THRIFT-247. cpp: Basic THttpServer implementation for C++

This does not integrate with any existing HTTP server.  Rather, it is a
transport that parses HTTP requests to allow any of the blocking
socket-based servers to handle HTTP.

git-svn-id: https://svn.apache.org/repos/asf/incubator/thrift/trunk@985031 13f79535-47bb-0310-9956-ffa450edef68

3 files changed, 28 insertions, 320 deletions

diff --git a/lib/cpp/Makefile.am b/lib/cpp/Makefile.am
index 579140cc5..bf2fe260d 100644
--- a/lib/cpp/Makefile.am
+++ b/lib/cpp/Makefile.am
@@ -58,7 +58,9 @@ libthrift_la_SOURCES = src/Thrift.cpp \
                        src/transport/TFDTransport.cpp \
                        src/transport/TFileTransport.cpp \
                        src/transport/TSimpleFileTransport.cpp \
+                       src/transport/THttpTransport.cpp \
                        src/transport/THttpClient.cpp \
+                       src/transport/THttpServer.cpp \
                        src/transport/TSocket.cpp \
                        src/transport/TSocketPool.cpp \
                        src/transport/TServerSocket.cpp \
@@ -121,7 +123,9 @@ include_transport_HEADERS = \
                          src/transport/TSimpleFileTransport.h \
                          src/transport/TServerSocket.h \
                          src/transport/TServerTransport.h \
+                         src/transport/THttpTransport.h \
                          src/transport/THttpClient.h \
+                         src/transport/THttpServer.h \
                          src/transport/TSocket.h \
                          src/transport/TSocketPool.h \
                          src/transport/TTransport.h \
diff --git a/lib/cpp/src/transport/THttpClient.cpp b/lib/cpp/src/transport/THttpClient.cpp
index 59f233968..400597e10 100644
--- a/lib/cpp/src/transport/THttpClient.cpp
+++ b/lib/cpp/src/transport/THttpClient.cpp
@@ -20,250 +20,38 @@
 #include <cstdlib>
 #include <sstream>
 
-#include "THttpClient.h"
-#include "TSocket.h"
+#include <transport/THttpClient.h>
+#include <transport/TSocket.h>
 
 namespace apache { namespace thrift { namespace transport {
 
 using namespace std;
 
-/**
- * Http client implementation.
- *
- */
-
-// Yeah, yeah, hacky to put these here, I know.
-static const char* CRLF = "\r\n";
-static const int CRLF_LEN = 2;
-
-THttpClient::THttpClient(boost::shared_ptr<TTransport> transport, string host, string path) :
-  transport_(transport),
-  host_(host),
-  path_(path),
-  readHeaders_(true),
-  chunked_(false),
-  chunkedDone_(false),
-  chunkSize_(0),
-  contentLength_(0),
-  httpBuf_(NULL),
-  httpPos_(0),
-  httpBufLen_(0),
-  httpBufSize_(1024) {
-  init();
+THttpClient::THttpClient(boost::shared_ptr<TTransport> transport, std::string host, std::string path) :
+  THttpTransport(transport), host_(host), path_(path) {
 }
 
 THttpClient::THttpClient(string host, int port, string path) :
-  host_(host),
-  path_(path),
-  readHeaders_(true),
-  chunked_(false),
-  chunkedDone_(false),
-  chunkSize_(0),
-  contentLength_(0),
-  httpBuf_(NULL),
-  httpPos_(0),
-  httpBufLen_(0),
-  httpBufSize_(1024) {
-  transport_ = boost::shared_ptr<TTransport>(new TSocket(host, port));
-  init();
-}
-
-void THttpClient::init() {
-  httpBuf_ = (char*)std::malloc(httpBufSize_+1);
-  if (httpBuf_ == NULL) {
-    throw TTransportException("Out of memory.");
-  }
-  httpBuf_[httpBufLen_] = '\0';
-}
-
-THttpClient::~THttpClient() {
-  if (httpBuf_ != NULL) {
-    std::free(httpBuf_);
-  }
-}
-
-uint32_t THttpClient::read(uint8_t* buf, uint32_t len) {
-  if (readBuffer_.available_read() == 0) {
-    readBuffer_.resetBuffer();
-    uint32_t got = readMoreData();
-    if (got == 0) {
-      return 0;
-    }
-  }
-  return readBuffer_.read(buf, len);
-}
-
-void THttpClient::readEnd() {
-  // Read any pending chunked data (footers etc.)
-  if (chunked_) {
-    while (!chunkedDone_) {
-      readChunked();
-    }
-  }
+  THttpTransport(boost::shared_ptr<TTransport>(new TSocket(host, port))), host_(host), path_(path) {
 }
 
