refactor(git): expand hook sync with skill scanning and multi-root discovery

Update sync module to support .claude/skills and .codex/skills roots,
add system/source tracking to discovered skills, and refactor
migration path for the new SQL-file based migrator.
This commit is contained in:
ZhenYi 2026-05-18 20:43:16 +08:00
parent 1d48cdc973
commit 3faaff6220

View File

@ -23,12 +23,16 @@ use crate::GitDomain;
use sha1::Digest;
const SKILL_ROOTS: &[(&str, &str)] = &[(".claude/skills", "claude"), (".codex/skills", "codex")];
const ROOT_SKILL_SYSTEM: &str = "root";
fn should_descend_dir(name: &str) -> bool {
name != ".git"
}
/// Recursively scan `base` for files named `SKILL.md`.
/// The skill slug is `{short_repo_id}/{parent_dir_name}` to ensure uniqueness across repos.
/// Recursively scan supported skill locations for files named `SKILL.md`.
/// Root-level skill packs keep the legacy slug `{short_repo_id}/{skill_dir}`.
/// System skills use `{short_repo_id}/{system}/{relative_skill_dir}`.
fn scan_skills_from_dir(
base: &Path,
repo_id: &RepoId,
@ -36,8 +40,34 @@ fn scan_skills_from_dir(
) -> Result<Vec<DiscoveredSkill>, std::io::Error> {
let repo_id_prefix = &repo_id.to_string()[..8];
let mut discovered = Vec::new();
let mut stack = vec![base.to_path_buf()];
for (root, system) in SKILL_ROOTS {
let root_path = base.join(root);
if root_path.exists() {
scan_skill_root_from_dir(
&root_path,
repo_id_prefix,
system,
root,
commit_sha,
&mut discovered,
);
}
}
scan_root_skill_pack_from_dir(base, repo_id_prefix, commit_sha, &mut discovered);
Ok(discovered)
}
fn scan_skill_root_from_dir(
root_path: &Path,
repo_id_prefix: &str,
system: &str,
root: &str,
commit_sha: &str,
discovered: &mut Vec<DiscoveredSkill>,
) {
let mut stack = vec![root_path.to_path_buf()];
while let Some(dir) = stack.pop() {
let entries = match std::fs::read_dir(&dir) {
Ok(e) => e,
@ -47,30 +77,82 @@ fn scan_skills_from_dir(
let path = entry.path();
if path.is_dir() {
stack.push(path);
} else if path
.file_name()
.and_then(|n| n.to_str())
.map(|x| x.to_lowercase())
== Some("skill.md".to_string())
{
if let Some(dir_name) = path
.parent()
.and_then(|p| p.file_name())
.and_then(|n| n.to_str())
{
let slug = format!("{}/{}", repo_id_prefix, dir_name);
if let Ok(raw) = std::fs::read(&path) {
let blob_hash = git_blob_hash(&raw);
let mut skill = parse_skill_content(&slug, &raw);
skill.commit_sha = Some(commit_sha.to_string());
skill.blob_hash = Some(blob_hash);
discovered.push(skill);
}
}
continue;
}
if !is_skill_file_name(&path) {
continue;
}
let Some(parent) = path.parent() else {
continue;
};
let relative_skill_dir = parent
.strip_prefix(root_path)
.ok()
.and_then(path_to_slug)
.filter(|s| !s.is_empty());
let Some(relative_skill_dir) = relative_skill_dir else {
continue;
};
let slug = format!("{}/{}/{}", repo_id_prefix, system, relative_skill_dir);
if let Ok(raw) = std::fs::read(&path) {
let blob_hash = git_blob_hash(&raw);
let mut skill = parse_skill_content(&slug, &raw);
skill.commit_sha = Some(commit_sha.to_string());
skill.blob_hash = Some(blob_hash);
skill.metadata = enrich_metadata(
skill.metadata,
system,
Some(&format!("{}/{}/SKILL.md", root, relative_skill_dir)),
);
discovered.push(skill);
}
}
}
Ok(discovered)
}
fn scan_root_skill_pack_from_dir(
base: &Path,
repo_id_prefix: &str,
commit_sha: &str,
discovered: &mut Vec<DiscoveredSkill>,
) {
