bool MaprFileSystem::IsDirectory(const std::string& uri){
std::string path = GetUriPathOrDie(uri);
std::string host = "default";
hdfsFS fs = hdfsConnect(host.c_str(), 0); // use default config file settings
CHECK(fs);
hdfsFileInfo* info = hdfsGetPathInfo(fs, path.c_str());
bool is_directory = false;
if (info){
is_directory = (info->mKind == kObjectKindDirectory);
hdfsFreeFileInfo(info,1);
}
else{
LOG(FATAL) << "uri does not exist: " << uri;
}
CHECK_EQ(hdfsDisconnect(fs), 0);
return is_directory;
}
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);
}
hdfsFS HdfsConnection::getConnection(std::string host, int port)
{
QString id = _makeId(host, port);
if (_connections.contains(id))
{
return _connections[id];
}
else
{
hdfsFS fs = hdfsConnect(host.c_str(), port);
if (fs == NULL)
{
throw std::ios_base::failure("Error connecting to HDFS.");
}
_connections[id] = fs;
return fs;
}
}
static PyObject *
pyhdfsFS_delete(char* filepath)
{
// Parse HDFS information
char* hostport = (char *)malloc(strlen(filepath)+1);
char* buf = NULL;
char* path;
char* portStr;
char* host;
int hdfsPort;
int ret = 0;
ret = sscanf(filepath, "hdfs://%s", hostport);
host = strtok_r(hostport, ":", &buf);
portStr = strtok_r(NULL, "/", &buf);
ret = sscanf(portStr, "%d", &hdfsPort);
hdfsFS fs = hdfsConnect(host, hdfsPort);
if (!fs)
{
PyErr_SetString(exception, "Cannot connect to host");
return exception;
}
path = (char*) malloc(strlen(buf)+2);
path[0] = '/';
memcpy(path+1,buf,strlen(buf));
path[strlen(buf)+1] = '\0';
if (hdfsExists(fs, path) == 0)
{
hdfsDelete(fs, path);
}
else
{
PyErr_SetString(exception, "Cannot delete file");
return NULL;
}
free(hostport);
free(path);
return Py_BuildValue("i",1);
}
int dfsList(const char* path){
hdfsFS fs = hdfsConnect("default", 0);
int i, entries;
hdfsFileInfo *files, *head;
/* Get the list info */
files = hdfsListDirectory(fs, path, &entries);
if (!files){
perror("Get directory info");
exit(-1);
}
head = files;
/* Print the info */
fprintf(stdout, "%s %-50s %-9s %s\n",
"Kind", "Name", "Size", "Replicas");
for (i = 0; i < entries; i++){
const char* unit[] = {" B", "KB", "MB", "GB", "TB", "PB"};
double size = files->mSize;
unsigned int u = 0;
while (size > 1024){
u++;
size /= 1024;
}
assert(u < 6);
fprintf(stdout, "%4c %-50s %-7.2lf%s %8d\n",
files->mKind, files->mName,
size, unit[u],
files->mReplication);
files += 1;
}
/* List ends */
hdfsFreeFileInfo(head, entries);
hdfsDisconnect(fs);
return 0;
}
int main(int argc, char **argv) {
if (argc != 4) {
fprintf(stderr, "Usage: hdfs_read <filename> <filesize> <buffersize>\n");
exit(-1);
}
hdfsFS fs = hdfsConnect("default", 0);
if (!fs) {
fprintf(stderr, "Oops! Failed to connect to hdfs!\n");
exit(-1);
}
const char* rfile = argv[1];
tSize fileTotalSize = strtoul(argv[2], NULL, 10);
tSize bufferSize = strtoul(argv[3], 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(-2);
}
// data to be written to the file
char* buffer = malloc(sizeof(char) * bufferSize);
if(buffer == NULL) {
return -2;
}
// read from the file
tSize curSize = bufferSize;
for (; curSize == bufferSize;) {
curSize = hdfsRead(fs, readFile, (void*)buffer, curSize);
}
free(buffer);
hdfsCloseFile(fs, readFile);
hdfsDisconnect(fs);
return 0;
}
qioerr hdfs_connect(void** fs_out, const char* pathname, int port)
{
qioerr err_out = 0;
hdfsFS fs;
hdfs_fs* ret = (hdfs_fs*)qio_calloc(sizeof(hdfs_fs), 1);
STARTING_SLOW_SYSCALL;
fs = hdfsConnect(pathname, port);
CREATE_ERROR((fs == NULL), err_out, ECONNREFUSED,"Unable to connect HDFS", error);
DONE_SLOW_SYSCALL;
ret->hfs = fs;
DO_INIT_REFCNT(ret);
ret->fs_name = pathname;
ret->fs_port = port;
*fs_out = ret;
error:
return err_out;
}
// Caller takes ownership of returned object and must delete it when done
google::protobuf::io::CodedInputStream*
MaprInputCodedBlockFile::CreateCodedStream(uint64 position, uint64 length) {
CHECK(is_open_);
// Seek to requested position (relative to start of file).
