From 9e2ddec4b31a57b32e4033b63a0f1827a000a777 Mon Sep 17 00:00:00 2001
From: trinity-1686a <trinity@quickwit.io>
Date: Sat, 13 Jul 2024 20:28:12 +0200
Subject: [PATCH] merge adjacent block when building delta for automaton

---
 sstable/Cargo.toml        |  1 +
 sstable/src/dictionary.rs | 11 +++++++++++
 2 files changed, 12 insertions(+)

diff --git a/sstable/Cargo.toml b/sstable/Cargo.toml
index cf840ec80..7a2702486 100644
--- a/sstable/Cargo.toml
+++ b/sstable/Cargo.toml
@@ -12,6 +12,7 @@ description = "sstables for tantivy"
 [dependencies]
 common = {version= "0.7", path="../common", package="tantivy-common"}
 futures-util = "0.3.30"
+itertools = "0.13.0"
 tantivy-bitpacker = { version= "0.6", path="../bitpacker" }
 tantivy-fst = "0.5"
 # experimental gives us access to Decompressor::upper_bound
diff --git a/sstable/src/dictionary.rs b/sstable/src/dictionary.rs
index e57748aa1..b98513788 100644
--- a/sstable/src/dictionary.rs
+++ b/sstable/src/dictionary.rs
@@ -8,6 +8,7 @@ use common::bounds::{transform_bound_inner_res, TransformBound};
 use common::file_slice::FileSlice;
 use common::{BinarySerializable, OwnedBytes};
 use futures_util::{stream, StreamExt, TryStreamExt};
+use itertools::Itertools;
 use tantivy_fst::automaton::AlwaysMatch;
 use tantivy_fst::Automaton;
 
@@ -254,6 +255,16 @@ impl<TSSTable: SSTable> Dictionary<TSSTable> {
             .get_block_for_automaton(automaton)
             .filter(move |(block_id, _)| block_range.contains(block_id))
             .map(|(_, block_addr)| block_addr)
+            .coalesce(|first, second| {
+                if first.byte_range.end == second.byte_range.start {
+                    Ok(BlockAddr {
+                        first_ordinal: first.first_ordinal,
+                        byte_range: first.byte_range.start..second.byte_range.end,
+                    })
+                } else {
+                    Err((first, second))
+                }
+            })
     }
 
     /// Opens a `TermDictionary`.