Compare commits

...

11 Commits

Author SHA1 Message Date
aweiker dc2e1ca5de Merge pull request 'feat: reject cross-host redirects and HTTPS→HTTP downgrades (#95)' (#111) from review-bot-issue-95 into main
CI / test (push) Successful in 17s
CI / review (anthropic--claude-4.6-sonnet, sonnet, SONNET_REVIEW_TOKEN) (push) Has been skipped
CI / review (gpt-5, gpt, GPT_REVIEW_TOKEN) (push) Has been skipped
CI / review (gpt-5, security, ., rodin/security-patterns, SECURITY_REVIEW.md, SECURITY_REVIEW_TOKEN) (push) Has been skipped
Reviewed-on: #111
Reviewed-by: security-review-bot <10+security-review-bot@noreply.gitea.weiker.me>
Reviewed-by: Aaron Weiker <aaron@weiker.org>
2026-05-13 16:58:49 +00:00
claw 7de6fdd9ec fix: address review feedback on redirect policy
PR Ready Gate / clear-labels (pull_request) Successful in 1s
CI / test (pull_request) Successful in 17s
CI / review (anthropic--claude-4.6-sonnet, sonnet, SONNET_REVIEW_TOKEN) (pull_request) Successful in 31s
CI / review (gpt-5, security, ., rodin/security-patterns, SECURITY_REVIEW.md, SECURITY_REVIEW_TOKEN) (pull_request) Successful in 52s
CI / review (gpt-5, gpt, GPT_REVIEW_TOKEN) (pull_request) Successful in 1m24s
- Replace Unicode arrows (→) with ASCII (->) in error messages and
  comments for log compatibility (gpt-review NITs #19626, #19628)
- Improve guard comment to clarify it exists for testability, not
  runtime safety (sonnet-review NIT #19619)
- Add cross-reference comments noting intentional duplication between
  gitea/client.go and github/client.go (sonnet-review #19618,
  gpt-review #19625, #19627)

Pushed back on:
- internal/ package for dedup: structural overhead not warranted for
  a single ~25-line function
- strings.EqualFold for scheme: Go's url.Parse normalizes schemes to
  lowercase, making case-insensitive comparison unnecessary
2026-05-13 09:28:46 -07:00
claw 1e0959b077 feat: reject cross-host redirects and HTTPS→HTTP downgrades (#95)
CI / test (pull_request) Successful in 16s
CI / review (anthropic--claude-4.6-sonnet, sonnet, SONNET_REVIEW_TOKEN) (pull_request) Successful in 37s
CI / review (gpt-5, gpt, GPT_REVIEW_TOKEN) (pull_request) Successful in 1m19s
CI / review (gpt-5, security, ., rodin/security-patterns, SECURITY_REVIEW.md, SECURITY_REVIEW_TOKEN) (pull_request) Successful in 1m47s
Add defaultCheckRedirect to both GitHub and Gitea clients that rejects:
- HTTPS→HTTP protocol downgrades (prevents plaintext leakage)
- Cross-host redirects entirely (prevents consuming untrusted responses)

Same-host, same-or-upgraded-scheme redirects remain allowed.

Both NewClient constructors wire the policy, and SetHTTPClient(nil)
restores it. Callers providing a non-nil client are responsible for
configuring their own safe redirect policy.

