Add experimental Go module proxy support (#358)

* feat: add experimental Go module support via pmg go

Adds Go modules as a proxy-guarded ecosystem, opt-in only: the command
runs solely when invoked explicitly as `pmg go ...` and is deliberately
excluded from setup aliases and PATH shims so existing users are
unaffected.

- packagemanager: goPackageManager with fail-safe command classification
  (vet/fix excluded from non-download since they can fetch on a cold
  cache) and pinned-version extraction where only canonical semver
  counts as explicit.
- GOPROXY normalization (fail-closed): effective GOPROXY read via
  `go env` (honors go env -w), rebuilt comma-joined with `direct`
  dropped so a 403 block is terminal and nothing silently falls back to
  unanalyzed VCS fetches. GOPRIVATE/GONOPROXY surface a warning;
  GOINSECURE is cleared. Contributed to the proxy flow through a new
  ProxyRoutingProvider hook (extra child env + dynamic MITM hosts).
- Go interceptor with dynamic host matching from the user's effective
  GOPROXY via InterceptorContext.GoProxyHosts. Malware analysis runs on
  .zip only (the sole endpoint that delivers code); .info/.mod/@latest/
  list pass through; /sumdb/ traffic and sum.golang.org are never
  touched so checksum-db verification stays intact; golang.org/toolchain
  is allowed on Go's own checksum verification.
- Dependency cooldown: publish time captured from .info responses
  (body unmodified), in-window .zip blocked with 403; fails open for
  cooldown only when the publish time was never observed.
- Cert gate: on macOS/Windows `pmg go` fails fast with actionable
  guidance unless the persisted PMG CA is OS-trusted (Go ignores
  SSL_CERT_FILE there); Linux works via the injected bundle.
- proxye2e: GOPROXY-protocol mock registry, Go driver and 10 hermetic
  cases (allow/block/confirm, case-escaped paths, cooldown block and
  fail-open, toolchain, sumdb passthrough).

Verified end-to-end on Linux: `pmg go get github.com/google/uuid@v1.6.0`
MITMs proxy.golang.org, analyzes the decoded module at the .zip fetch,
and go.sum verification succeeds through the tunneled checksum db.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014xuhBeTVpfU4SdqVaarvuK

* fix(go): address review findings on experimental Go support

- Drop fmt/clean from NonDownloadCommands: both load packages via go
  list and can download modules on a cold cache, which would bypass the
  proxy under install_only.
- Support GOPROXY entries with a base path (e.g. corp Athens/JFrog at
  https://corp/goproxy): the interceptor now receives host -> base URL
  and strips the path prefix before parsing module URLs, so verdicts
  and cooldown key on the real module path.
- Default unschemed GOPROXY entries to https, matching go's own
  behavior, so corp mirrors configured as bare hosts are intercepted
  instead of silently unanalyzed.
- Memoize the final verdict per module zip: go re-requests a failed
  zip during go get's load phase, which double-recorded stats (the
  report showed the same blocked module twice) and would have
  re-prompted on Confirm verdicts.
- Fetch .info out-of-band on a cooldown cache miss: go serves .info
  from its local module cache on any machine that used go before PMG,
  which silently disabled cooldown. Failure of the side-fetch still
  fails open for cooldown only.
- Move the noop package resolver into packagemanager.

Verified live: cold-cache cooldown block now records once; warm-cache
rerun is blocked via the side-fetch instead of failing open.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014xuhBeTVpfU4SdqVaarvuK

* docs: collapse Go proxy-mode details by default

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014xuhBeTVpfU4SdqVaarvuK

---------

Co-authored-by: Claude <noreply@anthropic.com>
This commit is contained in:
Abhisek Datta
2026-07-03 18:19:31 +05:30
committed by GitHub
co-authored by Claude Fable 5
parent 648adcbda4
commit b19473945b
24 changed files with 1941 additions and 15 deletions
+11 -1
View File
@@ -46,6 +46,16 @@ func (i *AuditLoggerInterceptor) HandleRequest(ctx *proxy.RequestContext) (*prox
return &proxy.InterceptorResponse{Action: proxy.ActionAllow}, nil
}
// wellKnownGoHosts are the default Go module-proxy and checksum-database
// hosts. Custom GOPROXY hosts are dynamic (known only to the Go interceptor's
// per-run config) and intentionally still surface here as observed hosts.
