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;
}
