fix(core/enrichment): credit changeset original creator (#174)

Author: goulvenclech

.sampo/changesets/dashing-baron-marjatta.md

@@ -0,0 +1,7 @@
+---
+cargo/sampo: patch
+cargo/sampo-core: patch
+cargo/sampo-github-action: patch
+---
+
+Changelog entries now correctly credit the original changeset author, even if the file was later edited by someone else.

crates/sampo-core/src/adapters/hex.rs

@@ -1,7 +1,7 @@
 use crate::errors::{Result, SampoError, WorkspaceError};
 use crate::types::PackageInfo;
-use reqwest::blocking::Client;
 use reqwest::StatusCode;
+use reqwest::blocking::Client;
 use std::collections::BTreeMap;
 use std::path::{Path, PathBuf};
 use std::sync::{Mutex, OnceLock};

crates/sampo-core/src/adapters/npm.rs

@@ -2,8 +2,8 @@ use crate::errors::{Result, SampoError, WorkspaceError};
 use crate::types::{PackageInfo, PackageKind};
 use reqwest::StatusCode;
 use serde::Deserialize;
-use serde_json::value::RawValue;
 use serde_json::Value as JsonValue;
+use serde_json::value::RawValue;
 use std::collections::{BTreeMap, BTreeSet, HashMap};
 use std::fs;
 use std::path::{Component, Path, PathBuf};
@@ -63,11 +63,7 @@ impl NpmAdapter {
     pub(super) fn is_publishable(&self, manifest_path: &Path) -> Result<bool> {
         let manifest = load_package_json(manifest_path)?;
         let info = parse_manifest_info(manifest_path, &manifest)?;
-        if info.private {
-            Ok(false)
-        } else {
-            Ok(true)
-        }
+        if info.private { Ok(false) } else { Ok(true) }
     }
 
     pub(super) fn version_exists(

crates/sampo-core/src/adapters/pypi.rs

@@ -1,7 +1,7 @@
 use crate::errors::{Result, SampoError, WorkspaceError};
 use crate::types::PackageInfo;
-use reqwest::blocking::Client;
 use reqwest::StatusCode;
+use reqwest::blocking::Client;
 use serde_json::Value as JsonValue;
 use std::collections::BTreeMap;
 use std::path::{Path, PathBuf};

crates/sampo-core/src/adapters/pypi/pip/pip_tests.rs

@@ -76,7 +76,10 @@ dependencies = []
     let packages = discover(root).unwrap();
     assert_eq!(packages.len(), 3);
 
-    let root_pkg = packages.iter().find(|p| p.name == "workspace-root").unwrap();
+    let root_pkg = packages
+        .iter()
+        .find(|p| p.name == "workspace-root")
+        .unwrap();
     assert!(root_pkg.internal_deps.contains("pypi/pkg-a"));
 
     let pkg_a = packages.iter().find(|p| p.name == "pkg-a").unwrap();

crates/sampo-core/src/enrichment.rs

@@ -48,6 +48,7 @@ pub fn get_commit_hash_for_path(repo_root: &Path, file_path: &Path) -> Option<St
         .current_dir(repo_root)
         .args([
             "log",
+            "--diff-filter=A",
             "-1",
             "--format=%H",
             "--",
@@ -546,6 +547,206 @@ mod tests {
         );
     }
 
