Compare commits

..

6 Commits

Author SHA1 Message Date
Rodin 24d4dcb751 chore(#130): mark self-review findings as addressed in TODO.md 2026-05-15 08:39:06 +00:00
Rodin f0ba8fe004 refactor(#130): move IsBlockedIP to internal/netutil to remove gitea import in validateurl.go
validateurl.go is VCS-generic but imported gitea.IsBlockedIP, creating an
unexpected generic→Gitea-specific dependency. Extract IsBlockedIP and its
CIDR list to internal/netutil/ipcheck.go (a neutral shared package).

- gitea/ipcheck.go becomes a thin forwarding wrapper (preserves API compat
  for callers within the gitea package)
- gitea/ipcheck_test.go replaced with a forwarding smoke test; full coverage
  moves to internal/netutil/ipcheck_test.go
- validateurl.go now imports internal/netutil directly
2026-05-15 08:38:48 +00:00
Rodin c5261b9b7f refactor(#130): rename vcsReviewComment.NewPosition to NewLine with clearer semantics
The field was named NewPosition with a misleading comment 'Gitea: absolute
line; GitHub: diff hunk position'. In reality both adapters use it as an
absolute new-file line number (Gitea maps it to new_position, GitHub maps it
to Line+Side:RIGHT). Rename to NewLine to match actual semantics and update
comments to explain per-adapter mapping.
2026-05-15 08:36:55 +00:00
Rodin 9a1410cc1a docs(#130): fix README CLI example and env var table for VCS-agnostic usage
- CLI example used $GITEA_TOKEN which is not an actual env var; rename to
  $REVIEWER_TOKEN (the correct env var the binary reads)
- Env var table referenced GITEA_REPO without noting GitHub support; add
  a note and include VCS_TYPE row so users know they can override detection
2026-05-15 08:36:16 +00:00
Rodin 5e20dba3a4 fix(#130): pass VCS_TYPE env var from action.yml Run review step
The binary detects VCS type from VCS_TYPE env var, but action.yml did not
pass it to the Run review step. This caused the binary to fall back to a
URL heuristic (github.com substring), which misclassifies GitHub Enterprise
Server hosts whose URL does not contain 'github'.

The 'Determine version' step already outputs vcs_type — wire it through to
the Run review env block so explicit VCS_TYPE always takes precedence.
2026-05-15 08:35:52 +00:00
claw d545abe392 fix(#130): enforce HTTPS scheme consistently in GitHub client write methods
PR Ready Gate / clear-labels (pull_request) Successful in 2s
CI / test (pull_request) Successful in 16s
CI / review (anthropic--claude-4.6-sonnet, sonnet, SONNET_REVIEW_TOKEN) (pull_request) Successful in 40s
CI / review (gpt-5, security, ., rodin/security-patterns, SECURITY_REVIEW.md, SECURITY_REVIEW_TOKEN) (pull_request) Successful in 1m38s
CI / review (gpt-5, gpt, GPT_REVIEW_TOKEN) (pull_request) Successful in 1m49s
PostReview, DeleteReview, and RequestReviewer were calling c.httpClient.Do
directly, bypassing the scheme check in doRequest that rejects http:// URLs
unless AllowInsecureHTTP is explicitly enabled.

Introduce doRequestWithBody(ctx, method, url, body) with the same HTTPS guard,
and refactor all three write methods to use it. This ensures tokens are never
sent over plaintext regardless of which API path is exercised.

Add scheme validation tests for each method.
