ssize_t HdfsFile::read(void *ptr, size_t count)
{
size_t offset = tell();
tSize ret = hdfsRead(m_fs, m_file, ptr, count);
// see comment in pread
if( ret < 0 )
{
reopen();
if (hdfsSeek(m_fs, m_file, offset) == 0)
ret = hdfsRead(m_fs, m_file, ptr, count);
}
else if ((size_t) ret < count) // ret >= 0
{
// retry
int retryCount = 0;
while (ret >= 0 && ret != (tSize) count && ++retryCount < 40)
{
ssize_t n = hdfsRead(m_fs, m_file, (char*)ptr+ret, count-ret);
if (n >= 0)
ret += n;
else
ret = n;
}
}
if( IDBLogger::isEnabled() )
IDBLogger::logRW("read", m_fname, this, offset, count, ret);
return ret;
}
qioerr hdfs_readv (void* file, const struct iovec *vector, int count, ssize_t* num_read_out, void* fs)
{
ssize_t got;
ssize_t got_total;
qioerr err_out = 0;
int i;
STARTING_SLOW_SYSCALL;
got_total = 0;
for( i = 0; i < count; i++) {
got = hdfsRead(to_hdfs_fs(fs)->hfs, to_hdfs_file(file)->file, (void*)vector[i].iov_base, vector[i].iov_len);
if( got != -1 ) {
got_total += got;
} else {
err_out = qio_mkerror_errno();
break;
}
if( got != sys_iov_total_bytes(&vector[i], i)) {
break;
}
}
if( err_out == 0 && got_total == 0 && sys_iov_total_bytes(vector, count) != 0 )
err_out = qio_int_to_err(EEOF);
*num_read_out = got_total;
DONE_SLOW_SYSCALL;
return err_out;
}
static PyObject *pyhdfsFS_read(pyhdfsFS *self, PyObject *args) {
int n;
PyArrayObject *result;
if (!PyArg_ParseTuple(args, "i", &n)) {
return NULL;
}
char *buffer = (char*) malloc(n);
int len = hdfsRead(self->fs, self->file, (void*)buffer, n);
if (len <= 0)
{
Py_INCREF(Py_None);
return Py_None;
}
int dim[1];
dim[0] = len;
result = (PyArrayObject *) PyArray_SimpleNew(1, dim, PyArray_CHAR);
memcpy(result->data,buffer,len);
free(buffer);
return PyArray_Return(result);
}
int read(char * start, size_t size)
{
int bytes_read = hdfsRead(fs.get(), fin, start, size);
if (bytes_read < 0)
throw Exception("Fail to read HDFS file: " + hdfs_uri.toString() + " " + std::string(hdfsGetLastError()),
ErrorCodes::NETWORK_ERROR);
return bytes_read;
}
int main(int argc, char **argv) {
const char* rfile;
tSize fileTotalSize, bufferSize, curSize, totalReadSize;
hdfsFS fs;
hdfsFile readFile;
char *buffer = NULL;
if (argc != 4) {
fprintf(stderr, "Usage: test_libwebhdfs_read"
" <filename> <filesize> <buffersize>\n");
exit(1);
}
fs = hdfsConnect("localhost", 50070);
if (!fs) {
fprintf(stderr, "Oops! Failed to connect to hdfs!\n");
exit(1);
}
rfile = argv[1];
fileTotalSize = strtoul(argv[2], NULL, 10);
bufferSize = strtoul(argv[3], NULL, 10);
readFile = hdfsOpenFile(fs, rfile, O_RDONLY, bufferSize, 0, 0);
if (!readFile) {
fprintf(stderr, "Failed to open %s for writing!\n", rfile);
exit(1);
}
// data to be written to the file
buffer = malloc(sizeof(char) * bufferSize);
if(buffer == NULL) {
fprintf(stderr, "Failed to allocate buffer.\n");
exit(1);
}
// read from the file
curSize = bufferSize;
totalReadSize = 0;
for (; (curSize = hdfsRead(fs, readFile, buffer, bufferSize)) == bufferSize; ) {
totalReadSize += curSize;
}
totalReadSize += curSize;
fprintf(stderr, "size of the file: %d; reading size: %d\n",
fileTotalSize, totalReadSize);
free(buffer);
hdfsCloseFile(fs, readFile);
hdfsDisconnect(fs);
return 0;
}
int main(int argc, char **argv) {
if (argc != 3) {
fprintf(stderr, "Usage: hdfs_read <filename> <buffersize>\n");
exit(-1);
}
hdfsFS fs = hdfsConnect("default", 0);
if (!fs) {
fprintf(stderr, "Oops! Failed to connect to hdfs!\n");
exit(-2);
}
const char* rfile = argv[1];
tSize bufferSize = strtoul(argv[2], NULL, 10);
hdfsFile readFile = hdfsOpenFile(fs, rfile, O_RDONLY, bufferSize, 0, 0);
if (!readFile) {
fprintf(stderr, "Failed to open %s for writing!\n", rfile);
exit(-3);
}
FILE *outf = fopen(rfile, "wb");
if (outf == NULL) {
printf("FILEIO error %d\n", errno);
exit(-4);
}
// data to be written to the file
char* buffer = malloc(sizeof(char) * bufferSize);
if(buffer == NULL) {
return -5;
}
//printf("buffersize is %d\n", bufferSize);
// read from the file
tSize curSize = bufferSize;
for (; curSize == bufferSize;) {
//printf("cursize before is %d\n", curSize);
curSize = hdfsRead(fs, readFile, (void*)buffer, curSize);
//printf("cursize is %d, errno is %d\n", curSize, errno);
fwrite((void *)buffer, sizeof(char), curSize, outf);
//printf("%.*s", bufferSize, buffer);
}
fclose(outf);
free(buffer);
hdfsCloseFile(fs, readFile);
hdfsDisconnect(fs);
return 0;
}
uint32_t THDFSFileTransport::read(uint8_t* buf, uint32_t len) {
tSize rv = hdfsRead(hdfsFile_->getFS()->getHandle(), (hdfsFile)hdfsFile_->getHandle(), buf, len);
if (rv < 0) {
int errno_copy = errno;
throw TTransportException(TTransportException::UNKNOWN,
"THDFSFileTransport::read()",
errno_copy);
} else if (rv == 0) {
throw TTransportException(TTransportException::END_OF_FILE,
"THDFSFileTransport::read()");
}
return rv;
}
int HDFSFileSplitter::read_block(const std::string& fn) {
file_ = hdfsOpenFile(fs_, fn.c_str(), O_RDONLY, 0, 0, 0);
assert(file_ != NULL);
hdfsSeek(fs_, file_, offset_);
size_t start = 0;
size_t nbytes = 0;
while (start < hdfs_block_size) {
// only 128KB per hdfsRead
nbytes = hdfsRead(fs_, file_, data_ + start, hdfs_block_size);
start += nbytes;
if (nbytes == 0)
break;
}
return start;
}
int main(int argc, char **argv) {
hdfsFS fs;
char* rfile;
int bufferSize;
hdfsFile readFile;
char* buffer;
int curSize;
if (argc != 4) {
fprintf(stderr, "Usage: hdfs_read <filename> <filesize> <buffersize>\n");
exit(-1);
}
fs = hdfsConnect("default", 0);
if (!fs) {
fprintf(stderr, "Oops! Failed to connect to hdfs!\n");
exit(-1);
}
rfile = argv[1];
bufferSize = strtoul(argv[3], NULL, 10);
readFile = hdfsOpenFile(fs, rfile, O_RDONLY, bufferSize, 0, 0);
if (!readFile) {
fprintf(stderr, "Failed to open %s for writing!\n", rfile);
exit(-2);
}
/* data to be written to the file */
buffer = malloc(sizeof(char) * bufferSize);
if(buffer == NULL) {
return -2;
}
/* read from the file */
curSize = bufferSize;
for (; curSize == bufferSize;) {
curSize = hdfsRead(fs, readFile, (void*)buffer, curSize);
}
free(buffer);
hdfsCloseFile(fs, readFile);
hdfsDisconnect(fs);
return 0;
}
void ClaimsHDFS::claimsRead(){
hdfsFS fs;
hdfsFile fd;
string filename="/home/casa/data/kmeans_data.txt";
fs=hdfsConnect("10.11.1.174",9000);
fd=hdfsOpenFile(fs,filename.c_str(),O_RDONLY,0,0,0);
if(!fd){
cout<<"failed to open hdfs file!!!"<<endl;
}
char array[72];
tSize bytes=hdfsRead(fs,fd,array,72);
cout<<"string is: "<<array<<endl;
hdfsCloseFile(fs,fd);
hdfsDisconnect(fs);
}
int libhdfsconnector::streamInFile(const char * rfile, int bufferSize)
{
if (!fs)
{
fprintf(stderr, "Could not connect to hdfs on");
return RETURN_FAILURE;
}
unsigned long fileTotalSize = 0;
hdfsFileInfo *fileInfo = NULL;
if ((fileInfo = hdfsGetPathInfo(fs, rfile)) != NULL)
{
fileTotalSize = fileInfo->mSize;
hdfsFreeFileInfo(fileInfo, 1);
}
else
{
fprintf(stderr, "Error: hdfsGetPathInfo for %s - FAILED!\n", rfile);
return RETURN_FAILURE;
}
hdfsFile readFile = hdfsOpenFile(fs, rfile, O_RDONLY, bufferSize, 0, 0);
if (!readFile)
{
fprintf(stderr, "Failed to open %s for writing!\n", rfile);
return RETURN_FAILURE;
}
unsigned char buff[bufferSize + 1];
buff[bufferSize] = '\0';
for (unsigned long bytes_read = 0; bytes_read < fileTotalSize;)
{
unsigned long read_length = hdfsRead(fs, readFile, buff, bufferSize);
bytes_read += read_length;
for (unsigned long i = 0; i < read_length; i++)
fprintf(stdout, "%c", buff[i]);
}
hdfsCloseFile(fs, readFile);
return 0;
}
bool FileManagerHdfs::readBlockOrBlob(const block_id block,
void *buffer,
const size_t length) {
DEBUG_ASSERT(buffer);
DEBUG_ASSERT(length % kSlotSizeBytes == 0);
string filename(blockFilename(block));
hdfsFile file_handle = hdfsOpenFile(hdfs_,
filename.c_str(),
O_RDONLY,
kSlotSizeBytes,
FLAGS_hdfs_num_replications,
kSlotSizeBytes);
if (file_handle == nullptr) {
LOG_WARNING("Failed to open file " << filename << " with error: " << strerror(errno));
return false;
}
size_t bytes_total = 0;
while (bytes_total < length) {
tSize bytes = hdfsRead(hdfs_, file_handle, static_cast<char*>(buffer) + bytes_total, length - bytes_total);
if (bytes > 0) {
bytes_total += bytes;
} else if (bytes == -1) {
if (errno != EINTR) {
LOG_WARNING("Failed to read file " << filename << " with error: " << strerror(errno));
break;
}
} else {
LOG_WARNING("Failed to read file " << filename << " since EOF was reached unexpectedly");
break;
}
}
if (hdfsCloseFile(hdfs_, file_handle) != 0) {
LOG_WARNING("Failed to close file " << filename << " with error: " << strerror(errno));
}
return (bytes_total == length);
}
tSize hdfsReadFully(hdfsFS fs, hdfsFile f, void* buffer, tSize length)
{
uint8_t *buf = buffer;
tSize ret, nread = 0;
while (length > 0) {
ret = hdfsRead(fs, f, buf, length);
if (ret < 0) {
if (errno != EINTR) {
return -1;
}
}
if (ret == 0) {
break;
}
nread += ret;
length -= ret;
buf += ret;
}
return nread;
}
int main(int argc, char* argv[]) {
if (argc < 4) {
printf("usage: hdfs_get <name node address> <name node port> <input file>\n");
return 1;
}
// Sleep for 100ms.
