From 91d9f431c5e0e49ee543c172a0b0ee4b48bb6c35 Mon Sep 17 00:00:00 2001
From: Mick Grove <mick.grove@mongodb.com>
Date: Mon, 18 May 2026 14:27:01 -0700
Subject: [PATCH] preparing for v1.100.0

---
 src/github.rs             | 26 ++++++++++++++++----------
 src/scanner/enumerate.rs  |  2 +-
 src/scanner/validation.rs | 19 +++++++++----------
 3 files changed, 26 insertions(+), 21 deletions(-)

diff --git a/src/github.rs b/src/github.rs
index added70..cd9f921 100644
--- a/src/github.rs
+++ b/src/github.rs
@@ -171,6 +171,10 @@ async fn ensure_github_success(resp: reqwest::Response, action: &str) -> Result<
     anyhow::bail!("GitHub API request failed while {action}: HTTP {status} ({url}): {body}");
 }
 
+fn is_github_soft_limit_status(status: StatusCode) -> bool {
+    matches!(status, StatusCode::FORBIDDEN | StatusCode::TOO_MANY_REQUESTS)
+}
+
 fn github_next_link(headers: &HeaderMap) -> Option<Url> {
     let raw = headers.get(reqwest::header::LINK)?.to_str().ok()?;
     raw.split(',').find_map(|part| {
@@ -267,11 +271,12 @@ pub async fn enumerate_contributor_repo_urls(
             .join(&format!("repos/{owner}/{repo}/contributors"))
             .context("Failed to build GitHub contributors URL")?;
         url.query_pairs_mut().append_pair("per_page", "100").append_pair("page", &page.to_string());
-        let resp = ensure_github_success(
-            github_get(&client, url, token.as_deref()).send().await?,
-            "listing contributors",
-        )
-        .await?;
+        let resp = github_get(&client, url, token.as_deref()).send().await?;
+        if is_github_soft_limit_status(resp.status()) {
+            warn_on_rate_limit("GitHub", resp.status(), "listing contributors");
+            break;
+        }
+        let resp = ensure_github_success(resp, "listing contributors").await?;
         let contributors: Vec<GitHubContributor> = resp.json().await?;
         if contributors.is_empty() {
             break;
@@ -324,11 +329,12 @@ pub async fn enumerate_contributor_repo_urls(
                 .append_pair("type", "all")
                 .append_pair("sort", "updated")
                 .append_pair("direction", "desc");
-            let resp = ensure_github_success(
-                github_get(&client, url, token.as_deref()).send().await?,
-                "listing user repositories",
-            )
-            .await?;
+            let resp = github_get(&client, url, token.as_deref()).send().await?;
+            if is_github_soft_limit_status(resp.status()) {
+                warn_on_rate_limit("GitHub", resp.status(), "listing user repositories");
+                break;
+            }
+            let resp = ensure_github_success(resp, "listing user repositories").await?;
             let repos: Vec<GitHubRepo> = resp.json().await?;
             if repos.is_empty() {
                 break;
diff --git a/src/scanner/enumerate.rs b/src/scanner/enumerate.rs
index d063ea0..4702711 100644
--- a/src/scanner/enumerate.rs
+++ b/src/scanner/enumerate.rs
@@ -715,7 +715,7 @@ fn try_extract_git_blob_archive(
         },
     };
 
-    Ok(Some(entries))
+    if entries.is_empty() { Ok(None) } else { Ok(Some(entries)) }
 }
 
 // A marker so the struct itself carries the lifetime.
diff --git a/src/scanner/validation.rs b/src/scanner/validation.rs
index c52ad8c..8e4149f 100644
--- a/src/scanner/validation.rs
+++ b/src/scanner/validation.rs
@@ -962,16 +962,6 @@ fn build_cache_key(
     dep_vars: &FxHashMap<String, Vec<(String, OffsetSpan)>>,
 ) -> String {
     // Build key
-    let dep_vars_str = dep_vars
-        .get(om.rule.id())
-        .map(|hm| {
-            let mut sorted: Vec<_> = hm.iter().collect();
-            sorted.sort_by(|(k, _), (k2, _)| k.cmp(k2));
-            sorted.into_iter().map(|(k, v)| format!("{}={}", k, v)).collect::<Vec<_>>().join("|")
-        })
-        .unwrap_or_default();
-    // For demonstration, we’ll do a simplistic approach
-    // You can adapt from your existing logic
     let capture0 = om.captures.captures.get(0).map_or(String::new(), |c| c.raw_value().to_string());
 
     if !om.rule.syntax().depends_on_rule.is_empty() {
@@ -985,6 +975,15 @@ fn build_cache_key(
         );
     }
 
+    let dep_vars_str = dep_vars
+        .get(om.rule.id())
+        .map(|hm| {
+            let mut sorted: Vec<_> = hm.iter().collect();
+            sorted.sort_by(|(k, _), (k2, _)| k.cmp(k2));
+            sorted.into_iter().map(|(k, v)| format!("{}={}", k, v)).collect::<Vec<_>>().join("|")
+        })
+        .unwrap_or_default();
+
     format!("{}|{}|{}", om.rule.name(), capture0, dep_vars_str)
 }