var wellKnownGoHosts = map[string]bool{
"proxy.golang.org": true,
"sum.golang.org": true,
}
func (i *AuditLoggerInterceptor) isKnownRegistryHost(hostname string) bool {
return npmRegistryDomains.ContainsHostname(hostname) || pypiRegistryDomains.ContainsHostname(hostname)
return npmRegistryDomains.ContainsHostname(hostname) ||
pypiRegistryDomains.ContainsHostname(hostname) ||
wellKnownGoHosts[hostname]
}
+17
View File
@@ -13,6 +13,13 @@ import (
// (analyzer, cache, stats), this holds context specific to the current run.
type InterceptorContext struct {
PinnedVersions map[string]string
// GoProxyBaseURLs maps module-proxy hostnames from the user's effective
// GOPROXY to their upstream base URL (scheme + host + optional path
// prefix). The Go interceptor MITMs and analyzes these hosts; Go is the
// only ecosystem whose registry hosts are user-configurable rather than
// fixed.
GoProxyBaseURLs map[string]string
}
// InterceptorFactory creates ecosystem-specific interceptors for the proxy
@@ -63,6 +70,15 @@ func (f *InterceptorFactory) CreateInterceptor(ecosystem packagev1.Ecosystem) (p
f.execContext,
), nil
case packagev1.Ecosystem_ECOSYSTEM_GO:
return NewGoRegistryInterceptor(
f.analyzer,
f.cache,
f.statsCollector,
f.confirmationChan,
f.execContext,
), nil
default:
return nil, fmt.Errorf("proxy-based interception not yet supported for ecosystem: %s", ecosystem.String())
}
@@ -73,6 +89,7 @@ func SupportedEcosystems() []packagev1.Ecosystem {
return []packagev1.Ecosystem{
packagev1.Ecosystem_ECOSYSTEM_NPM,
packagev1.Ecosystem_ECOSYSTEM_PYPI,
packagev1.Ecosystem_ECOSYSTEM_GO,
}
}
+189
View File
@@ -0,0 +1,189 @@
package interceptors
import (
"encoding/json"
"fmt"
"io"
"net/http"
"strings"
"sync"
"time"
packagev1 "buf.build/gen/go/safedep/api/protocolbuffers/go/safedep/messages/package/v1"
"github.com/safedep/dry/log"
pmgconfig "github.com/safedep/pmg/config"
"github.com/safedep/pmg/internal/audit"
"github.com/safedep/pmg/proxy"
gomodule "golang.org/x/mod/module"
)
// goCooldownHandler enforces dependency cooldown for Go modules. Unlike npm,
// there is no metadata to strip: the version go requests is already resolved
// by the time the proxy sees it. Instead the publish timestamp is captured
// from the .info response go fetches before each .zip, and an in-window .zip
// download is blocked with HTTP 403 — terminal, because the child GOPROXY is
// normalized to a comma-joined list with no direct fallback.
type goCooldownHandler struct {
statsCollector *AnalysisStatsCollector
mu sync.Mutex
publishTimes map[string]time.Time
}
func newGoCooldownHandler(statsCollector *AnalysisStatsCollector) *goCooldownHandler {
return &goCooldownHandler{
statsCollector: statsCollector,
publishTimes: map[string]time.Time{},
}
}
func goModuleVersionKey(module, version string) string {
return module + "@" + version
}
// HandleInfoRequest reads the .info response body without altering it and
// caches the version's publish time for the upcoming .zip request.
func (h *goCooldownHandler) HandleInfoRequest(ctx *proxy.RequestContext, module, version string) (*proxy.InterceptorResponse, error) {
// Force an uncompressed, non-conditional response so the body is parseable
// JSON rather than raw gzip bytes or an empty 304 (same as the npm
// metadata modifier).
