From 621971f0786e841de6fbf4afc92c5829b18918fe Mon Sep 17 00:00:00 2001
From: Pieter-Jan Briers <pieterjan.briers+git@gmail.com>
Date: Mon, 11 Apr 2022 16:06:28 +0200
Subject: [PATCH] Use some ::with_capacity when reading zip file.

Now with a proper benchmark
---
 Cargo.toml               |  4 ++++
 benches/read_metadata.rs | 41 ++++++++++++++++++++++++++++++++++++++++
 src/read.rs              |  4 ++--
 3 files changed, 47 insertions(+), 2 deletions(-)
 create mode 100644 benches/read_metadata.rs

diff --git a/Cargo.toml b/Cargo.toml
index 8a079446e..4870c3828 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -42,3 +42,7 @@ default = ["aes-crypto", "bzip2", "deflate", "time", "zstd"]
 [[bench]]
 name = "read_entry"
 harness = false
+
+[[bench]]
+name = "read_metadata"
+harness = false
diff --git a/benches/read_metadata.rs b/benches/read_metadata.rs
new file mode 100644
index 000000000..51f1f69e8
--- /dev/null
+++ b/benches/read_metadata.rs
@@ -0,0 +1,41 @@
+use bencher::{benchmark_group, benchmark_main};
+
+use std::io::{Cursor, Write};
+
+use bencher::Bencher;
+use zip::{ZipArchive, ZipWriter};
+
+const FILE_COUNT: usize = 15_000;
+const FILE_SIZE: usize = 1024;
+
+fn generate_random_archive(count_files: usize, file_size: usize) -> Vec<u8> {
+    let data = Vec::new();
+    let mut writer = ZipWriter::new(Cursor::new(data));
+    let options =
+        zip::write::FileOptions::default().compression_method(zip::CompressionMethod::Stored);
+
+    let bytes = vec![0u8; file_size];
+
+    for i in 0..count_files {
+        let name = format!(
+            "file_deadbeefdeadbeefdeadbeefdeadbeefdeadbeefdeadbeef_{}.dat",
+            i
+        );
+        writer.start_file(name, options).unwrap();
+        writer.write_all(&bytes).unwrap();
+    }
+
+    writer.finish().unwrap().into_inner()
+}
+
+fn read_metadata(bench: &mut Bencher) {
+    let bytes = generate_random_archive(FILE_COUNT, FILE_SIZE);
+
+    bench.iter(|| {
+        let archive = ZipArchive::new(Cursor::new(bytes.as_slice())).unwrap();
+        archive.len()
+    });
+}
+
+benchmark_group!(benches, read_metadata);
+benchmark_main!(benches);
diff --git a/src/read.rs b/src/read.rs
index 0d920032c..0c06db17c 100644
--- a/src/read.rs
+++ b/src/read.rs
@@ -408,8 +408,8 @@ impl<R: Read + io::Seek> ZipArchive<R> {
         let (archive_offset, directory_start, number_of_files) =
             Self::get_directory_counts(&mut reader, &footer, cde_start_pos)?;
 
-        let mut files = Vec::new();
-        let mut names_map = HashMap::new();
+        let mut files = Vec::with_capacity(number_of_files);
+        let mut names_map = HashMap::with_capacity(number_of_files);
 
         if reader.seek(io::SeekFrom::Start(directory_start)).is_err() {
             return Err(ZipError::InvalidArchive(