fix: SBOM multi-ecosystem support with correct package managers and licenses (#8)

compliance-agent/src/pipeline/orchestrator.rs

@@ -263,7 +263,15 @@ impl PipelineOrchestrator {
             }
         }
 
-        // Persist SBOM entries (upsert by repo_id + name + version)
+        // Remove stale SBOM entries for this repo before reinserting
+        if !sbom_entries.is_empty() {
+            self.db
+                .sbom_entries()
+                .delete_many(doc! { "repo_id": &repo.id })
+                .await?;
+        }
+
+        // Persist SBOM entries
         for entry in &sbom_entries {
             let filter = doc! {
                 "repo_id": &entry.repo_id,

compliance-agent/src/pipeline/sbom.rs

@@ -18,12 +18,18 @@ impl Scanner for SbomScanner {
     async fn scan(&self, repo_path: &Path, repo_id: &str) -> Result<ScanOutput, CoreError> {
         let mut entries = Vec::new();
 
+        // Generate missing lock files so Syft can resolve the full dependency tree
+        generate_lockfiles(repo_path).await;
+
         // Run syft for SBOM generation
         match run_syft(repo_path, repo_id).await {
             Ok(syft_entries) => entries.extend(syft_entries),
             Err(e) => tracing::warn!("syft failed: {e}"),
         }
 
+        // Enrich Cargo entries with license info from cargo metadata
+        enrich_cargo_licenses(repo_path, &mut entries).await;
+
         // Run cargo-audit for Rust-specific vulns
         match run_cargo_audit(repo_path, repo_id).await {
             Ok(vulns) => merge_audit_vulns(&mut entries, vulns),
@@ -37,10 +43,153 @@ impl Scanner for SbomScanner {
     }
 }
 