ctx.Headers.Set("Accept-Encoding", "identity")
ctx.Headers.Del("If-None-Match")
ctx.Headers.Del("If-Modified-Since")
modifier := func(statusCode int, headers http.Header, body []byte) (int, http.Header, []byte, error) {
if statusCode != http.StatusOK {
return statusCode, headers, body, nil
}
var info struct {
Time time.Time `json:"Time"`
}
if err := json.Unmarshal(body, &info); err != nil || info.Time.IsZero() {
log.Warnf("[%s] Cooldown: failed to parse publish time from .info for %s@%s", ctx.RequestID, module, version)
return statusCode, headers, body, nil
}
h.mu.Lock()
h.publishTimes[goModuleVersionKey(module, version)] = info.Time
h.mu.Unlock()
return statusCode, headers, body, nil
}
return &proxy.InterceptorResponse{
Action: proxy.ActionModifyResponse,
ResponseModifier: modifier,
}, nil
}
// CheckZipDownload blocks the module zip when its publish time is within the
// cooldown window. handled=false lets the request continue to malware
// analysis. When the publish time was not observed on the wire (go served
// .info from its local module cache, common on machines that used go before
// PMG), it is fetched out-of-band from the upstream proxy; only if that also
// fails does cooldown fail open — malware analysis still runs.
func (h *goCooldownHandler) CheckZipDownload(ctx *proxy.RequestContext, baseURL, module, version string, cooldownDays int) (*proxy.InterceptorResponse, bool) {
skip := pmgconfig.CooldownSkip(packagev1.Ecosystem_ECOSYSTEM_GO, module)
if skip.SkipAll || pmgconfig.IsTrustedPackageRef(packagev1.Ecosystem_ECOSYSTEM_GO, module, version) {
return nil, false
}
h.mu.Lock()
publishTime, ok := h.publishTimes[goModuleVersionKey(module, version)]
h.mu.Unlock()
if !ok {
publishTime, ok = h.fetchPublishTime(ctx, baseURL, module, version)
}
if !ok {
log.Warnf("[%s] Cooldown: no publish time available for %s@%s; cooldown not enforced for this download", ctx.RequestID, module, version)
return nil, false
}
within, daysAgo, daysLeft := cooldownIsWithinWindow(publishTime, cooldownDays)
if !within {
return nil, false
}
if skip.ExemptsVersion(version) {
auditCooldownSkips(ctx.RequestID, packagev1.Ecosystem_ECOSYSTEM_GO, module, cooldownExemptions{skipListed: []string{version}})
return nil, false
}
log.Infof("[%s] Cooldown: blocking %s@%s published %d day(s) ago (%d day cooldown, %d remaining)",
ctx.RequestID, module, version, daysAgo, cooldownDays, daysLeft)
if h.statsCollector != nil {
h.statsCollector.RecordCooldownBlocked(module, version, publishTime, daysAgo, daysLeft, cooldownDays)
}
pv := &packagev1.PackageVersion{}
pv.SetPackage(&packagev1.Package{})
pv.GetPackage().SetName(module)
pv.GetPackage().SetEcosystem(packagev1.Ecosystem_ECOSYSTEM_GO)
pv.SetVersion(version)
audit.LogDependencyCooldown(pv, publishTime, cooldownDays, daysAgo, daysLeft)
message := fmt.Sprintf("Package blocked by dependency cooldown: GO/%s@%s\n\nPublished %d day(s) ago; cooldown window is %d day(s) (%d remaining).",
module, version, daysAgo, cooldownDays, daysLeft)
return &proxy.InterceptorResponse{
Action: proxy.ActionBlock,
BlockCode: http.StatusForbidden,
BlockMessage: message,
}, true
}
// goInfoFetchClient fetches .info out-of-band, straight to the upstream proxy
// rather than back through PMG's own in-process proxy (which would
// re-intercept the request). It honors the process' own proxy environment,
// not the child's injected one.
var goInfoFetchClient = &http.Client{Timeout: 10 * time.Second}
// fetchPublishTime performs a one-shot authoritative $base/$module/@v/$version.info
// fetch and caches the result. Best-effort: any failure means no publish time.
