Future<Nothing> HadoopFetcherPlugin::fetch(
const URI& uri,
const string& directory)
{
// TODO(jieyu): Validate the given URI.
if (!uri.has_path()) {
return Failure("URI path is not specified");
}
Try<Nothing> mkdir = os::mkdir(directory);
if (mkdir.isError()) {
return Failure(
"Failed to create directory '" +
directory + "': " + mkdir.error());
}
// NOTE: We ignore the scheme prefix if the host in URI is not
// specified. This is the case when the host is set using the hadoop
// configuration file.
//
// TODO(jieyu): Allow user to specify the name of the output file.
return hdfs.get()->copyToLocal(
(uri.has_host() ? stringify(uri) : uri.path()),
path::join(directory, Path(uri.path()).basename()));
}
Future<Nothing> CurlFetcherPlugin::fetch(
const URI& uri,
const string& directory)
{
// TODO(jieyu): Validate the given URI.
if (!uri.has_path()) {
return Failure("URI path is not specified");
}
if (!os::exists(directory)) {
Try<Nothing> mkdir = os::mkdir(directory);
if (mkdir.isError()) {
return Failure(
"Failed to create directory '" +
directory + "': " + mkdir.error());
}
}
// TODO(jieyu): Allow user to specify the name of the output file.
const string output = path::join(directory, Path(uri.path()).basename());
const vector<string> argv = {
"curl",
"-s", // Don’t show progress meter or error messages.
"-S", // Makes curl show an error message if it fails.
"-L", // Follow HTTP 3xx redirects.
"-w", "%{http_code}", // Display HTTP response code on stdout.
"-o", output, // Write output to the file.
strings::trim(stringify(uri))
};
Try<Subprocess> s = subprocess(
"curl",
argv,
Subprocess::PATH("/dev/null"),
Subprocess::PIPE(),
Subprocess::PIPE());
if (s.isError()) {
return Failure("Failed to exec the curl subprocess: " + s.error());
}
return await(
s.get().status(),
io::read(s.get().out().get()),
io::read(s.get().err().get()))
.then(_fetch);
}
Future<Nothing> CopyFetcherPlugin::fetch(
const URI& uri,
const string& directory) const
{
// TODO(jojy): Validate the given URI.
if (!uri.has_path()) {
return Failure("URI path is not specified");
}
// TODO(jojy): Verify that the path is a file.
Try<Nothing> mkdir = os::mkdir(directory);
if (mkdir.isError()) {
return Failure(
"Failed to create directory '" +
directory + "': " + mkdir.error());
}
VLOG(1) << "Copying '" << uri.path() << "' to '" << directory << "'";
const vector<string> argv = {"cp", "-a", uri.path(), directory};
Try<Subprocess> s = subprocess(
"cp",
argv,
Subprocess::PATH(os::DEV_NULL),
Subprocess::PIPE(),
Subprocess::PIPE());
if (s.isError()) {
return Failure("Failed to exec the copy subprocess: " + s.error());
}
return await(
s.get().status(),
io::read(s.get().out().get()),
io::read(s.get().err().get()))
.then([](const tuple<
Future<Option<int>>,
Future<string>,
Future<string>>& t) -> Future<Nothing> {
Future<Option<int>> status = std::get<0>(t);
if (!status.isReady()) {
return Failure(
"Failed to get the exit status of the copy subprocess: " +
(status.isFailed() ? status.failure() : "discarded"));
}
if (status->isNone()) {
return Failure("Failed to reap the copy subprocess");
}
if (status->get() != 0) {
Future<string> error = std::get<2>(t);
if (!error.isReady()) {
return Failure(
"Failed to perform 'copy'. Reading stderr failed: " +
(error.isFailed() ? error.failure() : "discarded"));
}
return Failure("Failed to perform 'copy': " + error.get());
}
return Nothing();
});
}
Future<Nothing> CurlFetcherPlugin::fetch(
const URI& uri,
const string& directory)
{
// TODO(jieyu): Validate the given URI.
if (!uri.has_path()) {
return Failure("URI path is not specified");
}
Try<Nothing> mkdir = os::mkdir(directory);
if (mkdir.isError()) {
return Failure(
"Failed to create directory '" +
directory + "': " + mkdir.error());
}
// TODO(jieyu): Allow user to specify the name of the output file.
const string output = path::join(directory, Path(uri.path()).basename());
const vector<string> argv = {
"curl",
"-s", // Don’t show progress meter or error messages.
"-S", // Makes curl show an error message if it fails.
"-L", // Follow HTTP 3xx redirects.
"-w", "%{http_code}", // Display HTTP response code on stdout.
"-o", output, // Write output to the file.
strings::trim(stringify(uri))
};
Try<Subprocess> s = subprocess(
"curl",
argv,
Subprocess::PATH("/dev/null"),
Subprocess::PIPE(),
Subprocess::PIPE());
if (s.isError()) {
return Failure("Failed to exec the curl subprocess: " + s.error());
}
return await(
s.get().status(),
io::read(s.get().out().get()),
io::read(s.get().err().get()))
.then([](const tuple<
Future<Option<int>>,
Future<string>,
Future<string>>& t) -> Future<Nothing> {
Future<Option<int>> status = std::get<0>(t);
if (!status.isReady()) {
return Failure(
"Failed to get the exit status of the curl subprocess: " +
(status.isFailed() ? status.failure() : "discarded"));
}
if (status->isNone()) {
return Failure("Failed to reap the curl subprocess");
}
if (status->get() != 0) {
Future<string> error = std::get<2>(t);
if (!error.isReady()) {
return Failure(
"Failed to perform 'curl'. Reading stderr failed: " +
(error.isFailed() ? error.failure() : "discarded"));
}
return Failure("Failed to perform 'curl': " + error.get());
}
Future<string> output = std::get<1>(t);
if (!output.isReady()) {
return Failure(
"Failed to read stdout from 'curl': " +
(output.isFailed() ? output.failure() : "discarded"));
}
// Parse the output and get the HTTP response code.
Try<int> code = numify<int>(output.get());
if (code.isError()) {
return Failure("Unexpected output from 'curl': " + output.get());
}
if (code.get() != http::Status::OK) {
return Failure(
"Unexpected HTTP response code: " +
http::Status::string(code.get()));
}
return Nothing();
});
}
