From a3ec9bcddbdbadf70588697f3ee9f585a3263413 Mon Sep 17 00:00:00 2001
From: Alex Butler <alexheretic@gmail.com>
Date: Tue, 27 Feb 2024 14:40:31 +0000
Subject: [PATCH] Fix cache_releases_older_than

---
 CHANGELOG.md            |  2 +-
 src/config.rs           |  6 ++++--
 src/main.rs             |  2 +-
 src/providers/gitlab.rs | 46 ++++++++++++++++++++++++++++++----------------
 4 files changed, 32 insertions(+), 24 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 8860798..0870345 100644
--- a/CHANGELOG.md
+++ a/CHANGELOG.md
@@ -6,7 +6,7 @@
 - Add info logs for release & metadata fetch latency.
 - When fetching all releases handle 429 by backing off.
 - Improve fetch error logging.
-- Added crate eligibility cache.
+- Added crate eligibility cache. May be controlled with config `cache-releases-older-than`.
 - Introduce configurable cache backend with a RocksDB implementation (set `cache.type = "rocksdb"` and `cache.path = "cache"` to use it), defaults to `cache.type = "in-memory"`.
 - Support crate yanking by creating a `yanked` file on the release.
 - Add `bust-cache` command, invoked via `ssh [registry] -- bust-cache [project] [crate-name] [crate-version]` to remove eligibility cache (ie. after a crate has been yanked)
diff --git a/src/config.rs b/src/config.rs
index 3a9db39..6660b79 100644
--- a/src/config.rs
+++ a/src/config.rs
@@ -55,8 +55,10 @@
     #[serde(default)]
     pub metadata_format: MetadataFormat,
     /// Cache file checksum fetches for all release older than this value.
+    ///
+    /// Default zero (cache all releases).
     #[serde(default, with = "humantime_serde")]
-    pub cache_releases_older_than: Option<Duration>,
+    pub cache_releases_older_than: Duration,
 }
 
 impl GitlabConfig {
@@ -135,6 +137,6 @@
     assert_eq!(gitlab.metadata_format, MetadataFormat::JsonZst);
     assert_eq!(
         gitlab.cache_releases_older_than,
-        Some(Duration::from_secs(2 * 24 * 60 * 60))
+        Duration::from_secs(2 * 24 * 60 * 60)
     );
 }
diff --git a/src/main.rs b/src/main.rs
index d380ec0..2897836 100644
--- a/src/main.rs
+++ a/src/main.rs
@@ -251,7 +251,7 @@
             return Ok(cache);
         }
 
-        info!("Fetching metadata from GitLab");
+        debug!("Fetching metadata from GitLab");
 
         // fetch metadata from the provider
         let metadata = gitlab
diff --git a/src/providers/gitlab.rs b/src/providers/gitlab.rs
index f5ccc7b..c164636 100644
--- a/src/providers/gitlab.rs
+++ a/src/providers/gitlab.rs
@@ -1,10 +1,9 @@
 // blocks_in_conditions: didn't work with `#[instrument...`` usage
 #![allow(clippy::module_name_repetitions, clippy::blocks_in_conditions)]
-use crate::cache::{Cache, ConcreteCache, Yoked};
-use crate::providers::EligibilityCacheKey;
 use crate::{
+    cache::{Cache, ConcreteCache, Yoked},
     config::{GitlabConfig, MetadataFormat},
-    providers::{Release, User},
+    providers::{EligibilityCacheKey, Release, User},
 };
 use anyhow::Context;
 use async_trait::async_trait;
@@ -14,11 +13,10 @@
 use reqwest::{header, Certificate};
 use serde::{Deserialize, Serialize};
 use smol_str::SmolStr;
-use std::borrow::Cow;
-use std::sync::Arc;
+use std::{borrow::Cow, sync::Arc, time::Duration};
 use time::OffsetDateTime;
 use tokio::sync::Semaphore;
-use tracing::{debug, info, info_span, instrument, Instrument};
+use tracing::{debug, info_span, instrument, Instrument};
 use url::Url;
 use yoke::Yoke;
 
@@ -32,6 +30,7 @@
     metadata_format: MetadataFormat,
     admin_token: Option<String>,
     cache: ConcreteCache,
+    cache_checksums_older_than: Duration,
 }
 
 impl Gitlab {
@@ -51,6 +50,7 @@
             metadata_format: config.metadata_format,
             admin_token: config.admin_token.clone(),
             cache,
+            cache_checksums_older_than: config.cache_releases_older_than,
         })
     }
 
@@ -84,7 +84,7 @@
             return Ok(cached);
         }
 
-        info!("Fetching eligibility for release");
+        debug!("Fetching eligibility for release");
 
         let project = utf8_percent_encode(raw_project, NON_ALPHANUMERIC);
         let package_id = utf8_percent_encode(package_id, NON_ALPHANUMERIC);
@@ -119,21 +119,27 @@
         let expected_file_name = format!("{}-{}.crate", release.name, release.version);
 
         // grab the sha256 checksum of the .crate file itself
-        let release = package_files
+        let Some(package_file) = package_files
             .into_iter()
             .find(|package_file| package_file.file_name == expected_file_name)
-            .map(|package_file| Release {
-                name: Cow::Owned(release.name.to_string()),
-                version: Cow::Owned(release.version.clone()),
-                checksum: Cow::Owned(package_file.file_sha256),
-                project: Cow::Owned(raw_project.to_string()),
-                yanked,
-            });
-
-        self.cache
-            .put(cache_key, &release)
-            .await
-            .context("failed to write to cache")?;
+        else {
+            return Ok(Yoke::attach_to_cart(Vec::new(), |_| None));
+        };
+
+        let release = Some(Release {
+            name: Cow::Owned(release.name.to_string()),
+            version: Cow::Owned(release.version.clone()),
+            checksum: Cow::Owned(package_file.file_sha256),
+            project: Cow::Owned(raw_project.to_string()),
+            yanked,
+        });
+
+        if package_file.created_at + self.cache_checksums_older_than < OffsetDateTime::now_utc() {
+            self.cache
+                .put(cache_key, &release)
+                .await
+                .context("failed to write to cache")?;
+        }
 
         Ok(Yoke::attach_to_cart(Vec::new(), |_| release))
     }
--
rgit 0.1.3