func (h *goCooldownHandler) fetchPublishTime(ctx *proxy.RequestContext, baseURL, module, version string) (time.Time, bool) {
if baseURL == "" {
return time.Time{}, false
}
escapedPath, err := gomodule.EscapePath(module)
if err != nil {
return time.Time{}, false
}
escapedVersion, err := gomodule.EscapeVersion(version)
if err != nil {
return time.Time{}, false
}
infoURL := fmt.Sprintf("%s/%s/@v/%s.info", strings.TrimSuffix(baseURL, "/"), escapedPath, escapedVersion)
resp, err := goInfoFetchClient.Get(infoURL)
if err != nil {
log.Warnf("[%s] Cooldown: failed to fetch %s: %v", ctx.RequestID, infoURL, err)
return time.Time{}, false
}
defer func() { _ = resp.Body.Close() }()
if resp.StatusCode != http.StatusOK {
log.Warnf("[%s] Cooldown: fetching %s returned HTTP %d", ctx.RequestID, infoURL, resp.StatusCode)
return time.Time{}, false
}
var info struct {
Time time.Time `json:"Time"`
}
if err := json.NewDecoder(io.LimitReader(resp.Body, 1<<20)).Decode(&info); err != nil || info.Time.IsZero() {
log.Warnf("[%s] Cooldown: failed to parse publish time from %s", ctx.RequestID, infoURL)
return time.Time{}, false
}
h.mu.Lock()
h.publishTimes[goModuleVersionKey(module, version)] = info.Time
h.mu.Unlock()
return info.Time, true
}
+57
View File
@@ -0,0 +1,57 @@
package interceptors
import (
"encoding/json"
"net/http"
"net/http/httptest"
"testing"
"time"
"github.com/safedep/pmg/proxy"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
func TestGoCooldownCheckZipDownloadSideFetch(t *testing.T) {
publishTime := time.Now().Add(-24 * time.Hour)
server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, req *http.Request) {
switch req.URL.Path {
case "/example.com/fresh/@v/v1.1.0.info":
w.Header().Set("Content-Type", "application/json")
err := json.NewEncoder(w).Encode(map[string]string{
"Version": "v1.1.0",
"Time": publishTime.UTC().Format(time.RFC3339),
})
require.NoError(t, err)
default:
http.NotFound(w, req)
}
}))
defer server.Close()
ctx := &proxy.RequestContext{RequestID: "test"}
t.Run("blocks using out-of-band publish time on cache miss", func(t *testing.T) {
h := newGoCooldownHandler(NewAnalysisStatsCollector())
resp, handled := h.CheckZipDownload(ctx, server.URL, "example.com/fresh", "v1.1.0", 7)
require.True(t, handled)
assert.Equal(t, proxy.ActionBlock, resp.Action)
assert.Equal(t, http.StatusForbidden, resp.BlockCode)
})
t.Run("fails open when the out-of-band fetch fails", func(t *testing.T) {
h := newGoCooldownHandler(NewAnalysisStatsCollector())
_, handled := h.CheckZipDownload(ctx, server.URL, "example.com/unknown", "v9.9.9", 7)
assert.False(t, handled)
})
t.Run("fails open without a base URL", func(t *testing.T) {
h := newGoCooldownHandler(NewAnalysisStatsCollector())
_, handled := h.CheckZipDownload(ctx, "", "example.com/fresh", "v1.1.0", 7)
assert.False(t, handled)
})
}
+205
View File
@@ -0,0 +1,205 @@
package interceptors
import (
"net/url"
"strings"
"sync"
packagev1 "buf.build/gen/go/safedep/api/protocolbuffers/go/safedep/messages/package/v1"
"github.com/safedep/dry/log"
"github.com/safedep/pmg/analyzer"
pmgconfig "github.com/safedep/pmg/config"
"github.com/safedep/pmg/proxy"
)
// goToolchainModule is the module path Go uses to auto-download toolchains
// (GOTOOLCHAIN=auto). Toolchain zips are verified by go against the checksum
// database regardless of GOPRIVATE/GONOSUMDB, and downloads fail closed when
// GOSUMDB=off, so PMG passes them through on Go's own verification instead of
// treating them as ordinary (never-flagged) modules.