Closes #95
2026-05-13 08:51:36 -07:00
aweiker 67c3db70cb Merge pull request 'feat(github): support HTTP-date format in Retry-After header' (#110) from review-bot-issue-94 into main
CI / test (push) Successful in 18s
CI / review (anthropic--claude-4.6-sonnet, sonnet, SONNET_REVIEW_TOKEN) (push) Has been skipped
CI / review (gpt-5, gpt, GPT_REVIEW_TOKEN) (push) Has been skipped
CI / review (gpt-5, security, ., rodin/security-patterns, SECURITY_REVIEW.md, SECURITY_REVIEW_TOKEN) (push) Has been skipped
Reviewed-on: #110
Reviewed-by: security-review-bot <10+security-review-bot@noreply.gitea.weiker.me>
Reviewed-by: Aaron Weiker <aaron@weiker.org>
2026-05-13 15:34:25 +00:00
aweiker a845ce32eb Merge pull request 'feat(gitea): harden GetPullRequestDiff against unbounded diff size' (#109) from review-bot-issue-92 into main
CI / test (push) Successful in 17s
CI / review (anthropic--claude-4.6-sonnet, sonnet, SONNET_REVIEW_TOKEN) (push) Has been skipped
CI / review (gpt-5, gpt, GPT_REVIEW_TOKEN) (push) Has been skipped
CI / review (gpt-5, security, ., rodin/security-patterns, SECURITY_REVIEW.md, SECURITY_REVIEW_TOKEN) (push) Has been skipped
Reviewed-on: #109
Reviewed-by: security-review-bot <10+security-review-bot@noreply.gitea.weiker.me>
Reviewed-by: Aaron Weiker <aaron@weiker.org>
2026-05-13 14:01:22 +00:00
claw 49d6ca77a3 refactor(gitea): eliminate retry duplication, harden redactURL and MaxInt64 edge case
PR Ready Gate / clear-labels (pull_request) Successful in 2s
CI / test (pull_request) Successful in 17s
CI / review (anthropic--claude-4.6-sonnet, sonnet, SONNET_REVIEW_TOKEN) (pull_request) Successful in 32s
CI / review (gpt-5, security, ., rodin/security-patterns, SECURITY_REVIEW.md, SECURITY_REVIEW_TOKEN) (pull_request) Successful in 1m2s
CI / review (gpt-5, gpt, GPT_REVIEW_TOKEN) (pull_request) Successful in 2m21s
- Extract doGetWithReader to share retry/backoff logic between doGet and
  doGetLimited, eliminating ~60 lines of duplicated code (addresses MINOR
  finding from all reviewers).
- redactURL now strips userinfo credentials (user:pass@host) in addition
  to query parameters (addresses security-review-bot finding).
- GetPullRequestDiff treats MaxDiffSize == math.MaxInt64 as disabled,
  preventing the silent enforcement bypass where the overflow clamp makes
  the size check unreachable (addresses security-review-bot finding).
- Improved error message wording: 'response exceeds N bytes' (NIT fix).
2026-05-13 13:39:37 +00:00
claw 6ebf66aefb fix(gitea): address review feedback on diff size limiting
- Add concurrency safety note to MaxDiffSize field documentation,
  mirroring the existing note on RetryBackoff
- Consolidate six individual test functions into a single table-driven
  test (TestGetPullRequestDiff_SizeLimits) reducing repetition
- Add //nolint:errcheck annotation to test handler w.Write calls
2026-05-13 13:39:37 +00:00
claw 004343d05f fix(gitea): address review findings — clamp overflow, clarify maxSize doc
- Clamp maxBytes+1 to prevent integer overflow to negative when
  maxBytes == math.MaxInt64 (falls back to math.MaxInt64)
- Update MaxDiffSize doc: 'any negative value' disables the limit,
  matching actual behavior of 'maxSize < 0' check
2026-05-13 13:39:37 +00:00
claw 92b84976cf feat(gitea): harden GetPullRequestDiff against unbounded diff size
Add a configurable MaxDiffSize field to Client that limits how much
data GetPullRequestDiff will read into memory. The default is 10 MB
(DefaultMaxDiffSize). When the diff exceeds the limit, ErrDiffTooLarge
is returned, allowing callers to skip position translation gracefully.

Implementation uses io.LimitReader to read maxBytes+1, detecting
overflow without buffering the entire response. Setting MaxDiffSize
to -1 disables the limit entirely.

Closes #92
2026-05-13 13:39:37 +00:00
aweiker 881ce232eb Merge pull request 'docs(deps): update CONVENTIONS.md allowlist for go-yaml' (#108) from review-bot-issue-91 into main
CI / test (push) Successful in 17s
CI / review (anthropic--claude-4.6-sonnet, sonnet, SONNET_REVIEW_TOKEN) (push) Has been skipped
CI / review (gpt-5, gpt, GPT_REVIEW_TOKEN) (push) Has been skipped
CI / review (gpt-5, security, ., rodin/security-patterns, SECURITY_REVIEW.md, SECURITY_REVIEW_TOKEN) (push) Has been skipped
Reviewed-on: #108
Reviewed-by: security-review-bot <10+security-review-bot@noreply.gitea.weiker.me>
Reviewed-by: Aaron Weiker <aaron@weiker.org>
2026-05-13 13:16:40 +00:00
claw bf52fceea0 docs(deps): update CONVENTIONS.md allowlist for go-yaml
CI / test (pull_request) Successful in 20s
CI / review (anthropic--claude-4.6-sonnet, sonnet, SONNET_REVIEW_TOKEN) (pull_request) Successful in 20s
CI / review (gpt-5, security, ., rodin/security-patterns, SECURITY_REVIEW.md, SECURITY_REVIEW_TOKEN) (pull_request) Successful in 48s
CI / review (gpt-5, gpt, GPT_REVIEW_TOKEN) (pull_request) Successful in 51s
Update the approved dependency table to document go-yaml subpackage
usage (ast, parser) and remove the deviation comment now that the
proper allowlist process is being followed.

