void fillBuffer () {
size_t n = sizeof(buffer);
char* ptr = reinterpret_cast<char*>(&buffer);
while (0 < n) {
ssize_t r = TRI_READ(fd, ptr, n);
if (r == 0) {
LOGGER_FATAL << "read on random device failed: nothing read";
THROW_INTERNAL_ERROR("read on random device failed");
}
else if (errno == EWOULDBLOCK) {
LOGGER_INFO << "not enough entropy (got " << (sizeof(buffer) - n) << " bytes), switching to pseudo-random";
break;
}
else if (r < 0) {
LOGGER_FATAL << "read on random device failed: " << strerror(errno);
THROW_INTERNAL_ERROR("read on random device failed");
}
ptr += r;
n -= r;
rseed = buffer[0];
LOGGER_TRACE << "using seed " << rseed;
}
if (0 < n) {
unsigned long seed = (unsigned long) time(0);
#ifdef TRI_HAVE_GETTIMEOFDAY
struct timeval tv;
int result = gettimeofday(&tv, 0);
seed ^= static_cast<unsigned long>(tv.tv_sec);
seed ^= static_cast<unsigned long>(tv.tv_usec);
seed ^= static_cast<unsigned long>(result);
#endif
seed ^= static_cast<unsigned long>(Thread::currentProcessId());
mersenneDevice.seed(rseed ^ (uint32_t) seed);
while (0 < n) {
*ptr++ = randomGenerator();
--n;
}
}
pos = 0;
}
void fillBuffer () {
size_t n = sizeof(buffer);
char* ptr = reinterpret_cast<char*>(&buffer);
while (0 < n) {
ssize_t r = TRI_READ(fd, ptr, n);
if (r == 0) {
LOGGER_FATAL_AND_EXIT("read on random device failed: nothing read");
}
else if (r < 0) {
LOGGER_FATAL_AND_EXIT("read on random device failed: " << strerror(errno));
}
ptr += r;
n -= r;
}
pos = 0;
}
string slurp (string const& filename) {
int fd = TRI_OPEN(filename.c_str(), O_RDONLY);
if (fd == -1) {
TRI_set_errno(errno);
THROW_FILE_OPEN_ERROR("open", filename, "O_RDONLY", errno);
}
char buffer[10240];
StringBuffer result(TRI_CORE_MEM_ZONE);
while (true) {
ssize_t n = TRI_READ(fd, buffer, sizeof(buffer));
if (n == 0) {
break;
}
if (n < 0) {
TRI_set_errno(TRI_ERROR_SYS_ERROR);
LOG_TRACE("read failed for '%s' with %s and result %d on fd %d",
filename.c_str(),
strerror(errno),
(int) n,
fd);
TRI_CLOSE(fd);
THROW_FILE_FUNC_ERROR("read", "", errno);
}
result.appendText(buffer, n);
}
TRI_CLOSE(fd);
string r(result.c_str(), result.length());
return r;
}
void fillBuffer () {
size_t n = sizeof(buffer);
char* ptr = reinterpret_cast<char*>(&buffer);
while (0 < n) {
ssize_t r = TRI_READ(fd, ptr, (TRI_read_t) n);
if (r == 0) {
LOG_FATAL_AND_EXIT("read on random device failed: nothing read");
}
else if (errno == EWOULDBLOCK) {
LOG_INFO("not enough entropy (got %d), switching to pseudo-random", (int) (sizeof(buffer) - n));
break;
}
else if (r < 0) {
LOG_FATAL_AND_EXIT("read on random device failed: %s", strerror(errno));
}
ptr += r;
n -= r;
rseed = buffer[0];
LOG_TRACE("using seed %lu", (long unsigned int) rseed);
}
if (0 < n) {
std::mt19937 engine;
unsigned long seed = RandomDevice::getSeed();
engine.seed((uint32_t) (rseed ^ (uint32_t) seed));
while (0 < n) {
*ptr++ = engine();
--n;
}
}
pos = 0;
}
void fillBuffer () {
size_t n = sizeof(buffer);
char* ptr = reinterpret_cast<char*>(&buffer);
while (0 < n) {
ssize_t r = TRI_READ(fd, ptr, n);
if (r == 0) {
LOGGER_FATAL_AND_EXIT("read on random device failed: nothing read");
}
else if (errno == EWOULDBLOCK) {
LOGGER_INFO("not enough entropy (got " << (sizeof(buffer) - n) << " bytes), switching to pseudo-random");
break;
}
else if (r < 0) {
LOGGER_FATAL_AND_EXIT("read on random device failed: " << strerror(errno));
}
ptr += r;
n -= r;
rseed = buffer[0];
LOGGER_TRACE("using seed " << rseed);
}
if (0 < n) {
unsigned long seed = RandomDevice::getSeed();
TRI_SeedMersenneTwister((uint32_t) (rseed ^ (uint32_t) seed));
while (0 < n) {
*ptr++ = TRI_Int32MersenneTwister();
--n;
}
}
pos = 0;
}
void slurp (string const& filename, StringBuffer& result) {