usleep(100 * 1000);
struct hdfsBuilder* hdfs_builder = hdfsNewBuilder();
if (!hdfs_builder) {
printf("Could not create HDFS builder");
return 1;
}
hdfsBuilderSetNameNode(hdfs_builder, argv[1]);
int port = atoi(argv[2]);
hdfsBuilderSetNameNodePort(hdfs_builder, port);
hdfsBuilderConfSetStr(hdfs_builder, "dfs.client.read.shortcircuit", "false");
hdfsFS fs = hdfsBuilderConnect(hdfs_builder);
hdfsFreeBuilder(hdfs_builder);
if (!fs) {
printf("Could not connect to HDFS");
return 1;
}
hdfsFile file_in = hdfsOpenFile(fs, argv[3], O_RDONLY, 0, 0, 0);
char buffer[1048576];
int done = 0;
do {
done = hdfsRead(fs, file_in, &buffer, 1048576);
} while (done > 0);
if (done < 0) {
printf("Failed to read file: %s", hdfsGetLastError());
return 1;
}
hdfsCloseFile(fs, file_in);
hdfsDisconnect(fs);
return 0;
}
int libhdfsconnector::streamFlatFileOffset(const char * filename, unsigned long seekPos, unsigned long readlen,unsigned long bufferSize, int maxretries)
{
hdfsFile readFile = hdfsOpenFile(fs, filename, O_RDONLY, 0, 0, 0);
if (!readFile)
{
fprintf(stderr, "Failed to open %s for reading!\n", filename);
return EXIT_FAILURE;
}
if (hdfsSeek(fs, readFile, seekPos))
{
fprintf(stderr, "Failed to seek %s for reading!\n", filename);
return EXIT_FAILURE;
}
unsigned char buffer[bufferSize + 1];
unsigned long currentPos = seekPos;
fprintf(stderr, "\n--Start piping: %ld--\n", currentPos);
unsigned long bytesLeft = readlen;
while (hdfsAvailable(fs, readFile) && bytesLeft > 0)
{
tSize num_read_bytes = hdfsRead(fs, readFile, buffer, bytesLeft < bufferSize ? bytesLeft : bufferSize);
if (num_read_bytes <= 0)
break;
bytesLeft -= num_read_bytes;
for (int i = 0; i < num_read_bytes; i++, currentPos++)
fprintf(stdout, "%c", buffer[i]);
}
fprintf(stderr, "--\nStop Streaming: %ld--\n", currentPos);
hdfsCloseFile(fs, readFile);
return EXIT_SUCCESS;
}
boost::string_ref HDFSFileSplitter::fetch_block(bool is_next) {
int nbytes = 0;
if (is_next) {
// directly read the next block using the current file
nbytes = hdfsRead(fs_, file_, data_, hdfs_block_size);
if (nbytes == 0)
return "";
if (nbytes == -1) {
throw base::HuskyException("read next block error!");
}
} else {
// Ask the master for a new block
BinStream question;
question << url_ << husky::Context::get_param("hostname");
BinStream answer = husky::Context::get_coordinator()->ask_master(question, husky::TYPE_HDFS_BLK_REQ);
std::string fn;
answer >> fn;
answer >> offset_;
if (fn == "") {
// no more files
return "";
}
if (file_ != NULL) {
int rc = hdfsCloseFile(fs_, file_);
assert(rc == 0);
// Notice that "file" will be deleted inside hdfsCloseFile
file_ = NULL;
}
// read block
nbytes = read_block(fn);
}
return boost::string_ref(data_, nbytes);
}
std::streamsize read(char* strm_ptr, std::streamsize n) {
return hdfsRead(filesystem, file, strm_ptr, n);
} // end of read
/**
* hdfsCopy - Copy file from one filesystem to another.
*
* @param srcFS The handle to source filesystem.
* @param src The path of source file.
* @param dstFS The handle to destination filesystem.
* @param dst The path of destination file.
* @return Returns 0 on success, -1 on error.
*/
int
hdfsCopy(hdfsFS srcFS, const char* src, hdfsFS dstFS, const char* dst)
{
char *block = NULL, *src_abs, *dst_abs;
hdfsFileInfo *srcinfo = NULL;
int res = -1;
hdfsFile a = NULL, b = NULL;
tOffset twritten = 0;
src_abs = _makeabs(srcFS, src);
dst_abs = _makeabs(dstFS, dst);
if (hdfsExists(srcFS, src_abs) == -1) {
ERR(ENOENT, "'%s' doesn't exist on srcFS", src_abs);
goto out;
}
srcinfo = hdfsGetPathInfo(srcFS, src_abs);
if (!srcinfo) {
ERR(errno, "hdfsGetPathInfo failed");
goto out;
}
if (srcinfo->mKind == kObjectKindDirectory) {
ERR(ENOTSUP, "hdfsCopy can't do directories right now");
goto out;
}
a = hdfsOpenFile(srcFS, src_abs, O_RDONLY, 0, 0, 0);
if (!a) {
ERR(errno, "hdfsOpenFile failed");
goto out;
}
b = hdfsOpenFile(dstFS, dst_abs, O_WRONLY, 0, DEFAULT_REPLICATION,
DEFAULT_BLOCK_SIZE);
if (!b) {
ERR(errno, "hdfsOpenFile failed");
goto out;
}
block = malloc(DEFAULT_BLOCK_SIZE);
assert(block);
while (twritten < srcinfo->mSize) {
tSize toread, read, written;
toread = _imin(DEFAULT_BLOCK_SIZE, srcinfo->mSize - twritten);
read = hdfsRead(srcFS, a, block, toread);
if (read == -1) {
ERR(errno, "hdfsRead failed");
goto out;
}
written = hdfsWrite(dstFS, b, block, read);
if (written == -1) {
ERR(errno, "hdfsWrite failed");
goto out;
}
assert(written == read);
twritten += written;
}
res = 0;
out:
if (a)
hdfsCloseFile(srcFS, a);
if (b)
hdfsCloseFile(dstFS, b);
if (src_abs != src)
free(src_abs);
if (dst_abs != dst)
free(dst_abs);
if (block)
free(block);
if (srcinfo)
hdfsFreeFileInfo(srcinfo, 1);
return res;
}
int libhdfsconnector::readXMLOffset(const char * filename, unsigned long seekPos, unsigned long readlen,
const char * rowTag, const char * headerText, const char * footerText, unsigned long bufferSize)
{
string xmlizedxpath;
string elementname;
string rootelement;
xpath2xml(&xmlizedxpath, rowTag, true);
getLastXPathElement(&elementname, rowTag);
hdfsFile readFile = hdfsOpenFile(fs, filename, O_RDONLY, 0, 0, 0);
if (!readFile)
{
fprintf(stderr, "Failed to open %s for reading!\n", filename);
return EXIT_FAILURE;
}
if (hdfsSeek(fs, readFile, seekPos))
{
fprintf(stderr, "Failed to seek %s for reading!\n", filename);
return EXIT_FAILURE;
}
unsigned char buffer[bufferSize + 1];
bool firstRowfound = false;
string openRowTag("<");
openRowTag.append(elementname).append(1, '>');
string closeRowTag("</");
closeRowTag.append(elementname).append(1, '>');
string closeRootTag("</");
getLastXMLElement(&closeRootTag, footerText);
closeRootTag.append(1, '>');
unsigned long currentPos = seekPos + openRowTag.size();
string currentTag("");
bool withinRecord = false;
bool stopAtNextClosingTag = false;
bool parsingTag = false;
fprintf(stderr, "--Start looking <%s>: %ld--\n", elementname.c_str(), currentPos);
fprintf(stdout, "%s", xmlizedxpath.c_str());
unsigned long bytesLeft = readlen;
while (hdfsAvailable(fs, readFile) && bytesLeft > 0)
{
tSize numOfBytesRead = hdfsRead(fs, readFile, (void*) buffer, bufferSize);
if (numOfBytesRead <= 0)
{
fprintf(stderr, "\n--Hard Stop at: %ld--\n", currentPos);
break;
}
for (int buffIndex = 0; buffIndex < numOfBytesRead;)
{
char currChar = buffer[buffIndex];
if (currChar == '<' || parsingTag)
{
if (!parsingTag)
currentTag.clear();
int tagpos = buffIndex;
while (tagpos < numOfBytesRead)
{
currentTag.append(1, buffer[tagpos++]);
if (buffer[tagpos - 1] == '>')
break;
}
if (tagpos == numOfBytesRead && buffer[tagpos - 1] != '>')
{
fprintf(stderr, "\nTag accross buffer reads...\n");
currentPos += tagpos - buffIndex;
bytesLeft -= tagpos - buffIndex;
buffIndex = tagpos;
parsingTag = true;
if (bytesLeft <= 0)
{
bytesLeft = readlen; //not sure how much longer til next EOL read up readlen;
stopAtNextClosingTag = true;
}
break;
}
else
parsingTag = false;
if (!firstRowfound)
{
firstRowfound = strcmp(currentTag.c_str(), openRowTag.c_str()) == 0;
if (firstRowfound)
fprintf(stderr, "--start piping tag %s at %lu--\n", currentTag.c_str(), currentPos);
}
if (strcmp(currentTag.c_str(), closeRootTag.c_str()) == 0)
{
bytesLeft = 0;
break;
}
if (strcmp(currentTag.c_str(), openRowTag.c_str()) == 0)
withinRecord = true;
else if (strcmp(currentTag.c_str(), closeRowTag.c_str()) == 0)
withinRecord = false;
else if (firstRowfound && !withinRecord)
{
bytesLeft = 0;
fprintf(stderr, "Unexpected Tag found: %s at position %lu\n", currentTag.c_str(), currentPos);
break;
}
currentPos += tagpos - buffIndex;
bytesLeft -= tagpos - buffIndex;
buffIndex = tagpos;
if (bytesLeft <= 0 && !withinRecord)
stopAtNextClosingTag = true;
if (stopAtNextClosingTag && strcmp(currentTag.c_str(), closeRowTag.c_str()) == 0)
{
fprintf(stdout, "%s", currentTag.c_str());
fprintf(stderr, "--stop piping at %s %lu--\n", currentTag.c_str(), currentPos);
bytesLeft = 0;
break;
}
if (firstRowfound)
fprintf(stdout, "%s", currentTag.c_str());
else
fprintf(stderr, "skipping tag %s\n", currentTag.c_str());
if (buffIndex < numOfBytesRead)
currChar = buffer[buffIndex];
else
break;
}
if (firstRowfound)
fprintf(stdout, "%c", currChar);
buffIndex++;
currentPos++;
bytesLeft--;
if (bytesLeft <= 0)
{
if (withinRecord)
{
fprintf(stderr, "\n--Looking for last closing row tag: %ld--\n", currentPos);
bytesLeft = readlen; //not sure how much longer til next EOL read up readlen;
stopAtNextClosingTag = true;
}
else
break;
}
}
}
xmlizedxpath.clear();
xpath2xml(&xmlizedxpath, rowTag, false);
fprintf(stdout, "%s", xmlizedxpath.c_str());
return EXIT_SUCCESS;
}
int libhdfsconnector::streamCSVFileOffset(const char * filename, unsigned long seekPos, unsigned long readlen,
const char * eolseq, unsigned long bufferSize, bool outputTerminator, unsigned long recLen,
unsigned long maxLen, const char * quote, int maxretries)
{
fprintf(stderr, "CSV terminator: \'%s\' and quote: \'%c\'\n", eolseq, quote[0]);
unsigned long recsFound = 0;
hdfsFile readFile = hdfsOpenFile(fs, filename, O_RDONLY, 0, 0, 0);
if (!readFile)
{
fprintf(stderr, "Failed to open %s for reading!\n", filename);
return EXIT_FAILURE;
}
unsigned eolseqlen = strlen(eolseq);
if (seekPos > eolseqlen)
seekPos -= eolseqlen; //read back sizeof(EOL) in case the seekpos happens to be a the first char after an EOL
if (hdfsSeek(fs, readFile, seekPos))
{
fprintf(stderr, "Failed to seek %s for reading!\n", filename);
return EXIT_FAILURE;
}
bool withinQuote = false;
unsigned char buffer[bufferSize + 1];
bool stopAtNextEOL = false;
bool firstEOLfound = seekPos == 0 ? true : false;
unsigned long currentPos = seekPos;
fprintf(stderr, "--Start looking: %ld--\n", currentPos);
unsigned long bytesLeft = readlen;
while (hdfsAvailable(fs, readFile) && bytesLeft > 0)
{
tSize num_read_bytes = hdfsRead(fs, readFile, (void*) buffer, bufferSize);
if (num_read_bytes <= 0)
{
fprintf(stderr, "\n--Hard Stop at: %ld--\n", currentPos);
break;
}
for (int bufferIndex = 0; bufferIndex < num_read_bytes; bufferIndex++, currentPos++)
{
char currChar = buffer[bufferIndex];
if (currChar == EOF)
break;
if (currChar == quote[0])
{
fprintf(stderr, "found quote char at pos: %ld\n", currentPos);
withinQuote = !withinQuote;
}
if (currChar == eolseq[0] && !withinQuote)
{
bool eolfound = true;
tSize extraNumOfBytesRead = 0;
string tmpstr("");
if (eolseqlen > 1)
{
int eoli = bufferIndex;
while (eoli < num_read_bytes && eoli - bufferIndex < eolseqlen)
{
tmpstr.append(1, buffer[eoli++]);
}
if (eoli == num_read_bytes && tmpstr.size() < eolseqlen)
{
//looks like we have to do a remote read, but before we do, let's make sure the substring matches
if (strncmp(eolseq, tmpstr.c_str(), tmpstr.size()) == 0)
{
unsigned char tmpbuffer[eolseqlen - tmpstr.size() + 1];
//TODO have to make a read... of eolseqlen - tmpstr.size is it worth it?