+    #[test]
+    fn get_commit_hash_returns_creation_commit_not_modification() {
+        use std::fs;
+        use tempfile::TempDir;
+
+        let temp_dir = TempDir::new().unwrap();
+        let repo_path = temp_dir.path();
+
+        // Initialize a git repo
+        std::process::Command::new("git")
+            .arg("init")
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        // Configure git user for initial commit
+        std::process::Command::new("git")
+            .args(["config", "user.name", "Original Author"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        std::process::Command::new("git")
+            .args(["config", "user.email", "[email protected]"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        // Create and commit file
+        let test_file = repo_path.join("changeset.md");
+        fs::write(&test_file, "initial content").unwrap();
+
+        std::process::Command::new("git")
+            .args(["add", "changeset.md"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        std::process::Command::new("git")
+            .args(["commit", "-m", "create changeset"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        // Record the creation commit hash
+        let creation_commit =
+            get_commit_hash_for_path(repo_path, &test_file).expect("Should find creation commit");
+
+        // Now modify the file with a different author
+        std::process::Command::new("git")
+            .args(["config", "user.name", "Editor"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        std::process::Command::new("git")
+            .args(["config", "user.email", "[email protected]"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        fs::write(&test_file, "modified content").unwrap();
+
+        std::process::Command::new("git")
+            .args(["add", "changeset.md"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        std::process::Command::new("git")
+            .args(["commit", "-m", "edit changeset"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        // After modification, get_commit_hash_for_path should still return the creation commit
+        let hash_after_modification = get_commit_hash_for_path(repo_path, &test_file)
+            .expect("Should find commit after modification");
+
+        assert_eq!(
+            hash_after_modification, creation_commit,
+            "Should return creation commit, not modification commit"
+        );
+
+        // Verify we can get the correct author from the creation commit
+        let commit_info =
+            get_commit_info_for_hash(repo_path, &creation_commit).expect("Should get commit info");
+        assert_eq!(
+            commit_info.author_name, "Original Author",
+            "Should credit original author, not editor"
+        );
+    }
+
+    #[test]
+    fn get_commit_hash_returns_latest_creation_after_delete_recreate() {
+        use std::fs;
+        use tempfile::TempDir;
+
+        let temp_dir = TempDir::new().unwrap();
+        let repo_path = temp_dir.path();
+
+        // Initialize a git repo
+        std::process::Command::new("git")
+            .arg("init")
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        // Configure git user for first creation
+        std::process::Command::new("git")
+            .args(["config", "user.name", "First Author"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        std::process::Command::new("git")
+            .args(["config", "user.email", "[email protected]"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        // Create and commit file
+        let test_file = repo_path.join("changeset.md");
+        fs::write(&test_file, "first version").unwrap();
+
+        std::process::Command::new("git")
+            .args(["add", "changeset.md"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        std::process::Command::new("git")
+            .args(["commit", "-m", "first creation"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        let first_creation_commit = get_commit_hash_for_path(repo_path, &test_file)
+            .expect("Should find first creation commit");
+
+        // Delete the file
+        fs::remove_file(&test_file).unwrap();
+
+        std::process::Command::new("git")
+            .args(["add", "changeset.md"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        std::process::Command::new("git")
+            .args(["commit", "-m", "delete changeset"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        // Re-create the file with a different author
+        std::process::Command::new("git")
+            .args(["config", "user.name", "Second Author"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        std::process::Command::new("git")
+            .args(["config", "user.email", "[email protected]"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        fs::write(&test_file, "second version").unwrap();
+
+        std::process::Command::new("git")
+            .args(["add", "changeset.md"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        std::process::Command::new("git")
+            .args(["commit", "-m", "recreate changeset"])
+            .current_dir(repo_path)
+            .output()
+            .unwrap();
+
+        // Should return the latest creation commit (second author's)
+        let latest_creation_commit = get_commit_hash_for_path(repo_path, &test_file)
+            .expect("Should find latest creation commit");
+
+        assert_ne!(
+            latest_creation_commit, first_creation_commit,
+            "Should return different commit after delete/recreate"
+        );
+
+        // Verify we credit the second author (current file creator)
+        let commit_info = get_commit_info_for_hash(repo_path, &latest_creation_commit)
+            .expect("Should get commit info");
+        assert_eq!(
+            commit_info.author_name, "Second Author",
+            "Should credit the author who re-created the file"
+        );
+    }
+
     #[tokio::test]
     async fn test_github_api_get_with_invalid_token() {
         // Test with invalid token should return None (graceful failure)

crates/sampo-core/src/lib.rs

@@ -18,20 +18,20 @@ pub const USER_AGENT: &str = concat!("sampo-core/", env!("CARGO_PKG_VERSION"));
 // Re-export commonly used items
 pub use adapters::ManifestMetadata;
 pub use changeset::{
-    load_changesets, parse_changeset, render_changeset_markdown,
-    render_changeset_markdown_with_tags, ChangesetInfo,
+    ChangesetInfo, load_changesets, parse_changeset, render_changeset_markdown,
+    render_changeset_markdown_with_tags,
 };
 pub use config::Config;
 pub use enrichment::{
-    detect_github_repo_slug, detect_github_repo_slug_with_config, enrich_changeset_message,
-    get_commit_hash_for_path, CommitInfo, GitHubUserInfo,
+    CommitInfo, GitHubUserInfo, detect_github_repo_slug, detect_github_repo_slug_with_config,
+    enrich_changeset_message, get_commit_hash_for_path,
 };
 pub use errors::{Result, SampoError, WorkspaceError};
 pub use filters::{filter_members, list_visible_packages, should_ignore_package, wildcard_match};
 pub use git::current_branch;
 pub use markdown::format_markdown_list_item;
 pub use prerelease::{
-    enter_prerelease, exit_prerelease, restore_preserved_changesets, VersionChange,
+    VersionChange, enter_prerelease, exit_prerelease, restore_preserved_changesets,
 };
 pub use publish::{run_publish, tag_published_crate, topo_order};
 pub use release::{

crates/sampo/README.md

@@ -80,7 +80,7 @@ At the root of each published package, a human-readable file listing all changes
 ... previous entries ...
 ```
 
-Sampo generates changelog entries from consumed changesets and enriches them with commit hash links and author acknowledgments (can be disabled in config). 
+Sampo generates changelog entries from consumed changesets and enriches them with commit hash links and author acknowledgments (can be disabled in config).
 
 Any intro content or custom main header before the first `##` section is preserved. You can also manually edit the previously released entries, and Sampo will keep them intact.
 
@@ -164,6 +164,9 @@ linked = [["cargo/pkg-e", "cargo/pkg-f"], ["cargo/pkg-g", "cargo/pkg-h"]]
 
 ### `[changelog]` section
 
+> [!WARNING]
+> Commit hash links and author acknowledgments won't work in shallow clones or CI environments where the full git history is not available.
+
 `show_commit_hash`: Whether to include commit hash links in changelog entries (default: `true`). When enabled, changelog entries include clickable commit hash links that point to the commit on GitHub.
 
 `show_acknowledgments`: Whether to include author acknowledgments in changelog entries (default: `true`). When enabled, changelog entries include author acknowledgments with special messages for first-time contributors.