2026-05-14 14:11:14 -07:00
12 changed files with 392 additions and 277 deletions
+1
View File
@@ -476,6 +476,7 @@ runs:
shell: bash shell: bash
env: env:
VCS_URL: ${{ steps.version.outputs.server_url }} VCS_URL: ${{ steps.version.outputs.server_url }}
VCS_TYPE: ${{ steps.version.outputs.vcs_type }}
GITEA_REPO: ${{ inputs.repo || github.repository }} GITEA_REPO: ${{ inputs.repo || github.repository }}
PR_NUMBER: ${{ inputs.pr-number || github.event.pull_request.number }} PR_NUMBER: ${{ inputs.pr-number || github.event.pull_request.number }}
REVIEWER_TOKEN: ${{ inputs.reviewer-token }} REVIEWER_TOKEN: ${{ inputs.reviewer-token }}
+3 -2
View File
@@ -285,7 +285,7 @@ review-bot \
--vcs-url https://gitea.example.com \ --vcs-url https://gitea.example.com \
--repo owner/name \ --repo owner/name \
--pr 42 \ --pr 42 \
--reviewer-token "$GITEA_TOKEN" \ --reviewer-token "$REVIEWER_TOKEN" \
--reviewer-name "code-review" \ --reviewer-name "code-review" \
--llm-base-url https://api.openai.com/v1 \ --llm-base-url https://api.openai.com/v1 \
--llm-api-key "$OPENAI_API_KEY" \ --llm-api-key "$OPENAI_API_KEY" \
@@ -300,7 +300,8 @@ All flags have environment variable equivalents:
| Flag | Env Var | | Flag | Env Var |
|------|---------| |------|---------|
| `--vcs-url` | `VCS_URL` (fallback: `GITEA_URL`) | | `--vcs-url` | `VCS_URL` (fallback: `GITEA_URL`) |
| `--repo` | `GITEA_REPO` | | `--vcs-type` | `VCS_TYPE` (auto-detected from URL if not set; `gitea` or `github`) |
| `--repo` | `GITEA_REPO` (also accepted: set `GITEA_REPO` for Gitea; VCS-agnostic `REPO` coming) |
| `--pr` | `PR_NUMBER` | | `--pr` | `PR_NUMBER` |
| `--reviewer-token` | `REVIEWER_TOKEN` | | `--reviewer-token` | `REVIEWER_TOKEN` |
| `--reviewer-name` | `REVIEWER_NAME` | | `--reviewer-name` | `REVIEWER_NAME` |
+9
View File
@@ -77,3 +77,12 @@ All old worktrees cleaned up. Ready for new issue work.
- **Cron ID:** 5342ac81-4bbc-4e4c-a123-347a7788d50c - **Cron ID:** 5342ac81-4bbc-4e4c-a123-347a7788d50c
- **Scheduled:** Every 4 hours - **Scheduled:** Every 4 hours
- **Last health check:** 2026-05-14 20:10 UTC (✅ all healthy) - **Last health check:** 2026-05-14 20:10 UTC (✅ all healthy)
## Self-Review: review-bot-issue-130-work — 2026-05-14
### Verdict: NEEDS_WORK
- [x] [completeness] `VCS_TYPE` env var is detected in `main.go` but **not passed from `action.yml`'s `Run review` step** to the binary. The binary falls back to a URL heuristic (`github.com` / `github.concur.com` substrings), which will silently misclassify any GitHub Enterprise Server whose hostname does not contain `github`. The action already has `vcs_type` as a step output — it should be passed as `VCS_TYPE: ${{ steps.version.outputs.vcs_type }}` in the `Run review` env block.
- [x] [completeness] `README.md` still references `GITEA_REPO` and `$GITEA_TOKEN` in the CLI example (line ~288) and the env var table (line ~303). Now that the binary supports GitHub too, these should be renamed to VCS-agnostic names (`VCS_REPO`/`REPO`, `VCS_TOKEN`/`REVIEWER_TOKEN`) or at minimum the env var table should note GitHub support.
- [x] [fit] `vcsReviewComment.NewPosition` field comment says "Gitea: absolute line; GitHub: diff hunk position" but in `githubVCSAdapter.PostReview` it is actually mapped to `Line` (absolute line) + `Side: "RIGHT"`, not to `Position` (hunk position). The field name and comment are now slightly misleading — the field means "new line number" for Gitea and is repurposed as "absolute line" for GitHub. Consider renaming to `Line` or `NewLine` with a clearer comment explaining the per-backend semantics.
- [x] [instinct] `validateurl.go` imports `gitea` package for `gitea.IsBlockedIP` — this creates a dependency from the GitHub-generic `validateurl.go` on the Gitea-specific package. The `IsBlockedIP` function is a general networking utility. Consider moving it to a shared `internal/netutil` package (or keeping it in `gitea` and accepting the coupling), but the current import is an unexpected relationship.