//extraNumOfBytesRead = hdfsRead(*fs, readFile, (void*) tmpbuffer,
extraNumOfBytesRead = hdfsRead(fs, readFile, (void*) tmpbuffer, eolseqlen - tmpstr.size());
for (int y = 0; y < extraNumOfBytesRead; y++)
tmpstr.append(1, tmpbuffer[y]);
}
}
if (strcmp(tmpstr.c_str(), eolseq) != 0)
eolfound = false;
}
if (eolfound)
{
if (!firstEOLfound)
{
bufferIndex = bufferIndex + eolseqlen - 1;
currentPos = currentPos + eolseqlen - 1;
bytesLeft = bytesLeft - eolseqlen;
fprintf(stderr, "\n--Start reading: %ld--\n", currentPos);
firstEOLfound = true;
continue;
}
if (outputTerminator)
{
//if (currentPos > seekPos) //Don't output first EOL
fprintf(stdout, "%s", eolseq);
bufferIndex += eolseqlen;
currentPos += eolseqlen;
bytesLeft -= eolseqlen;
}
recsFound++;
if (stopAtNextEOL)
{
fprintf(stderr, "\n--Stop piping: %ld--\n", currentPos);
bytesLeft = 0;
break;
}
if (bufferIndex < num_read_bytes)
currChar = buffer[bufferIndex];
else
break;
}
else if (extraNumOfBytesRead > 0)
{
if (hdfsSeek(fs, readFile, hdfsTell(fs, readFile) - extraNumOfBytesRead))
{
fprintf(stderr, "Error while attempting to correct seek position\n");
return EXIT_FAILURE;
}
}
}
//don't pipe until we're beyond the first EOL (if offset = 0 start piping ASAP)
if (firstEOLfound)
{
fprintf(stdout, "%c", currChar);
bytesLeft--;
}
else
{
fprintf(stderr, "%c", currChar);
bytesLeft--;
if (maxLen > 0 && currentPos - seekPos > maxLen * 10)
{
fprintf(stderr, "\nFirst EOL was not found within the first %lu bytes", currentPos - seekPos);
return EXIT_FAILURE;
}
}
if (stopAtNextEOL)
fprintf(stderr, "%c", currChar);
// ok, so if bytesLeft <= 0 at this point, we need to keep piping
// IF the last char read was not an EOL char
if (bytesLeft <= 0 && currChar != eolseq[0])
{
if (!firstEOLfound)
{
fprintf(stderr,
"\n--Reached end of readlen before finding first record start at: %ld (breaking out)--\n",
currentPos);
break;
}
fprintf(stderr, "\n--Looking for Last EOL: %ld--\n", currentPos);
bytesLeft = readlen; //not sure how much longer until next EOL read up readlen;
stopAtNextEOL = true;
}
}
}
fprintf(stderr, "\nCurrentPos: %ld, RecsFound: %ld\n", currentPos, recsFound);
hdfsCloseFile(fs, readFile);
return EXIT_SUCCESS;
}
int main(int argc, char **argv)
{
char buffer[32];
tSize num_written_bytes;
const char* slashTmp = "/tmp";
int nnPort;
char *rwTemplate, *rwTemplate2, *newDirTemplate,
*appendTemplate, *userTemplate, *rwPath = NULL;
const char* fileContents = "Hello, World!";
const char* nnHost = NULL;
if (argc != 2) {
fprintf(stderr, "usage: test_libwebhdfs_ops <username>\n");
exit(1);
}
struct NativeMiniDfsConf conf = {
.doFormat = 1, .webhdfsEnabled = 1, .namenodeHttpPort = 50070,
};
cluster = nmdCreate(&conf);
if (!cluster) {
fprintf(stderr, "Failed to create the NativeMiniDfsCluster.\n");
exit(1);
}
if (nmdWaitClusterUp(cluster)) {
fprintf(stderr, "Error when waiting for cluster to be ready.\n");
exit(1);
}
if (nmdGetNameNodeHttpAddress(cluster, &nnPort, &nnHost)) {
fprintf(stderr, "Error when retrieving namenode host address.\n");
exit(1);
}
hdfsFS fs = hdfsConnectAsUserNewInstance(nnHost, nnPort, argv[1]);
if(!fs) {
fprintf(stderr, "Oops! Failed to connect to hdfs!\n");
exit(-1);
}
{
// Write tests
rwTemplate = strdup("/tmp/helloWorldXXXXXX");
if (!rwTemplate) {
fprintf(stderr, "Failed to create rwTemplate!\n");
exit(1);
}
rwPath = mktemp(rwTemplate);
// hdfsOpenFile
hdfsFile writeFile = hdfsOpenFile(fs, rwPath,
O_WRONLY|O_CREAT, 0, 0, 0);
if(!writeFile) {
fprintf(stderr, "Failed to open %s for writing!\n", rwPath);
exit(1);
}
fprintf(stderr, "Opened %s for writing successfully...\n", rwPath);
// hdfsWrite
num_written_bytes = hdfsWrite(fs, writeFile, (void*)fileContents,
(int) strlen(fileContents) + 1);
if (num_written_bytes != strlen(fileContents) + 1) {
fprintf(stderr, "Failed to write correct number of bytes - "
"expected %d, got %d\n",
(int)(strlen(fileContents) + 1), (int) num_written_bytes);
exit(1);
}
fprintf(stderr, "Wrote %d bytes\n", num_written_bytes);
// hdfsTell
tOffset currentPos = -1;
if ((currentPos = hdfsTell(fs, writeFile)) == -1) {
fprintf(stderr,
"Failed to get current file position correctly. Got %"
PRId64 "!\n", currentPos);
exit(1);
}
fprintf(stderr, "Current position: %" PRId64 "\n", currentPos);
hdfsCloseFile(fs, writeFile);
// Done test write
}
sleep(1);
{
//Read tests
int available = 0, exists = 0;
// hdfsExists
exists = hdfsExists(fs, rwPath);
if (exists) {
fprintf(stderr, "Failed to validate existence of %s\n", rwPath);
exists = hdfsExists(fs, rwPath);
if (exists) {
fprintf(stderr,
"Still failed to validate existence of %s\n", rwPath);
exit(1);
}
}
hdfsFile readFile = hdfsOpenFile(fs, rwPath, O_RDONLY, 0, 0, 0);
if (!readFile) {
fprintf(stderr, "Failed to open %s for reading!\n", rwPath);
exit(1);
}
if (!hdfsFileIsOpenForRead(readFile)) {
fprintf(stderr, "hdfsFileIsOpenForRead: we just opened a file "
"with O_RDONLY, and it did not show up as 'open for "
"read'\n");
exit(1);
}
available = hdfsAvailable(fs, readFile);
fprintf(stderr, "hdfsAvailable: %d\n", available);
// hdfsSeek, hdfsTell
tOffset seekPos = 1;
if(hdfsSeek(fs, readFile, seekPos)) {
fprintf(stderr, "Failed to seek %s for reading!\n", rwPath);
exit(1);
}
tOffset currentPos = -1;
if((currentPos = hdfsTell(fs, readFile)) != seekPos) {
fprintf(stderr,
"Failed to get current file position correctly! Got %"
PRId64 "!\n", currentPos);
exit(1);
}
fprintf(stderr, "Current position: %" PRId64 "\n", currentPos);
if(hdfsSeek(fs, readFile, 0)) {
fprintf(stderr, "Failed to seek %s for reading!\n", rwPath);
exit(1);
}
// hdfsRead
memset(buffer, 0, sizeof(buffer));
tSize num_read_bytes = hdfsRead(fs, readFile, buffer, sizeof(buffer));
if (strncmp(fileContents, buffer, strlen(fileContents)) != 0) {
fprintf(stderr, "Failed to read (direct). "
"Expected %s but got %s (%d bytes)\n",
fileContents, buffer, num_read_bytes);
exit(1);
}
fprintf(stderr, "Read following %d bytes:\n%s\n",
num_read_bytes, buffer);
if (hdfsSeek(fs, readFile, 0L)) {
fprintf(stderr, "Failed to seek to file start!\n");
exit(1);
}
// hdfsPread
memset(buffer, 0, strlen(fileContents + 1));
num_read_bytes = hdfsPread(fs, readFile, 0, buffer, sizeof(buffer));
fprintf(stderr, "Read following %d bytes:\n%s\n",
num_read_bytes, buffer);
hdfsCloseFile(fs, readFile);
// Done test read
}
int totalResult = 0;
int result = 0;
{
//Generic file-system operations
char *srcPath = rwPath;
char buffer[256];
const char *resp;
rwTemplate2 = strdup("/tmp/helloWorld2XXXXXX");
if (!rwTemplate2) {
fprintf(stderr, "Failed to create rwTemplate2!\n");
exit(1);
}
char *dstPath = mktemp(rwTemplate2);
newDirTemplate = strdup("/tmp/newdirXXXXXX");
if (!newDirTemplate) {
fprintf(stderr, "Failed to create newDirTemplate!\n");
exit(1);
}
char *newDirectory = mktemp(newDirTemplate);
// hdfsRename
fprintf(stderr, "hdfsRename: %s\n",
((result = hdfsRename(fs, rwPath, dstPath)) ?
"Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsRename back: %s\n",
((result = hdfsRename(fs, dstPath, srcPath)) ?