Closes #91
2026-05-13 02:56:06 -07:00
6 changed files with 495 additions and 22 deletions
+1 -3
View File
@@ -9,7 +9,7 @@
| Package | Use Case | Scope |
|---------|----------|-------|
| `github.com/goccy/go-yaml` | YAML parsing (persona files, config) | production |
| `github.com/goccy/go-yaml` | YAML parsing and AST inspection (subpkgs: `ast`, `parser`) | production |
| `github.com/google/go-cmp` | Test comparisons (`cmp.Diff`) | test only |
**Any import not in this table or the Go standard library is forbidden.**
@@ -21,8 +21,6 @@ To request a new dependency:
2. Requires explicit approval from Aaron
3. After merge, a separate PR may use the package
<!-- Deviation from step 1+3 for go-yaml migration: see #91 for rationale. -->
*Enforcement: `scripts/check-deps.sh` parses this table — update only here.*
## Error Handling
+128 -16
View File
@@ -11,6 +11,7 @@ import (
"fmt"
"io"
"log/slog"
"math"
"net"
"net/http"
"net/url"
@@ -47,6 +48,12 @@ func IsServerError(err error) bool {
return errors.As(err, &apiErr) && apiErr.StatusCode >= 500 && apiErr.StatusCode < 600
}
// DefaultMaxDiffSize is the default maximum diff size in bytes (10 MB).
const DefaultMaxDiffSize = 10 * 1024 * 1024
// ErrDiffTooLarge is returned when a PR diff exceeds the configured MaxDiffSize.
var ErrDiffTooLarge = errors.New("diff size exceeds maximum allowed size")
// Client interacts with the Gitea API.
// A Client is safe for concurrent use by multiple goroutines.
type Client struct {
@@ -61,6 +68,42 @@ type Client struct {
// This field must be configured before the first request is made.
// Modifying it while requests are in flight is not safe.
RetryBackoff []time.Duration
// MaxDiffSize is the maximum number of bytes allowed when fetching a PR diff.
// If zero, defaults to DefaultMaxDiffSize (10 MB). Set to any negative value
// (or math.MaxInt64) to disable the limit.
//
// This field must be configured before the first request is made.
// Modifying it while requests are in flight is not safe.
MaxDiffSize int64
}
// defaultCheckRedirect is the redirect policy used by NewClient.
// NOTE: This function is intentionally duplicated in github/client.go (and vice versa)
// because the packages are separate. Changes here must be mirrored there.
// It rejects HTTPS->HTTP protocol downgrades (to prevent plaintext leakage)
// and cross-host redirects (to prevent following responses from untrusted
// endpoints). Same-host, same-or-upgraded-scheme redirects are allowed.
func defaultCheckRedirect(req *http.Request, via []*http.Request) error {
if len(via) >= 10 {
return fmt.Errorf("stopped after 10 redirects")
}
// Guard for direct invocation in tests and any future callers;
// net/http guarantees len(via) >= 1 during actual redirects.
if len(via) == 0 {
return nil
}
prev := via[len(via)-1]
// Reject protocol downgrade: HTTPS->HTTP leaks request metadata over plaintext.
if prev.URL.Scheme == "https" && req.URL.Scheme == "http" {
return fmt.Errorf("refusing redirect: HTTPS to HTTP downgrade (%s -> %s)", prev.URL.Host, req.URL.Host)
}
// Reject cross-host redirect entirely to avoid consuming responses
// from untrusted endpoints.
if req.URL.Host != prev.URL.Host {
return fmt.Errorf("refusing redirect: cross-host (%s -> %s)", prev.URL.Host, req.URL.Host)
}
return nil
}
// NewClient creates a new Gitea API client.