const goToolchainModule = "golang.org/toolchain"
// GoRegistryInterceptor intercepts Go module proxy requests and analyzes
// module zips for malware. Unlike npm/PyPI, the registry hosts are not fixed:
// they come from the user's effective GOPROXY via
// InterceptorContext.GoProxyBaseURLs. sum.golang.org is never in that set, so
// checksum-database traffic is tunneled, not MITM'd.
type GoRegistryInterceptor struct {
baseRegistryInterceptor
domains registryConfigMap
baseURLs map[string]string
cooldownHandler *goCooldownHandler
// zipVerdicts memoizes the final response per module zip. go re-requests
// a failed zip (once more during go get's load phase), and without this
// the repeat would double-record stats — the report would show the same
// blocked module twice — and re-prompt the user on a Confirm verdict.
zipVerdictsMu sync.Mutex
zipVerdicts map[string]*proxy.InterceptorResponse
}
var _ proxy.Interceptor = (*GoRegistryInterceptor)(nil)
var _ proxy.MITMDecider = (*GoRegistryInterceptor)(nil)
func NewGoRegistryInterceptor(
analyzer analyzer.PackageVersionAnalyzer,
cache AnalysisCache,
statsCollector *AnalysisStatsCollector,
confirmationChan chan *ConfirmationRequest,
execContext InterceptorContext,
) *GoRegistryInterceptor {
domains := registryConfigMap{}
baseURLs := map[string]string{}
for host, baseURL := range execContext.GoProxyBaseURLs {
basePath := ""
if u, err := url.Parse(baseURL); err == nil {
basePath = strings.TrimSuffix(u.Path, "/")
}
domains[host] = &registryConfig{
Host: host,
SupportedForAnalysis: true,
Parser: goProxyParser{basePath: basePath},
}
baseURLs[host] = baseURL
}
return &GoRegistryInterceptor{
baseRegistryInterceptor: baseRegistryInterceptor{
analyzer: analyzer,
cache: cache,
statsCollector: statsCollector,
confirmationChan: confirmationChan,
circuitBreaker: newAnalyzerCircuitBreaker("malysis-analyzer-go"),
execContext: execContext,
},
domains: domains,
baseURLs: baseURLs,
cooldownHandler: newGoCooldownHandler(statsCollector),
zipVerdicts: map[string]*proxy.InterceptorResponse{},
}
}
func (i *GoRegistryInterceptor) Name() string {
return "go-registry-interceptor"
}
func (i *GoRegistryInterceptor) ShouldMITM(ctx *proxy.RequestContext) bool {
config := i.domains.GetConfigForHostname(ctx.Hostname)
if config == nil {
return false
}
return config.SupportedForAnalysis
}
func (i *GoRegistryInterceptor) ShouldIntercept(ctx *proxy.RequestContext) bool {
return i.domains.ContainsHostname(ctx.Hostname)
}
// HandleRequest processes the request and returns response action.
// We take a fail-open approach here, allowing requests that we can't parse the
// package information from the URL — but an unparseable .zip means an
// unanalyzed source download, so that case is logged loudly.