"Failed!" : "Success!"));
totalResult += result;
// hdfsCreateDirectory
fprintf(stderr, "hdfsCreateDirectory: %s\n",
((result = hdfsCreateDirectory(fs, newDirectory)) ?
"Failed!" : "Success!"));
totalResult += result;
// hdfsSetReplication
fprintf(stderr, "hdfsSetReplication: %s\n",
((result = hdfsSetReplication(fs, srcPath, 1)) ?
"Failed!" : "Success!"));
totalResult += result;
// hdfsGetWorkingDirectory, hdfsSetWorkingDirectory
fprintf(stderr, "hdfsGetWorkingDirectory: %s\n",
((resp = hdfsGetWorkingDirectory(fs, buffer, sizeof(buffer))) ?
buffer : "Failed!"));
totalResult += (resp ? 0 : 1);
const char* path[] = {"/foo", "/foo/bar", "foobar", "//foo/bar//foobar",
"foo//bar", "foo/bar///", "/", "////"};
int i;
for (i = 0; i < 8; i++) {
fprintf(stderr, "hdfsSetWorkingDirectory: %s, %s\n",
((result = hdfsSetWorkingDirectory(fs, path[i])) ?
"Failed!" : "Success!"),
hdfsGetWorkingDirectory(fs, buffer, sizeof(buffer)));
totalResult += result;
}
fprintf(stderr, "hdfsSetWorkingDirectory: %s\n",
((result = hdfsSetWorkingDirectory(fs, slashTmp)) ?
"Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsGetWorkingDirectory: %s\n",
((resp = hdfsGetWorkingDirectory(fs, buffer, sizeof(buffer))) ?
buffer : "Failed!"));
totalResult += (resp ? 0 : 1);
// hdfsGetPathInfo
hdfsFileInfo *fileInfo = NULL;
if((fileInfo = hdfsGetPathInfo(fs, slashTmp)) != NULL) {
fprintf(stderr, "hdfsGetPathInfo - SUCCESS!\n");
fprintf(stderr, "Name: %s, ", fileInfo->mName);
fprintf(stderr, "Type: %c, ", (char)(fileInfo->mKind));
fprintf(stderr, "Replication: %d, ", fileInfo->mReplication);
fprintf(stderr, "BlockSize: %"PRId64", ", fileInfo->mBlockSize);
fprintf(stderr, "Size: %"PRId64", ", fileInfo->mSize);
fprintf(stderr, "LastMod: %s", ctime(&fileInfo->mLastMod));
fprintf(stderr, "Owner: %s, ", fileInfo->mOwner);
fprintf(stderr, "Group: %s, ", fileInfo->mGroup);
char permissions[10];
permission_disp(fileInfo->mPermissions, permissions);
fprintf(stderr, "Permissions: %d (%s)\n",
fileInfo->mPermissions, permissions);
hdfsFreeFileInfo(fileInfo, 1);
} else {
totalResult++;
fprintf(stderr, "hdfsGetPathInfo for %s - FAILED!\n", slashTmp);
}
// hdfsListDirectory
hdfsFileInfo *fileList = 0;
int numEntries = 0;
if((fileList = hdfsListDirectory(fs, slashTmp, &numEntries)) != NULL) {
int i = 0;
for(i=0; i < numEntries; ++i) {
fprintf(stderr, "Name: %s, ", fileList[i].mName);
fprintf(stderr, "Type: %c, ", (char)fileList[i].mKind);
fprintf(stderr, "Replication: %d, ", fileList[i].mReplication);
fprintf(stderr, "BlockSize: %"PRId64", ", fileList[i].mBlockSize);
fprintf(stderr, "Size: %"PRId64", ", fileList[i].mSize);
fprintf(stderr, "LastMod: %s", ctime(&fileList[i].mLastMod));
fprintf(stderr, "Owner: %s, ", fileList[i].mOwner);
fprintf(stderr, "Group: %s, ", fileList[i].mGroup);
char permissions[10];
permission_disp(fileList[i].mPermissions, permissions);
fprintf(stderr, "Permissions: %d (%s)\n",
fileList[i].mPermissions, permissions);
}
hdfsFreeFileInfo(fileList, numEntries);
} else {
if (errno) {
totalResult++;
fprintf(stderr, "waah! hdfsListDirectory - FAILED!\n");
} else {
fprintf(stderr, "Empty directory!\n");
}
}
char *newOwner = "root";
// Setting tmp dir to 777 so later when connectAsUser nobody,
// we can write to it
short newPerm = 0666;
// hdfsChown
fprintf(stderr, "hdfsChown: %s\n",
((result = hdfsChown(fs, rwPath, NULL, "users")) ?
"Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsChown: %s\n",
((result = hdfsChown(fs, rwPath, newOwner, NULL)) ?
"Failed!" : "Success!"));
totalResult += result;
// hdfsChmod
fprintf(stderr, "hdfsChmod: %s\n",
((result = hdfsChmod(fs, rwPath, newPerm)) ?
"Failed!" : "Success!"));
totalResult += result;
sleep(2);
tTime newMtime = time(NULL);
tTime newAtime = time(NULL);
// utime write
fprintf(stderr, "hdfsUtime: %s\n",
((result = hdfsUtime(fs, rwPath, newMtime, newAtime)) ?
"Failed!" : "Success!"));
totalResult += result;
// chown/chmod/utime read
hdfsFileInfo *finfo = hdfsGetPathInfo(fs, rwPath);
fprintf(stderr, "hdfsChown read: %s\n",
((result = (strcmp(finfo->mOwner, newOwner) != 0)) ?
"Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsChmod read: %s\n",
((result = (finfo->mPermissions != newPerm)) ?
"Failed!" : "Success!"));
totalResult += result;
// will later use /tmp/ as a different user so enable it
fprintf(stderr, "hdfsChmod: %s\n",
((result = hdfsChmod(fs, slashTmp, 0777)) ?
"Failed!" : "Success!"));
totalResult += result;
fprintf(stderr,"newMTime=%ld\n",newMtime);
fprintf(stderr,"curMTime=%ld\n",finfo->mLastMod);
fprintf(stderr, "hdfsUtime read (mtime): %s\n",
((result = (finfo->mLastMod != newMtime / 1000)) ?
"Failed!" : "Success!"));
totalResult += result;
// Clean up
hdfsFreeFileInfo(finfo, 1);
fprintf(stderr, "hdfsDelete: %s\n",
((result = hdfsDelete(fs, newDirectory, 1)) ?
"Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsDelete: %s\n",
((result = hdfsDelete(fs, srcPath, 1)) ?
"Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsExists: %s\n",
((result = hdfsExists(fs, newDirectory)) ?
"Success!" : "Failed!"));
totalResult += (result ? 0 : 1);
// Done test generic operations
}
{
// Test Appends
appendTemplate = strdup("/tmp/appendsXXXXXX");
if (!appendTemplate) {
fprintf(stderr, "Failed to create appendTemplate!\n");
exit(1);
}
char *appendPath = mktemp(appendTemplate);
const char* helloBuffer = "Hello,";
hdfsFile writeFile = NULL;
// Create
writeFile = hdfsOpenFile(fs, appendPath, O_WRONLY, 0, 0, 0);
if(!writeFile) {
fprintf(stderr, "Failed to open %s for writing!\n", appendPath);
exit(1);
}
fprintf(stderr, "Opened %s for writing successfully...\n", appendPath);
num_written_bytes = hdfsWrite(fs, writeFile, helloBuffer,
(int) strlen(helloBuffer));
fprintf(stderr, "Wrote %d bytes\n", num_written_bytes);
hdfsCloseFile(fs, writeFile);
fprintf(stderr, "hdfsSetReplication: %s\n",
((result = hdfsSetReplication(fs, appendPath, 1)) ?
"Failed!" : "Success!"));
totalResult += result;
// Re-Open for Append
writeFile = hdfsOpenFile(fs, appendPath, O_WRONLY | O_APPEND, 0, 0, 0);
if(!writeFile) {
fprintf(stderr, "Failed to open %s for writing!\n", appendPath);
exit(1);
}
fprintf(stderr, "Opened %s for appending successfully...\n",
appendPath);
helloBuffer = " World";
num_written_bytes = hdfsWrite(fs, writeFile, helloBuffer,
(int)strlen(helloBuffer) + 1);
fprintf(stderr, "Wrote %d bytes\n", num_written_bytes);
hdfsCloseFile(fs, writeFile);
// Check size
hdfsFileInfo *finfo = hdfsGetPathInfo(fs, appendPath);
fprintf(stderr, "fileinfo->mSize: == total %s\n",
((result = (finfo->mSize == strlen("Hello, World") + 1)) ?
"Success!" : "Failed!"));
totalResult += (result ? 0 : 1);
// Read and check data
hdfsFile readFile = hdfsOpenFile(fs, appendPath, O_RDONLY, 0, 0, 0);
if (!readFile) {
fprintf(stderr, "Failed to open %s for reading!\n", appendPath);
exit(1);
}
tSize num_read_bytes = hdfsRead(fs, readFile, buffer, sizeof(buffer));
fprintf(stderr, "Read following %d bytes:\n%s\n",
num_read_bytes, buffer);
fprintf(stderr, "read == Hello, World %s\n",
(result = (strcmp(buffer, "Hello, World") == 0)) ?
"Success!" : "Failed!");
hdfsCloseFile(fs, readFile);
// Cleanup
fprintf(stderr, "hdfsDelete: %s\n",
((result = hdfsDelete(fs, appendPath, 1)) ?
"Failed!" : "Success!"));
totalResult += result;
// Done test appends
}
totalResult += (hdfsDisconnect(fs) != 0);
{
//
// Now test as connecting as a specific user
// This only meant to test that we connected as that user, not to test
// the actual fs user capabilities. Thus just create a file and read
// the owner is correct.
const char *tuser = "******";
userTemplate = strdup("/tmp/usertestXXXXXX");
if (!userTemplate) {
fprintf(stderr, "Failed to create userTemplate!\n");
exit(1);
}
char* userWritePath = mktemp(userTemplate);
hdfsFile writeFile = NULL;
fs = hdfsConnectAsUserNewInstance("default", 50070, tuser);
if(!fs) {
fprintf(stderr,
"Oops! Failed to connect to hdfs as user %s!\n",tuser);
exit(1);
}
writeFile = hdfsOpenFile(fs, userWritePath, O_WRONLY|O_CREAT, 0, 0, 0);
if(!writeFile) {
fprintf(stderr, "Failed to open %s for writing!\n", userWritePath);
exit(1);
}
fprintf(stderr, "Opened %s for writing successfully...\n",
userWritePath);
num_written_bytes = hdfsWrite(fs, writeFile, fileContents,
(int)strlen(fileContents) + 1);
fprintf(stderr, "Wrote %d bytes\n", num_written_bytes);
hdfsCloseFile(fs, writeFile);
hdfsFileInfo *finfo = hdfsGetPathInfo(fs, userWritePath);
if (finfo) {
fprintf(stderr, "hdfs new file user is correct: %s\n",
((result = (strcmp(finfo->mOwner, tuser) != 0)) ?
"Failed!" : "Success!"));
} else {
fprintf(stderr,
"hdfsFileInfo returned by hdfsGetPathInfo is NULL\n");
result = -1;
}
totalResult += result;
// Cleanup
fprintf(stderr, "hdfsDelete: %s\n",
((result = hdfsDelete(fs, userWritePath, 1)) ?
