Method vault_scan_working_copy

atomic-repository/src/repository/vault.rs:554–643

Scan the vault working copy for changed files. Walks `.vault/` and compares each markdown file against its stored entry in redb. Returns a list of paths that are new or modified. Files are identified as changed if: - They don't exist in redb (new file) - Their body hash or user frontmatter differs from the stored entry Ignores non-`.md` files and `_manifest.json`.

(&self)

    ///
    /// Ignores non-`.md` files and `_manifest.json`.
    pub fn vault_scan_working_copy(&self) -> Result<Vec<VaultFileChange>, RepositoryError> {
        let vault_dir = self.vault_dir();
        if !vault_dir.exists() {
            return Ok(Vec::new());
        }

        let mut changes = Vec::new();

        // Walk the vault directory
        for entry in walkdir::WalkDir::new(&vault_dir)
            .into_iter()
            .filter_map(|e| e.ok())
        {
            let path = entry.path();

            // Skip directories
            if path.is_dir() {
                continue;
            }

            // Skip non-markdown files
            let ext = path.extension().and_then(|e| e.to_str()).unwrap_or("");
            if ext != "md" {
                continue;
            }

            // Skip manifest
            if path.file_name().and_then(|n| n.to_str()) == Some("_manifest.json") {
                continue;
            }

            // Get vault-relative path
            let rel_path =
                path.strip_prefix(&vault_dir)
                    .map_err(|_| RepositoryError::InvalidOperation {
                        message: format!("path not under vault: {}", path.display()),
                    })?;
            let rel_path_str = rel_path.to_string_lossy().to_string();
            // Normalize to forward slashes (for Windows compat)
            let rel_path_str = rel_path_str.replace('\\', "/");

            // Read file content
            let file_content = std::fs::read_to_string(path)?;

            // Parse into frontmatter + body
            let (frontmatter_json, body) = parse_markdown_frontmatter(&file_content);

            // Body hashes remain useful for content/embedding caches, but
            // frontmatter is also knowledge: status, labels, identity, and
            // descriptions affect retrieval and KG extraction.
            let body_bytes = body.as_bytes();
            let new_hash = Hash::of(body_bytes);

            // Check if this is new or changed
            let existing = self.vault_retrieve(&rel_path_str)?;
            let change_type = match &existing {
                Some(entry)
                    if Hash::from_bytes(entry.content_hash) == new_hash