-uint32_t THttpClient::readMoreData() {
-  // Get more data!
-  refill();
-
-  if (readHeaders_) {
-    readHeaders();
-  }
-
-  if (chunked_) {
-    return readChunked();
-  } else {
-    return readContent(contentLength_);
-  }
-}
-
-uint32_t THttpClient::readChunked() {
-  uint32_t length = 0;
-
-  char* line = readLine();
-  uint32_t chunkSize = parseChunkSize(line);
-  if (chunkSize == 0) {
-    readChunkedFooters();
-  } else {
-    // Read data content
-    length += readContent(chunkSize);
-    // Read trailing CRLF after content
-    readLine();
-  }
-  return length;
-}
-
-void THttpClient::readChunkedFooters() {
-  // End of data, read footer lines until a blank one appears
-  while (true) {
-    char* line = readLine();
-    if (strlen(line) == 0) {
-      chunkedDone_ = true;
-      break;
-    }
-  }
-}
-
-uint32_t THttpClient::parseChunkSize(char* line) {
-  char* semi = strchr(line, ';');
-  if (semi != NULL) {
-    *semi = '\0';
-  }
-  int size = 0;
-  sscanf(line, "%x", &size);
-  return (uint32_t)size;
-}
-
-uint32_t THttpClient::readContent(uint32_t size) {
-  uint32_t need = size;
-  while (need > 0) {
-    uint32_t avail = httpBufLen_ - httpPos_;
-    if (avail == 0) {
-      // We have given all the data, reset position to head of the buffer
-      httpPos_ = 0;
-      httpBufLen_ = 0;
-      refill();
-
-      // Now have available however much we read
-      avail = httpBufLen_;
-    }
-    uint32_t give = avail;
-    if (need < give) {
-      give = need;
-    }
-    readBuffer_.write((uint8_t*)(httpBuf_+httpPos_), give);
-    httpPos_ += give;
-    need -= give;
-  }
-  return size;
-}
-
-char* THttpClient::readLine() {
-  while (true) {
-    char* eol = NULL;
-
-    eol = strstr(httpBuf_+httpPos_, CRLF);
-
-    // No CRLF yet?
-    if (eol == NULL) {
-      // Shift whatever we have now to front and refill
-      shift();
-      refill();
-    } else {
-      // Return pointer to next line
-      *eol = '\0';
-      char* line = httpBuf_+httpPos_;
-      httpPos_ = (eol-httpBuf_) + CRLF_LEN;
-      return line;
-    }
-  }
+THttpClient::~THttpClient() {}
 
-}
-
-void THttpClient::shift() {
-  if (httpBufLen_ > httpPos_) {
-    // Shift down remaining data and read more
-    uint32_t length = httpBufLen_ - httpPos_;
-    memmove(httpBuf_, httpBuf_+httpPos_, length);
-    httpBufLen_ = length;
-  } else {
-    httpBufLen_ = 0;
-  }
-  httpPos_ = 0;
-  httpBuf_[httpBufLen_] = '\0';
-}
-
-void THttpClient::refill() {
-  uint32_t avail = httpBufSize_ - httpBufLen_;
-  if (avail <= (httpBufSize_ / 4)) {
-    httpBufSize_ *= 2;
-    httpBuf_ = (char*)std::realloc(httpBuf_, httpBufSize_+1);
-    if (httpBuf_ == NULL) {
-      throw TTransportException("Out of memory.");
-    }
-  }
-
-  // Read more data
-  uint32_t got = transport_->read((uint8_t*)(httpBuf_+httpBufLen_), httpBufSize_-httpBufLen_);
-  httpBufLen_ += got;
-  httpBuf_[httpBufLen_] = '\0';
-
-  if (got == 0) {
-    throw TTransportException("Could not refill buffer");
+void THttpClient::parseHeader(char* header) {
+  char* colon = strchr(header, ':');
+  if (colon == NULL) {
+    return;
   }
-}
-
-void THttpClient::readHeaders() {
-  // Initialize headers state variables
-  contentLength_ = 0;
-  chunked_ = false;
-  chunkedDone_ = false;
-  chunkSize_ = 0;
-
-  // Control state flow
-  bool statusLine = true;
-  bool finished = false;
-
-  // Loop until headers are finished
-  while (true) {
-    char* line = readLine();
+  uint32_t sz = colon - header;
+  char* value = colon+1;
 