"Failed!" : "Success!"));
totalResult += result;
// Done test specific user
}
totalResult += (hdfsDisconnect(fs) != 0);
// Shutdown the native minidfscluster
nmdShutdown(cluster);
nmdFree(cluster);
fprintf(stderr, "totalResult == %d\n", totalResult);
if (totalResult != 0) {
return -1;
} else {
return 0;
}
}
int main(int argc, char **argv) {
const char *writePath = "/tmp/testfile.txt";
const char *fileContents = "Hello, World!";
const char *readPath = "/tmp/testfile.txt";
const char *srcPath = "/tmp/testfile.txt";
const char *dstPath = "/tmp/testfile2.txt";
const char *slashTmp = "/tmp";
const char *newDirectory = "/tmp/newdir";
const char *newOwner = "root";
const char *tuser = "******";
const char *appendPath = "/tmp/appends";
const char *userPath = "/tmp/usertestfile.txt";
char buffer[32], buffer2[256], rdbuffer[32];
tSize num_written_bytes, num_read_bytes;
hdfsFS fs, lfs;
hdfsFile writeFile, readFile, localFile, appendFile, userFile;
tOffset currentPos, seekPos;
int exists, totalResult, result, numEntries, i, j;
const char *resp;
hdfsFileInfo *fileInfo, *fileList, *finfo;
char *buffer3;
char permissions[10];
char ***hosts;
short newPerm = 0666;
tTime newMtime, newAtime;
fs = hdfsConnectNewInstance("default", 0);
if(!fs) {
fprintf(stderr, "Oops! Failed to connect to hdfs!\n");
exit(-1);
}
lfs = hdfsConnectNewInstance(NULL, 0);
if(!lfs) {
fprintf(stderr, "Oops! Failed to connect to 'local' hdfs!\n");
exit(-1);
}
{
//Write tests
writeFile = hdfsOpenFile(fs, writePath, O_WRONLY|O_CREAT, 0, 0, 0);
if(!writeFile) {
fprintf(stderr, "Failed to open %s for writing!\n", writePath);
exit(-1);
}
fprintf(stderr, "Opened %s for writing successfully...\n", writePath);
num_written_bytes =
hdfsWrite(fs, writeFile, (void*)fileContents,
(tSize)(strlen(fileContents)+1));
if (num_written_bytes != strlen(fileContents) + 1) {
fprintf(stderr, "Failed to write correct number of bytes - expected %d, got %d\n",
(int)(strlen(fileContents) + 1), (int)num_written_bytes);
exit(-1);
}
fprintf(stderr, "Wrote %d bytes\n", num_written_bytes);
currentPos = -1;
if ((currentPos = hdfsTell(fs, writeFile)) == -1) {
fprintf(stderr,
"Failed to get current file position correctly! Got %" PRId64 "!\n",
currentPos);
exit(-1);
}
fprintf(stderr, "Current position: %" PRId64 "\n", currentPos);
if (hdfsFlush(fs, writeFile)) {
fprintf(stderr, "Failed to 'flush' %s\n", writePath);
exit(-1);
}
fprintf(stderr, "Flushed %s successfully!\n", writePath);
if (hdfsHFlush(fs, writeFile)) {
fprintf(stderr, "Failed to 'hflush' %s\n", writePath);
exit(-1);
}
fprintf(stderr, "HFlushed %s successfully!\n", writePath);
hdfsCloseFile(fs, writeFile);
}
{
//Read tests
exists = hdfsExists(fs, readPath);
if (exists) {
fprintf(stderr, "Failed to validate existence of %s\n", readPath);
exit(-1);
}
readFile = hdfsOpenFile(fs, readPath, O_RDONLY, 0, 0, 0);
if (!readFile) {
fprintf(stderr, "Failed to open %s for reading!\n", readPath);
exit(-1);
}
if (!hdfsFileIsOpenForRead(readFile)) {
fprintf(stderr, "hdfsFileIsOpenForRead: we just opened a file "
"with O_RDONLY, and it did not show up as 'open for "
"read'\n");
exit(-1);
}
fprintf(stderr, "hdfsAvailable: %d\n", hdfsAvailable(fs, readFile));
seekPos = 1;
if(hdfsSeek(fs, readFile, seekPos)) {
fprintf(stderr, "Failed to seek %s for reading!\n", readPath);
exit(-1);
}
currentPos = -1;
if((currentPos = hdfsTell(fs, readFile)) != seekPos) {
fprintf(stderr,
"Failed to get current file position correctly! Got %" PRId64 "!\n",
currentPos);
exit(-1);
}
fprintf(stderr, "Current position: %" PRId64 "\n", currentPos);
if (!hdfsFileUsesDirectRead(readFile)) {
fprintf(stderr, "Direct read support incorrectly not detected "
"for HDFS filesystem\n");
exit(-1);
}
fprintf(stderr, "Direct read support detected for HDFS\n");
// Test the direct read path
if(hdfsSeek(fs, readFile, 0)) {
fprintf(stderr, "Failed to seek %s for reading!\n", readPath);
exit(-1);
}
memset(buffer, 0, sizeof(buffer));
num_read_bytes = hdfsRead(fs, readFile, (void*)buffer,
sizeof(buffer));
if (strncmp(fileContents, buffer, strlen(fileContents)) != 0) {
fprintf(stderr, "Failed to read (direct). Expected %s but got %s (%d bytes)\n",
fileContents, buffer, num_read_bytes);
exit(-1);
}
fprintf(stderr, "Read (direct) following %d bytes:\n%s\n",
num_read_bytes, buffer);
if (hdfsSeek(fs, readFile, 0L)) {
fprintf(stderr, "Failed to seek to file start!\n");
exit(-1);
}
// Disable the direct read path so that we really go through the slow
// read path
hdfsFileDisableDirectRead(readFile);
num_read_bytes = hdfsRead(fs, readFile, (void*)buffer,
sizeof(buffer));
fprintf(stderr, "Read following %d bytes:\n%s\n",
num_read_bytes, buffer);
memset(buffer, 0, strlen(fileContents + 1));
num_read_bytes = hdfsPread(fs, readFile, 0, (void*)buffer,
sizeof(buffer));
fprintf(stderr, "Read following %d bytes:\n%s\n",
num_read_bytes, buffer);
hdfsCloseFile(fs, readFile);
// Test correct behaviour for unsupported filesystems
localFile = hdfsOpenFile(lfs, writePath, O_WRONLY|O_CREAT, 0, 0, 0);
if(!localFile) {
fprintf(stderr, "Failed to open %s for writing!\n", writePath);
exit(-1);
}
num_written_bytes = hdfsWrite(lfs, localFile, (void*)fileContents,
(tSize)(strlen(fileContents) + 1));
hdfsCloseFile(lfs, localFile);
localFile = hdfsOpenFile(lfs, writePath, O_RDONLY, 0, 0, 0);
if (hdfsFileUsesDirectRead(localFile)) {
fprintf(stderr, "Direct read support incorrectly detected for local "
"filesystem\n");
exit(-1);
}
hdfsCloseFile(lfs, localFile);
}
totalResult = 0;
result = 0;
{
//Generic file-system operations
fprintf(stderr, "hdfsCopy(remote-local): %s\n", ((result = hdfsCopy(fs, srcPath, lfs, srcPath)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsCopy(remote-remote): %s\n", ((result = hdfsCopy(fs, srcPath, fs, dstPath)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsMove(local-local): %s\n", ((result = hdfsMove(lfs, srcPath, lfs, dstPath)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsMove(remote-local): %s\n", ((result = hdfsMove(fs, srcPath, lfs, srcPath)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsRename: %s\n", ((result = hdfsRename(fs, dstPath, srcPath)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsCopy(remote-remote): %s\n", ((result = hdfsCopy(fs, srcPath, fs, dstPath)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsCreateDirectory: %s\n", ((result = hdfsCreateDirectory(fs, newDirectory)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsSetReplication: %s\n", ((result = hdfsSetReplication(fs, srcPath, 2)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsGetWorkingDirectory: %s\n", ((resp = hdfsGetWorkingDirectory(fs, buffer2, sizeof(buffer2))) != 0 ? buffer2 : "Failed!"));
totalResult += (resp ? 0 : 1);
fprintf(stderr, "hdfsSetWorkingDirectory: %s\n", ((result = hdfsSetWorkingDirectory(fs, slashTmp)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsGetWorkingDirectory: %s\n", ((resp = hdfsGetWorkingDirectory(fs, buffer2, sizeof(buffer2))) != 0 ? buffer2 : "Failed!"));
totalResult += (resp ? 0 : 1);
fprintf(stderr, "hdfsGetDefaultBlockSize: %" PRId64 "\n", hdfsGetDefaultBlockSize(fs));
fprintf(stderr, "hdfsGetCapacity: %" PRId64 "\n", hdfsGetCapacity(fs));
fprintf(stderr, "hdfsGetUsed: %" PRId64 "\n", hdfsGetUsed(fs));
fileInfo = NULL;
if((fileInfo = hdfsGetPathInfo(fs, slashTmp)) != NULL) {
fprintf(stderr, "hdfsGetPathInfo - SUCCESS!\n");
fprintf(stderr, "Name: %s, ", fileInfo->mName);
fprintf(stderr, "Type: %c, ", (char)(fileInfo->mKind));
fprintf(stderr, "Replication: %d, ", fileInfo->mReplication);
fprintf(stderr, "BlockSize: %" PRId64 ", ", fileInfo->mBlockSize);
fprintf(stderr, "Size: %" PRId64 ", ", fileInfo->mSize);
fprintf(stderr, "LastMod: %s", ctime(&fileInfo->mLastMod));
fprintf(stderr, "Owner: %s, ", fileInfo->mOwner);
fprintf(stderr, "Group: %s, ", fileInfo->mGroup);
permission_disp(fileInfo->mPermissions, permissions);
fprintf(stderr, "Permissions: %d (%s)\n", fileInfo->mPermissions, permissions);
hdfsFreeFileInfo(fileInfo, 1);
} else {
totalResult++;
fprintf(stderr, "waah! hdfsGetPathInfo for %s - FAILED!\n", slashTmp);
}
fileList = 0;
fileList = hdfsListDirectory(fs, newDirectory, &numEntries);
if (!(fileList == NULL && numEntries == 0 && !errno)) {
fprintf(stderr, "waah! hdfsListDirectory for empty %s - FAILED!\n", newDirectory);
totalResult++;
} else {
fprintf(stderr, "hdfsListDirectory for empty %s - SUCCESS!\n", newDirectory);
}
fileList = 0;
if((fileList = hdfsListDirectory(fs, slashTmp, &numEntries)) != NULL) {