@@ -68,13 +111,30 @@ func NewClient(baseURL, token string) *Client {
return &Client{
baseURL: strings.TrimRight(baseURL, "/"),
token: token,
http: &http.Client{Timeout: 30 * time.Second},
http: &http.Client{
Timeout: 30 * time.Second,
CheckRedirect: defaultCheckRedirect,
},
}
}
// SetHTTPClient sets the underlying HTTP client used for requests.
// This is intended for testing to inject mock transports.
// This is intended for test setup only to inject mock transports; it must be
// called before any goroutines issue requests.
//
// Passing nil restores the default client (30s timeout + redirect-rejecting
// CheckRedirect policy matching NewClient).
//
// Callers providing a non-nil client are responsible for configuring a safe
// CheckRedirect policy. Without one, the default net/http behavior will follow
// redirects and may forward the Authorization header to untrusted hosts.
func (c *Client) SetHTTPClient(hc *http.Client) {
if hc == nil {
hc = &http.Client{
Timeout: 30 * time.Second,
CheckRedirect: defaultCheckRedirect,
}
}
c.http = hc
}
@@ -125,9 +185,28 @@ func (c *Client) GetPullRequest(ctx context.Context, owner, repo string, number
}
// GetPullRequestDiff fetches the unified diff for a PR.
// It enforces MaxDiffSize to prevent unbounded memory allocation.
// Returns ErrDiffTooLarge if the diff exceeds the configured limit.
func (c *Client) GetPullRequestDiff(ctx context.Context, owner, repo string, number int) (string, error) {
reqURL := fmt.Sprintf("%s/api/v1/repos/%s/%s/pulls/%d.diff", c.baseURL, url.PathEscape(owner), url.PathEscape(repo), number)
body, err := c.doGet(ctx, reqURL)
maxSize := c.MaxDiffSize
if maxSize == 0 {
maxSize = DefaultMaxDiffSize
}
// When the limit is disabled (negative) or set to math.MaxInt64 (which
// would overflow the +1 detection and silently disable enforcement),
// use the standard unlimited doGet path.
if maxSize < 0 || maxSize == math.MaxInt64 {
body, err := c.doGet(ctx, reqURL)
if err != nil {
return "", fmt.Errorf("fetch diff: %w", err)
}
return string(body), nil
}
body, err := c.doGetLimited(ctx, reqURL, maxSize)
if err != nil {
return "", fmt.Errorf("fetch diff: %w", err)
}
@@ -292,9 +371,9 @@ func isRetriableSyscallError(err error) bool {
return true
}
// redactURL strips query parameters from a URL for safe logging.
// This prevents accidental exposure of sensitive data that future callers
// might pass via query strings.
// redactURL strips query parameters and userinfo credentials from a URL for
// safe logging. This prevents accidental exposure of sensitive data (tokens in
// query strings, or user:pass in the authority) in log output.
func redactURL(rawURL string) string {
parsed, err := url.Parse(rawURL)
if err != nil {
@@ -302,6 +381,9 @@ func redactURL(rawURL string) string {
// potentially logging something sensitive.
return "[invalid URL]"
}
if parsed.User != nil {
parsed.User = url.User("REDACTED")
}
if parsed.RawQuery != "" {
parsed.RawQuery = "[redacted]"
}
@@ -322,10 +404,12 @@ func sanitizeErrorForLog(err error) string {
return err.Error()
}
// doGet performs an HTTP GET request with retry on 5xx errors and temporary
// network errors. Retries up to 3 times with exponential backoff (1s, 2s delays
// by default; configurable via Client.RetryBackoff for testing).
func (c *Client) doGet(ctx context.Context, reqURL string) ([]byte, error) {
// doGetWithReader performs an HTTP GET request with retry on 5xx errors and
// temporary network errors. Retries up to 3 times with exponential backoff
// (1s, 2s delays by default; configurable via Client.RetryBackoff for testing).
// The readBody function is called with the response body on success (2xx) and
// is responsible for reading and closing it.
func (c *Client) doGetWithReader(ctx context.Context, reqURL string, readBody func(io.ReadCloser) ([]byte, error)) ([]byte, error) {
const maxAttempts = 3
// backoff[i] is the delay before attempt i+1 (i.e., after attempt i fails).
