pkg-proxy/internal/handler/npm.go

package handler

import (
	"encoding/json"
	"errors"
	"fmt"
	"net/http"
	"net/url"
	"sort"
	"strings"
	"time"

	"github.com/git-pkgs/purl"
)

const (
	npmUpstream      = "https://registry.npmjs.org"
	npmAbbreviatedCT = "application/vnd.npm.install-v1+json"
	scopedParts      = 2 // scope + name in scoped packages
)

// NPMHandler handles npm registry protocol requests.
type NPMHandler struct {
	proxy       *Proxy
	upstreamURL string
	proxyURL    string // URL where this proxy is hosted
}

// NewNPMHandler creates a new npm protocol handler.
func NewNPMHandler(proxy *Proxy, proxyURL string) *NPMHandler {
	return &NPMHandler{
		proxy:       proxy,
		upstreamURL: npmUpstream,
		proxyURL:    strings.TrimSuffix(proxyURL, "/"),
	}
}

// Routes returns the HTTP handler for npm requests.
// Mount this at /npm on your router.
func (h *NPMHandler) Routes() http.Handler {
	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if r.Method != http.MethodGet {
			http.Error(w, "method not allowed", http.StatusMethodNotAllowed)
			return
		}

		path := strings.TrimPrefix(r.URL.Path, "/")

		// Check if this is a tarball download (contains /-/)
		if strings.Contains(path, "/-/") {
			h.handleDownload(w, r)
			return
		}

		// Otherwise it's a metadata request
		h.handlePackageMetadata(w, r)
	})
}

// handlePackageMetadata proxies package metadata from upstream and rewrites tarball URLs.
func (h *NPMHandler) handlePackageMetadata(w http.ResponseWriter, r *http.Request) {
	packageName := h.extractPackageName(r)
	if packageName == "" {
		JSONError(w, http.StatusBadRequest, "invalid package name")
		return
	}

	h.proxy.Logger.Info("npm metadata request", "package", packageName)

	upstreamURL := fmt.Sprintf("%s/%s", h.upstreamURL, url.PathEscape(packageName))

	// Use abbreviated metadata when cooldown is disabled — it's much smaller
	// (e.g. drizzle-orm: 4MB vs 92MB) but lacks the time map needed for cooldown.
	accept := npmAbbreviatedCT
	if h.proxy.Cooldown != nil && h.proxy.Cooldown.Enabled() {
		accept = contentTypeJSON
	}

	body, _, err := h.proxy.FetchOrCacheMetadata(r.Context(), "npm", packageName, upstreamURL, accept)
	if err != nil {
		if errors.Is(err, ErrUpstreamNotFound) {
			JSONError(w, http.StatusNotFound, "package not found")
			return
		}
		h.proxy.Logger.Error("failed to fetch npm metadata", "error", err)
		JSONError(w, http.StatusBadGateway, "failed to fetch from upstream")
		return
	}

	rewritten, err := h.rewriteMetadata(packageName, body)
	if err != nil {
		// If rewriting fails, just proxy the original
		h.proxy.Logger.Warn("failed to rewrite metadata, proxying original", "error", err)
		w.Header().Set("Content-Type", contentTypeJSON)
		w.WriteHeader(http.StatusOK)
		_, _ = w.Write(body)
		return
	}

	w.Header().Set("Content-Type", contentTypeJSON)
	w.WriteHeader(http.StatusOK)
	_, _ = w.Write(rewritten)
}

// rewriteMetadata rewrites tarball URLs in npm package metadata to point at this proxy.
// If cooldown is enabled, versions published too recently are filtered out.
func (h *NPMHandler) rewriteMetadata(packageName string, body []byte) ([]byte, error) {
	var metadata map[string]any
	if err := json.Unmarshal(body, &metadata); err != nil {
		return nil, err
	}

	// Rewrite tarball URLs in versions
	versions, ok := metadata["versions"].(map[string]any)
	if !ok {
		return body, nil // No versions to rewrite
	}

	h.applyCooldownFiltering(metadata, versions, packageName)
	h.rewriteTarballURLs(versions, packageName)

	return json.Marshal(metadata)
}

// applyCooldownFiltering removes versions that are too recently published,
// and updates dist-tags.latest if the current latest was filtered out.
func (h *NPMHandler) applyCooldownFiltering(metadata map[string]any, versions map[string]any, packageName string) {
	if h.proxy.Cooldown == nil || !h.proxy.Cooldown.Enabled() {
		return
	}

	timeMap, _ := metadata["time"].(map[string]any)
	if timeMap == nil {
		return
	}

	packagePURL := purl.MakePURLString("npm", packageName, "")

	for version := range versions {
		publishedStr, ok := timeMap[version].(string)
		if !ok {
			continue
		}
		publishedAt, err := time.Parse(time.RFC3339, publishedStr)
		if err != nil {
			continue
		}
		if !h.proxy.Cooldown.IsAllowed("npm", packagePURL, publishedAt) {
			h.proxy.Logger.Info("cooldown: filtering npm version",
				"package", packageName, "version", version,
				"published", publishedStr)
			delete(versions, version)
			delete(timeMap, version)
		}
	}

	h.updateDistTagsLatest(metadata, versions, timeMap)
}

// updateDistTagsLatest updates the dist-tags.latest field if the current latest
// version was removed by cooldown filtering.
func (h *NPMHandler) updateDistTagsLatest(metadata, versions, timeMap map[string]any) {
	distTags, ok := metadata["dist-tags"].(map[string]any)
	if !ok {
		return
	}

	latest, ok := distTags["latest"].(string)
	if !ok {
		return
	}

	if _, exists := versions[latest]; exists {
		return
	}

	if newLatest := h.findNewestVersion(versions, timeMap); newLatest != "" {
		distTags["latest"] = newLatest
	}
}