for(i=0; i < numEntries; ++i) {
fprintf(stderr, "Name: %s, ", fileList[i].mName);
fprintf(stderr, "Type: %c, ", (char)fileList[i].mKind);
fprintf(stderr, "Replication: %d, ", fileList[i].mReplication);
fprintf(stderr, "BlockSize: %" PRId64 ", ", fileList[i].mBlockSize);
fprintf(stderr, "Size: %" PRId64 ", ", fileList[i].mSize);
fprintf(stderr, "LastMod: %s", ctime(&fileList[i].mLastMod));
fprintf(stderr, "Owner: %s, ", fileList[i].mOwner);
fprintf(stderr, "Group: %s, ", fileList[i].mGroup);
permission_disp(fileList[i].mPermissions, permissions);
fprintf(stderr, "Permissions: %d (%s)\n", fileList[i].mPermissions, permissions);
}
hdfsFreeFileInfo(fileList, numEntries);
} else {
if (errno) {
totalResult++;
fprintf(stderr, "waah! hdfsListDirectory - FAILED!\n");
} else {
fprintf(stderr, "Empty directory!\n");
}
}
hosts = hdfsGetHosts(fs, srcPath, 0, 1);
if(hosts) {
fprintf(stderr, "hdfsGetHosts - SUCCESS! ... \n");
i=0;
while(hosts[i]) {
j = 0;
while(hosts[i][j]) {
fprintf(stderr,
"\thosts[%d][%d] - %s\n", i, j, hosts[i][j]);
++j;
}
++i;
}
} else {
totalResult++;
fprintf(stderr, "waah! hdfsGetHosts - FAILED!\n");
}
// setting tmp dir to 777 so later when connectAsUser nobody, we can write to it
// chown write
fprintf(stderr, "hdfsChown: %s\n", ((result = hdfsChown(fs, writePath, NULL, "users")) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsChown: %s\n", ((result = hdfsChown(fs, writePath, newOwner, NULL)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
// chmod write
fprintf(stderr, "hdfsChmod: %s\n", ((result = hdfsChmod(fs, writePath, newPerm)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
sleep(2);
newMtime = time(NULL);
newAtime = time(NULL);
// utime write
fprintf(stderr, "hdfsUtime: %s\n", ((result = hdfsUtime(fs, writePath, newMtime, newAtime)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
// chown/chmod/utime read
finfo = hdfsGetPathInfo(fs, writePath);
fprintf(stderr, "hdfsChown read: %s\n", ((result = (strcmp(finfo->mOwner, newOwner))) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsChmod read: %s\n", ((result = (finfo->mPermissions != newPerm)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
// will later use /tmp/ as a different user so enable it
fprintf(stderr, "hdfsChmod: %s\n", ((result = hdfsChmod(fs, "/tmp/", 0777)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr,"newMTime=%ld\n",newMtime);
fprintf(stderr,"curMTime=%ld\n",finfo->mLastMod);
fprintf(stderr, "hdfsUtime read (mtime): %s\n", ((result = (finfo->mLastMod != newMtime)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
// No easy way to turn on access times from hdfs_test right now
// fprintf(stderr, "hdfsUtime read (atime): %s\n", ((result = (finfo->mLastAccess != newAtime)) != 0 ? "Failed!" : "Success!"));
// totalResult += result;
hdfsFreeFileInfo(finfo, 1);
// Clean up
fprintf(stderr, "hdfsDelete: %s\n", ((result = hdfsDelete(fs, newDirectory, 1)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsDelete: %s\n", ((result = hdfsDelete(fs, srcPath, 1)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsDelete: %s\n", ((result = hdfsDelete(lfs, srcPath, 1)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsDelete: %s\n", ((result = hdfsDelete(lfs, dstPath, 1)) != 0 ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsExists: %s\n", ((result = hdfsExists(fs, newDirectory)) != 0 ? "Success!" : "Failed!"));
totalResult += (result ? 0 : 1);
}
{
// TEST APPENDS
// CREATE
appendFile = hdfsOpenFile(fs, appendPath, O_WRONLY, 0, 0, 0);
if(!appendFile) {
fprintf(stderr, "Failed to open %s for writing!\n", appendPath);
exit(-1);
}
fprintf(stderr, "Opened %s for writing successfully...\n", appendPath);
buffer3 = "Hello,";
num_written_bytes = hdfsWrite(fs, appendFile, (void*)buffer3,
(tSize)strlen(buffer3));
fprintf(stderr, "Wrote %d bytes\n", num_written_bytes);
if (hdfsFlush(fs, appendFile)) {
fprintf(stderr, "Failed to 'flush' %s\n", appendPath);
exit(-1);
}
fprintf(stderr, "Flushed %s successfully!\n", appendPath);
hdfsCloseFile(fs, appendFile);
// RE-OPEN
appendFile = hdfsOpenFile(fs, appendPath, O_WRONLY|O_APPEND, 0, 0, 0);
if(!appendFile) {
fprintf(stderr, "Failed to open %s for writing!\n", appendPath);
exit(-1);
}
fprintf(stderr, "Opened %s for writing successfully...\n", appendPath);
buffer3 = " World";
num_written_bytes = hdfsWrite(fs, appendFile, (void*)buffer3,
(tSize)(strlen(buffer3) + 1));
fprintf(stderr, "Wrote %d bytes\n", num_written_bytes);
if (hdfsFlush(fs, appendFile)) {
fprintf(stderr, "Failed to 'flush' %s\n", appendPath);
exit(-1);
}
fprintf(stderr, "Flushed %s successfully!\n", appendPath);
hdfsCloseFile(fs, appendFile);
// CHECK size
finfo = hdfsGetPathInfo(fs, appendPath);
fprintf(stderr, "fileinfo->mSize: == total %s\n", ((result = (finfo->mSize == (tOffset)(strlen("Hello, World") + 1))) == 1 ? "Success!" : "Failed!"));
totalResult += (result ? 0 : 1);
// READ and check data
readFile = hdfsOpenFile(fs, appendPath, O_RDONLY, 0, 0, 0);
if (!readFile) {
fprintf(stderr, "Failed to open %s for reading!\n", appendPath);
exit(-1);
}
num_read_bytes = hdfsRead(fs, readFile, (void*)rdbuffer, sizeof(rdbuffer));
fprintf(stderr, "Read following %d bytes:\n%s\n",
num_read_bytes, rdbuffer);
fprintf(stderr, "read == Hello, World %s\n", ((result = (strcmp(rdbuffer, "Hello, World"))) == 0 ? "Success!" : "Failed!"));
hdfsCloseFile(fs, readFile);
// DONE test appends
}
totalResult += (hdfsDisconnect(fs) != 0);
{
//
// Now test as connecting as a specific user
// This is only meant to test that we connected as that user, not to test
// the actual fs user capabilities. Thus just create a file and read
// the owner is correct.
fs = hdfsConnectAsUserNewInstance("default", 0, tuser);
if(!fs) {
fprintf(stderr, "Oops! Failed to connect to hdfs as user %s!\n",tuser);
exit(-1);
}
userFile = hdfsOpenFile(fs, userPath, O_WRONLY|O_CREAT, 0, 0, 0);
if(!userFile) {
fprintf(stderr, "Failed to open %s for writing!\n", userPath);
exit(-1);
}
fprintf(stderr, "Opened %s for writing successfully...\n", userPath);
num_written_bytes = hdfsWrite(fs, userFile, (void*)fileContents,
(tSize)(strlen(fileContents)+1));
fprintf(stderr, "Wrote %d bytes\n", num_written_bytes);
if (hdfsFlush(fs, userFile)) {
fprintf(stderr, "Failed to 'flush' %s\n", userPath);
exit(-1);
}
fprintf(stderr, "Flushed %s successfully!\n", userPath);
hdfsCloseFile(fs, userFile);
finfo = hdfsGetPathInfo(fs, userPath);
fprintf(stderr, "hdfs new file user is correct: %s\n", ((result = (strcmp(finfo->mOwner, tuser))) != 0 ? "Failed!" : "Success!"));
totalResult += result;
}
totalResult += (hdfsDisconnect(fs) != 0);
if (totalResult != 0) {
return -1;
} else {
return 0;
}
}
ssize_t read(void* data, size_t size) final {
return hdfsRead(fs_, file_, data, size);
}
int libhdfsconnector::mergeFile()
{
if (nodeID == 0)
{
if (!fs)
{
fprintf(stderr, "Could not connect to hdfs on");
return RETURN_FAILURE;
}
fprintf(stderr, "merging %d file(s) into %s\n", clusterCount, fileName);
fprintf(stderr, "Opening %s for writing!\n", fileName);
hdfsFile writeFile = hdfsOpenFile(fs, fileName, O_CREAT | O_WRONLY, 0, filereplication, 0);
if (!writeFile)
{
fprintf(stderr, "Failed to open %s for writing!\n", fileName);
return EXIT_FAILURE;
}
tSize totalBytesWritten = 0;
for (unsigned node = 0; node < clusterCount; node++)
{
if (node > 0)
{
writeFile = hdfsOpenFile(fs, fileName, O_WRONLY | O_APPEND, 0, filereplication, 0);
fprintf(stderr, "Re-opening %s for append!\n", fileName);
}
unsigned bytesWrittenSinceLastFlush = 0;
string filepartname;
createFilePartName(&filepartname, fileName, node, clusterCount);
if (hdfsExists(fs, filepartname.c_str()) == 0)
{
fprintf(stderr, "Opening readfile %s\n", filepartname.c_str());
hdfsFile readFile = hdfsOpenFile(fs, filepartname.c_str(), O_RDONLY, 0, 0, 0);
if (!readFile)
{
fprintf(stderr, "Failed to open %s for reading!\n", fileName);
return EXIT_FAILURE;
}
unsigned char buffer[bufferSize + 1];
while (hdfsAvailable(fs, readFile))
{
tSize num_read_bytes = hdfsRead(fs, readFile, buffer, bufferSize);
if (num_read_bytes <= 0)
break;
tSize bytesWritten = 0;
try
{
bytesWritten = hdfsWrite(fs, writeFile, (void*) buffer, num_read_bytes);
totalBytesWritten += bytesWritten;
bytesWrittenSinceLastFlush += bytesWritten;
if (bytesWrittenSinceLastFlush >= flushThreshold)
{
if (hdfsFlush(fs, writeFile))
{
fprintf(stderr, "Failed to 'flush' %s\n", fileName);
return EXIT_FAILURE;
}
bytesWrittenSinceLastFlush = 0;
}
} catch (...)
