public abstract class CascadingUtils {
private static final String MAPPING_NAMES = "es.mapping.names";
private static final boolean CASCADING_22_AVAILABLE =
ObjectUtils.isClassPresent("cascading.tuple.type.CoercibleType", Tap.class.getClassLoader());
static Settings addDefaultsToSettings(
Properties flowProperties, Properties tapProperties, Log log) {
Settings settings =
HadoopSettingsManager.loadFrom(CascadingUtils.extractOriginalProperties(flowProperties))
.merge(tapProperties);
InitializationUtils.discoverNodesIfNeeded(settings, log);
InitializationUtils.filterNonClientNodesIfNeeded(settings, log);
InitializationUtils.discoverEsVersion(settings, log);
InitializationUtils.setValueWriterIfNotSet(settings, CascadingValueWriter.class, log);
InitializationUtils.setValueReaderIfNotSet(settings, JdkValueReader.class, log);
InitializationUtils.setBytesConverterIfNeeded(
settings, CascadingLocalBytesConverter.class, log);
InitializationUtils.setFieldExtractorIfNotSet(settings, CascadingFieldExtractor.class, log);
return settings;
}
static void addSerializationToken(Object config) {
Configuration cfg = (Configuration) config;
String tokens = cfg.get(TupleSerializationProps.SERIALIZATION_TOKENS);
String lmw = LinkedMapWritable.class.getName();
// if no tokens are defined, add one starting with 140
if (tokens == null) {
cfg.set(TupleSerializationProps.SERIALIZATION_TOKENS, "140=" + lmw);
LogFactory.getLog(EsTap.class)
.trace(String.format("Registered Cascading serialization token %s for %s", 140, lmw));
} else {
// token already registered
if (tokens.contains(lmw)) {
return;
}
// find token id
Map<Integer, String> mapping = new LinkedHashMap<Integer, String>();
tokens = tokens.replaceAll("\\s", ""); // allow for whitespace in token set
for (String pair : tokens.split(",")) {
String[] elements = pair.split("=");
mapping.put(Integer.parseInt(elements[0]), elements[1]);
}
for (int id = 140; id < 255; id++) {
if (!mapping.containsKey(Integer.valueOf(id))) {
cfg.set(
TupleSerializationProps.SERIALIZATION_TOKENS,
Util.join(",", Util.removeNulls(tokens, id + "=" + lmw)));
LogFactory.getLog(EsTap.class)
.trace(String.format("Registered Cascading serialization token %s for %s", id, lmw));
return;
}
}
}
}
static FieldAlias alias(Settings settings) {
return new FieldAlias(SettingsUtils.aliases(settings.getProperty(MAPPING_NAMES), false), false);
}
static List<String> asStrings(Fields fields) {
if (fields == null || !fields.isDefined()) {
// use auto-generated name
return Collections.emptyList();
}
int size = fields.size();
List<String> names = new ArrayList<String>(size);
for (int fieldIndex = 0; fieldIndex < size; fieldIndex++) {
names.add(fields.get(fieldIndex).toString());
}
return names;
}
static Collection<String> fieldToAlias(Settings settings, Fields fields) {
FieldAlias fa = alias(settings);
List<String> names = asStrings(fields);
for (int i = 0; i < names.size(); i++) {
String original = names.get(i);
String alias = fa.toES(original);
if (alias != null) {
names.set(i, alias);
}
}
return names;
}
static Properties extractOriginalProperties(Properties copy) {
Field field = ReflectionUtils.findField(Properties.class, "defaults", Properties.class);
ReflectionUtils.makeAccessible(field);
return ReflectionUtils.getField(field, copy);
}
static Settings init(
Settings settings, String nodes, int port, String resource, String query, boolean read) {
if (StringUtils.hasText(nodes)) {
settings.setHosts(nodes);
}
if (port > 0) {
settings.setPort(port);
}
if (StringUtils.hasText(query)) {
settings.setQuery(query);
}
if (StringUtils.hasText(resource)) {
if (read) {
settings.setResourceRead(resource);
} else {
settings.setResourceWrite(resource);
}
}
return settings;
}
private abstract static class CoercibleOps {
static void setObject(TupleEntry entry, Comparable<?> field, Object object) {
if (object != null && entry.getFields().getType(field) instanceof CoercibleType) {
entry.setObject(field, object.toString());
} else {
entry.setObject(field, object);
}
}
static Tuple coerceToString(SinkCall<?, ?> sinkCall) {
TupleEntry entry = sinkCall.getOutgoingEntry();
Fields fields = entry.getFields();
Tuple tuple = entry.getTuple();
if (fields.hasTypes()) {
Type types[] = new Type[fields.size()];
for (int index = 0; index < fields.size(); index++) {
Type type = fields.getType(index);
if (type instanceof CoercibleType<?>) {
types[index] = String.class;
} else {
types[index] = type;
}
}
tuple = entry.getCoercedTuple(types);
}
return tuple;
}
}
private abstract static class LegacyOps {
static void setObject(TupleEntry entry, Comparable<?> field, Object object) {
entry.setObject(field, object);
}
static Tuple coerceToString(SinkCall<?, ?> sinkCall) {
return sinkCall.getOutgoingEntry().getTuple();
}
}
static void setObject(TupleEntry entry, Comparable<?> field, Object object) {
if (CASCADING_22_AVAILABLE) {
CoercibleOps.setObject(entry, field, object);
} else {
LegacyOps.setObject(entry, field, object);
}
}
static Tuple coerceToString(SinkCall<?, ?> sinkCall) {
return (CASCADING_22_AVAILABLE
? CoercibleOps.coerceToString(sinkCall)
: LegacyOps.coerceToString(sinkCall));
}
@SuppressWarnings("rawtypes")
public static Tap hadoopTap(
String host, int port, String path, String query, Fields fields, Properties props) {
return new EsHadoopTap(host, port, path, query, fields, props);
}
}
abstract class HiveUtils {
// Date type available since Hive 0.12
static final boolean DATE_WRITABLE_AVAILABLE =
ObjectUtils.isClassPresent(
HiveConstants.DATE_WRITABLE, TimestampWritable.class.getClassLoader());
static StandardStructObjectInspector structObjectInspector(Properties tableProperties) {
// extract column info - don't use Hive constants as they were renamed in 0.9 breaking
// compatibility
// the column names are saved as the given inspector to #serialize doesn't preserves them (maybe
// because it's an external table)
// use the class since StructType requires it ...