// rewriteTarballURLs rewrites all tarball URLs in version entries to point at this proxy.
func (h *NPMHandler) rewriteTarballURLs(versions map[string]any, packageName string) {
	for version, vdata := range versions {
		vmap, ok := vdata.(map[string]any)
		if !ok {
			continue
		}

		dist, ok := vmap["dist"].(map[string]any)
		if !ok {
			continue
		}

		tarball, ok := dist["tarball"].(string)
		if !ok {
			continue
		}

		filename := tarball
		if idx := strings.LastIndex(tarball, "/"); idx >= 0 {
			filename = tarball[idx+1:]
		}

		escapedName := url.PathEscape(packageName)
		newTarball := fmt.Sprintf("%s/npm/%s/-/%s", h.proxyURL, escapedName, filename)
		dist["tarball"] = newTarball

		h.proxy.Logger.Debug("rewrote tarball URL",
			"package", packageName, "version", version,
			"old", tarball, "new", newTarball)
	}
}

// findNewestVersion returns the version string with the most recent timestamp
// from the remaining versions, using the time map.
func (h *NPMHandler) findNewestVersion(versions map[string]any, timeMap map[string]any) string {
	if timeMap == nil {
		return ""
	}

	type versionTime struct {
		version string
		t       time.Time
	}

	var vts []versionTime
	for v := range versions {
		if ts, ok := timeMap[v].(string); ok {
			if t, err := time.Parse(time.RFC3339, ts); err == nil {
				vts = append(vts, versionTime{v, t})
			}
		}
	}

	if len(vts) == 0 {
		return ""
	}

	sort.Slice(vts, func(i, j int) bool {
		return vts[i].t.After(vts[j].t)
	})

	return vts[0].version
}

// handleDownload serves a package tarball, fetching and caching from upstream if needed.
func (h *NPMHandler) handleDownload(w http.ResponseWriter, r *http.Request) {
	packageName, filename := h.parseDownloadPath(r.URL.Path)

	if packageName == "" || filename == "" {
		JSONError(w, http.StatusBadRequest, "invalid request")
		return
	}

	// Extract version from filename (e.g., "lodash-4.17.21.tgz" -> "4.17.21")
	version := h.extractVersionFromFilename(packageName, filename)
	if version == "" {
		JSONError(w, http.StatusBadRequest, "could not determine version from filename")
		return
	}

	h.proxy.Logger.Info("npm download request",
		"package", packageName, "version", version, "filename", filename)

	result, err := h.proxy.GetOrFetchArtifact(r.Context(), "npm", packageName, version, filename)
	if err != nil {
		h.proxy.Logger.Error("failed to get artifact", "error", err)
		JSONError(w, http.StatusBadGateway, "failed to fetch package")
		return
	}

	ServeArtifact(w, result)
}

// extractPackageName extracts the package name from the request path.
// Handles both scoped (@scope/name) and unscoped (name) packages.
func (h *NPMHandler) extractPackageName(r *http.Request) string {
	path := strings.TrimPrefix(r.URL.Path, "/")

	// Remove /-/filename suffix if present
	if idx := strings.Index(path, "/-/"); idx >= 0 {
		path = path[:idx]
	}

	// URL decode the path (handles %40 -> @, %2f -> /)
	decoded, err := url.PathUnescape(path)
	if err != nil {
		return path
	}

	return decoded
}

// parseDownloadPath extracts package name and filename from a download path.
// Path format: /@scope/name/-/filename.tgz or /name/-/filename.tgz
func (h *NPMHandler) parseDownloadPath(path string) (packageName, filename string) {
	path = strings.TrimPrefix(path, "/")

	idx := strings.Index(path, "/-/")
	if idx < 0 {
		return "", ""
	}

	packageName = path[:idx]
	filename = path[idx+3:] // skip "/-/"

	// URL decode package name
	if decoded, err := url.PathUnescape(packageName); err == nil {
		packageName = decoded
	}

	return packageName, filename
}

// extractVersionFromFilename extracts version from npm tarball filename.
// e.g., "lodash-4.17.21.tgz" -> "4.17.21"
// e.g., "core-7.23.0.tgz" for @babel/core -> "7.23.0"
func (h *NPMHandler) extractVersionFromFilename(packageName, filename string) string {
	// Remove .tgz extension
	if !strings.HasSuffix(filename, ".tgz") {
		return ""
	}
	base := strings.TrimSuffix(filename, ".tgz")

	// For scoped packages, the filename uses the short name
	shortName := packageName
	if strings.Contains(packageName, "/") {
		parts := strings.SplitN(packageName, "/", scopedParts)
		shortName = parts[1]
	}

	// Expected format: {shortName}-{version}
	prefix := shortName + "-"
	if !strings.HasPrefix(base, prefix) {
		return ""
	}

	return strings.TrimPrefix(base, prefix)
}