func (i *GoRegistryInterceptor) HandleRequest(ctx *proxy.RequestContext) (*proxy.InterceptorResponse, error) {
log.Debugf("[%s] Handling Go module proxy request: %s", ctx.RequestID, ctx.URL.Path)
config := i.domains.GetConfigForHostname(ctx.Hostname)
if config == nil {
log.Warnf("[%s] No registry config found for hostname: %s", ctx.RequestID, ctx.Hostname)
return &proxy.InterceptorResponse{Action: proxy.ActionAllow}, nil
}
pkgInfo, err := config.Parser.ParseURL(ctx.URL.Path)
if err != nil {
if strings.HasSuffix(ctx.URL.Path, ".zip") {
log.Warnf("[%s] Failed to parse Go module proxy zip URL %s: %v — download allowed without analysis",
ctx.RequestID, ctx.URL.Path, err)
} else {
log.Debugf("[%s] Failed to parse Go module proxy URL %s: %v", ctx.RequestID, ctx.URL.Path, err)
}
return &proxy.InterceptorResponse{Action: proxy.ActionAllow}, nil
}
info, ok := pkgInfo.(*goModuleInfo)
if !ok {
return &proxy.InterceptorResponse{Action: proxy.ActionAllow}, nil
}
if info.requestType == goRequestSumDB {
log.Debugf("[%s] Allowing proxied checksum-database request: %s", ctx.RequestID, ctx.URL.Path)
return &proxy.InterceptorResponse{Action: proxy.ActionAllow}, nil
}
if info.name == goToolchainModule {
if info.IsFileDownload() {
log.Infof("[%s] Allowing Go toolchain download %s@%s (verified by Go's checksum database)",
ctx.RequestID, info.name, info.version)
}
return &proxy.InterceptorResponse{Action: proxy.ActionAllow}, nil
}
depCooldownConfig := pmgconfig.Get().Config.DependencyCooldown
if !info.IsFileDownload() {
if info.requestType == goRequestInfo && info.version != "" && depCooldownConfig.Enabled {
return i.cooldownHandler.HandleInfoRequest(ctx, info.name, info.version)
}
log.Debugf("[%s] Skipping analysis for metadata request: %s", ctx.RequestID, info.name)
return &proxy.InterceptorResponse{Action: proxy.ActionAllow}, nil
}
key := goModuleVersionKey(info.name, info.version)
i.zipVerdictsMu.Lock()
memo := i.zipVerdicts[key]
i.zipVerdictsMu.Unlock()
if memo != nil {
log.Debugf("[%s] Reusing verdict for repeated zip request: %s", ctx.RequestID, key)
return memo, nil
}
resp, memoize, err := i.handleZipDownload(ctx, config, info, depCooldownConfig)
if err != nil {
return resp, err
}
if memoize {
i.zipVerdictsMu.Lock()
i.zipVerdicts[key] = resp
i.zipVerdictsMu.Unlock()
}
return resp, nil
}
// handleZipDownload runs the security controls for a module source download:
// dependency cooldown, then trusted/insecure fast-allow, then malware
// analysis. memoize is false only when the outcome is a fail-open allow after
// an analyzer error, so a retried request gets another chance to be analyzed.
func (i *GoRegistryInterceptor) handleZipDownload(
ctx *proxy.RequestContext,
config *registryConfig,
info *goModuleInfo,
depCooldownConfig pmgconfig.DependencyCooldownConfig,
) (*proxy.InterceptorResponse, bool, error) {
if depCooldownConfig.Enabled {
if resp, handled := i.cooldownHandler.CheckZipDownload(ctx, i.baseURLs[config.Host], info.name, info.version, depCooldownConfig.Days); handled {
return resp, true, nil
}
}
if resp, ok := i.fastAllow(ctx, packagev1.Ecosystem_ECOSYSTEM_GO, info.name, info.version); ok {
return resp, true, nil
}
result, err := i.analyzePackage(ctx, packagev1.Ecosystem_ECOSYSTEM_GO, info.name, info.version)
if err != nil {
log.Errorf("[%s] Failed to analyze package %s@%s: %v", ctx.RequestID, info.name, info.version, err)
return &proxy.InterceptorResponse{Action: proxy.ActionAllow}, false, nil
}
resp, err := i.handleAnalysisResult(ctx, packagev1.Ecosystem_ECOSYSTEM_GO, info.name, info.version, result)
return resp, err == nil, err
}
+43
View File
@@ -0,0 +1,43 @@
package interceptors
import (
"testing"
"github.com/safedep/pmg/proxy"
"github.com/stretchr/testify/assert"
)
func TestGoRegistryInterceptorHostMatching(t *testing.T) {
interceptor := NewGoRegistryInterceptor(nil, nil, nil, nil, InterceptorContext{
GoProxyBaseURLs: map[string]string{
"proxy.golang.org": "https://proxy.golang.org",
"corp.example.com": "https://corp.example.com/goproxy",
},
})
cases := []struct {
hostname string
wantIntercept bool
wantMITM bool
}{
{"proxy.golang.org", true, true},
{"corp.example.com", true, true},
{"sum.golang.org", false, false},
{"github.com", false, false},
{"registry.npmjs.org", false, false},
}
for _, tc := range cases {
ctx := &proxy.RequestContext{Hostname: tc.hostname}
assert.Equal(t, tc.wantIntercept, interceptor.ShouldIntercept(ctx), "ShouldIntercept(%s)", tc.hostname)
assert.Equal(t, tc.wantMITM, interceptor.ShouldMITM(ctx), "ShouldMITM(%s)", tc.hostname)
}
}
func TestGoRegistryInterceptorNoHosts(t *testing.T) {
interceptor := NewGoRegistryInterceptor(nil, nil, nil, nil, InterceptorContext{})
ctx := &proxy.RequestContext{Hostname: "proxy.golang.org"}
assert.False(t, interceptor.ShouldIntercept(ctx))
assert.False(t, interceptor.ShouldMITM(ctx))
}
+122
View File
@@ -0,0 +1,122 @@
package interceptors
import (
"fmt"
"strings"
"golang.org/x/mod/module"
)
const (
goRequestLatest = "latest"
goRequestList = "list"
goRequestInfo = "info"
goRequestMod = "mod"
goRequestZip = "zip"
// goRequestSumDB is checksum-database traffic proxied through the module
// proxy ($GOPROXY/sumdb/...). It is passed through unmodified so go's
// signature verification stays intact.