let entries = match std::fs::read_dir(base) {
Ok(e) => e,
Err(_) => return,
};
for entry in entries.flatten() {
let path = entry.path();
if !path.is_dir() {
continue;
}
let Some(dir_name) = path.file_name().and_then(|n| n.to_str()) else {
continue;
};
if dir_name == ".git" || dir_name == ".claude" || dir_name == ".codex" {
continue;
}
let skill_file = path.join("SKILL.md");
if !skill_file.exists() {
continue;
}
let relative_skill_dir = slugify_segment(dir_name);
if relative_skill_dir.is_empty() {
continue;
}
let slug = format!("{}/{}", repo_id_prefix, relative_skill_dir);
if let Ok(raw) = std::fs::read(&skill_file) {
let blob_hash = git_blob_hash(&raw);
let mut skill = parse_skill_content(&slug, &raw);
skill.commit_sha = Some(commit_sha.to_string());
skill.blob_hash = Some(blob_hash);
skill.metadata = enrich_metadata(
skill.metadata,
ROOT_SKILL_SYSTEM,
Some(&format!("{}/SKILL.md", relative_skill_dir)),
);
discovered.push(skill);
}
}
}
fn git_blob_hash(content: &[u8]) -> String {
@ -126,6 +208,57 @@ struct DiscoveredSkill {
blob_hash: Option<String>,
}
fn is_skill_file_name(path: &Path) -> bool {
path.file_name()
.and_then(|n| n.to_str())
.is_some_and(|name| name.eq_ignore_ascii_case("SKILL.md"))
}
fn path_to_slug(path: &Path) -> Option<String> {
let parts: Vec<String> = path
.components()
.filter_map(|c| c.as_os_str().to_str())
.map(slugify_segment)
.filter(|s| !s.is_empty())
.collect();
(!parts.is_empty()).then(|| parts.join("/"))
}
fn slugify_segment(input: &str) -> String {
let mut out = String::with_capacity(input.len());
let mut last_dash = false;
for ch in input.chars() {
let ch = ch.to_ascii_lowercase();
if ch.is_ascii_alphanumeric() || ch == '_' || ch == '-' {
out.push(ch);
last_dash = false;
} else if !last_dash {
out.push('-');
last_dash = true;
}
}
out.trim_matches('-').to_string()
}
fn enrich_metadata(
mut metadata: serde_json::Value,
system: &str,
relative_path: Option<&str>,
) -> serde_json::Value {
if !metadata.is_object() {
metadata = serde_json::json!({});
}
if let Some(obj) = metadata.as_object_mut() {
obj.entry("system")
.or_insert_with(|| serde_json::Value::String(system.to_string()));
if let Some(relative_path) = relative_path {
obj.entry("path")
.or_insert_with(|| serde_json::Value::String(relative_path.to_string()));
}
}
metadata
}
fn extract_frontmatter(raw: &str) -> (Option<&str>, &str) {
let trimmed = raw.trim_start();
if !trimmed.starts_with("---") {
@ -175,20 +308,25 @@ fn scan_skills_from_tree(
}
}
}
Some(git2::ObjectType::Blob) if name.to_lowercase() == "skill.md" => {
let dir_name = std::path::Path::new(&entry_path)
.parent()
.and_then(|p| p.file_name())
.and_then(|n| n.to_str());
let Some(dir_name) = dir_name else { continue };
Some(git2::ObjectType::Blob) if name.eq_ignore_ascii_case("SKILL.md") => {
let Some((system, relative_skill_dir, legacy_slug)) =
skill_location_from_path(&entry_path)
else {
continue;
};
let slug = format!("{}/{}", repo_id_prefix, dir_name);
let slug = if legacy_slug {
format!("{}/{}", repo_id_prefix, relative_skill_dir)
} else {
format!("{}/{}/{}", repo_id_prefix, system, relative_skill_dir)
};
if let Ok(blob) = entry.to_object(git_repo).and_then(|o| o.peel_to_blob()) {
let raw = blob.content();
let blob_hash = git_blob_hash(raw);
let mut skill = parse_skill_content(&slug, raw);
skill.commit_sha = Some(commit_sha.to_string());
skill.blob_hash = Some(blob_hash);