int fd = TRI_OPEN(filename.c_str(), O_RDONLY);
if (fd == -1) {
THROW_FILE_OPEN_ERROR("open", filename, "O_RDONLY", errno);
}
// reserve space in the output buffer
off_t fileSize = size(filename);
if (fileSize > 0) {
result.reserve((size_t) fileSize);
}
char buffer[10240];
while (true) {
ssize_t n = TRI_READ(fd, buffer, sizeof(buffer));
if (n == 0) {
break;
}
if (n < 0) {
TRI_CLOSE(fd);
LOG_TRACE("read failed for '%s' with %s and result %d on fd %d",
filename.c_str(),
strerror(errno),
(int) n,
fd);
THROW_FILE_FUNC_ERROR("read", "", errno);
}
result.appendText(buffer, n);
}
TRI_CLOSE(fd);
}
bool ImportHelper::importDelimited (string const& collectionName,
string const& fileName,
DelimitedImportType typeImport) {
_collectionName = collectionName;
_firstLine = "";
_numberLines = 0;
_numberOk = 0;
_numberError = 0;
_outputBuffer.clear();
_lineBuffer.clear();
_errorMessage = "";
_hasError = false;
// read and convert
int fd;
int64_t totalLength;
if (fileName == "-") {
// we don't have a filesize
totalLength = 0;
fd = STDIN_FILENO;
}
else {
// read filesize
totalLength = TRI_SizeFile(fileName.c_str());
fd = TRI_OPEN(fileName.c_str(), O_RDONLY);
if (fd < 0) {
_errorMessage = TRI_LAST_ERROR_STR;
return false;
}
}
// progress display control variables
int64_t totalRead = 0;
double nextProgress = ProgressStep;
size_t separatorLength;
char* separator = TRI_UnescapeUtf8StringZ(TRI_UNKNOWN_MEM_ZONE, _separator.c_str(), _separator.size(), &separatorLength);
if (separator == 0) {
if (fd != STDIN_FILENO) {
TRI_CLOSE(fd);
}
_errorMessage = "out of memory";
return false;
}
TRI_csv_parser_t parser;
TRI_InitCsvParser(&parser,
TRI_UNKNOWN_MEM_ZONE,
ProcessCsvBegin,
ProcessCsvAdd,
ProcessCsvEnd);
TRI_SetSeparatorCsvParser(&parser, separator[0]);
// in csv, we'll use the quote char if set
// in tsv, we do not use the quote char
if (typeImport == ImportHelper::CSV && _quote.size() > 0) {
TRI_SetQuoteCsvParser(&parser, _quote[0], true);
}
else {
TRI_SetQuoteCsvParser(&parser, '\0', false);
}
parser._dataAdd = this;
_rowOffset = 0;
_rowsRead = 0;
char buffer[32768];
while (! _hasError) {
ssize_t n = TRI_READ(fd, buffer, sizeof(buffer));
if (n < 0) {
TRI_Free(TRI_UNKNOWN_MEM_ZONE, separator);
TRI_DestroyCsvParser(&parser);
if (fd != STDIN_FILENO) {
TRI_CLOSE(fd);
}
_errorMessage = TRI_LAST_ERROR_STR;
return false;
}
else if (n == 0) {
break;
}
totalRead += (int64_t) n;
reportProgress(totalLength, totalRead, nextProgress);
TRI_ParseCsvString2(&parser, buffer, n);
}
if (_outputBuffer.length() > 0) {
sendCsvBuffer();
}
TRI_DestroyCsvParser(&parser);
TRI_Free(TRI_UNKNOWN_MEM_ZONE, separator);
if (fd != STDIN_FILENO) {
TRI_CLOSE(fd);
}
_outputBuffer.clear();
return !_hasError;
}
bool ImportHelper::importJson (const string& collectionName, const string& fileName) {
_collectionName = collectionName;
_firstLine = "";
_numberLines = 0;
_numberOk = 0;
_numberError = 0;
_outputBuffer.clear();
_errorMessage = "";
_hasError = false;
// read and convert
int fd;
int64_t totalLength;
if (fileName == "-") {
// we don't have a filesize
totalLength = 0;
fd = STDIN_FILENO;
}
else {
// read filesize
totalLength = TRI_SizeFile(fileName.c_str());
fd = TRI_OPEN(fileName.c_str(), O_RDONLY);
if (fd < 0) {
_errorMessage = TRI_LAST_ERROR_STR;
return false;
}
}
bool isArray = false;
bool checkedFront = false;
// progress display control variables
int64_t totalRead = 0;
double nextProgress = ProgressStep;
static const int BUFFER_SIZE = 32768;
while (! _hasError) {
// reserve enough room to read more data
if (_outputBuffer.reserve(BUFFER_SIZE) == TRI_ERROR_OUT_OF_MEMORY) {
_errorMessage = TRI_errno_string(TRI_ERROR_OUT_OF_MEMORY);