+1 -1
View File
@@ -466,7 +466,7 @@ func main() {
if f.File != "" && f.Line > 0 && diffRanges.Contains(f.File, f.Line) { if f.File != "" && f.Line > 0 && diffRanges.Contains(f.File, f.Line) {
inlineComments = append(inlineComments, vcsReviewComment{ inlineComments = append(inlineComments, vcsReviewComment{
Path: f.File, Path: f.File,
NewPosition: int64(f.Line), NewLine: int64(f.Line),
Body: fmt.Sprintf("**[%s]** %s", f.Severity, f.Finding), Body: fmt.Sprintf("**[%s]** %s", f.Severity, f.Finding),
}) })
} }
+2 -2
View File
@@ -9,7 +9,7 @@ import (
"strings" "strings"
"time" "time"
"gitea.weiker.me/rodin/review-bot/gitea" "gitea.weiker.me/rodin/review-bot/internal/netutil"
) )
// runValidateURL implements the `review-bot validate-url <url>` subcommand. // runValidateURL implements the `review-bot validate-url <url>` subcommand.
@@ -114,7 +114,7 @@ func validateURL(rawURL string) error {
} }
for _, a := range addrs { for _, a := range addrs {
if gitea.IsBlockedIP(a.IP) { if netutil.IsBlockedIP(a.IP) {
return &validateError{ return &validateError{
code: 1, code: 1,
message: fmt.Sprintf("blocked: %q resolves to private/reserved IP %s", host, a.IP), message: fmt.Sprintf("blocked: %q resolves to private/reserved IP %s", host, a.IP),
+5 -7
View File
@@ -84,7 +84,7 @@ type vcsCommitStatus struct {
// vcsReviewComment is an inline review comment. // vcsReviewComment is an inline review comment.
type vcsReviewComment struct { type vcsReviewComment struct {
Path string Path string
NewPosition int64 // Gitea: absolute line; GitHub: diff hunk position NewLine int64 // absolute line number on the new (right) side of the diff, used by both Gitea and GitHub adapters
Body string Body string
} }
@@ -176,7 +176,7 @@ func (a *giteaVCSAdapter) GetAllFilesInPath(ctx context.Context, owner, repo, pa
func (a *giteaVCSAdapter) PostReview(ctx context.Context, owner, repo string, number int, event, body, commitID string, comments []vcsReviewComment) (*vcsReview, error) { func (a *giteaVCSAdapter) PostReview(ctx context.Context, owner, repo string, number int, event, body, commitID string, comments []vcsReviewComment) (*vcsReview, error) {
gc := make([]gitea.ReviewComment, len(comments)) gc := make([]gitea.ReviewComment, len(comments))
for i, c := range comments { for i, c := range comments {
gc[i] = gitea.ReviewComment{Path: c.Path, NewPosition: c.NewPosition, Body: c.Body} gc[i] = gitea.ReviewComment{Path: c.Path, NewPosition: c.NewLine, Body: c.Body}
} }
r, err := a.c.PostReview(ctx, owner, repo, number, event, body, commitID, gc) r, err := a.c.PostReview(ctx, owner, repo, number, event, body, commitID, gc)
if err != nil { if err != nil {
@@ -311,14 +311,12 @@ func (a *githubVCSAdapter) GetAllFilesInPath(ctx context.Context, owner, repo, p
func (a *githubVCSAdapter) PostReview(ctx context.Context, owner, repo string, number int, event, body, commitID string, comments []vcsReviewComment) (*vcsReview, error) { func (a *githubVCSAdapter) PostReview(ctx context.Context, owner, repo string, number int, event, body, commitID string, comments []vcsReviewComment) (*vcsReview, error) {
gc := make([]github.ReviewComment, len(comments)) gc := make([]github.ReviewComment, len(comments))
for i, c := range comments { for i, c := range comments {
// GitHub inline comments use diff hunk "position", not absolute line numbers. // GitHub inline comments use Line+Side (absolute line on the RIGHT side).
// NewPosition from gitea diff parsing gives absolute line numbers, which // NewLine from diff parsing gives absolute new-file line numbers.