CHECK_LT(position, size_);
CHECK_LE(position+length, size_);
// Starting with MapR V2.1.1, sometimes seek fails and the hdfs connection
// must be reset in order to try again. Not sure why this transient error
// happens with MapR V2.1.1 but not earlier versions.
bool success = false;
for (int i=0; i < 10; ++i){
if (hdfsSeek(fs_, file_, position) == 0){
success = true;
break;
}
//LOG(INFO) << "seek attempt failed: " << i;
//LOG(INFO) << "path:" << path_ << "\n position: " << position << "\n length: " << length << "\n size: " << size_; // success if returns 0
CHECK_EQ(hdfsCloseFile(fs_, file_), 0);
CHECK_EQ(hdfsDisconnect(fs_), 0);
std::string host = "default";
fs_ = hdfsConnect(host.c_str(), 0); // use default config file settings
CHECK(fs_) << "error connecting to maprfs";
file_ = hdfsOpenFile(fs_, path_.c_str(), O_RDONLY, 0, 0, 0);
CHECK(file_ != NULL);
sleep(2*i);
}
CHECK(success);
// Create a coded stream (hold it in a scoped ptr to manage deleting).
limiting_stream_.reset(NULL); // the destructor references the copying_stream_, so must destroy it before destroying it
copying_stream_.reset(new google::protobuf::io::CopyingInputStreamAdaptor(copying_input_stream_.get()));
limiting_stream_.reset(new google::protobuf::io::LimitingInputStream(copying_stream_.get(), length));
return new google::protobuf::io::CodedInputStream(limiting_stream_.get());
}
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;
}
}
HdfsConnector::HdfsConnector(vector<vector <string> > hdfs_writepath):Connector(hdfs_writepath) {
hdfsFS hdfsfs = hdfsConnect(Config::hdfs_master_ip.c_str(),Config::hdfs_master_port);
fs = hdfsfs;
}
int uploadFile(const char *path){
hdfsFS fs = hdfsConnect("default", 0);
hdfsFile fd_w;
int fd;
unsigned long size, i;
struct stat fd_s;
char *buf_r;
const char *filename;
/* Get the file size */
if ((fd = open(path, O_RDONLY)) < 0){
perror("Open file failed");
exit(1);
}
if (fstat(fd, &fd_s) < 0){
perror("Get file stat failed");
exit(1);
}
size = fd_s.st_size;
/* Open the file at hdfs */
filename = strrchr(path, '/');
if (!filename){
filename = path;
}else{
filename += 1;
}
fd_w = hdfsOpenFile(fs, filename, O_WRONLY|O_CREAT, 0, 0, 0);
if (!fd_w){
perror("Failed to create file to upload");
exit(1);
}
/* Write the file */
for (i = 0; i + UPLOAD_BLOCK < size; i += UPLOAD_BLOCK){
buf_r = (char *)mmap(NULL,\
UPLOAD_BLOCK,\
PROT_READ,\
MAP_SHARED,\
fd, i);
if (buf_r == MAP_FAILED){
perror("Failed to map memory");
exit(1);
}
hdfsWrite(fs, fd_w, (void*)buf_r, UPLOAD_BLOCK);
if (hdfsFlush(fs, fd_w)){
perror("Failed to flush");
exit(1);
}
if ((munmap(buf_r, UPLOAD_BLOCK)) < 0){
perror("memory unmap error");
exit(1);
}
}
if (size){
/* To avoid size == 0 */
buf_r = (char *)mmap(NULL,\
size - i,\
PROT_READ,\
MAP_SHARED,\
fd, i);
if (buf_r == MAP_FAILED){
perror("Failed to map memory");
exit(1);
}
hdfsWrite(fs, fd_w, (void*)buf_r, size - i);
if (hdfsFlush(fs, fd_w)){
perror("Failed to flush");
exit(1);
}
if ((munmap(buf_r, size - i)) < 0){
perror("memory unmap error");
exit(1);
}
}
/* Upload end */
close(fd);
hdfsCloseFile(fs, fd_w);
hdfsDisconnect(fs);
return 0;
}
/**
* Open a connection to the filesystem. The default arguments
* should be sufficient for most uses
*/
hdfs(const std::string& host = "default", tPort port = 0) {
filesystem = hdfsConnect(host.c_str(), port);
ASSERT_TRUE(filesystem != NULL);
} // end of constructor
/* constructor */
static PyObject *
pyhdfsFS_open(char* filepath,char mode)
{
pyhdfsFS *object = PyObject_NEW(pyhdfsFS, &pyhdfsFSType);
if (object != NULL)
{
// Parse HDFS information
char* hostport = (char *)malloc(strlen(filepath)+1);
char* buf = NULL;
char* path;
char* portStr;
char* host;
int hdfsPort;
int ret = 0;
ret = sscanf(filepath, "hdfs://%s", hostport);
host = strtok_r(hostport, ":", &buf);
portStr = strtok_r(NULL, "/", &buf);
ret = sscanf(portStr, "%d", &hdfsPort);
object->fs = hdfsConnect(host, hdfsPort);