// First attempt (i=0) has no delay; retries wait 1s then 2s by default.
@@ -390,12 +474,7 @@ func (c *Client) doGet(ctx context.Context, reqURL string) ([]byte, error) {
return nil, lastErr
}
if resp.StatusCode >= 200 && resp.StatusCode < 300 {
body, err := io.ReadAll(resp.Body)
resp.Body.Close()
if err != nil {
return nil, err
}
return body, nil
return readBody(resp.Body)
}
// Error path: limit how much we read from potentially malicious server
@@ -413,6 +492,39 @@ func (c *Client) doGet(ctx context.Context, reqURL string) ([]byte, error) {
return nil, lastErr
}
// doGet performs an HTTP GET request with retry, reading the full response body.
func (c *Client) doGet(ctx context.Context, reqURL string) ([]byte, error) {
return c.doGetWithReader(ctx, reqURL, func(body io.ReadCloser) ([]byte, error) {
defer body.Close()
return io.ReadAll(body)
})
}
// doGetLimited performs an HTTP GET request with retry but enforces a maximum
// response body size. Returns ErrDiffTooLarge if the response exceeds maxBytes.
// It reads maxBytes+1 (clamped to avoid overflow) to detect truncation without
// buffering the entire body.
func (c *Client) doGetLimited(ctx context.Context, reqURL string, maxBytes int64) ([]byte, error) {
return c.doGetWithReader(ctx, reqURL, func(body io.ReadCloser) ([]byte, error) {
defer body.Close()
// Read up to maxBytes+1 to detect overflow.
// Clamp to prevent integer overflow when maxBytes == math.MaxInt64.
limitBytes := maxBytes + 1
if limitBytes <= 0 {
limitBytes = math.MaxInt64
}
limited := io.LimitReader(body, limitBytes)
data, err := io.ReadAll(limited)
if err != nil {
return nil, err
}
if int64(len(data)) > maxBytes {
return nil, fmt.Errorf("%w: response exceeds %d bytes", ErrDiffTooLarge, maxBytes)
}
return data, nil
})
}
// escapePath escapes each segment of a relative file path for use in URLs.
// Slashes are preserved as path separators; other special characters are escaped.
// Input should be a relative path (no leading slash). Already-encoded segments
+117
View File
@@ -9,6 +9,7 @@ import (
"net"
"net/http"
"net/http/httptest"
"net/url"
"strings"
"sync/atomic"
"syscall"
@@ -1092,6 +1093,21 @@ func TestRedactURL(t *testing.T) {
input: "",
want: "",
},
{
name: "with userinfo - redacts credentials",
input: "https://admin:secret@gitea.example.com/api/v1/repos",
want: "https://REDACTED@gitea.example.com/api/v1/repos",
},
{
name: "with userinfo and query params",
input: "https://user:pass@example.com/path?token=abc",
want: "https://REDACTED@example.com/path?[redacted]",
},
{
name: "username only - no password",
input: "https://user@example.com/path",
want: "https://REDACTED@example.com/path",
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
@@ -1144,3 +1160,104 @@ func TestSanitizeErrorForLog(t *testing.T) {
})
}
}
func TestNewClient_HasCheckRedirect(t *testing.T) {
c := NewClient("https://gitea.example.com", "token")
if c.http.CheckRedirect == nil {
t.Fatal("expected CheckRedirect to be set")
}
}
func TestDefaultCheckRedirect_RejectsHTTPSToHTTP(t *testing.T) {
prev := &http.Request{URL: &url.URL{Scheme: "https", Host: "gitea.example.com", Path: "/foo"}}
req := &http.Request{
URL: &url.URL{Scheme: "http", Host: "gitea.example.com", Path: "/foo"},
Header: http.Header{"Authorization": []string{"token abc"}},
}
err := defaultCheckRedirect(req, []*http.Request{prev})
if err == nil {
t.Fatal("expected error on HTTPS->HTTP redirect")
}
if !strings.Contains(err.Error(), "HTTPS to HTTP downgrade") {
t.Errorf("unexpected error message: %v", err)
}
}