// will not match GitHub's position values. For initial GitHub support, we
// attach comments with Line+Side (absolute line on the RIGHT side) instead.
// Comments that cannot be mapped will be omitted (GitHub rejects invalid positions). // Comments that cannot be mapped will be omitted (GitHub rejects invalid positions).
gc[i] = github.ReviewComment{ gc[i] = github.ReviewComment{
Path: c.Path, Path: c.Path,
Line: c.NewPosition, Line: c.NewLine,
Side: "RIGHT", Side: "RIGHT",
Body: c.Body, Body: c.Body,
} }
+12 -81
View File
@@ -1,91 +1,22 @@
// Package gitea provides a client for the Gitea API. // Package gitea provides a client for the Gitea API.
// ipcheck.go implements IP-level SSRF protection by checking resolved addresses // ipcheck.go re-exports the IsBlockedIP function from internal/netutil for use
// against known blocked CIDR ranges (RFC1918, loopback, link-local, etc.). // by this package's safe dialer (client.go) and for backward compatibility with
// any callers that previously imported it from here.
//
// The implementation has moved to internal/netutil so it can be shared with the
// validate-url subcommand (cmd/review-bot/validateurl.go) without creating a
// dependency from VCS-generic code on the Gitea-specific package.
package gitea package gitea
import ( import (
"fmt"
"net" "net"
"gitea.weiker.me/rodin/review-bot/internal/netutil"
) )
// blockedCIDRStrings is the canonical list of CIDR strings that should never
// be contacted by review-bot. See IsBlockedIP for the full list of covered
// address families.
//
// These are hard-coded literals: any parse failure is a programming error.
// Validity is verified by TestBlockedCIDRsValid in ipcheck_test.go.
var blockedCIDRStrings = []string{
// IPv4 loopback
"127.0.0.0/8",
// IPv4 unspecified / "this network"
"0.0.0.0/8",
// RFC1918 private ranges
"10.0.0.0/8",
"172.16.0.0/12",
"192.168.0.0/16",
// IPv4 link-local (APIPA, also used by AWS instance metadata 169.254.169.254)
"169.254.0.0/16",
// IPv4 shared address space (RFC6598, carrier-grade NAT)
"100.64.0.0/10",
// IPv4 multicast
"224.0.0.0/4",
// IPv4 reserved / broadcast
"240.0.0.0/4",
// IPv6 loopback
"::1/128",
// IPv6 unspecified
"::/128",
// IPv6 link-local
"fe80::/10",
// IPv6 unique local (ULA) — RFC4193
"fc00::/7",
// IPv6 multicast
"ff00::/8",
}
// blockedCIDRs is the parsed form of blockedCIDRStrings.
// Any entry that fails to parse is recorded in blockedCIDRParseErrors instead
// of panicking; tests verify this slice is always empty via TestBlockedCIDRsValid.
var (
blockedCIDRs []*net.IPNet
blockedCIDRParseErrors []string
)
func init() {
blockedCIDRs = make([]*net.IPNet, 0, len(blockedCIDRStrings))
for _, r := range blockedCIDRStrings {
_, cidr, err := net.ParseCIDR(r)
if err != nil {
// Record the error rather than panicking; TestBlockedCIDRsValid
// will catch this during tests, and the CI build will fail.
blockedCIDRParseErrors = append(blockedCIDRParseErrors,
fmt.Sprintf("ipcheck: invalid built-in CIDR %q: %v", r, err))
continue
}
blockedCIDRs = append(blockedCIDRs, cidr)
}
}
// IsBlockedIP reports whether ip is in a blocked address range. // IsBlockedIP reports whether ip is in a blocked address range.
// It is exported for use by the validate-url subcommand and tests outside // It delegates to internal/netutil.IsBlockedIP; see that function for the full
// this package. // list of blocked ranges and IPv6-mapped IPv4 normalization behavior.
//
// IPv6-mapped IPv4 addresses (e.g. ::ffff:192.168.1.1) are normalized to their
// IPv4 form before checking so that IPv4 CIDRs catch them.