+/// Generate missing lock files so Syft can resolve the full dependency tree.
+/// This handles repos that gitignore their lock files (common for Rust libraries).
+async fn generate_lockfiles(repo_path: &Path) {
+    // Cargo: generate Cargo.lock if Cargo.toml exists without it
+    if repo_path.join("Cargo.toml").exists() && !repo_path.join("Cargo.lock").exists() {
+        tracing::info!("generating Cargo.lock for SBOM scan");
+        let result = tokio::process::Command::new("cargo")
+            .args(["generate-lockfile"])
+            .current_dir(repo_path)
+            .output()
+            .await;
+        match result {
+            Ok(o) if o.status.success() => tracing::info!("Cargo.lock generated"),
+            Ok(o) => tracing::warn!(
+                "cargo generate-lockfile failed: {}",
+                String::from_utf8_lossy(&o.stderr)
+            ),
+            Err(e) => tracing::warn!("cargo generate-lockfile error: {e}"),
+        }
+    }
+
+    // pip: generate a requirements lock if only pyproject.toml / setup.py exists
+    let has_pip_manifest = repo_path.join("pyproject.toml").exists()
+        || repo_path.join("setup.py").exists()
+        || repo_path.join("setup.cfg").exists();
+    let has_pip_lock = repo_path.join("requirements.txt").exists()
+        || repo_path.join("requirements-lock.txt").exists()
+        || repo_path.join("poetry.lock").exists()
+        || repo_path.join("Pipfile.lock").exists();
+    if has_pip_manifest && !has_pip_lock {
+        // Try pip-compile (pip-tools) first, fall back to pip freeze approach
+        tracing::info!("attempting to generate pip requirements for SBOM scan");
+        if repo_path.join("pyproject.toml").exists() {
+            let result = tokio::process::Command::new("pip-compile")
+                .args([
+                    "--quiet",
+                    "--output-file",
+                    "requirements.txt",
+                    "pyproject.toml",
+                ])
+                .current_dir(repo_path)
+                .output()
+                .await;
+            match result {
+                Ok(o) if o.status.success() => {
+                    tracing::info!("requirements.txt generated via pip-compile")
+                }
+                _ => tracing::warn!(
+                    "pip-compile not available or failed, Syft will parse pyproject.toml directly"
+                ),
+            }
+        }
+    }
+
+    // npm: generate package-lock.json if package.json exists without it
+    let has_npm_lock = repo_path.join("package-lock.json").exists()
+        || repo_path.join("yarn.lock").exists()
+        || repo_path.join("pnpm-lock.yaml").exists();
+    if repo_path.join("package.json").exists() && !has_npm_lock {
+        tracing::info!("generating package-lock.json for SBOM scan");
+        let result = tokio::process::Command::new("npm")
+            .args(["install", "--package-lock-only", "--ignore-scripts"])
+            .current_dir(repo_path)
+            .output()
+            .await;
+        match result {
+            Ok(o) if o.status.success() => tracing::info!("package-lock.json generated"),
+            Ok(o) => tracing::warn!(
+                "npm install --package-lock-only failed: {}",
+                String::from_utf8_lossy(&o.stderr)
+            ),
+            Err(e) => tracing::warn!("npm lock generation error: {e}"),
+        }
+    }
+}
+
+/// Enrich Cargo SBOM entries with license info from `cargo metadata`.
+/// Syft doesn't read license data from Cargo.lock, so we fill it in.
+async fn enrich_cargo_licenses(repo_path: &Path, entries: &mut [SbomEntry]) {
+    if !repo_path.join("Cargo.toml").exists() {
+        return;
+    }
+
+    let has_cargo_entries = entries.iter().any(|e| e.package_manager == "cargo");
+    if !has_cargo_entries {
+        return;
+    }
+
+    let output = match tokio::process::Command::new("cargo")
+        .args(["metadata", "--format-version", "1"])
+        .current_dir(repo_path)
+        .output()
+        .await
+    {
+        Ok(o) if o.status.success() => o,
+        Ok(o) => {
+            tracing::warn!(
+                "cargo metadata failed: {}",
+                String::from_utf8_lossy(&o.stderr)
+            );
+            return;
+        }
+        Err(e) => {
+            tracing::warn!("cargo metadata error: {e}");
+            return;
+        }
+    };
+
+    let meta: CargoMetadata = match serde_json::from_slice(&output.stdout) {
+        Ok(m) => m,
+        Err(e) => {
+            tracing::warn!("failed to parse cargo metadata: {e}");
+            return;
+        }
+    };
+
+    // Build a lookup: (name, version) -> license
+    let license_map: std::collections::HashMap<(&str, &str), &str> = meta
+        .packages
+        .iter()
+        .filter_map(|p| {
+            p.license
+                .as_deref()
+                .map(|l| (p.name.as_str(), p.version.as_str(), l))
+        })
+        .map(|(n, v, l)| ((n, v), l))
+        .collect();
+
+    for entry in entries.iter_mut() {
+        if entry.package_manager != "cargo" || entry.license.is_some() {
+            continue;
+        }
+        if let Some(license) = license_map.get(&(entry.name.as_str(), entry.version.as_str())) {
+            entry.license = Some(license.to_string());
+        }
+    }
+}
+
 async fn run_syft(repo_path: &Path, repo_id: &str) -> Result<Vec<SbomEntry>, CoreError> {
     let output = tokio::process::Command::new("syft")
         .arg(repo_path)
         .args(["-o", "cyclonedx-json"])
+        // Enable remote license lookups for all ecosystems
+        .env("SYFT_GOLANG_SEARCH_REMOTE_LICENSES", "true")
+        .env("SYFT_JAVASCRIPT_SEARCH_REMOTE_LICENSES", "true")
+        .env("SYFT_PYTHON_SEARCH_REMOTE_LICENSES", "true")
+        .env("SYFT_JAVA_USE_NETWORK", "true")
         .output()
         .await
         .map_err(|e| CoreError::Scanner {
@@ -62,22 +211,19 @@ async fn run_syft(repo_path: &Path, repo_id: &str) -> Result<Vec<SbomEntry>, Cor
         .unwrap_or_default()
         .into_iter()
         .map(|c| {
+            let package_manager = c
+                .purl
+                .as_deref()
+                .and_then(extract_ecosystem_from_purl)
+                .unwrap_or_else(|| "unknown".to_string());
             let mut entry = SbomEntry::new(
                 repo_id.to_string(),
                 c.name,
                 c.version.unwrap_or_else(|| "unknown".to_string()),
-                c.component_type.unwrap_or_else(|| "library".to_string()),
+                package_manager,
             );
             entry.purl = c.purl;
-            entry.license = c.licenses.and_then(|ls| {
-                ls.first().and_then(|l| {
-                    l.license.as_ref().map(|lic| {
-                        lic.id
-                            .clone()
-                            .unwrap_or_else(|| lic.name.clone().unwrap_or_default())
-                    })
-                })
-            });
+            entry.license = c.licenses.and_then(|ls| extract_license(&ls));
             entry
         })
         .collect();
@@ -144,6 +290,7 @@ struct CdxComponent {
     name: String,
     version: Option<String>,
     #[serde(rename = "type")]
+    #[allow(dead_code)]
     component_type: Option<String>,
     purl: Option<String>,
     licenses: Option<Vec<CdxLicenseWrapper>>,
@@ -152,6 +299,8 @@ struct CdxComponent {
 #[derive(serde::Deserialize)]
 struct CdxLicenseWrapper {
     license: Option<CdxLicense>,
+    /// SPDX license expression (e.g. "MIT OR Apache-2.0")
+    expression: Option<String>,
 }
 