skill.metadata = enrich_metadata(skill.metadata, system, Some(&entry_path));
discovered.push(skill);
}
}
@ -200,6 +338,39 @@ fn scan_skills_from_tree(
Ok(discovered)
}
fn skill_location_from_path(path: &str) -> Option<(&'static str, String, bool)> {
let normalized = path.replace('\\', "/");
for (root, system) in SKILL_ROOTS {
let prefix = format!("{}/", root);
let suffix = "/SKILL.md";
if normalized.starts_with(&prefix) && normalized.ends_with(suffix) {
let relative = &normalized[prefix.len()..normalized.len() - suffix.len()];
let slug = relative
.split('/')
.map(slugify_segment)
.filter(|s| !s.is_empty())
.collect::<Vec<_>>()
.join("/");
if !slug.is_empty() {
return Some((*system, slug, false));
}
}
}
let suffix = "/SKILL.md";
if normalized.ends_with(suffix) && !normalized.starts_with('.') {
let relative = &normalized[..normalized.len() - suffix.len()];
if !relative.contains('/') {
let slug = slugify_segment(relative);
if !slug.is_empty() {
return Some((ROOT_SKILL_SYSTEM, slug, true));
}
}
}
None
}
#[derive(Clone)]
pub struct HookMetaDataSync {
pub db: AppDatabase,
@ -414,40 +585,27 @@ impl HookMetaDataSync {
}
};
// Deduplicate by {repo_id}+{blob_hash}, keep latest by commit_sha
// Deduplicate by stable slug. Blob hash changes when content changes and must not be the
// upsert key because project_skill has a unique (project_uuid, slug) constraint.
let mut deduped: std::collections::HashMap<String, DiscoveredSkill> =
std::collections::HashMap::new();
for skill in discovered {
let key = if let Some(ref hash) = skill.blob_hash {
format!("{}:{}", self.repo.id, hash)
} else {
format!("{}:{}:slug", self.repo.id, skill.slug)
};
match deduped.get(&key) {
match deduped.get(&skill.slug) {
Some(existing) => {
if skill.commit_sha.as_ref().unwrap_or(&String::new())
> existing.commit_sha.as_ref().unwrap_or(&String::new())
{
deduped.insert(key, skill);
deduped.insert(skill.slug.clone(), skill);
}
}
None => {
deduped.insert(key, skill);
deduped.insert(skill.slug.clone(), skill);
}
}
}
let existing_by_hash: HashMap<_, _> = existing
.into_iter()
.map(|s| {
let key = format!(
"{}:{}",
s.repo_id.unwrap_or_default(),
s.blob_hash.clone().unwrap_or_default()
);
(key, s)
})
.collect();
let existing_by_slug: HashMap<_, _> =
existing.into_iter().map(|s| (s.slug.clone(), s)).collect();
let mut seen_keys = std::collections::HashSet::new();
@ -455,12 +613,17 @@ impl HookMetaDataSync {
seen_keys.insert(key.clone());
let json_meta = serde_json::to_value(&skill.metadata).unwrap_or_default();
if let Some(existing_skill) = existing_by_hash.get(&key) {
if let Some(existing_skill) = existing_by_slug.get(&key) {
if existing_skill.content != skill.content
|| existing_skill.metadata != json_meta
|| existing_skill.commit_sha != skill.commit_sha
|| existing_skill.blob_hash != skill.blob_hash
|| existing_skill.name != skill.name
|| existing_skill.description != skill.description
{
let mut active: SkillActiveModel = existing_skill.clone().into();
active.name = Set(skill.name);
active.description = Set(skill.description);
active.content = Set(skill.content);
active.metadata = Set(json_meta);
active.commit_sha = Set(skill.commit_sha);
@ -494,7 +657,7 @@ impl HookMetaDataSync {
}
}
for (key, old_skill) in existing_by_hash {
for (key, old_skill) in existing_by_slug {
if !seen_keys.contains(&key) {
if SkillEntity::delete_by_id(old_skill.id)
.exec(&self.db)