//
// Based on:
// - RFC1918 private ranges
// - RFC5735 / RFC4193 special-use IPv4/IPv6 ranges
// - RFC4291 IPv6 link-local / loopback
func IsBlockedIP(ip net.IP) bool { func IsBlockedIP(ip net.IP) bool {
// Normalize IPv6-mapped IPv4 addresses (::ffff:x.x.x.x) to plain IPv4. return netutil.IsBlockedIP(ip)
if v4 := ip.To4(); v4 != nil {
ip = v4
}
for _, cidr := range blockedCIDRs {
if cidr.Contains(ip) {
return true
}
}
return false
} }
+19 -126
View File
@@ -3,142 +3,35 @@ package gitea
import ( import (
"net" "net"
"testing" "testing"
"gitea.weiker.me/rodin/review-bot/internal/netutil"
) )
func TestIsBlockedIP(t *testing.T) { // TestIsBlockedIPForwarding verifies that gitea.IsBlockedIP correctly forwards
blocked := []struct { // to internal/netutil.IsBlockedIP. Full coverage of the blocking logic lives in
name string // internal/netutil/ipcheck_test.go.
func TestIsBlockedIPForwarding(t *testing.T) {
cases := []struct {
ip string ip string
blocked bool
}{ }{
// IPv4 loopback {"127.0.0.1", true}, // loopback — must be blocked
{"loopback 127.0.0.1", "127.0.0.1"}, {"192.168.1.1", true}, // RFC1918 — must be blocked
{"loopback 127.0.0.2", "127.0.0.2"}, {"8.8.8.8", false}, // public — must not be blocked
{"loopback 127.255.255.255", "127.255.255.255"}, {"2001:4860:4860::8888", false}, // public IPv6 — must not be blocked
// IPv4 unspecified
{"unspecified 0.0.0.0", "0.0.0.0"},
{"unspecified 0.1.2.3", "0.1.2.3"},
// RFC1918
{"RFC1918 10.0.0.1", "10.0.0.1"},
{"RFC1918 10.255.255.255", "10.255.255.255"},
{"RFC1918 172.16.0.1", "172.16.0.1"},
{"RFC1918 172.31.255.255", "172.31.255.255"},
{"RFC1918 192.168.0.1", "192.168.0.1"},
{"RFC1918 192.168.255.255", "192.168.255.255"},
// Link-local (APIPA / AWS metadata)
{"link-local 169.254.0.1", "169.254.0.1"},
{"link-local 169.254.169.254", "169.254.169.254"},
// Shared address space (carrier-grade NAT)
{"CGN 100.64.0.1", "100.64.0.1"},
{"CGN 100.127.255.255", "100.127.255.255"},
// Multicast
{"multicast 224.0.0.1", "224.0.0.1"},
{"multicast 239.255.255.255", "239.255.255.255"},
// Reserved
{"reserved 240.0.0.1", "240.0.0.1"},
{"broadcast 255.255.255.255", "255.255.255.255"},
// IPv6 loopback
{"IPv6 loopback ::1", "::1"},
// IPv6 unspecified
{"IPv6 unspecified ::", "::"},
// IPv6 link-local
{"IPv6 link-local fe80::1", "fe80::1"},
{"IPv6 link-local fe80::dead:beef", "fe80::dead:beef"},
// IPv6 ULA
{"IPv6 ULA fc00::1", "fc00::1"},
{"IPv6 ULA fd00::1", "fd00::1"},
// IPv6 multicast
{"IPv6 multicast ff02::1", "ff02::1"},
} }
for _, tc := range cases {
for _, tc := range blocked {
t.Run(tc.name, func(t *testing.T) {
ip := net.ParseIP(tc.ip) ip := net.ParseIP(tc.ip)
if ip == nil { if ip == nil {
t.Fatalf("failed to parse IP %q", tc.ip) t.Fatalf("failed to parse IP %q", tc.ip)
} }
if !IsBlockedIP(ip) { got := IsBlockedIP(ip)
t.Errorf("IsBlockedIP(%q) = false, want true", tc.ip) want := netutil.IsBlockedIP(ip)
if got != want {
t.Errorf("gitea.IsBlockedIP(%q) = %v, netutil.IsBlockedIP = %v: forwarding mismatch", tc.ip, got, want)
} }
}) if got != tc.blocked {
} t.Errorf("gitea.IsBlockedIP(%q) = %v, want %v", tc.ip, got, tc.blocked)
allowed := []struct {
name string
ip string
}{
{"public 8.8.8.8", "8.8.8.8"},
{"public 1.1.1.1", "1.1.1.1"},
{"public 198.51.100.1", "198.51.100.1"}, // RFC5737 TEST-NET-2 — a documentation-only range;
// not assigned to any real host, but intentionally left unblocked here because
// it has no special routing treatment (unlike RFC1918/loopback/link-local) and
// blocking it would require tracking every RFC5737 range without meaningful
// security benefit (no server should ever listen on a TEST-NET address).
