WIP: Multiset for verifying no duplicate chunks

tests/collections.rs

@@ -0,0 +1,51 @@
+//! Custom collections used in our tests.
+//!
+//! Ideally, we would eventually move these to their own crate.
+//! The only reason we have this in the `tests` directory for now,
+//! is that it is only used in tests, so should not be compiled into
+//! production code.
+
+use std::collections::HashMap;
+use std::hash::Hash;
+
+/// An unordered collection of items that allows duplicates.
+/// This is implemented as a `HashMap` where the keys are the items
+/// and the values are the counts of the items.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct HashMultiSet<T>
+where
+    T: Hash + Eq,
+{
+    pub inner: HashMap<T, usize>,
+}
+
+impl<T> HashMultiSet<T>
+where
+    T: Hash + Eq,
+{
+    pub fn new() -> Self {
+        Self {
+            inner: HashMap::new(),
+        }
+    }
+
+    pub fn insert(&mut self, item: T) {
+        *self.inner.entry(item).or_default() += 1;
+    }
+}
+
+impl<T> FromIterator<T> for HashMultiSet<T>
+where
+    T: Hash + Eq,
+{
+    fn from_iter<I>(iter: I) -> Self
+    where
+        I: IntoIterator<Item = T>,
+    {
+        let mut set = Self::new();
+        for item in iter {
+            set.insert(item);
+        }
+        set
+    }
+}

tests/integration/test_utils/chunk_upload.rs

@@ -1,12 +1,13 @@
 //! Utilities for chunk upload tests.
-use std::collections::HashSet;
 use std::error::Error;
 use std::str;
 use std::sync::LazyLock;
 
 use mockito::Request;
 use regex::bytes::Regex;
 
+use crate::collections::HashMultiSet;
+
 /// This regex is used to extract the boundary from the content-type header.
 /// We need to match the boundary, since it changes with each request.
 /// The regex matches the format as specified in
@@ -41,7 +42,7 @@ impl HeaderContainer for Request {
 pub fn split_chunk_body<'b>(
     body: &'b [u8],
     boundary: &str,
-) -> Result<HashSet<&'b [u8]>, Box<dyn Error>> {
+) -> Result<HashMultiSet<&'b [u8]>, Box<dyn Error>> {
     let escaped_boundary = regex::escape(boundary);
 
     let inner_body = entire_body_regex(&escaped_boundary)

tests/mod.rs

@@ -1 +1,2 @@
+mod collections;
 mod integration;