From 562435ec39ce0a3ea4eea5bf98c5783f9dd1fd59 Mon Sep 17 00:00:00 2001
From: Ritchie Vink <ritchie46@gmail.com>
Date: Sun, 26 Jun 2022 15:43:38 +0200
Subject: [PATCH] reuse scratch when reading dictionaries

---
 src/io/ipc/append/mod.rs  |  3 ++-
 src/io/ipc/read/common.rs |  1 +
 src/io/ipc/read/reader.rs | 10 +++++++---
 3 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/src/io/ipc/append/mod.rs b/src/io/ipc/append/mod.rs
index 4a225c4db61..9e084a45896 100644
--- a/src/io/ipc/append/mod.rs
+++ b/src/io/ipc/append/mod.rs
@@ -32,7 +32,8 @@ impl<R: Read + Seek + Write> FileWriter<R> {
             ));
         }
 
-        let dictionaries = read::read_file_dictionaries(&mut writer, &metadata)?;
+        let dictionaries =
+            read::read_file_dictionaries(&mut writer, &metadata, &mut Default::default())?;
 
         let last_block = metadata.blocks.last().ok_or_else(|| {
             Error::oos("An Arrow IPC file must have at least 1 message (the schema message)")
diff --git a/src/io/ipc/read/common.rs b/src/io/ipc/read/common.rs
index 04504c0f6d7..dededdb817d 100644
--- a/src/io/ipc/read/common.rs
+++ b/src/io/ipc/read/common.rs
@@ -306,6 +306,7 @@ impl ReadBuffer {
     /// method on `Vec` this is `safe` because this function guarantees that
     /// the underlying data always is initialized.
     pub fn set_len(&mut self, length: usize) {
+        dbg!(length, self.data.len(), self.data.capacity());
         if length > self.data.capacity() {
             self.data = vec![0; length];
         } else if length > self.data.len() {
diff --git a/src/io/ipc/read/reader.rs b/src/io/ipc/read/reader.rs
index c40adbdd694..9f7ab43346f 100644
--- a/src/io/ipc/read/reader.rs
+++ b/src/io/ipc/read/reader.rs
@@ -116,9 +116,9 @@ fn read_dictionary_block<R: Read + Seek>(
 pub fn read_file_dictionaries<R: Read + Seek>(
     reader: &mut R,
     metadata: &FileMetadata,
+    scratch: &mut ReadBuffer,
 ) -> Result<Dictionaries> {
     let mut dictionaries = Default::default();
-    let mut data = vec![].into();
 
     let blocks = if let Some(blocks) = metadata.dictionaries.as_deref() {
         blocks
@@ -127,7 +127,7 @@ pub fn read_file_dictionaries<R: Read + Seek>(
     };
 
     for block in blocks {
-        read_dictionary_block(reader, metadata, block, &mut dictionaries, &mut data)?;
+        read_dictionary_block(reader, metadata, block, &mut dictionaries, scratch)?;
     }
     Ok(dictionaries)
 }
@@ -343,7 +343,11 @@ impl<R: Read + Seek> FileReader<R> {
 
     fn read_dictionaries(&mut self) -> Result<()> {
         if self.dictionaries.is_none() {
-            self.dictionaries = Some(read_file_dictionaries(&mut self.reader, &self.metadata)?);
+            self.dictionaries = Some(read_file_dictionaries(
+                &mut self.reader,
+                &self.metadata,
+                &mut self.buffer,
+            )?);
         };
         Ok(())
     }