{"public 151.101.1.1", "151.101.1.1"}, // Fastly
{"public IPv6 2001:4860:4860::8888", "2001:4860:4860::8888"}, // Google DNS
{"public IPv6 2606:4700:4700::1111", "2606:4700:4700::1111"}, // Cloudflare DNS
}
for _, tc := range allowed {
t.Run(tc.name, func(t *testing.T) {
ip := net.ParseIP(tc.ip)
if ip == nil {
t.Fatalf("failed to parse IP %q", tc.ip)
}
if IsBlockedIP(ip) {
t.Errorf("IsBlockedIP(%q) = true, want false", tc.ip)
}
})
}
}
func TestIsBlockedIPv6MappedIPv4(t *testing.T) {
// ::ffff:192.168.1.1 is an IPv6-mapped IPv4 address — should be blocked as RFC1918.
// Construct it manually as a 16-byte IP.
mapped := net.IP{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0xff, 0xff, 192, 168, 1, 1}
if !IsBlockedIP(mapped) {
t.Errorf("IsBlockedIP(::ffff:192.168.1.1) = false, want true (IPv6-mapped IPv4 must be normalized)")
}
// ::ffff:8.8.8.8 — IPv6-mapped public IP — should be allowed.
mappedPublic := net.IP{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0xff, 0xff, 8, 8, 8, 8}
if IsBlockedIP(mappedPublic) {
t.Errorf("IsBlockedIP(::ffff:8.8.8.8) = true, want false")
}
}
func TestIsBlockedIPEdgeCases(t *testing.T) {
// The boundary between RFC1918 and public ranges.
// 172.15.255.255 is NOT private (just below 172.16.0.0/12).
notPrivate := net.ParseIP("172.15.255.255")
if IsBlockedIP(notPrivate) {
t.Errorf("IsBlockedIP(172.15.255.255) = true, want false (outside 172.16.0.0/12)")
}
// 172.32.0.0 is NOT private (just above 172.31.255.255).
notPrivate2 := net.ParseIP("172.32.0.0")
if IsBlockedIP(notPrivate2) {
t.Errorf("IsBlockedIP(172.32.0.0) = true, want false (outside 172.16.0.0/12)")
}
// CGN: 100.63.255.255 is NOT in 100.64.0.0/10.
notCGN := net.ParseIP("100.63.255.255")
if IsBlockedIP(notCGN) {
t.Errorf("IsBlockedIP(100.63.255.255) = true, want false (outside 100.64.0.0/10)")
}
// CGN: 100.128.0.0 is NOT in 100.64.0.0/10.
notCGN2 := net.ParseIP("100.128.0.0")
if IsBlockedIP(notCGN2) {
t.Errorf("IsBlockedIP(100.128.0.0) = true, want false (outside 100.64.0.0/10)")
}
}
// TestBlockedCIDRsValid verifies that all entries in blockedCIDRStrings parse
// successfully. This catches programming errors in the CIDR list without
// requiring a startup panic. The init() function records parse failures in
// blockedCIDRParseErrors rather than panicking; this test makes those failures
// visible as test failures during CI.
func TestBlockedCIDRsValid(t *testing.T) {
if len(blockedCIDRParseErrors) > 0 {
for _, msg := range blockedCIDRParseErrors {
t.Errorf("CIDR parse error: %s", msg)
} }
} }
} }
+55 -48
View File
@@ -376,6 +376,57 @@ func (c *Client) doGet(ctx context.Context, url string) ([]byte, error) {
return c.doRequest(ctx, http.MethodGet, url, "") return c.doRequest(ctx, http.MethodGet, url, "")
} }
// doRequestWithBody performs an HTTP request with an optional body, applying the
// same HTTPS enforcement as doRequest. It is used by write methods (POST, PUT,
// DELETE) that bypass the retry loop in doRequest because write operations are
// not idempotent.