{
fprintf(stderr, "Issue detected during HDFSWrite\n");
fprintf(stderr, "Bytes written in current iteration: %d\n", bytesWritten);
return EXIT_FAILURE;
}
}
if (hdfsFlush(fs, writeFile))
{
fprintf(stderr, "Failed to 'flush' %s\n", fileName);
return EXIT_FAILURE;
}
fprintf(stderr, "Closing readfile %s\n", filepartname.c_str());
hdfsCloseFile(fs, readFile);
if (cleanmerge)
{
#ifdef HADOOP_GT_21
hdfsDelete(fs, filepartname.c_str(), 0);
#else
hdfsDelete(fs, filepartname.c_str());
#endif
}
}
else
{
fprintf(stderr, "Could not merge, part %s was not located\n", filepartname.c_str());
return EXIT_FAILURE;
}
fprintf(stderr, "Closing writefile %s\n", fileName);
if (hdfsCloseFile(fs, writeFile) != 0)
fprintf(stderr, "Could not close writefile %s\n", fileName);
}
if (cleanmerge)
{
string filecontainer;
filecontainer.assign(fileName);
filecontainer.append("-parts");
#ifdef HADOOP_GT_21
hdfsDelete(fs, filecontainer.c_str(), 0);
#else
hdfsDelete(fs, filecontainer.c_str());
#endif
}
}
return EXIT_SUCCESS;
}
qioerr hdfs_preadv (void* file, const struct iovec *vector, int count, off_t offset, ssize_t* num_read_out, void* fs)
{
ssize_t got;
ssize_t got_total;
qioerr err_out = 0;
int i;
STARTING_SLOW_SYSCALL;
#ifdef HDFS3
const hdfs_file orig_hfl = *to_hdfs_file(file);
const hdfs_fs orig_hfs = *to_hdfs_fs(fs);
hdfsFS hfs = hdfsConnect(orig_hfs.fs_name, orig_hfs.fs_port);
hdfsFile hfl = hdfsOpenFile(hfs, orig_hfl.pathnm, O_RDONLY, 0, 0, 0);
//assert connection
CREATE_ERROR((hfs == NULL), err_out, ECONNREFUSED, "Unable to read HDFS file", error);
if(hfl == NULL) {
err_out = qio_mkerror_errno();
goto error;
}
#endif
err_out = 0;
got_total = 0;
for(i = 0; i < count; i++) {
#ifdef HDFS3
hdfsSeek(hfs, hfl, offset+got_total);
got = hdfsRead(hfs, hfl, (void*)vector[i].iov_base, vector[i].iov_len);
#else
got = hdfsPread(to_hdfs_fs(fs)->hfs, to_hdfs_file(file)->file, offset + got_total, (void*)vector[i].iov_base, vector[i].iov_len);
#endif
if( got != -1 ) {
got_total += got;
} else {
err_out = qio_mkerror_errno();
break;
}
if(got != (ssize_t)vector[i].iov_len ) {
break;
}
}
if( err_out == 0 && got_total == 0 && sys_iov_total_bytes(vector, count) != 0 )
err_out = qio_int_to_err(EEOF);
*num_read_out = got_total;
#ifdef HDFS3
got = hdfsCloseFile(hfs, hfl);
if(got == -1) { err_out = qio_mkerror_errno(); }
got = hdfsDisconnect(hfs);
if(got == -1) { err_out = qio_mkerror_errno(); }
#endif
DONE_SLOW_SYSCALL;
#ifdef HDFS3
error:
#endif
return err_out;
}
static int doTestHdfsOperations(struct tlhThreadInfo *ti, hdfsFS fs,
const struct tlhPaths *paths)
{
char tmp[4096];
hdfsFile file;
int ret, expected, numEntries;
hdfsFileInfo *fileInfo;
struct hdfsReadStatistics *readStats = NULL;
if (hdfsExists(fs, paths->prefix) == 0) {
EXPECT_ZERO(hdfsDelete(fs, paths->prefix, 1));
}
EXPECT_ZERO(hdfsCreateDirectory(fs, paths->prefix));
EXPECT_ZERO(doTestGetDefaultBlockSize(fs, paths->prefix));
/* There should be no entry in the directory. */
errno = EACCES; // see if errno is set to 0 on success
EXPECT_NULL_WITH_ERRNO(hdfsListDirectory(fs, paths->prefix, &numEntries), 0);
if (numEntries != 0) {
fprintf(stderr, "hdfsListDirectory set numEntries to "
"%d on empty directory.", numEntries);
}
/* There should not be any file to open for reading. */
EXPECT_NULL(hdfsOpenFile(fs, paths->file1, O_RDONLY, 0, 0, 0));
/* hdfsOpenFile should not accept mode = 3 */
EXPECT_NULL(hdfsOpenFile(fs, paths->file1, 3, 0, 0, 0));
file = hdfsOpenFile(fs, paths->file1, O_WRONLY, 0, 0, 0);
EXPECT_NONNULL(file);
/* TODO: implement writeFully and use it here */
expected = (int)strlen(paths->prefix);
ret = hdfsWrite(fs, file, paths->prefix, expected);
if (ret < 0) {
ret = errno;
fprintf(stderr, "hdfsWrite failed and set errno %d\n", ret);
return ret;
}
if (ret != expected) {
fprintf(stderr, "hdfsWrite was supposed to write %d bytes, but "
"it wrote %d\n", ret, expected);
return EIO;
}
EXPECT_ZERO(hdfsFlush(fs, file));
EXPECT_ZERO(hdfsHSync(fs, file));
EXPECT_ZERO(hdfsCloseFile(fs, file));
/* There should be 1 entry in the directory. */
EXPECT_NONNULL(hdfsListDirectory(fs, paths->prefix, &numEntries));
if (numEntries != 1) {
fprintf(stderr, "hdfsListDirectory set numEntries to "
"%d on directory containing 1 file.", numEntries);
}
/* Let's re-open the file for reading */
file = hdfsOpenFile(fs, paths->file1, O_RDONLY, 0, 0, 0);
EXPECT_NONNULL(file);
EXPECT_ZERO(hdfsFileGetReadStatistics(file, &readStats));
errno = 0;
EXPECT_UINT64_EQ(UINT64_C(0), readStats->totalBytesRead);
EXPECT_UINT64_EQ(UINT64_C(0), readStats->totalLocalBytesRead);
EXPECT_UINT64_EQ(UINT64_C(0), readStats->totalShortCircuitBytesRead);
hdfsFileFreeReadStatistics(readStats);
/* TODO: implement readFully and use it here */
ret = hdfsRead(fs, file, tmp, sizeof(tmp));
if (ret < 0) {
ret = errno;
fprintf(stderr, "hdfsRead failed and set errno %d\n", ret);
return ret;
}
if (ret != expected) {
fprintf(stderr, "hdfsRead was supposed to read %d bytes, but "
"it read %d\n", ret, expected);
return EIO;
}
EXPECT_ZERO(hdfsFileGetReadStatistics(file, &readStats));
errno = 0;
EXPECT_UINT64_EQ((uint64_t)expected, readStats->totalBytesRead);
hdfsFileFreeReadStatistics(readStats);
EXPECT_ZERO(hdfsFileClearReadStatistics(file));
EXPECT_ZERO(hdfsFileGetReadStatistics(file, &readStats));
EXPECT_UINT64_EQ((uint64_t)0, readStats->totalBytesRead);
hdfsFileFreeReadStatistics(readStats);
EXPECT_ZERO(memcmp(paths->prefix, tmp, expected));
EXPECT_ZERO(hdfsCloseFile(fs, file));
// TODO: Non-recursive delete should fail?
//EXPECT_NONZERO(hdfsDelete(fs, prefix, 0));
EXPECT_ZERO(hdfsCopy(fs, paths->file1, fs, paths->file2));
EXPECT_ZERO(hdfsChown(fs, paths->file2, NULL, NULL));
EXPECT_ZERO(hdfsChown(fs, paths->file2, NULL, "doop"));
fileInfo = hdfsGetPathInfo(fs, paths->file2);
EXPECT_NONNULL(fileInfo);
EXPECT_ZERO(strcmp("doop", fileInfo->mGroup));
EXPECT_ZERO(hdfsFileIsEncrypted(fileInfo));
hdfsFreeFileInfo(fileInfo, 1);
EXPECT_ZERO(hdfsChown(fs, paths->file2, "ha", "doop2"));
fileInfo = hdfsGetPathInfo(fs, paths->file2);
EXPECT_NONNULL(fileInfo);
EXPECT_ZERO(strcmp("ha", fileInfo->mOwner));
EXPECT_ZERO(strcmp("doop2", fileInfo->mGroup));
hdfsFreeFileInfo(fileInfo, 1);
EXPECT_ZERO(hdfsChown(fs, paths->file2, "ha2", NULL));
fileInfo = hdfsGetPathInfo(fs, paths->file2);
EXPECT_NONNULL(fileInfo);
EXPECT_ZERO(strcmp("ha2", fileInfo->mOwner));
EXPECT_ZERO(strcmp("doop2", fileInfo->mGroup));
hdfsFreeFileInfo(fileInfo, 1);
snprintf(tmp, sizeof(tmp), "%s/nonexistent-file-name", paths->prefix);
EXPECT_NEGATIVE_ONE_WITH_ERRNO(hdfsChown(fs, tmp, "ha3", NULL), ENOENT);
return 0;
}
static int doTestHdfsOperations(struct tlhThreadInfo *ti, hdfsFS fs)
{
char prefix[256], tmp[256];
hdfsFile file;
int ret, expected;
hdfsFileInfo *fileInfo;
snprintf(prefix, sizeof(prefix), "/tlhData%04d", ti->threadIdx);
if (hdfsExists(fs, prefix) == 0) {
EXPECT_ZERO(hdfsDelete(fs, prefix, 1));
}
EXPECT_ZERO(hdfsCreateDirectory(fs, prefix));
snprintf(tmp, sizeof(tmp), "%s/file", prefix);
EXPECT_NONNULL(hdfsOpenFile(fs, tmp, O_RDONLY, 0, 0, 0));
file = hdfsOpenFile(fs, tmp, O_WRONLY, 0, 0, 0);
EXPECT_NONNULL(file);
/* TODO: implement writeFully and use it here */
expected = (int)strlen(prefix);
ret = hdfsWrite(fs, file, prefix, expected);
if (ret < 0) {
ret = errno;
fprintf(stderr, "hdfsWrite failed and set errno %d\n", ret);
return ret;
}
if (ret != expected) {
fprintf(stderr, "hdfsWrite was supposed to write %d bytes, but "
"it wrote %d\n", ret, expected);
return EIO;
}
EXPECT_ZERO(hdfsFlush(fs, file));
EXPECT_ZERO(hdfsCloseFile(fs, file));
/* Let's re-open the file for reading */
file = hdfsOpenFile(fs, tmp, O_RDONLY, 0, 0, 0);
EXPECT_NONNULL(file);
/* TODO: implement readFully and use it here */
ret = hdfsRead(fs, file, tmp, sizeof(tmp));
if (ret < 0) {
ret = errno;
fprintf(stderr, "hdfsRead failed and set errno %d\n", ret);
return ret;
}
if (ret != expected) {
fprintf(stderr, "hdfsRead was supposed to read %d bytes, but "
"it read %d\n", ret, expected);
return EIO;
}
EXPECT_ZERO(memcmp(prefix, tmp, expected));
EXPECT_ZERO(hdfsCloseFile(fs, file));
snprintf(tmp, sizeof(tmp), "%s/file", prefix);
EXPECT_NONZERO(hdfsChown(fs, tmp, NULL, NULL));
EXPECT_ZERO(hdfsChown(fs, tmp, NULL, "doop"));
fileInfo = hdfsGetPathInfo(fs, tmp);
EXPECT_NONNULL(fileInfo);
EXPECT_ZERO(strcmp("doop", fileInfo->mGroup));
hdfsFreeFileInfo(fileInfo, 1);
EXPECT_ZERO(hdfsChown(fs, tmp, "ha", "doop2"));
fileInfo = hdfsGetPathInfo(fs, tmp);
EXPECT_NONNULL(fileInfo);
EXPECT_ZERO(strcmp("ha", fileInfo->mOwner));
EXPECT_ZERO(strcmp("doop2", fileInfo->mGroup));
hdfsFreeFileInfo(fileInfo, 1);
EXPECT_ZERO(hdfsChown(fs, tmp, "ha2", NULL));
fileInfo = hdfsGetPathInfo(fs, tmp);
EXPECT_NONNULL(fileInfo);
EXPECT_ZERO(strcmp("ha2", fileInfo->mOwner));
EXPECT_ZERO(strcmp("doop2", fileInfo->mGroup));
hdfsFreeFileInfo(fileInfo, 1);
EXPECT_ZERO(hdfsDelete(fs, prefix, 1));
return 0;
}
int main(int argc, char **argv) {
hdfsFS fs = hdfsConnect("default", 0);
if(!fs) {
fprintf(stderr, "Oops! Failed to connect to hdfs!\n");
exit(-1);
}
hdfsFS lfs = hdfsConnect(NULL, 0);
if(!lfs) {
fprintf(stderr, "Oops! Failed to connect to 'local' hdfs!\n");
exit(-1);
}
const char* writePath = "/tmp/testfile.txt";
{
//Write tests
hdfsFile writeFile = hdfsOpenFile(fs, writePath, O_WRONLY|O_CREAT, 0, 0, 0);
if(!writeFile) {
fprintf(stderr, "Failed to open %s for writing!\n", writePath);
exit(-1);
}
fprintf(stderr, "Opened %s for writing successfully...\n", writePath);
char* buffer = "Hello, World!";
tSize num_written_bytes = hdfsWrite(fs, writeFile, (void*)buffer, strlen(buffer)+1);
fprintf(stderr, "Wrote %d bytes\n", num_written_bytes);
tOffset currentPos = -1;