goRequestSumDB = "sumdb"
)
// goModuleInfo is parsed module information from a Go module proxy URL.
type goModuleInfo struct {
name string
version string
requestType string
}
var _ packageInfo = (*goModuleInfo)(nil)
func (g *goModuleInfo) GetName() string { return g.name }
func (g *goModuleInfo) GetVersion() string { return g.version }
// IsFileDownload is true only for .zip: the single endpoint that downloads
// module source. .info/.mod are fetched for the entire candidate graph during
// version selection, including modules never selected into the build, so
// analyzing them would block builds over code that never lands in the cache.
func (g *goModuleInfo) IsFileDownload() bool { return g.requestType == goRequestZip }
// goProxyParser parses Go module proxy URLs per the GOPROXY protocol
// (https://go.dev/ref/mod#goproxy-protocol):
//
// /<module>/@latest -> latest version metadata
// /<module>/@v/list -> version list
// /<module>/@v/<version>.info -> version metadata JSON (publish time)
// /<module>/@v/<version>.mod -> go.mod file
// /<module>/@v/<version>.zip -> module source archive
// /sumdb/<name>/... -> proxied checksum-database traffic
//
// Uppercase letters in module path and version arrive escaped as '!'+lowercase
// and are decoded before use as Malysis query keys.
//
// basePath is the path prefix of the GOPROXY entry (e.g. "/goproxy" for
// GOPROXY=https://corp.example.com/goproxy): go sends requests under that
// base, so it is stripped before the module path is parsed.
type goProxyParser struct {
basePath string
}
var _ registryURLParser = goProxyParser{}
func (g goProxyParser) ParseURL(urlPath string) (packageInfo, error) {
if g.basePath != "" {
rest, ok := strings.CutPrefix(urlPath, g.basePath)
if !ok || (rest != "" && rest[0] != '/') {
return nil, fmt.Errorf("go proxy URL %q is outside proxy base path %q", urlPath, g.basePath)
}
urlPath = rest
}
p := strings.TrimPrefix(urlPath, "/")
if p == "" {
return nil, fmt.Errorf("empty go proxy URL path")
}
if p == "sumdb" || strings.HasPrefix(p, "sumdb/") {
return &goModuleInfo{requestType: goRequestSumDB}, nil
}
if escaped, ok := strings.CutSuffix(p, "/@latest"); ok {
name, err := module.UnescapePath(escaped)
if err != nil {
return nil, fmt.Errorf("invalid module path in go proxy URL: %w", err)
}
return &goModuleInfo{name: name, requestType: goRequestLatest}, nil
}
escapedPath, versionPart, ok := strings.Cut(p, "/@v/")
if !ok {
return nil, fmt.Errorf("go proxy URL missing /@v/ or /@latest marker")
}
name, err := module.UnescapePath(escapedPath)
if err != nil {
return nil, fmt.Errorf("invalid module path in go proxy URL: %w", err)
}
if versionPart == "list" {
return &goModuleInfo{name: name, requestType: goRequestList}, nil
}
dot := strings.LastIndex(versionPart, ".")
