + } + if (idxHash > idxPathEnd) + { + a_Query = a_Url.substr(idxPathEnd + 1, idxHash - idxPathEnd - 1); + } + a_Fragment = a_Url.substr(idxHash + 1); + return std::make_pair(true, AString()); +} + + + + + diff --git a/src/HTTP/UrlParser.h b/src/HTTP/UrlParser.h new file mode 100644 index 000000000..15a63e05d --- /dev/null +++ b/src/HTTP/UrlParser.h @@ -0,0 +1,58 @@ + +// UrlParser.h + +// Declares the cUrlParser class that parses string URL into individual parts + + + + + +#pragma once + + + + + +class cUrlParser +{ +public: + /** Returns true if the specified scheme (http, ftp, mailto, ...) is recognized by the URL parser. + Is case sensitive, known schemes are always lowercase. */ + static bool IsKnownScheme(const AString & a_Scheme) { return (GetDefaultPort(a_Scheme) > 0); } + + /** Returns the default port used by the specified scheme / protocol. + If the scheme is not known, 0 is returned. */ + static UInt16 GetDefaultPort(const AString & a_Scheme); + + /** Parses the given Authority part of an URL into individual components. + Returns true on success, + returns false and error message on failure. */ + static std::pair<bool, AString> ParseAuthorityPart( + const AString & a_AuthorityPart, + AString & a_Username, + AString & a_Password, + AString & a_Host, + UInt16 & a_Port + ); + + /** Parses the given URL into individual components. + Returns true on success, + returns false and error message on failure. + Fails if the scheme (protocol) is not known. + If port is missing, the default port for the specific scheme is applied. */ + static std::pair<bool, AString> Parse( + const AString & a_Url, + AString & a_Scheme, + AString & a_Username, + AString & a_Password, + AString & a_Host, + UInt16 & a_Port, + AString & a_Path, + AString & a_Query, + AString & a_Fragment + ); +}; + + + + |