PARQUET-212: Read non-thrift files if a Thrift class is supplied.

Parquet-thrift can now read files not written by parquet-thrift if an
appropriate Thrift class is supplied. This adds a check to derive the
necessary StructType from a class. Previously, attempting to read a file
without Thrift metadata in its properties would return a null
ThriftMetaData and throw NPE.

This also updates the logic in ThriftMetaData.fromExtraMetaData to avoid
NPE when the class is present by the descriptor property is not.

Author: rdblue

parquet-thrift/src/main/java/org/apache/parquet/hadoop/thrift/ThriftReadSupport.java

@@ -203,17 +203,16 @@ private void initThriftClassFromMultipleFiles(Map<String, Set<String>> fileMetad
   }
 
   @SuppressWarnings("unchecked")
-  private void initThriftClass(Map<String, String> fileMetadata, Configuration conf) throws ClassNotFoundException {
+  private void initThriftClass(ThriftMetaData metadata, Configuration conf) throws ClassNotFoundException {
     if (thriftClass != null) {
       return;
     }
     String className = conf.get(THRIFT_READ_CLASS_KEY, null);
     if (className == null) {
-      final ThriftMetaData metaData = ThriftMetaData.fromExtraMetaData(fileMetadata);
-      if (metaData == null) {
+      if (metadata == null) {
         throw new ParquetDecodingException("Could not read file as the Thrift class is not provided and could not be resolved from the file");
       }
-      thriftClass = (Class<T>)metaData.getThriftClass();
+      thriftClass = (Class<T>)metadata.getThriftClass();
     } else {
       thriftClass = (Class<T>)Class.forName(className);
     }
@@ -225,7 +224,12 @@ public RecordMaterializer<T> prepareForRead(Configuration configuration,
       org.apache.parquet.hadoop.api.ReadSupport.ReadContext readContext) {
     ThriftMetaData thriftMetaData = ThriftMetaData.fromExtraMetaData(keyValueMetaData);
     try {
-      initThriftClass(keyValueMetaData, configuration);
+      initThriftClass(thriftMetaData, configuration);
+
+      // if there was not metadata in the file, get it from requested class
+      if (thriftMetaData == null) {
+        thriftMetaData = ThriftMetaData.fromThriftClass(thriftClass);
+      }
 
       String converterClassName = configuration.get(RECORD_CONVERTER_CLASS_KEY, RECORD_CONVERTER_DEFAULT);
       @SuppressWarnings("unchecked")

parquet-thrift/src/main/java/org/apache/parquet/thrift/ThriftMetaData.java

@@ -23,6 +23,7 @@
 import org.apache.parquet.hadoop.BadConfigurationException;
 import org.apache.parquet.thrift.struct.ThriftType;
 import org.apache.parquet.thrift.struct.ThriftType.StructType;
+import org.apache.thrift.TBase;
 
 /**
  *
@@ -86,19 +87,35 @@ public StructType getDescriptor() {
    * Reads ThriftMetadata from the parquet file footer.
    *
    * @param extraMetaData  extraMetaData field of the parquet footer
-   * @return
+   * @return the ThriftMetaData used to write a data file
    */
   public static ThriftMetaData fromExtraMetaData(
       Map<String, String> extraMetaData) {
     final String thriftClassName = extraMetaData.get(THRIFT_CLASS);
     final String thriftDescriptorString = extraMetaData.get(THRIFT_DESCRIPTOR);
-    if (thriftClassName == null && thriftDescriptorString == null) {
+    if (thriftClassName == null || thriftDescriptorString == null) {
       return null;
     }
     final StructType descriptor = parseDescriptor(thriftDescriptorString);
     return new ThriftMetaData(thriftClassName, descriptor);
   }
 
+  /**
+   * Creates ThriftMetaData from a Thrift-generated class.
+   *
+   * @param thriftClass a Thrift-generated class
+   * @return ThriftMetaData for the given class
+   */
+  @SuppressWarnings("unchecked")
+  public static ThriftMetaData fromThriftClass(Class<?> thriftClass) {
+    if (thriftClass != null && TBase.class.isAssignableFrom(thriftClass)) {
+      Class<? extends TBase<?, ?>> tClass = (Class<? extends TBase<?, ?>>) thriftClass;
+      StructType descriptor = new ThriftSchemaConverter().toStructType(tClass);
+      return new ThriftMetaData(thriftClass.getName(), descriptor);
+    }
+    return null;
+  }
+
   private static StructType parseDescriptor(String json) {
     try {
       return (StructType)ThriftType.fromJSON(json);