func TestDefaultCheckRedirect_RejectsCrossHost(t *testing.T) {
prev := &http.Request{URL: &url.URL{Scheme: "https", Host: "gitea.example.com", Path: "/foo"}}
req := &http.Request{
URL: &url.URL{Scheme: "https", Host: "cdn.example.com", Path: "/bar"},
Header: http.Header{"Authorization": []string{"token abc"}},
}
err := defaultCheckRedirect(req, []*http.Request{prev})
if err == nil {
t.Fatal("expected error on cross-host redirect")
}
if !strings.Contains(err.Error(), "cross-host") {
t.Errorf("unexpected error message: %v", err)
}
}
func TestDefaultCheckRedirect_AllowsSameHost(t *testing.T) {
prev := &http.Request{URL: &url.URL{Scheme: "https", Host: "gitea.example.com", Path: "/foo"}}
req := &http.Request{
URL: &url.URL{Scheme: "https", Host: "gitea.example.com", Path: "/bar"},
Header: http.Header{"Authorization": []string{"token abc"}},
}
err := defaultCheckRedirect(req, []*http.Request{prev})
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
if auth := req.Header.Get("Authorization"); auth != "token abc" {
t.Errorf("expected Authorization to be preserved, got %q", auth)
}
}
func TestDefaultCheckRedirect_AllowsSameHostHTTPToHTTP(t *testing.T) {
prev := &http.Request{URL: &url.URL{Scheme: "http", Host: "localhost:3000", Path: "/foo"}}
req := &http.Request{
URL: &url.URL{Scheme: "http", Host: "localhost:3000", Path: "/bar"},
Header: http.Header{},
}
err := defaultCheckRedirect(req, []*http.Request{prev})
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
}
func TestDefaultCheckRedirect_RejectsTooManyRedirects(t *testing.T) {
via := make([]*http.Request, 10)
for i := range via {
via[i] = &http.Request{URL: &url.URL{Scheme: "https", Host: "gitea.example.com", Path: "/"}}
}
req := &http.Request{URL: &url.URL{Scheme: "https", Host: "gitea.example.com", Path: "/final"}}
err := defaultCheckRedirect(req, via)
if err == nil {
t.Fatal("expected error after 10 redirects")
}
if !strings.Contains(err.Error(), "10 redirects") {
t.Errorf("unexpected error message: %v", err)
}
}
func TestDefaultCheckRedirect_EmptyViaAllowed(t *testing.T) {
req := &http.Request{URL: &url.URL{Scheme: "https", Host: "gitea.example.com", Path: "/foo"}}
err := defaultCheckRedirect(req, nil)
if err != nil {
t.Fatalf("unexpected error with empty via: %v", err)
}
}
func TestSetHTTPClient_NilRestoresDefault(t *testing.T) {
c := NewClient("https://gitea.example.com", "token")
c.SetHTTPClient(nil)
if c.http == nil {
t.Fatal("expected non-nil http client after SetHTTPClient(nil)")
}
if c.http.Timeout != 30*time.Second {
t.Errorf("expected 30s timeout, got %v", c.http.Timeout)
}
if c.http.CheckRedirect == nil {
t.Fatal("expected CheckRedirect policy after SetHTTPClient(nil)")
}
}
+97
View File
@@ -0,0 +1,97 @@
package gitea
import (
"context"
"errors"
"math"
"net/http"
"net/http/httptest"
"strings"
"testing"
"time"
)
func TestGetPullRequestDiff_SizeLimits(t *testing.T) {
tests := []struct {
name string
diff string
maxDiffSize int64
wantErr error
wantDiff string
}{
{
name: "exceeds max size",
diff: strings.Repeat("+ added line\n", 1000), // ~13 KB
maxDiffSize: 100,
wantErr: ErrDiffTooLarge,
},
{
name: "within max size",
diff: "diff --git a/f.go b/f.go\n--- a/f.go\n+++ b/f.go\n@@ -1 +1 @@\n-old\n+new\n",
maxDiffSize: 1024,
wantDiff: "diff --git a/f.go b/f.go\n--- a/f.go\n+++ b/f.go\n@@ -1 +1 @@\n-old\n+new\n",
},
{
name: "exactly at limit",
diff: strings.Repeat("x", 50),
maxDiffSize: 50,
wantDiff: strings.Repeat("x", 50),
},
{
name: "one byte over limit",
diff: strings.Repeat("x", 51),
maxDiffSize: 50,
wantErr: ErrDiffTooLarge,
},
{
