YT-23398: More robust checks in dictionary compression

commit_hash:7cf2c37d325706e6180cee7e15e7193f760f3716

3 files changed, 19 insertions, 6 deletions

diff --git a/yt/yt/client/table_client/config.h b/yt/yt/client/table_client/config.h
index 2f91bcd032..bdb92d90be 100644
--- a/yt/yt/client/table_client/config.h
+++ b/yt/yt/client/table_client/config.h
@@ -249,10 +249,6 @@ public:
     //! Recommended to be ~100 times less than weight of samples for that column.
     i64 ColumnDictionarySize;
 
-    //! Level of compression algorithm.
-    //! Applied to digested compression dictionary upon its construction.
-    int CompressionLevel;
-
     //! Subset of all dictionary building policies.
     //! Will build and apply dictionaries only from this subset.
     //! Upon each chunk compression will independently decide which dictionary fits best.
diff --git a/yt/yt/core/compression/dictionary_codec.h b/yt/yt/core/compression/dictionary_codec.h
index a394d452e6..fc36b4812d 100644
--- a/yt/yt/core/compression/dictionary_codec.h
+++ b/yt/yt/core/compression/dictionary_codec.h
@@ -95,8 +95,9 @@ struct IDictionaryCompressionCodec
     virtual IDictionaryDecompressorPtr CreateDictionaryDecompressor(
         const IDigestedDecompressionDictionaryPtr& digestedDecompressionDictionary) const = 0;
 
-    // NB: Raw #compressionDictionary data will be copied and stored within digested dictionary in a preprocessed form.
+    //! NB: Raw #compressionDictionary data will be copied and stored within digested dictionary in a preprocessed form.
     //! #compressionLevel determines compression level that will be applied for each compression with that dictionary later on.
+    //! These methods may throw.
     virtual IDigestedCompressionDictionaryPtr CreateDigestedCompressionDictionary(
         const TSharedRef& compressionDictionary,
         int compressionLevel) const = 0;
@@ -104,6 +105,7 @@ struct IDictionaryCompressionCodec
         const TSharedRef& compressionDictionary) const = 0;
 
     //! Parses header of compressed frame #input and returns specified frame info.
+    //! This method may throw.
     virtual TDictionaryCompressionFrameInfo GetFrameInfo(TRef input) const = 0;
 };
 
diff --git a/yt/yt/core/compression/zstd.cpp b/yt/yt/core/compression/zstd.cpp
index 5d64d69f80..8772de735b 100644
--- a/yt/yt/core/compression/zstd.cpp
+++ b/yt/yt/core/compression/zstd.cpp
@@ -220,7 +220,12 @@ TDictionaryCompressionFrameInfo ZstdGetFrameInfo(TRef input)
         input.Begin(),
         input.Size(),
         ZSTD_f_zstd1_magicless);
-    YT_VERIFY(result == 0);
+    if (result != 0) {
+        THROW_ERROR_EXCEPTION("Failed to get frame header")
+            << TErrorAttribute("code", result)
+            << TErrorAttribute("is_error", ZSTD_isError(result))
+            << TErrorAttribute("error", ZSTD_getErrorName(result));
+    }
 
     return {
         .ContentSize = frameHeader.frameContentSize,
@@ -488,11 +493,17 @@ IDigestedCompressionDictionaryPtr ZstdCreateDigestedCompressionDictionary(
     int compressionLevel)
 {
     YT_VERIFY(compressionDictionary);
+    YT_VERIFY(compressionLevel >= 0 && compressionLevel <= ZstdGetMaxCompressionLevel());
 
     auto* digestedDictionary = ZSTD_createCDict(
         compressionDictionary.Begin(),
         compressionDictionary.Size(),
         compressionLevel);
+    if (!digestedDictionary) {
+        THROW_ERROR_EXCEPTION("Failed to create digested compression dictionary")
+            << TErrorAttribute("compression_level", compressionLevel)
+            << TErrorAttribute("dictionary_size", compressionDictionary.Size());
+    }
     return New<TDigestedCompressionDictionary>(digestedDictionary);
 }
 
@@ -504,6 +515,10 @@ IDigestedDecompressionDictionaryPtr ZstdCreateDigestedDecompressionDictionary(
     auto* digestedDictionary = ZSTD_createDDict(
         compressionDictionary.Begin(),
         compressionDictionary.Size());
+    if (!digestedDictionary) {
+        THROW_ERROR_EXCEPTION("Failed to create digested decompression dictionary")
+            << TErrorAttribute("dictionary_size", compressionDictionary.Size());
+    }
     return New<TDigestedDecompressionDictionary>(digestedDictionary);
 }