From 808bcba975b1e994bc9dd9e318c6cf855f26f1a2 Mon Sep 17 00:00:00 2001
From: Tishj <t_b@live.nl>
Date: Tue, 14 Jan 2025 12:46:41 +0100
Subject: [PATCH] optimize scan to flat vector

---
 .../compression/dict_fsst/decompression.cpp   | 31 ++++++++++++++-----
 .../compression/dict_fsst/fetch_row.test      |  4 +--
 2 files changed, 25 insertions(+), 10 deletions(-)

diff --git a/src/storage/compression/dict_fsst/decompression.cpp b/src/storage/compression/dict_fsst/decompression.cpp
index f0a957596be..0df5f3d43bd 100644
--- a/src/storage/compression/dict_fsst/decompression.cpp
+++ b/src/storage/compression/dict_fsst/decompression.cpp
@@ -106,15 +106,30 @@ void CompressedStringScanState::ScanToFlatVector(Vector &result, idx_t result_of
 
 	BitpackingPrimitives::UnPackBuffer<sel_t>(data_ptr_cast(sel_vec_ptr), src, decompress_count, current_width);
 
-	for (idx_t i = 0; i < scan_count; i++) {
-		// Lookup dict offset in index buffer
-		auto string_number = sel_vec->get_index(i + start_offset);
-		if (string_number == 0) {
-			validity.SetInvalid(result_offset + i);
+	if (dictionary) {
+		// We have prepared the full dictionary, we can reference these strings directly
+		auto dictionary_values = FlatVector::GetData<string_t>(*dictionary);
+		for (idx_t i = 0; i < scan_count; i++) {
+			// Lookup dict offset in index buffer
+			auto string_number = sel_vec->get_index(i + start_offset);
+			if (string_number == 0) {
+				validity.SetInvalid(result_offset + i);
+			}
+			result_data[result_offset + i] = dictionary_values[string_number];
+		}
+	} else {
+		// This path is taken for fetch, where we don't want to decompress the full dictionary
+		for (idx_t i = 0; i < scan_count; i++) {
+			// Lookup dict offset in index buffer
+			auto string_number = sel_vec->get_index(i + start_offset);
+			if (string_number == 0) {
+				validity.SetInvalid(result_offset + i);
+			}
+			auto dict_offset = index_buffer_ptr[string_number];
+			auto str_len = GetStringLength(UnsafeNumericCast<sel_t>(string_number));
+			result_data[result_offset + i] =
+			    FetchStringFromDict(result, UnsafeNumericCast<int32_t>(dict_offset), str_len);
 		}
-		auto dict_offset = index_buffer_ptr[string_number];
-		auto str_len = GetStringLength(UnsafeNumericCast<sel_t>(string_number));
-		result_data[result_offset + i] = FetchStringFromDict(result, UnsafeNumericCast<int32_t>(dict_offset), str_len);
 	}
 }
 
diff --git a/test/sql/storage/compression/dict_fsst/fetch_row.test b/test/sql/storage/compression/dict_fsst/fetch_row.test
index 1d3f85bffe0..e60eb440ca0 100644
--- a/test/sql/storage/compression/dict_fsst/fetch_row.test
+++ b/test/sql/storage/compression/dict_fsst/fetch_row.test
@@ -24,7 +24,7 @@ SELECT
 		WHEN 0 THEN 'aaaa'
 		WHEN 1 THEN 'bbbb'
 		WHEN 2 THEN 'cccc'
-		WHEN 3 THEN 'dddd'
+		WHEN 3 THEN 'this is not an inlined string'
 		WHEN 4 THEN NULL
 	END AS b
 FROM range(10_000) t(x);
@@ -48,5 +48,5 @@ select distinct b from test order by a % 5;
 aaaa
 bbbb
 cccc
-dddd
+this is not an inlined string
 NULL