name: "disabled limit",
diff: strings.Repeat("x", 10000),
maxDiffSize: -1,
wantDiff: strings.Repeat("x", 10000),
},
{
name: "math.MaxInt64 treated as disabled",
diff: strings.Repeat("x", 10000),
maxDiffSize: math.MaxInt64,
wantDiff: strings.Repeat("x", 10000),
},
{
name: "default limit",
diff: "diff content",
maxDiffSize: 0, // zero means use DefaultMaxDiffSize
wantDiff: "diff content",
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
w.Write([]byte(tt.diff)) //nolint:errcheck // test handler
}))
defer server.Close()
client := NewClient(server.URL, "test-token")
client.MaxDiffSize = tt.maxDiffSize
client.RetryBackoff = []time.Duration{}
got, err := client.GetPullRequestDiff(context.Background(), "owner", "repo", 1)
if tt.wantErr != nil {
if err == nil {
t.Fatal("expected error, got nil")
}
if !errors.Is(err, tt.wantErr) {
t.Errorf("expected %v, got: %v", tt.wantErr, err)
}
return
}
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
if got != tt.wantDiff {
t.Errorf("diff mismatch: got length %d, want length %d", len(got), len(tt.wantDiff))
}
})
}
}
+48 -3
View File
@@ -107,6 +107,34 @@ type Client struct {
now func() time.Time
}
// defaultCheckRedirect is the redirect policy used by NewClient.
// NOTE: This function is intentionally duplicated in gitea/client.go (and vice versa)
// because the packages are separate. Changes here must be mirrored there.
// It rejects HTTPS->HTTP protocol downgrades (to prevent plaintext leakage)
// and cross-host redirects (to prevent following responses from untrusted
// endpoints). Same-host, same-or-upgraded-scheme redirects are allowed.
func defaultCheckRedirect(req *http.Request, via []*http.Request) error {
if len(via) >= 10 {
return fmt.Errorf("stopped after 10 redirects")
}
// Guard for direct invocation in tests and any future callers;
// net/http guarantees len(via) >= 1 during actual redirects.
if len(via) == 0 {
return nil
}
prev := via[len(via)-1]
// Reject protocol downgrade: HTTPS->HTTP leaks request metadata over plaintext.
if prev.URL.Scheme == "https" && req.URL.Scheme == "http" {
return fmt.Errorf("refusing redirect: HTTPS to HTTP downgrade (%s -> %s)", prev.URL.Host, req.URL.Host)
}
// Reject cross-host redirect entirely to avoid consuming responses
// from untrusted endpoints.
if req.URL.Host != prev.URL.Host {
return fmt.Errorf("refusing redirect: cross-host (%s -> %s)", prev.URL.Host, req.URL.Host)
}
return nil
}
// NewClient creates a new GitHub API client.
// If baseURL is empty, it defaults to https://api.github.com.
// For GitHub Enterprise, pass the API base URL (e.g. https://github.concur.com/api/v3).
@@ -117,14 +145,31 @@ func NewClient(token, baseURL string) *Client {
return &Client{
baseURL: strings.TrimRight(baseURL, "/"),
token: token,
httpClient: &http.Client{Timeout: 30 * time.Second},
now: time.Now,
httpClient: &http.Client{
Timeout: 30 * time.Second,
CheckRedirect: defaultCheckRedirect,
},
now: time.Now,
}
}
// SetHTTPClient sets the underlying HTTP client used for requests.
// This is intended for testing to inject mock transports.
// This is intended for test setup only to inject mock transports; it must be
// called before any goroutines issue requests.
//
// Passing nil restores the default client (30s timeout + redirect-rejecting
// CheckRedirect policy matching NewClient).
//
// Callers providing a non-nil client are responsible for configuring a safe
// CheckRedirect policy. Without one, the default net/http behavior will follow
// redirects and may forward the Authorization header to untrusted hosts.