if dot <= 0 {
return nil, fmt.Errorf("go proxy URL has no version suffix: %q", versionPart)
}
requestType := versionPart[dot+1:]
switch requestType {
case goRequestInfo, goRequestMod, goRequestZip:
default:
return nil, fmt.Errorf("unrecognized go proxy version suffix: %q", requestType)
}
version, err := module.UnescapeVersion(versionPart[:dot])
if err != nil {
return nil, fmt.Errorf("invalid version in go proxy URL: %w", err)
}
return &goModuleInfo{name: name, version: version, requestType: requestType}, nil
}
+132
View File
@@ -0,0 +1,132 @@
package interceptors
import (
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
func TestGoProxyParserParseURL(t *testing.T) {
cases := []struct {
name string
path string
wantName string
wantVersion string
wantType string
wantIsDownload bool
wantErrContains string
}{
{
name: "zip download",
path: "/github.com/x/y/@v/v1.2.3.zip",
wantName: "github.com/x/y",
wantVersion: "v1.2.3",
wantType: goRequestZip,
wantIsDownload: true,
},
{
name: "info metadata",
path: "/github.com/x/y/@v/v1.2.3.info",
wantName: "github.com/x/y",
wantVersion: "v1.2.3",
wantType: goRequestInfo,
},
{
name: "mod metadata",
path: "/github.com/x/y/@v/v1.2.3.mod",
wantName: "github.com/x/y",
wantVersion: "v1.2.3",
wantType: goRequestMod,
},
{
name: "version list",
path: "/github.com/x/y/@v/list",
wantName: "github.com/x/y",
wantType: goRequestList,
},
{
name: "latest metadata",
path: "/github.com/x/y/@latest",
wantName: "github.com/x/y",
wantType: goRequestLatest,
},
{
name: "case-escaped module path and version are decoded",
path: "/github.com/!burnt!sushi/toml/@v/!v1.0.0-!rc1.zip",
wantName: "github.com/BurntSushi/toml",
wantVersion: "V1.0.0-Rc1",
wantType: goRequestZip,
wantIsDownload: true,
},
{
name: "pseudo-version",
path: "/example.com/m/@v/v0.0.0-20191109021931-daa7c04131f5.zip",
wantName: "example.com/m",
wantVersion: "v0.0.0-20191109021931-daa7c04131f5",
wantType: goRequestZip,
wantIsDownload: true,
},
{
name: "incompatible version",
path: "/github.com/x/y/@v/v4.1.2+incompatible.zip",
wantName: "github.com/x/y",
wantVersion: "v4.1.2+incompatible",
wantType: goRequestZip,
wantIsDownload: true,
},
{
name: "proxied checksum database traffic",
path: "/sumdb/sum.golang.org/lookup/example.com/m@v1.0.0",
wantType: goRequestSumDB,
},
{
name: "sumdb capability check",
path: "/sumdb/sum.golang.org/supported",
wantType: goRequestSumDB,
},
{
name: "missing marker",
path: "/github.com/x/y",
wantErrContains: "missing /@v/",
},
{
name: "unknown suffix",
path: "/github.com/x/y/@v/v1.2.3.tar",
wantErrContains: "unrecognized go proxy version suffix",
},
{
name: "empty path",
path: "/",
wantErrContains: "empty go proxy URL path",
},
{
name: "version without suffix",
path: "/github.com/x/y/@v/v123",
wantErrContains: "no version suffix",
},
}
parser := goProxyParser{}
for _, tc := range cases {
t.Run(tc.name, func(t *testing.T) {
info, err := parser.ParseURL(tc.path)
if tc.wantErrContains != "" {
require.Error(t, err)
assert.Contains(t, err.Error(), tc.wantErrContains)
return
}
require.NoError(t, err)
assert.Equal(t, tc.wantName, info.GetName())
assert.Equal(t, tc.wantVersion, info.GetVersion())
assert.Equal(t, tc.wantIsDownload, info.IsFileDownload())
goInfo, ok := info.(*goModuleInfo)
require.True(t, ok)
assert.Equal(t, tc.wantType, goInfo.requestType)
})
}
}