Date: Monday, December 19, 2005 @ 16:11:14
Author: marc
Path: /cvsroot/carob/carob
Modified: include/JavaSocket.hpp (1.16 -> 1.17) src/DriverSocket.cpp (1.10
-> 1.11) src/JavaSocket.cpp (1.22 -> 1.23)
Fix CAROB-49: now sending/receiving real UTF-8 to the controller: no more
dependency on the global
locale, and also supporting nulls inside strings. Controller does not yet
support nulls inside strings,
see SEQUOIA-133.
------------------------+
include/JavaSocket.hpp | 12 +++---
src/DriverSocket.cpp | 5 ++
src/JavaSocket.cpp | 88 +++++++++++++++++------------------------------
3 files changed, 43 insertions(+), 62 deletions(-)
Index: carob/include/JavaSocket.hpp
diff -u carob/include/JavaSocket.hpp:1.16 carob/include/JavaSocket.hpp:1.17
--- carob/include/JavaSocket.hpp:1.16 Tue Dec 13 16:08:30 2005
+++ carob/include/JavaSocket.hpp Mon Dec 19 16:11:14 2005
@@ -85,21 +85,23 @@
virtual bool closeSocket() throw (SocketIOException, UnexpectedException);
/**
- * Writes a string to the socket
+ * Writes a UTF-8 encoded wstring to the socket. This is NOT Java's
+ * modified UTF-8, see SEQUOIA-133.
* @param str string to write
* @return the number of bytes send
* @throws SocketIOException
*/
size_t writeJavaUTF(const std::wstring& str) const
- throw (SocketIOException, UnexpectedException);
+ throw (SocketIOException, CodecException, UnexpectedException);
/**
- * Reads java encoded string from socket.
+ * Reads UTF-8 encoded string from socket. This is NOT Java's
+ * modified UTF-8, see SEQUOIA-133.
* @param str string to read
* @return the length read in bytes
* @throws SocketIOException
*/
- size_t readJavaUTF(std::wstring& str) const throw (SocketIOException,
- UnexpectedException);
+ size_t readJavaUTF(std::wstring& str) const
+ throw (SocketIOException, CodecException, UnexpectedException);
/**
* Writes 32bits-integer to socket
* @param i integer to send
Index: carob/src/DriverSocket.cpp
diff -u carob/src/DriverSocket.cpp:1.10 carob/src/DriverSocket.cpp:1.11
--- carob/src/DriverSocket.cpp:1.10 Tue Dec 13 16:08:30 2005
+++ carob/src/DriverSocket.cpp Mon Dec 19 16:11:14 2005
@@ -50,6 +50,7 @@
size_t strLen = s.size();
*this<<(int32_t)strLen;
size_t totalWritten = 0;
+ // FIXME: catch CodecException here
while (totalWritten < strLen)
{
totalWritten += writeJavaUTF(s);
@@ -74,7 +75,7 @@
*this>>ack;
if (ack)
{
- //First reads the size of the upcomming string, then the string itself.
+ //First reads the size of the upcoming string, then the string itself.
int strSize;
*this>>strSize;
if (strSize > 0)
@@ -82,8 +83,10 @@
s.reserve(strSize+1);
size_t sizeRead = 0;
//We must read strSize character, which can be done in several passes
+ // Actually not! - MH
while ((int)sizeRead<strSize)
{
+ // FIXME: catch CodecException here
sizeRead += readJavaUTF(s);
}
}
Index: carob/src/JavaSocket.cpp
diff -u carob/src/JavaSocket.cpp:1.22 carob/src/JavaSocket.cpp:1.23
--- carob/src/JavaSocket.cpp:1.22 Fri Dec 16 21:27:26 2005
+++ carob/src/JavaSocket.cpp Mon Dec 19 16:11:14 2005
@@ -185,54 +185,34 @@
// /usr/share/doc/gcc-4.0-base/libstdc++/html/documentation.html
size_t JavaSocket::writeJavaUTF(const wstring& str) const
- throw (SocketIOException, UnexpectedException)
+ throw (SocketIOException, CodecException, UnexpectedException)
{
wstring fctName(L"JavaSocket::writeJavaUTF");
- // Size of the wide string in bytes, including the wide terminating null
- uint16_t strlenPlusOne = (str.length()+1)*sizeof(wchar_t);
-
- // FIXME: we don't support inline zeros. We should get rid of c_str
- const wchar_t* oriStr = str.c_str();
- // guessing the max converted size like this is hopelessly wrong. By chance
it's usually enough.
