Add TSF spectrum propagation and improve cloud parsing

Cargo.toml

@@ -13,20 +13,20 @@ unreachable_pub = "warn"
 
 [workspace.dependencies]
 # crates
-filemanager = { path = "crates/filemanager", version = "0.1.3", default-features = false }
-timsrust = { path = "crates/timsrust", version = "0.5.3", default-features = false }
-timsrust-centroid = { path = "crates/timsrust-centroid", version = "0.1.3", default-features = false }
-timsrust-core = { path = "crates/timsrust-core", version = "0.1.3", default-features = false }
-timsrust-mgf = { path = "crates/timsrust-mgf", version = "0.1.3", default-features = false }
-timsrust-minitdf = { path = "crates/timsrust-minitdf", version = "0.1.3", default-features = false }
-timsrust-parquet-spectra = { path = "crates/timsrust-parquet-spectra", version = "0.1.3", default-features = false }
-timsrust-sdk = { path = "crates/timsrust-sdk", version = "0.1.3", default-features = false }
-timsrust-tdf = { path = "crates/timsrust-tdf", version = "0.1.3", default-features = false }
-timsrust-tsf = { path = "crates/timsrust-tsf", version = "0.1.3", default-features = false }
-timsrust-utils = { path = "crates/timsrust-utils", version = "0.1.3", default-features = false }
+filemanager = { path = "crates/filemanager", version = "0.1.4", default-features = false }
+timsrust = { path = "crates/timsrust", version = "0.5.4", default-features = false }
+timsrust-centroid = { path = "crates/timsrust-centroid", version = "0.1.4", default-features = false }
+timsrust-core = { path = "crates/timsrust-core", version = "0.1.4", default-features = false }
+timsrust-mgf = { path = "crates/timsrust-mgf", version = "0.1.4", default-features = false }
+timsrust-minitdf = { path = "crates/timsrust-minitdf", version = "0.1.4", default-features = false }
+timsrust-parquet-spectra = { path = "crates/timsrust-parquet-spectra", version = "0.1.4", default-features = false }
+timsrust-sdk = { path = "crates/timsrust-sdk", version = "0.1.4", default-features = false }
+timsrust-tdf = { path = "crates/timsrust-tdf", version = "0.1.4", default-features = false }
+timsrust-tsf = { path = "crates/timsrust-tsf", version = "0.1.4", default-features = false }
+timsrust-utils = { path = "crates/timsrust-utils", version = "0.1.4", default-features = false }
 timsrust-patched = { version = "0.1.1", default-features = false }
 # clis
-timsrust-cli-core = { path = "clis/timsrust-cli-core", version = "0.1.3", default-features = false }
+timsrust-cli-core = { path = "clis/timsrust-cli-core", version = "0.1.4", default-features = false }
 # external
 arrow = { version = "57", default-features = false }
 bytemuck = { version = "1", default-features = false }

clis/timsrust-centroid-cli/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "timsrust-centroid-cli"
-version = "0.1.3"
+version = "0.1.4"
 edition = "2024"
 publish = false
 

clis/timsrust-cli-core/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "timsrust-cli-core"
-version = "0.1.3"
+version = "0.1.4"
 edition = "2024"
 publish = false
 

clis/timsrust-mgf-cli/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "timsrust-mgf-cli"
-version = "0.1.3"
+version = "0.1.4"
 edition = "2024"
 publish = false
 

clis/timsrust-parquet-spectra-cli/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "timsrust-parquet-spectra-cli"
-version = "0.1.3"
+version = "0.1.4"
 edition = "2024"
 publish = false
 

crates/filemanager/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "filemanager"
-version = "0.1.3"
+version = "0.1.4"
 edition = "2021"
 description = "Unified file abstraction for local and cloud files with caching options and support for multiple formats."
 license = "Apache-2.0"

crates/filemanager/src/cache.rs

@@ -132,6 +132,23 @@ impl FileCache {
         }
     }
 
+    /// Returns the local cache URI for `uri` without checking existence.
+    ///
+    /// Like [`cached_local`](Self::cached_local) but never touches the
+    /// filesystem. Local URIs pass through unchanged. For cloud URIs,
+    /// returns the path the file *would* live at inside the cache
+    /// directory (whether or not it has been downloaded yet), or `None`
+    /// when the URI has no key or no cache directory is configured.
+    pub fn local_path(&self, uri: impl Into<Uri>) -> Option<Uri> {
+        let uri = uri.into();
+        if uri.is_local() {
+            return Some(uri);
+        }
+        let dir = self.dir.as_ref()?;
+        let key = uri.key()?;
+        Some(Uri::from(dir.join(key)))
+    }
+
     pub fn invalidate(&self, uri: impl Into<Uri>) -> Result<(), CacheError> {
         match &self.dir {
             Some(dir) => {
@@ -186,4 +203,22 @@ impl Uri {
     pub fn cached_local(&self) -> Option<Uri> {
         crate::global_cache().cached_local(self.clone())
     }
+
+    /// Returns the local-filesystem representation of this URI, without
+    /// performing any network or existence checks.
+    ///
+    /// Local URIs are returned unchanged. Cloud URIs are mapped to the
+    /// path they would occupy inside the configured cache directory
+    /// (regardless of whether the file has been downloaded). If no cache
+    /// directory is configured or the URI has no key, the URI is returned
+    /// unchanged.
+    ///
+    /// Useful as a fast pre-check: callers can probe the returned URI
+    /// with local filesystem operations and only fall back to network
+    /// access if those probes fail.
+    pub fn local_representation(&self) -> Uri {
+        crate::global_cache()
+            .local_path(self.clone())
+            .unwrap_or_else(|| self.clone())
+    }
 }

crates/filemanager/src/cloud_store/cloud/authenticate.rs

@@ -1,4 +1,5 @@
-use std::sync::Arc;
+use std::collections::HashMap;
+use std::sync::{Arc, OnceLock, RwLock};
 
 use object_store::{path::Path as ObjectPath, ObjectStore};
 use url::Url;
@@ -7,6 +8,69 @@ use crate::{
     cloud_store::CloudProvider, uri::URI_SCHEME_SEPARATOR, CloudError,
 };
 