 #[derive(serde::Deserialize)]
@@ -188,3 +337,62 @@ struct AuditVuln {
     id: String,
     url: String,
 }
+
+// Cargo metadata types
+#[derive(serde::Deserialize)]
+struct CargoMetadata {
+    packages: Vec<CargoPackage>,
+}
+
+#[derive(serde::Deserialize)]
+struct CargoPackage {
+    name: String,
+    version: String,
+    license: Option<String>,
+}
+
+/// Extract the best license string from CycloneDX license entries.
+/// Handles three formats: expression ("MIT OR Apache-2.0"), license.id ("MIT"), license.name ("MIT License").
+fn extract_license(entries: &[CdxLicenseWrapper]) -> Option<String> {
+    // First pass: look for SPDX expressions (most precise for dual-licensed packages)
+    for entry in entries {
+        if let Some(ref expr) = entry.expression {
+            if !expr.is_empty() {
+                return Some(expr.clone());
+            }
+        }
+    }
+    // Second pass: collect license.id or license.name from all entries
+    let parts: Vec<String> = entries
+        .iter()
+        .filter_map(|e| {
+            e.license.as_ref().and_then(|lic| {
+                lic.id
+                    .clone()
+                    .or_else(|| lic.name.clone())
+                    .filter(|s| !s.is_empty())
+            })
+        })
+        .collect();
+    if parts.is_empty() {
+        return None;
+    }
+    Some(parts.join(" OR "))
+}
+
+/// Extract the ecosystem/package-manager from a PURL string.
+/// e.g. "pkg:npm/lodash@4.17.21" → "npm", "pkg:cargo/serde@1.0" → "cargo"
+fn extract_ecosystem_from_purl(purl: &str) -> Option<String> {
+    let rest = purl.strip_prefix("pkg:")?;
+    let ecosystem = rest.split('/').next()?;
+    if ecosystem.is_empty() {
+        return None;
+    }
+    // Normalise common PURL types to user-friendly names
+    let normalised = match ecosystem {
+        "golang" => "go",
+        "pypi" => "pip",
+        _ => ecosystem,
+    };
+    Some(normalised.to_string())
+}