25.8 Stable Backport of #96118 - Fix Gorilla codec decompression when data_bytes_size exceeds data size

src/Compression/CompressionCodecGorilla.cpp

@@ -438,10 +438,14 @@ UInt32 CompressionCodecGorilla::doDecompressData(const char * source, UInt32 sou
     if (static_cast<UInt32>(2 + bytes_to_skip) > source_size)
         throw Exception(ErrorCodes::CANNOT_DECOMPRESS, "Cannot decompress Gorilla-encoded data. File has wrong header");
 
-    if (bytes_to_skip >= uncompressed_size)
+    if (bytes_to_skip > uncompressed_size)
         throw Exception(ErrorCodes::CANNOT_DECOMPRESS, "Cannot decompress Gorilla-encoded data. File has wrong header");
 
     memcpy(dest, &source[2], bytes_to_skip);
+
+    /// All data is in the skip section when data_bytes_size > uncompressed_size.
+    if (bytes_to_skip == uncompressed_size)
+        return uncompressed_size;
     UInt32 source_size_no_header = source_size - bytes_to_skip - 2;
     UInt32 uncompressed_size_left = uncompressed_size - bytes_to_skip;
     switch (bytes_size)

tests/queries/0_stateless/03823_gorilla_codec_mismatched_size.reference

@@ -0,0 +1,4 @@
+1
+42
+100
+7

tests/queries/0_stateless/03823_gorilla_codec_mismatched_size.sql

@@ -0,0 +1,31 @@
+-- Gorilla codec with data_bytes_size larger than the column type.
+-- When all data fits in the "skip" section (uncompressed_size < data_bytes_size),
+-- decompression should still work correctly.
+
+SET allow_suspicious_codecs = 1;
+
+DROP TABLE IF EXISTS t_gorilla_mismatched;
+
+CREATE TABLE t_gorilla_mismatched (c0 Int16 CODEC(Gorilla(4))) ENGINE = MergeTree() ORDER BY tuple();
+INSERT INTO t_gorilla_mismatched (c0) VALUES (1);
+SELECT * FROM t_gorilla_mismatched;
+
+-- Also test with codec pipeline
+DROP TABLE IF EXISTS t_gorilla_mismatched;
+CREATE TABLE t_gorilla_mismatched (c0 Int16 CODEC(Gorilla(4), ZSTD)) ENGINE = MergeTree() ORDER BY tuple();
+INSERT INTO t_gorilla_mismatched (c0) VALUES (42);
+SELECT * FROM t_gorilla_mismatched;
+
+-- Test with Gorilla(8) on Int16
+DROP TABLE IF EXISTS t_gorilla_mismatched;
+CREATE TABLE t_gorilla_mismatched (c0 Int16 CODEC(Gorilla(8))) ENGINE = MergeTree() ORDER BY tuple();
+INSERT INTO t_gorilla_mismatched (c0) VALUES (100);
+SELECT * FROM t_gorilla_mismatched;
+
+-- Test with compact parts (the original issue scenario)
+DROP TABLE IF EXISTS t_gorilla_mismatched;
+CREATE TABLE t_gorilla_mismatched (c0 Int16 CODEC(Gorilla(4))) ENGINE = MergeTree() ORDER BY tuple() SETTINGS min_bytes_for_wide_part = 10000000000;
+INSERT INTO t_gorilla_mismatched (c0) VALUES (7);
+SELECT * FROM t_gorilla_mismatched;
+
+DROP TABLE IF EXISTS t_gorilla_mismatched;