func (c *Client) SetHTTPClient(hc *http.Client) {
if hc == nil {
hc = &http.Client{
Timeout: 30 * time.Second,
CheckRedirect: defaultCheckRedirect,
}
}
c.httpClient = hc
}
+104
View File
@@ -5,6 +5,8 @@ import (
"errors"
"net/http"
"net/http/httptest"
"net/url"
"strings"
"testing"
"time"
)
@@ -407,3 +409,105 @@ func TestAPIError_Error_NewlineSanitized(t *testing.T) {
}
}
}
func TestNewClient_HasCheckRedirect(t *testing.T) {
c := NewClient("secret-token", "https://api.github.com")
if c.httpClient.CheckRedirect == nil {
t.Fatal("expected CheckRedirect to be set")
}
}
func TestDefaultCheckRedirect_RejectsHTTPSToHTTP(t *testing.T) {
prev := &http.Request{URL: &url.URL{Scheme: "https", Host: "api.github.com", Path: "/foo"}}
req := &http.Request{
URL: &url.URL{Scheme: "http", Host: "api.github.com", Path: "/foo"},
Header: http.Header{"Authorization": []string{"Bearer token"}},
}
err := defaultCheckRedirect(req, []*http.Request{prev})
if err == nil {
t.Fatal("expected error on HTTPS->HTTP redirect")
}
if !strings.Contains(err.Error(), "HTTPS to HTTP downgrade") {
t.Errorf("unexpected error message: %v", err)
}
}
func TestDefaultCheckRedirect_RejectsCrossHost(t *testing.T) {
prev := &http.Request{URL: &url.URL{Scheme: "https", Host: "api.github.com", Path: "/foo"}}
req := &http.Request{
URL: &url.URL{Scheme: "https", Host: "objects.githubusercontent.com", Path: "/bar"},
Header: http.Header{"Authorization": []string{"Bearer token"}},
}
err := defaultCheckRedirect(req, []*http.Request{prev})
if err == nil {
t.Fatal("expected error on cross-host redirect")
}
if !strings.Contains(err.Error(), "cross-host") {
t.Errorf("unexpected error message: %v", err)
}
}
func TestDefaultCheckRedirect_AllowsSameHost(t *testing.T) {
prev := &http.Request{URL: &url.URL{Scheme: "https", Host: "api.github.com", Path: "/foo"}}
req := &http.Request{
URL: &url.URL{Scheme: "https", Host: "api.github.com", Path: "/bar"},
Header: http.Header{"Authorization": []string{"Bearer token"}},
}
err := defaultCheckRedirect(req, []*http.Request{prev})
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
// Auth should be preserved on same-host redirect
if auth := req.Header.Get("Authorization"); auth != "Bearer token" {
t.Errorf("expected Authorization to be preserved, got %q", auth)
}
}
func TestDefaultCheckRedirect_AllowsSameHostHTTPToHTTP(t *testing.T) {
prev := &http.Request{URL: &url.URL{Scheme: "http", Host: "localhost:8080", Path: "/foo"}}
req := &http.Request{
URL: &url.URL{Scheme: "http", Host: "localhost:8080", Path: "/bar"},
Header: http.Header{},
}
err := defaultCheckRedirect(req, []*http.Request{prev})
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
}
func TestDefaultCheckRedirect_RejectsTooManyRedirects(t *testing.T) {
via := make([]*http.Request, 10)
for i := range via {
via[i] = &http.Request{URL: &url.URL{Scheme: "https", Host: "api.github.com", Path: "/"}}
}
req := &http.Request{URL: &url.URL{Scheme: "https", Host: "api.github.com", Path: "/final"}}
err := defaultCheckRedirect(req, via)
if err == nil {
t.Fatal("expected error after 10 redirects")
}
if !strings.Contains(err.Error(), "10 redirects") {
t.Errorf("unexpected error message: %v", err)
}
}
func TestDefaultCheckRedirect_EmptyViaAllowed(t *testing.T) {
req := &http.Request{URL: &url.URL{Scheme: "https", Host: "api.github.com", Path: "/foo"}}
err := defaultCheckRedirect(req, nil)
if err != nil {
t.Fatalf("unexpected error with empty via: %v", err)
}
}
func TestSetHTTPClient_NilRestoresDefault(t *testing.T) {
c := NewClient("token", "https://api.github.com")
c.SetHTTPClient(nil)
if c.httpClient == nil {
t.Fatal("expected non-nil httpClient after SetHTTPClient(nil)")
}
if c.httpClient.Timeout != 30*time.Second {
t.Errorf("expected 30s timeout, got %v", c.httpClient.Timeout)
}
if c.httpClient.CheckRedirect == nil {
t.Fatal("expected CheckRedirect policy after SetHTTPClient(nil)")
}
}