//
// body may be nil for requests that carry no payload (e.g. DELETE).
// When body is non-nil, Content-Type is set to application/json.
func (c *Client) doRequestWithBody(ctx context.Context, method, reqURL string, body []byte) ([]byte, error) {
if !c.allowInsecureHTTP {
parsed, err := url.Parse(reqURL)
if err != nil {
return nil, fmt.Errorf("parse request URL: %w", err)
}
if strings.EqualFold(parsed.Scheme, "http") {
return nil, fmt.Errorf("refusing HTTP request to %s: use HTTPS or set AllowInsecureHTTP option", redactURL(reqURL))
}
}
var reqBody io.Reader
if body != nil {
reqBody = bytes.NewReader(body)
}
req, err := http.NewRequestWithContext(ctx, method, reqURL, reqBody)
if err != nil {
return nil, fmt.Errorf("create request: %w", err)
}
req.Header.Set("Authorization", "Bearer "+c.token)
req.Header.Set("Accept", "application/vnd.github+json")
if body != nil {
req.Header.Set("Content-Type", "application/json")
}
resp, err := c.httpClient.Do(req)
if err != nil {
return nil, fmt.Errorf("do request: %w", err)
}
defer resp.Body.Close()
if resp.StatusCode >= 200 && resp.StatusCode < 300 {
respBody, err := io.ReadAll(io.LimitReader(resp.Body, maxResponseBodyBytes))
if err != nil {
return nil, fmt.Errorf("read response body: %w", err)
}
return respBody, nil
}
errBody, _ := io.ReadAll(io.LimitReader(resp.Body, maxErrorBodyBytes))
return nil, &APIError{StatusCode: resp.StatusCode, Body: string(errBody)}
}
// --- API types --- // --- API types ---
// PullRequest holds relevant PR metadata. // PullRequest holds relevant PR metadata.
@@ -677,29 +728,11 @@ func (c *Client) PostReview(ctx context.Context, owner, repo string, number int,
return nil, fmt.Errorf("marshal review payload: %w", err) return nil, fmt.Errorf("marshal review payload: %w", err)
} }
req, err := http.NewRequestWithContext(ctx, http.MethodPost, reqURL, bytes.NewReader(data)) respBody, err := c.doRequestWithBody(ctx, http.MethodPost, reqURL, data)
if err != nil {
return nil, fmt.Errorf("create review request: %w", err)
}
req.Header.Set("Authorization", "Bearer "+c.token)
req.Header.Set("Content-Type", "application/json")
req.Header.Set("Accept", "application/vnd.github+json")
resp, err := c.httpClient.Do(req)
if err != nil { if err != nil {
return nil, fmt.Errorf("post review: %w", err) return nil, fmt.Errorf("post review: %w", err)
} }
defer resp.Body.Close()
if resp.StatusCode < 200 || resp.StatusCode >= 300 {
respBody, _ := io.ReadAll(io.LimitReader(resp.Body, maxErrorBodyBytes))
return nil, &APIError{StatusCode: resp.StatusCode, Body: string(respBody)}
}
respBody, err := io.ReadAll(io.LimitReader(resp.Body, maxResponseBodyBytes))
if err != nil {
return nil, fmt.Errorf("read review response: %w", err)
}
var review Review var review Review
if err := json.Unmarshal(respBody, &review); err != nil { if err := json.Unmarshal(respBody, &review); err != nil {
return nil, fmt.Errorf("parse review response: %w", err) return nil, fmt.Errorf("parse review response: %w", err)
@@ -740,23 +773,11 @@ func (c *Client) DeleteReview(ctx context.Context, owner, repo string, number in
reqURL := fmt.Sprintf("%s/repos/%s/%s/pulls/%d/reviews/%d", reqURL := fmt.Sprintf("%s/repos/%s/%s/pulls/%d/reviews/%d",
c.baseURL, url.PathEscape(owner), url.PathEscape(repo), number, reviewID) c.baseURL, url.PathEscape(owner), url.PathEscape(repo), number, reviewID)
req, err := http.NewRequestWithContext(ctx, http.MethodDelete, reqURL, nil) // nil body: the GitHub DELETE endpoint for reviews requires no request body.