if (!object->fs)
{
PyErr_SetString(exception, "Cannot connect to host");
return NULL;
}
path = (char*) malloc(strlen(buf)+2);
path[0] = '/';
memcpy(path+1,buf,strlen(buf));
path[strlen(buf)+1] = '\0';
if (mode == 'r')
{
if (hdfsExists(object->fs, path) == 0)
{
object->file = hdfsOpenFile(object->fs, path, O_RDONLY, 0, 0, 0);
}
else
{
char* error_message = (char *)calloc(strlen("Cannot open file for read : ") + strlen(path) + 1,
sizeof(char));
strcat(error_message, "Cannot open file for read : ");
strcat(error_message, path);
PyErr_SetString(exception, error_message);
free(error_message);
return NULL;
}
}
else if (mode == 'w')
{
if (hdfsExists(object->fs, path) == 0)
{
char* error_message = (char *)calloc(strlen("Cannot open file for write : ") + strlen(path) + 1,
sizeof(char));
strcat(error_message, "Cannot open file for write : ");
strcat(error_message, path);
PyErr_SetString(exception, error_message);
free(error_message);
return NULL;
}
else
object->file = hdfsOpenFile(object->fs, path, O_WRONLY, 0, 0, 0);
}
else
{
PyErr_SetString(exception, "Invalid mode");
return NULL;
}
if(!object->file)
{
char* error_message = (char *)calloc(strlen("Cannot open file : ") + strlen(path) + 1,
sizeof(char));
strcat(error_message, "Cannot open file : ");
strcat(error_message, path);
PyErr_SetString(exception, error_message);
free(error_message);
return NULL;
}
free(hostport);
free(path);
}
return (PyObject *)object;
}
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;
}
MaprOutputCodedBlockFile::MaprOutputCodedBlockFile() {
fs_ = hdfsConnect("default", 0); // use default config file settings
CHECK(fs_) << "error connecting to hdfs";
}
int main(int argc, char **argv) {
if (argc != 4) {
fprintf(stderr, "Usage: hdfs_write <filename> <filesize> <buffersize>\n");
exit(-1);
}
hdfsFS fs = hdfsConnect("default", 0);
if (!fs) {
fprintf(stderr, "Oops! Failed to connect to hdfs!\n");
exit(-1);
}
const char* writeFileName = argv[1];
off_t fileTotalSize = strtoul(argv[2], NULL, 10);
long long tmpBufferSize = strtoul(argv[3], NULL, 10);
// sanity check
if(fileTotalSize == ULONG_MAX && errno == ERANGE) {
fprintf(stderr, "invalid file size %s - must be <= %lu\n", argv[2], ULONG_MAX);
exit(-3);
}
// currently libhdfs writes are of tSize which is int32
if(tmpBufferSize > INT_MAX) {
fprintf(stderr, "invalid buffer size libhdfs API write chunks must be <= %d\n",INT_MAX);
exit(-3);
}
tSize bufferSize = tmpBufferSize;
hdfsFile writeFile = hdfsOpenFile(fs, writeFileName, O_WRONLY, bufferSize, 0, 0);
if (!writeFile) {
fprintf(stderr, "Failed to open %s for writing!\n", writeFileName);
exit(-2);
}
// data to be written to the file
char* buffer = malloc(sizeof(char) * bufferSize);
if(buffer == NULL) {
fprintf(stderr, "Could not allocate buffer of size %d\n", bufferSize);
return -2;
}
int i = 0;
for (i=0; i < bufferSize; ++i) {
buffer[i] = 'a' + (i%26);
}
// write to the file
off_t nrRemaining;
for (nrRemaining = fileTotalSize; nrRemaining > 0; nrRemaining -= bufferSize ) {
tSize curSize = ( bufferSize < nrRemaining ) ? bufferSize : (tSize)nrRemaining;
tSize written;
if ((written = hdfsWrite(fs, writeFile, (void*)buffer, curSize)) != curSize) {
fprintf(stderr, "ERROR: hdfsWrite returned an error on write: %d\n", written);
exit(-3);
}
}
free(buffer);
hdfsCloseFile(fs, writeFile);
hdfsDisconnect(fs);
return 0;
}
hdfsFS hdfsConnectNewInstance(const char* nn, tPort port)
{
return hdfsConnect(nn, port);
}
///////////////////////////////////////////////////////////////////////////////
// MaprInputCodedBlockFile
///////////////////////////////////////////////////////////////////////////////
MaprInputCodedBlockFile::MaprInputCodedBlockFile() {
std::string host = "default";
fs_ = hdfsConnect(host.c_str(), 0); // use default config file settings
CHECK(fs_) << "error connecting to maprfs";
copying_input_stream_.reset(new MaprCopyingInputStream(this));
}