Start caching path dependencies, but compute the hash of their conten…

…ts to avoid serving stale data

libs/Cargo.toml

@@ -57,6 +57,7 @@ libtest-mimic = "0.7.0"
 liquid = "0.26.4"
 liquid-core = "0.26.4"
 matchit = "0.7"
+merkle_hash = "3.7.0"
 miette = "7.2.0"
 mime = "0.3"
 num_cpus = "1.15.0"

libs/pavexc/Cargo.toml

@@ -52,6 +52,7 @@ semver = { workspace = true }
 persist_if_changed = { path = "../persist_if_changed", version = "0.1.48" }
 matchit = { version = "0.7", package = "pavex_matchit" }
 relative-path = { workspace = true }
+merkle_hash = { workspace = true }
 
 # Sqlite cache
 xdg-home = { workspace = true }

libs/pavexc/src/rustdoc/compute/cache.rs

@@ -7,6 +7,7 @@ use guppy::{
     PackageId,
 };
 use itertools::Itertools;
+use merkle_hash::Encodable;
 use r2d2_sqlite::SqliteConnectionManager;
 use rusqlite::params;
 use tracing::instrument;
@@ -376,6 +377,7 @@ impl ThirdPartyCrateCache {
                 WHERE crate_name = ? AND 
                     crate_source = ? AND 
                     crate_version = ? AND 
+                    crate_hash = ? AND 
                     cargo_fingerprint = ? AND 
                     rustdoc_options = ? AND 
                     default_feature_is_enabled = ? AND 
@@ -387,6 +389,7 @@ impl ThirdPartyCrateCache {
             cache_key.crate_name,
             cache_key.crate_source,
             cache_key.crate_version,
+            cache_key.crate_hash,
             cache_key.cargo_fingerprint,
             cache_key.rustdoc_options,
             cache_key.default_feature_is_enabled,
@@ -450,6 +453,7 @@ impl ThirdPartyCrateCache {
                 crate_name,
                 crate_source,
                 crate_version,
+                crate_hash,
                 cargo_fingerprint,
                 rustdoc_options,
                 default_feature_is_enabled,
@@ -464,12 +468,13 @@ impl ThirdPartyCrateCache {
                 id2private_import_paths,
                 import_path2id,
                 re_exports
-            ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
+            ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
         )?;
         stmt.execute(params![
             cache_key.crate_name,
             cache_key.crate_source,
             cache_key.crate_version,
+            cache_key.crate_hash,
             cache_key.cargo_fingerprint,
             cache_key.rustdoc_options,
             cache_key.default_feature_is_enabled,
@@ -494,6 +499,7 @@ impl ThirdPartyCrateCache {
                 crate_name TEXT NOT NULL,
                 crate_source TEXT NOT NULL,
                 crate_version TEXT NOT NULL,
+                crate_hash TEXT,
                 cargo_fingerprint TEXT NOT NULL,
                 rustdoc_options TEXT NOT NULL,
                 default_feature_is_enabled INTEGER NOT NULL,
@@ -508,7 +514,7 @@ impl ThirdPartyCrateCache {
                 id2private_import_paths BLOB NOT NULL,
                 import_path2id BLOB NOT NULL,
                 re_exports BLOB NOT NULL,
-                PRIMARY KEY (crate_name, crate_source, crate_version, cargo_fingerprint, rustdoc_options, default_feature_is_enabled, active_named_features)
+                PRIMARY KEY (crate_name, crate_source, crate_version, crate_hash, cargo_fingerprint, rustdoc_options, default_feature_is_enabled, active_named_features)
             )",
             []
         )?;
@@ -640,6 +646,9 @@ pub(super) struct ThirdPartyCrateCacheKey<'a> {
     pub crate_name: &'a str,
     pub crate_source: &'a str,
     pub crate_version: String,
+    /// The hash of the crate's source code, computed via BLAKE3.
+    /// It is only populated for path dependencies.
+    pub crate_hash: Option<String>,
     pub cargo_fingerprint: &'a str,
     pub rustdoc_options: String,
     pub default_feature_is_enabled: bool,
@@ -652,10 +661,39 @@ impl<'a> ThirdPartyCrateCacheKey<'a> {
         package_metadata: &'a PackageMetadata<'a>,
         cargo_fingerprint: &'a str,
     ) -> Option<ThirdPartyCrateCacheKey<'a>> {
-        // We don't want to cache the docs for workspace crates and path dependencies.
-        let Some(source) = package_metadata.source().external_source() else {
-            return None;
+        let source = match package_metadata.source() {
+            guppy::graph::PackageSource::Workspace(_) => {
+                // We don't want to cache the docs for workspace crates.
+                return None;
+            }
+            guppy::graph::PackageSource::Path(p) => p.as_str(),
+            guppy::graph::PackageSource::External(e) => e,
         };
+        let crate_hash =
+            if let guppy::graph::PackageSource::Path(package_path) = package_metadata.source() {
+                // We need to compute the hash of the package's contents,
+                // to invalidate the cache when the package changes.
+                // This is only relevant for path dependencies.
+                // We don't need to do this for external dependencies,
+                // since they are assumed to be immutable.
+                // We visit all files in the package directory, traversing
+                // subdirectories recursively, and hash the contents of each file.
+                let Ok(tree) = merkle_hash::MerkleTree::builder(package_path)
+                    .algorithm(merkle_hash::Algorithm::Blake3)
+                    .hash_names(true)
+                    .build()
+                else {
+                    tracing::warn!(
+                        "Failed to compute the hash of the package at {:?}.
+                        I won't cache its JSON documentation to avoid serving stale data.",
+                        package_metadata.id()
+                    );
+                    return None;
+                };
+                Some(tree.root.item.hash.to_hex_string())
+            } else {
+                None
+            };
         let features = package_metadata
             .to_feature_set(StandardFeatures::Default)
             .features_for(package_metadata.id())
@@ -669,6 +707,7 @@ impl<'a> ThirdPartyCrateCacheKey<'a> {
             crate_name: package_metadata.name(),
             crate_source: source,
             crate_version: package_metadata.version().to_string(),
+            crate_hash,
             cargo_fingerprint,
             default_feature_is_enabled,
             // SQLite doesn't support arrays, so we have to serialize these two collections as strings.