- uint8_t* utfStr = new uint8_t[strlenPlusOne];
- size_t countConverted;
- mbstate_t mbstate = {0};
-
if (isVerboseEnabled())
logVerbose(fctName, L"Converting string...");
- // FIXME: don't use this locale converter, we need to send UTF-8!
- // we should use: toUTF8() instead
- countConverted = wcsrtombs((char*)utfStr, &oriStr,
- strlenPlusOne, &mbstate);
- if (countConverted == (size_t)-1)
- {
- throw SocketIOException(fctName + L"An encoding error occured while
converting the string.");
- }
- else if (countConverted > 0) //don't send anything if empty string
- {
- if (isVerboseEnabled())
- logVerbose(fctName, L"String was successfuly converted, sending string
length");
- uint16_t netlen = htons(countConverted);
+ std::string utf8str(toUTF8(str));
- //First write number of bytes to follow as
- if (sendToSocket(fctName, L"UTF string", &netlen, sizeof(netlen),
MSG_NOSIGNAL))
- {
- if (isVerboseEnabled())
- logVerbose(fctName, L"Ok. Sending string");
- // Send the encoded string (wcsrtombs() does not count the terminating
zero)
- sendToSocket(fctName, L"UTF string", utfStr, countConverted,
MSG_NOSIGNAL);
- }
+ // FIXME: move netlen to uint32_t once SEQUOIA-133 is complete
+ uint16_t netlen = htons(utf8str.length());
+
+ //First write number of bytes to follow as
+ if (sendToSocket(fctName, L"UTF string", &netlen, sizeof(netlen),
MSG_NOSIGNAL))
+ {
+ if (isVerboseEnabled())
+ logVerbose(fctName, L"Ok. Sending string");
+ sendToSocket(fctName, L"UTF string", utf8str.data(), utf8str.length(),
MSG_NOSIGNAL);
}
- delete[] utfStr;
- return countConverted;
+
+ return utf8str.length();
}
-size_t JavaSocket::readJavaUTF(wstring& s) const throw (SocketIOException,
- UnexpectedException)
+size_t JavaSocket::readJavaUTF(wstring& s) const
+ throw (SocketIOException, CodecException, UnexpectedException)
{
wstring fctName(L"JavaSocket::readJavaUTF");
+ // FIXME: move lenRecNet to uint32_t once SEQUOIA-133 is complete
//the size read on the network
uint16_t lenRecNet;
//the converted size
@@ -242,30 +222,18 @@
if (receiveFromSocket(fctName, L"UTF string size", &lenRecNet,
sizeof(lenRecNet), 0))
{
lenRec = ntohs(lenRecNet); // number of bytes to come
-
- uint8_t* utfStr = new uint8_t[lenRec+1]; // add a zero to C-terminate it
- utfStr[lenRec] = 0;
+
+ uint8_t* utfStr = new uint8_t[lenRec];
if (receiveFromSocket(fctName, L"UTF string", utfStr, lenRec, 0))
{
- // FIXME: we don't support inline zeros. We should get rid of c_str
- std::string received((const char*)utfStr);
+ const std::string received((const char*)utfStr, lenRec);
- // FIXME: don't use this locale converter cause received is ALWAYS UTF-8!
- // this is correct:
- // s = fromUTF8(received);
- s = fromString(received);
-
-#if 0 // FIXME catch conversion exception
- {
- delete[] utfStr;
- throw SocketIOException(fctName
- + L"An error occured while converting the string.");
+ try {
+ s = fromUTF8(received);
+ } catch (CodecException) {
+ delete[] utfStr; throw;
}
-#endif
-
- sizeRead = (size_t)lenRec;
-
-
+ sizeRead = (size_t)lenRec;
}
delete[] utfStr;
}
@@ -394,3 +362,11 @@
return true;
}
}
+
+/*
+ * Local Variables:
+ * c-file-style: "bsd"
+ * c-basic-offset: 2
+ * indent-tabs-mode: nil
+ * End:
+ */
_______________________________________________
Carob-commits mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/carob-commits