-    if (strlen(line) == 0) {
-      if (finished) {
-        readHeaders_ = false;
-        return;
-      } else {
-        // Must have been an HTTP 100, keep going for another status line
-        statusLine = true;
-      }
-    } else {
-      if (statusLine) {
-        statusLine = false;
-        finished = parseStatusLine(line);
-      } else {
-        parseHeader(line);
-      }
+  if (strncmp(header, "Transfer-Encoding", sz) == 0) {
+    if (strstr(value, "chunked") != NULL) {
+      chunked_ = true;
     }
+  } else if (strncmp(header, "Content-Length", sz) == 0) {
+    chunked_ = false;
+    contentLength_ = atoi(value);
   }
 }
 
@@ -295,28 +83,6 @@ bool THttpClient::parseStatusLine(char* status) {
   }
 }
 
-void THttpClient::parseHeader(char* header) {
-  char* colon = strchr(header, ':');
-  if (colon == NULL) {
-    return;
-  }
-  uint32_t sz = colon - header;
-  char* value = colon+1;
-
-  if (strncmp(header, "Transfer-Encoding", sz) == 0) {
-    if (strstr(value, "chunked") != NULL) {
-      chunked_ = true;
-    }
-  } else if (strncmp(header, "Content-Length", sz) == 0) {
-    chunked_ = false;
-    contentLength_ = atoi(value);
-  }
-}
-
-void THttpClient::write(const uint8_t* buf, uint32_t len) {
-  writeBuffer_.write(buf, len);
-}
-
 void THttpClient::flush() {
   // Fetch the contents of the write buffer
   uint8_t* buf;
@@ -331,7 +97,7 @@ void THttpClient::flush() {
     "Content-Type: application/x-thrift" << CRLF <<
     "Content-Length: " << len << CRLF <<
     "Accept: application/x-thrift" << CRLF <<
-    "User-Agent: C++/THttpClient" << CRLF <<
+    "User-Agent: Thrift/" << VERSION << " (C++/THttpClient)" << CRLF <<
     CRLF;
   string header = h.str();
 
diff --git a/lib/cpp/src/transport/THttpClient.h b/lib/cpp/src/transport/THttpClient.h
index f4be4c1a6..142063d0c 100644
--- a/lib/cpp/src/transport/THttpClient.h
+++ b/lib/cpp/src/transport/THttpClient.h
@@ -20,19 +20,11 @@
 #ifndef _THRIFT_TRANSPORT_THTTPCLIENT_H_
 #define _THRIFT_TRANSPORT_THTTPCLIENT_H_ 1
 
-#include <transport/TBufferTransports.h>
+#include <transport/THttpTransport.h>
 
 namespace apache { namespace thrift { namespace transport {
 
-/**
- * HTTP client implementation of the thrift transport. This was irritating
- * to write, but the alternatives in C++ land are daunting. Linking CURL
- * requires 23 dynamic libraries last time I checked (WTF?!?). All we have
- * here is a VERY basic HTTP/1.1 client which supports HTTP 100 Continue,
- * chunked transfer encoding, keepalive, etc. Tested against Apache.
- *
- */
-class THttpClient : public TTransport {
+class THttpClient : public THttpTransport {
  public:
   THttpClient(boost::shared_ptr<TTransport> transport, std::string host, std::string path="");
 
@@ -40,69 +32,15 @@ class THttpClient : public TTransport {
 
   virtual ~THttpClient();
 
-  void open() {
-    transport_->open();
-  }
-
-  bool isOpen() {
-    return transport_->isOpen();
-  }
-
-  bool peek() {
-    return transport_->peek();
-  }
-
-  void close() {
-    transport_->close();
-  }
-
-  uint32_t read(uint8_t* buf, uint32_t len);
-
-  void readEnd();
-
-  void write(const uint8_t* buf, uint32_t len);
-
-  void flush();
-
- private:
-  void init();
+  virtual void flush();
 
  protected:
 
-  boost::shared_ptr<TTransport> transport_;
-
-  TMemoryBuffer writeBuffer_;
-  TMemoryBuffer readBuffer_;
-
   std::string host_;
   std::string path_;
 
-  bool readHeaders_;
-  bool chunked_;
-  bool chunkedDone_;
-  uint32_t chunkSize_;
-  uint32_t contentLength_;
-
-  char* httpBuf_;
-  uint32_t httpPos_;
-  uint32_t httpBufLen_;
-  uint32_t httpBufSize_;
-
-  uint32_t readMoreData();
-  char* readLine();
-
-  void readHeaders();
-  void parseHeader(char* header);
-  bool parseStatusLine(char* status);
-
-  uint32_t readChunked();
-  void readChunkedFooters();
-  uint32_t parseChunkSize(char* line);
-
-  uint32_t readContent(uint32_t size);
-
-  void refill();
-  void shift();
+  virtual void parseHeader(char* header);
+  virtual bool parseStatusLine(char* status);
 
 };