commit f349dc65af43911f8b839d25bacbd39bcc3dddd9
author Quanlong Huang <huangquanlong@gmail.com> Mon Jan 20 16:08:10 2020 +0800
committer Gang Wu <gangwu@apache.org> Mon Jan 20 16:08:10 2020 +0800
tree d1bc44ef0ffcd79d58cee1e100beb365024e7e39
parent a4d6e845576960f9f55c0a43082a8a6cdecd6338

ORC-591: [C++] Check missing blob stream for StringDictionaryColumnReader

This fixes #471 

c++/src/ColumnReader.cc

diff --git a/c++/src/ColumnReader.cc b/c++/src/ColumnReader.cc
index e01a24b..016aed8 100644
--- a/c++/src/ColumnReader.cc
+++ b/c++/src/ColumnReader.cc
@@ -589,14 +589,15 @@
       }
       lengthArray[i] += lengthArray[i - 1];
     }
-    dictionary->dictionaryBlob.resize(
-                                static_cast<uint64_t>(lengthArray[dictSize]));
+    int64_t blobSize = lengthArray[dictSize];
+    dictionary->dictionaryBlob.resize(static_cast<uint64_t>(blobSize));
     std::unique_ptr<SeekableInputStream> blobStream =
       stripe.getStream(columnId, proto::Stream_Kind_DICTIONARY_DATA, false);
-    readFully(
-              dictionary->dictionaryBlob.data(),
-              lengthArray[dictSize],
-              blobStream.get());
+    if (blobSize > 0 && blobStream == nullptr) {
+      throw ParseError(
+          "DICTIONARY_DATA stream not found in StringDictionaryColumn");
+    }
+    readFully(dictionary->dictionaryBlob.data(), blobSize, blobStream.get());
   }
 
   StringDictionaryColumnReader::~StringDictionaryColumnReader() {

examples/corrupt/missing_blob_stream_in_string_dict.orc

diff --git a/examples/corrupt/missing_blob_stream_in_string_dict.orc b/examples/corrupt/missing_blob_stream_in_string_dict.orc
new file mode 100644
index 0000000..1c7f742
--- /dev/null
+++ b/examples/corrupt/missing_blob_stream_in_string_dict.orc
Binary files differ

tools/test/TestFileScan.cc

diff --git a/tools/test/TestFileScan.cc b/tools/test/TestFileScan.cc
index 8c783d7..de2f91d 100644
--- a/tools/test/TestFileScan.cc
+++ b/tools/test/TestFileScan.cc
@@ -149,7 +149,9 @@
   checkForError(findExample("corrupt/stripe_footer_bad_column_encodings.orc"),
       "bad number of ColumnEncodings in StripeFooter: expected=6, actual=0");
   checkForError(findExample("corrupt/negative_dict_entry_lengths.orc"),
-        "Negative dictionary entry length");
+      "Negative dictionary entry length");
   checkForError(findExample("corrupt/missing_length_stream_in_string_dict.orc"),
-        "LENGTH stream not found in StringDictionaryColumn");
+      "LENGTH stream not found in StringDictionaryColumn");
+  checkForError(findExample("corrupt/missing_blob_stream_in_string_dict.orc"),
+      "DICTIONARY_DATA stream not found in StringDictionaryColumn");
 }