if ((currentPos = hdfsTell(fs, writeFile)) == -1) {
fprintf(stderr,
"Failed to get current file position correctly! Got %ld!\n",
currentPos);
exit(-1);
}
fprintf(stderr, "Current position: %ld\n", currentPos);
if (hdfsFlush(fs, writeFile)) {
fprintf(stderr, "Failed to 'flush' %s\n", writePath);
exit(-1);
}
fprintf(stderr, "Flushed %s successfully!\n", writePath);
hdfsCloseFile(fs, writeFile);
}
{
//Read tests
const char* readPath = "/tmp/testfile.txt";
int exists = hdfsExists(fs, readPath);
if (exists) {
fprintf(stderr, "Failed to validate existence of %s\n", readPath);
exit(-1);
}
hdfsFile readFile = hdfsOpenFile(fs, readPath, O_RDONLY, 0, 0, 0);
if (!readFile) {
fprintf(stderr, "Failed to open %s for reading!\n", readPath);
exit(-1);
}
fprintf(stderr, "hdfsAvailable: %d\n", hdfsAvailable(fs, readFile));
tOffset seekPos = 1;
if(hdfsSeek(fs, readFile, seekPos)) {
fprintf(stderr, "Failed to seek %s for reading!\n", readPath);
exit(-1);
}
tOffset currentPos = -1;
if((currentPos = hdfsTell(fs, readFile)) != seekPos) {
fprintf(stderr,
"Failed to get current file position correctly! Got %ld!\n",
currentPos);
exit(-1);
}
fprintf(stderr, "Current position: %ld\n", currentPos);
static char buffer[32];
tSize num_read_bytes = hdfsRead(fs, readFile, (void*)buffer,
sizeof(buffer));
fprintf(stderr, "Read following %d bytes:\n%s\n",
num_read_bytes, buffer);
num_read_bytes = hdfsPread(fs, readFile, 0, (void*)buffer,
sizeof(buffer));
fprintf(stderr, "Read following %d bytes:\n%s\n",
num_read_bytes, buffer);
hdfsCloseFile(fs, readFile);
}
int totalResult = 0;
int result = 0;
{
//Generic file-system operations
const char* srcPath = "/tmp/testfile.txt";
const char* dstPath = "/tmp/testfile2.txt";
fprintf(stderr, "hdfsCopy(remote-local): %s\n", ((result = hdfsCopy(fs, srcPath, lfs, srcPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsCopy(remote-remote): %s\n", ((result = hdfsCopy(fs, srcPath, fs, dstPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsMove(local-local): %s\n", ((result = hdfsMove(lfs, srcPath, lfs, dstPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsMove(remote-local): %s\n", ((result = hdfsMove(fs, srcPath, lfs, srcPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsRename: %s\n", ((result = hdfsRename(fs, dstPath, srcPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsCopy(remote-remote): %s\n", ((result = hdfsCopy(fs, srcPath, fs, dstPath)) ? "Failed!" : "Success!"));
totalResult += result;
const char* slashTmp = "/tmp";
const char* newDirectory = "/tmp/newdir";
fprintf(stderr, "hdfsCreateDirectory: %s\n", ((result = hdfsCreateDirectory(fs, newDirectory)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsSetReplication: %s\n", ((result = hdfsSetReplication(fs, srcPath, 2)) ? "Failed!" : "Success!"));
totalResult += result;
char buffer[256];
const char *resp;
fprintf(stderr, "hdfsGetWorkingDirectory: %s\n", ((resp = hdfsGetWorkingDirectory(fs, buffer, sizeof(buffer))) ? buffer : "Failed!"));
totalResult += (resp ? 0 : 1);
fprintf(stderr, "hdfsSetWorkingDirectory: %s\n", ((result = hdfsSetWorkingDirectory(fs, slashTmp)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsGetWorkingDirectory: %s\n", ((resp = hdfsGetWorkingDirectory(fs, buffer, sizeof(buffer))) ? buffer : "Failed!"));
totalResult += (resp ? 0 : 1);
fprintf(stderr, "hdfsGetDefaultBlockSize: %ld\n", hdfsGetDefaultBlockSize(fs));
fprintf(stderr, "hdfsGetCapacity: %ld\n", hdfsGetCapacity(fs));
fprintf(stderr, "hdfsGetUsed: %ld\n", hdfsGetUsed(fs));
hdfsFileInfo *fileInfo = NULL;
if((fileInfo = hdfsGetPathInfo(fs, slashTmp)) != NULL) {
fprintf(stderr, "hdfsGetPathInfo - SUCCESS!\n");
fprintf(stderr, "Name: %s, ", fileInfo->mName);
fprintf(stderr, "Type: %c, ", (char)(fileInfo->mKind));
fprintf(stderr, "Replication: %d, ", fileInfo->mReplication);
fprintf(stderr, "BlockSize: %ld, ", fileInfo->mBlockSize);
fprintf(stderr, "Size: %ld, ", fileInfo->mSize);
fprintf(stderr, "LastMod: %s", ctime(&fileInfo->mLastMod));
fprintf(stderr, "Owner: %s, ", fileInfo->mOwner);
fprintf(stderr, "Group: %s, ", fileInfo->mGroup);
char permissions[10];
permission_disp(fileInfo->mPermissions, permissions);
fprintf(stderr, "Permissions: %d (%s)\n", fileInfo->mPermissions, permissions);
hdfsFreeFileInfo(fileInfo, 1);
} else {
totalResult++;
fprintf(stderr, "waah! hdfsGetPathInfo for %s - FAILED!\n", slashTmp);
}
hdfsFileInfo *fileList = 0;
int numEntries = 0;
if((fileList = hdfsListDirectory(fs, slashTmp, &numEntries)) != NULL) {
int i = 0;
for(i=0; i < numEntries; ++i) {
fprintf(stderr, "Name: %s, ", fileList[i].mName);
fprintf(stderr, "Type: %c, ", (char)fileList[i].mKind);
fprintf(stderr, "Replication: %d, ", fileList[i].mReplication);
fprintf(stderr, "BlockSize: %ld, ", fileList[i].mBlockSize);
fprintf(stderr, "Size: %ld, ", fileList[i].mSize);
fprintf(stderr, "LastMod: %s", ctime(&fileList[i].mLastMod));
fprintf(stderr, "Owner: %s, ", fileList[i].mOwner);
fprintf(stderr, "Group: %s, ", fileList[i].mGroup);
char permissions[10];
permission_disp(fileList[i].mPermissions, permissions);
fprintf(stderr, "Permissions: %d (%s)\n", fileList[i].mPermissions, permissions);
}
hdfsFreeFileInfo(fileList, numEntries);
} else {
if (errno) {
totalResult++;
fprintf(stderr, "waah! hdfsListDirectory - FAILED!\n");
} else {
fprintf(stderr, "Empty directory!\n");
}
}
char*** hosts = hdfsGetHosts(fs, srcPath, 0, 1);
if(hosts) {
fprintf(stderr, "hdfsGetHosts - SUCCESS! ... \n");
int i=0;
while(hosts[i]) {
int j = 0;
while(hosts[i][j]) {
fprintf(stderr,
"\thosts[%d][%d] - %s\n", i, j, hosts[i][j]);
++j;
}
++i;
}
} else {
totalResult++;
fprintf(stderr, "waah! hdfsGetHosts - FAILED!\n");
}
char *newOwner = "root";
// setting tmp dir to 777 so later when connectAsUser nobody, we can write to it
short newPerm = 0666;
// chown write
fprintf(stderr, "hdfsChown: %s\n", ((result = hdfsChown(fs, writePath, NULL, "users")) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsChown: %s\n", ((result = hdfsChown(fs, writePath, newOwner, NULL)) ? "Failed!" : "Success!"));
totalResult += result;
// chmod write
fprintf(stderr, "hdfsChmod: %s\n", ((result = hdfsChmod(fs, writePath, newPerm)) ? "Failed!" : "Success!"));
totalResult += result;
sleep(2);
tTime newMtime = time(NULL);
tTime newAtime = time(NULL);
// utime write
fprintf(stderr, "hdfsUtime: %s\n", ((result = hdfsUtime(fs, writePath, newMtime, newAtime)) ? "Failed!" : "Success!"));
totalResult += result;
// chown/chmod/utime read
hdfsFileInfo *finfo = hdfsGetPathInfo(fs, writePath);
fprintf(stderr, "hdfsChown read: %s\n", ((result = (strcmp(finfo->mOwner, newOwner) != 0)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsChmod read: %s\n", ((result = (finfo->mPermissions != newPerm)) ? "Failed!" : "Success!"));
totalResult += result;
// will later use /tmp/ as a different user so enable it
fprintf(stderr, "hdfsChmod: %s\n", ((result = hdfsChmod(fs, "/tmp/", 0777)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr,"newMTime=%ld\n",newMtime);
fprintf(stderr,"curMTime=%ld\n",finfo->mLastMod);
fprintf(stderr, "hdfsUtime read (mtime): %s\n", ((result = (finfo->mLastMod != newMtime)) ? "Failed!" : "Success!"));
totalResult += result;
// No easy way to turn on access times from hdfs_test right now
// fprintf(stderr, "hdfsUtime read (atime): %s\n", ((result = (finfo->mLastAccess != newAtime)) ? "Failed!" : "Success!"));
// totalResult += result;
hdfsFreeFileInfo(finfo, 1);
// Clean up
fprintf(stderr, "hdfsDelete: %s\n", ((result = hdfsDelete(fs, newDirectory)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsDelete: %s\n", ((result = hdfsDelete(fs, srcPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsDelete: %s\n", ((result = hdfsDelete(lfs, srcPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsDelete: %s\n", ((result = hdfsDelete(lfs, dstPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsExists: %s\n", ((result = hdfsExists(fs, newDirectory)) ? "Success!" : "Failed!"));
totalResult += (result ? 0 : 1);
}
totalResult += (hdfsDisconnect(fs) != 0);
{
//
// Now test as connecting as a specific user
// This is only meant to test that we connected as that user, not to test
// the actual fs user capabilities. Thus just create a file and read
// the owner is correct.
const char *tuser = "******";
const char* writePath = "/tmp/usertestfile.txt";
const char **groups = (const char**)malloc(sizeof(char*)* 2);
groups[0] = "users";
groups[1] = "nobody";
fs = hdfsConnectAsUser("default", 0, tuser, groups, 2);
if(!fs) {
fprintf(stderr, "Oops! Failed to connect to hdfs as user %s!\n",tuser);
exit(-1);
}
hdfsFile writeFile = hdfsOpenFile(fs, writePath, O_WRONLY|O_CREAT, 0, 0, 0);
if(!writeFile) {
fprintf(stderr, "Failed to open %s for writing!\n", writePath);
exit(-1);
}
fprintf(stderr, "Opened %s for writing successfully...\n", writePath);
char* buffer = "Hello, World!";
tSize num_written_bytes = hdfsWrite(fs, writeFile, (void*)buffer, strlen(buffer)+1);
fprintf(stderr, "Wrote %d bytes\n", num_written_bytes);
if (hdfsFlush(fs, writeFile)) {
fprintf(stderr, "Failed to 'flush' %s\n", writePath);
exit(-1);
}
fprintf(stderr, "Flushed %s successfully!\n", writePath);
hdfsCloseFile(fs, writeFile);
hdfsFileInfo *finfo = hdfsGetPathInfo(fs, writePath);
fprintf(stderr, "hdfs new file user is correct: %s\n", ((result = (strcmp(finfo->mOwner, tuser) != 0)) ? "Failed!" : "Success!"));
totalResult += result;
}
totalResult += (hdfsDisconnect(fs) != 0);
if (totalResult != 0) {
return -1;
} else {
return 0;
}
}
int MaprCopyingInputStream::Read(void * buffer, int size){
CHECK(input_->is_open_);
return hdfsRead(input_->fs_, input_->file_, buffer, size);
}
