Verify cached digest against file

Author: luca020400

data/src/cache.rs

@@ -5,23 +5,29 @@ use std::path::{Path, PathBuf};
 use chrono::Utc;
 use derive_more::AsRef;
 use serde::{Deserialize, Serialize, de::DeserializeOwned};
+use sha2::{Digest, Sha256};
 use tokio::fs;
+use tokio_stream::StreamExt;
+use tokio_util::io::ReaderStream;
 use url::Url;
 
 pub use trim::TrimConfig;
 
 /// SHA256 digest of cache content.
-#[derive(Debug, Clone, Serialize, Deserialize, AsRef)]
-pub struct Digest(String);
+#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, AsRef)]
+pub struct HexDigest(String);
 
-impl Digest {
+impl HexDigest {
     pub fn new(data: &[u8]) -> Self {
         Self(hex::encode(data))
     }
 }
 
+#[derive(Debug, Clone, Copy)]
+pub struct Asset<'a>(pub &'a Path, pub &'a HexDigest);
+
 pub trait CachedAsset {
-    fn paths(&self) -> Vec<&Path>;
+    fn assets(&self) -> Vec<Asset<'_>>;
 }
 
 #[derive(Debug, Clone, Serialize, Deserialize)]
@@ -51,10 +57,14 @@ impl FileCache {
         let state: CacheState<T> = serde_json::from_slice(&bytes).ok()?;
 
         if let CacheState::Ok(ref asset) = state {
-            let any_missing = asset.paths().iter().any(|p| !p.exists());
-            if any_missing {
-                // If any of the asset's files are missing, treat the cache as invalid.
-                return None;
+            for Asset(path, digest) in asset.assets() {
+                // Check if we can open the file
+                let file = fs::File::open(path).await.ok()?;
+
+                // Check if the file content matches the expected digest
+                if hash_file(file).await? != *digest {
+                    return None;
+                }
             }
         }
 
@@ -90,7 +100,7 @@ impl FileCache {
             .join(format!("{hash}.json"))
     }
 
-    pub fn blob_path(&self, digest: &Digest, ext: &str) -> PathBuf {
+    pub fn blob_path(&self, digest: &HexDigest, ext: &str) -> PathBuf {
         let hash = digest.as_ref();
 
         blob_dir_from_root(&self.root)
@@ -113,3 +123,15 @@ impl FileCache {
 pub fn blob_dir_from_root(root: &Path) -> PathBuf {
     root.join("blobs")
 }
+
+async fn hash_file(file: fs::File) -> Option<HexDigest> {
+    let mut stream = ReaderStream::new(file);
+    let mut hasher = Sha256::new();
+
+    while let Some(item) = stream.next().await {
+        let chunk = item.ok()?;
+        hasher.update(&chunk);
+    }
+
+    Some(HexDigest::new(&hasher.finalize()))
+}

data/src/preview.rs

@@ -1,6 +1,5 @@
 use std::collections::HashMap;
 use std::io;
-use std::path::Path;
 use std::sync::{Arc, LazyLock, OnceLock};
 use std::time::Duration;
 
@@ -19,7 +18,7 @@ use url::Url;
 
 pub use self::card::Card;
 pub use self::image::Image;
-use crate::cache::{self, CacheState, CachedAsset, FileCache};
+use crate::cache::{self, Asset, CacheState, CachedAsset, FileCache};
 use crate::message::Source;
 use crate::server::Server;
 use crate::target::{self, TargetRef};
@@ -125,10 +124,12 @@ impl Preview {
 }
 
 impl CachedAsset for Preview {
-    fn paths(&self) -> Vec<&Path> {
+    fn assets(&self) -> Vec<Asset<'_>> {
         match self {
-            Preview::Card(c) => vec![c.image.path.as_path()],
-            Preview::Image(i) => vec![i.path.as_path()],
+            Preview::Card(c) => {
+                vec![Asset(c.image.path.as_path(), &c.image.digest)]
+            }
+            Preview::Image(i) => vec![Asset(i.path.as_path(), &i.digest)],
         }
     }
 }
@@ -321,7 +322,7 @@ async fn fetch(
                     written += chunk.len();
                 }
 
-                let digest = cache::Digest::new(&hasher.finalize());
+                let digest = cache::HexDigest::new(&hasher.finalize());
                 let image_path =
                     cache.blob_path(&digest, format.extensions_str()[0]);
 

data/src/preview/image.rs

@@ -3,7 +3,7 @@ use std::path::PathBuf;
 use serde::{Deserialize, Serialize};
 use url::Url;
 
-use crate::cache::Digest;
+use crate::cache::HexDigest;
 
 pub type Format = image::ImageFormat;
 pub type Error = image::ImageError;
@@ -13,15 +13,15 @@ pub struct Image {
     #[serde(with = "serde_format")]
     pub format: Format,
     pub url: Url,
-    pub digest: Digest,
+    pub digest: HexDigest,
     pub path: PathBuf,
 }
 
 impl Image {
     pub fn new(
         format: Format,
         url: Url,
-        digest: Digest,
+        digest: HexDigest,
         path: PathBuf,
     ) -> Self {
         Self {