List<String> columnNames =
StringUtils.tokenize(tableProperties.getProperty(HiveConstants.COLUMNS), ",");
List<TypeInfo> colTypes =
TypeInfoUtils.getTypeInfosFromTypeString(
tableProperties.getProperty(HiveConstants.COLUMNS_TYPES));
// create a standard writable Object Inspector - used later on by serialization/deserialization
List<ObjectInspector> inspectors = new ArrayList<ObjectInspector>();
for (TypeInfo typeInfo : colTypes) {
inspectors.add(TypeInfoUtils.getStandardWritableObjectInspectorFromTypeInfo(typeInfo));
}
return ObjectInspectorFactory.getStandardStructObjectInspector(columnNames, inspectors);
}
static StructTypeInfo typeInfo(StructObjectInspector inspector) {
return (StructTypeInfo) TypeInfoUtils.getTypeInfoFromObjectInspector(inspector);
}
static Collection<String> columnToAlias(Settings settings) {
FieldAlias fa = alias(settings);
List<String> columnNames =
StringUtils.tokenize(settings.getProperty(HiveConstants.COLUMNS), ",");
// eliminate virtual columns
// we can't use virtual columns since some distro don't have this field...
// for (VirtualColumn vc : VirtualColumn.VIRTUAL_COLUMNS) {
// columnNames.remove(vc.getName());
// }
for (String vc : HiveConstants.VIRTUAL_COLUMNS) {
columnNames.remove(vc);
}
for (int i = 0; i < columnNames.size(); i++) {
String original = columnNames.get(i);
String alias = fa.toES(original);
if (alias != null) {
columnNames.set(i, alias);
}
}
return columnNames;
}
static FieldAlias alias(Settings settings) {
Map<String, String> aliasMap =
SettingsUtils.aliases(settings.getProperty(HiveConstants.MAPPING_NAMES));
// add default aliases for serialization (_colX -> mapping name)
Map<String, String> columnMap = columnMap(settings);
for (Entry<String, String> entry : columnMap.entrySet()) {
String columnName = entry.getKey();
String columnIndex = entry.getValue();
if (!aliasMap.isEmpty()) {
String alias = aliasMap.get(columnName);
if (alias != null) {
columnName = alias;
}
}
aliasMap.put(columnIndex, columnName);
}
return new FieldAlias(aliasMap);
}
static Map<String, String> columnMap(Settings settings) {
return columnMap(settings.getProperty(HiveConstants.COLUMNS));
}
// returns a map of {<column-name>:_colX}
private static Map<String, String> columnMap(String columnString) {
// add default aliases for serialization (mapping name -> _colX)
List<String> columnNames = StringUtils.tokenize(columnString, ",");
if (columnNames.isEmpty()) {
return Collections.emptyMap();
}
Map<String, String> columns = new LinkedHashMap<String, String>();
for (int i = 0; i < columnNames.size(); i++) {
columns.put(columnNames.get(i), HiveConstants.UNNAMED_COLUMN_PREFIX + i);
}
return columns;
}
static void init(Settings settings, Log log) {
InitializationUtils.checkIdForOperation(settings);
InitializationUtils.setFieldExtractorIfNotSet(settings, HiveFieldExtractor.class, log);
try {
InitializationUtils.discoverEsVersion(settings, log);
} catch (IOException ex) {
throw new EsHadoopIllegalStateException("Cannot discover Elasticsearch version", ex);
}
}
static void fixHive13InvalidComments(Settings settings, Properties tbl) {
if (Booleans.parseBoolean(settings.getProperty("es.hive.disable.columns.comments.fix"))) {
return;
}
settings.setProperty(HiveConstants.COLUMN_COMMENTS, "");
tbl.remove(HiveConstants.COLUMN_COMMENTS);
}
}