+/// Process-wide cache of authenticated [`ObjectStore`] instances.
+///
+/// Building an `ObjectStore` involves env-var lookups and constructing an
+/// HTTP client, both of which are needlessly repeated when many URIs under
+/// the same bucket/container are accessed. The cache is keyed by the parts
+/// of the URI that determine store identity (scheme, host, and — for
+/// virtual-hosted Azure HTTPS URLs — the container path segment).
+fn store_cache() -> &'static RwLock<HashMap<String, Arc<dyn ObjectStore>>> {
+    static CACHE: OnceLock<RwLock<HashMap<String, Arc<dyn ObjectStore>>>> =
+        OnceLock::new();
+    CACHE.get_or_init(|| RwLock::new(HashMap::new()))
+}
+
+/// Computes the cache key for a parsed URL + provider.
+///
+/// For Azure HTTPS URLs the container is encoded into the bound store, so
+/// it must be part of the key; for every other case `scheme://host` is
+/// sufficient.
+fn store_cache_key(url: &Url, provider: CloudProvider) -> String {
+    let scheme = url.scheme();
+    let host = url.host_str().unwrap_or("");
+    if provider == CloudProvider::Azure && scheme == "https" {
+        let container =
+            url.path_segments().and_then(|mut s| s.next()).unwrap_or("");
+        format!("{scheme}://{host}/{container}")
+    } else {
+        format!("{scheme}://{host}")
+    }
+}
+
+fn build_store(
+    url: &Url,
+    host: &str,
+    provider: CloudProvider,
+) -> Result<Arc<dyn ObjectStore>, CloudError> {
+    match provider {
+        CloudProvider::S3 => build_s3(url, host),
+        CloudProvider::Azure => build_azure(url, host),
+        CloudProvider::Gcs => build_gcs(url, host),
+    }
+}
+
+fn cached_or_build_store(
+    url: &Url,
+    host: &str,
+    provider: CloudProvider,
+) -> Result<Arc<dyn ObjectStore>, CloudError> {
+    let key = store_cache_key(url, provider);
+    let cache = store_cache();
+    if let Some(store) = cache
+        .read()
+        .unwrap_or_else(|poisoned| poisoned.into_inner())
+        .get(&key)
+    {
+        return Ok(store.clone());
+    }
+    let store = build_store(url, host, provider)?;
+    let mut guard = cache
+        .write()
+        .unwrap_or_else(|poisoned| poisoned.into_inner());
+    Ok(guard.entry(key).or_insert(store).clone())
+}
+
 // --- credential detection ---
 
 fn env_any(vars: &[&str]) -> bool {
@@ -169,18 +233,14 @@ impl CloudProvider {
         let url =
             Url::parse(raw).map_err(|e| CloudError::ThirdParty(Box::new(e)))?;
         let host = url.host_str().unwrap_or("");
-        let provider = CloudProvider::parse(&url);
-        let store = match provider {
-            Some(CloudProvider::S3) => build_s3(&url, host),
-            Some(CloudProvider::Azure) => build_azure(&url, host),
-            Some(CloudProvider::Gcs) => build_gcs(&url, host),
-            None => Err(CloudError::NotACloudUri(url.to_string())),
-        }?;
+        let provider = CloudProvider::parse(&url)
+            .ok_or_else(|| CloudError::NotACloudUri(url.to_string()))?;
+        let store = cached_or_build_store(&url, host, provider)?;
         // For HTTPS Azure URLs the first path segment is the container, already baked into the
         // ObjectStore. Strip it so the resulting path is relative to the container.
         let path_str = url.path().strip_prefix('/').unwrap_or("");
         let path_str = if url.scheme() == "https"
-            && matches!(provider, Some(CloudProvider::Azure))
+            && matches!(provider, CloudProvider::Azure)
         {
             path_str.split_once('/').map(|(_, rest)| rest).unwrap_or("")
         } else {

crates/timsrust-centroid/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "timsrust-centroid"
-version = "0.1.3"
+version = "0.1.4"
 edition = "2024"
 description = "Centroiding algorithms for timsTOF ion mobility spectra"
 license = "Apache-2.0"

crates/timsrust-core/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "timsrust-core"
-version = "0.1.3"
+version = "0.1.4"
 edition = "2024"
 description = "Core data types and abstractions for timsTOF data"
 license = "Apache-2.0"

crates/timsrust-mgf/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "timsrust-mgf"
-version = "0.1.3"
+version = "0.1.4"
 edition = "2024"
 description = "MGF (Mascot Generic Format) writer for timsTOF spectra"
 license = "Apache-2.0"

crates/timsrust-minitdf/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "timsrust-minitdf"
-version = "0.1.3"
+version = "0.1.4"
 edition = "2024"
 description = "Reader for the ProteoScape miniTDF timsTOF file format"
 license = "Apache-2.0"