if (fd != STDIN_FILENO) {
TRI_CLOSE(fd);
}
return false;
}
// read directly into string buffer
ssize_t n = TRI_READ(fd, _outputBuffer.end(), BUFFER_SIZE - 1);
if (n < 0) {
_errorMessage = TRI_LAST_ERROR_STR;
if (fd != STDIN_FILENO) {
TRI_CLOSE(fd);
}
return false;
}
else if (n == 0) {
// we're done
break;
}
// adjust size of the buffer by the size of the chunk we just read
_outputBuffer.increaseLength(n);
if (! checkedFront) {
// detect the import file format (single lines with individual JSON objects
// or a JSON array with all documents)
char const* p = _outputBuffer.begin();
char const* e = _outputBuffer.end();
while (p < e &&
(*p == ' ' || *p == '\r' || *p == '\n' || *p == '\t' || *p == '\f' || *p == '\b')) {
++p;
}
isArray = (*p == '[');
checkedFront = true;
}
totalRead += (int64_t) n;
reportProgress(totalLength, totalRead, nextProgress);
if (_outputBuffer.length() > _maxUploadSize) {
if (isArray) {
if (fd != STDIN_FILENO) {
TRI_CLOSE(fd);
}
_errorMessage = "import file is too big. please increase the value of --batch-size (currently " + StringUtils::itoa(_maxUploadSize) + ")";
return false;
}
// send all data before last '\n'
char const* first = _outputBuffer.c_str();
char* pos = (char*) memrchr(first, '\n', _outputBuffer.length());
if (pos != 0) {
size_t len = pos - first + 1;
sendJsonBuffer(first, len, isArray);
_outputBuffer.erase_front(len);
}
}
}
if (_outputBuffer.length() > 0) {
sendJsonBuffer(_outputBuffer.c_str(), _outputBuffer.length(), isArray);
}
_numberLines = _numberError + _numberOk;
if (fd != STDIN_FILENO) {
TRI_CLOSE(fd);
}
_outputBuffer.clear();
return ! _hasError;
}
bool ImportHelper::importJson (const string& collectionName, const string& fileName) {
_collectionName = collectionName;
_firstLine = "";
_numberLines = 0;
_numberOk = 0;
_numberError = 0;
_outputBuffer.clear();
_errorMessage = "";
_hasError = false;
// read and convert
int fd;
int64_t totalLength;
if (fileName == "-") {
// we don't have a filesize
totalLength = 0;
fd = STDIN_FILENO;
}
else {
// read filesize
totalLength = TRI_SizeFile(fileName.c_str());
fd = TRI_OPEN(fileName.c_str(), O_RDONLY);
if (fd < 0) {
_errorMessage = TRI_LAST_ERROR_STR;
return false;
}
}
char buffer[32768];
bool isArray = false;
bool checkedFront = false;
// progress display control variables
int64_t totalRead = 0;
double nextProgress = ProgressStep;
while (! _hasError) {
ssize_t n = TRI_READ(fd, buffer, sizeof(buffer));
if (n < 0) {
_errorMessage = TRI_LAST_ERROR_STR;
if (fd != STDIN_FILENO) {
TRI_CLOSE(fd);
}
return false;
}
else if (n == 0) {
// we're done
break;
}
if (! checkedFront) {
// detect the import file format (single lines with individual JSON objects
// or a JSON array with all documents)
const string firstChar = StringUtils::lTrim(string(buffer, n), "\r\n\t\f\b ").substr(0, 1);
isArray = (firstChar == "[");
checkedFront = true;
}
_outputBuffer.appendText(buffer, n);
totalRead += (int64_t) n;
reportProgress(totalLength, totalRead, nextProgress);
if (_outputBuffer.length() > _maxUploadSize) {
if (isArray) {
if (fd != STDIN_FILENO) {
TRI_CLOSE(fd);
}
_errorMessage = "import file is too big.";
return false;
}
// send all data before last '\n'
const char* first = _outputBuffer.c_str();
char* pos = (char*) memrchr(first, '\n', _outputBuffer.length());
if (pos != 0) {
size_t len = pos - first + 1;
sendJsonBuffer(first, len, isArray);
_outputBuffer.erase_front(len);
}
}
}
if (_outputBuffer.length() > 0) {
sendJsonBuffer(_outputBuffer.c_str(), _outputBuffer.length(), isArray);
}
_numberLines = _numberError + _numberOk;
if (fd != STDIN_FILENO) {
TRI_CLOSE(fd);
}
_outputBuffer.clear();
return ! _hasError;
}