if err != nil { _, err := c.doRequestWithBody(ctx, http.MethodDelete, reqURL, nil)
return fmt.Errorf("create delete request: %w", err)
}
req.Header.Set("Authorization", "Bearer "+c.token)
req.Header.Set("Accept", "application/vnd.github+json")
resp, err := c.httpClient.Do(req)
if err != nil { if err != nil {
return fmt.Errorf("delete review: %w", err) return fmt.Errorf("delete review: %w", err)
} }
defer resp.Body.Close()
if resp.StatusCode < 200 || resp.StatusCode >= 300 {
respBody, _ := io.ReadAll(io.LimitReader(resp.Body, maxErrorBodyBytes))
return &APIError{StatusCode: resp.StatusCode, Body: string(respBody)}
}
return nil return nil
} }
@@ -790,24 +811,10 @@ func (c *Client) RequestReviewer(ctx context.Context, owner, repo string, number
return fmt.Errorf("marshal reviewer request: %w", err) return fmt.Errorf("marshal reviewer request: %w", err)
} }
req, err := http.NewRequestWithContext(ctx, http.MethodPost, reqURL, bytes.NewReader(data)) _, err = c.doRequestWithBody(ctx, http.MethodPost, reqURL, data)
if err != nil {
return fmt.Errorf("create reviewer request: %w", err)
}
req.Header.Set("Authorization", "Bearer "+c.token)
req.Header.Set("Content-Type", "application/json")
req.Header.Set("Accept", "application/vnd.github+json")
resp, err := c.httpClient.Do(req)
if err != nil { if err != nil {
return fmt.Errorf("request reviewer: %w", err) return fmt.Errorf("request reviewer: %w", err)
} }
defer resp.Body.Close()
if resp.StatusCode != http.StatusOK && resp.StatusCode != http.StatusCreated && resp.StatusCode != http.StatusNoContent {
respBody, _ := io.ReadAll(io.LimitReader(resp.Body, 256))
return &APIError{StatusCode: resp.StatusCode, Body: string(respBody)}
}
return nil return nil
} }
+36
View File
@@ -1077,6 +1077,42 @@ func TestRequestReviewer_Success(t *testing.T) {
} }
} }
func TestPostReview_RejectsHTTP(t *testing.T) {
// PostReview must reject http:// base URLs — tokens must not be sent in plaintext.
c := NewClient("tok", "http://127.0.0.1:1")
_, err := c.PostReview(context.Background(), "owner", "repo", 1, "APPROVE", "body", "", nil)
if err == nil {
t.Fatal("expected error for HTTP base URL in PostReview")
}
if !strings.Contains(err.Error(), "refusing HTTP request") {
t.Errorf("unexpected error message: %v", err)
}
}
func TestDeleteReview_RejectsHTTP(t *testing.T) {
// DeleteReview must reject http:// base URLs — tokens must not be sent in plaintext.
c := NewClient("tok", "http://127.0.0.1:1")
err := c.DeleteReview(context.Background(), "owner", "repo", 1, 42)
if err == nil {
t.Fatal("expected error for HTTP base URL in DeleteReview")
}
if !strings.Contains(err.Error(), "refusing HTTP request") {
t.Errorf("unexpected error message: %v", err)
}
}
func TestRequestReviewer_RejectsHTTP(t *testing.T) {
// RequestReviewer must reject http:// base URLs — tokens must not be sent in plaintext.
c := NewClient("tok", "http://127.0.0.1:1")
err := c.RequestReviewer(context.Background(), "owner", "repo", 1, "reviewer1")
if err == nil {
t.Fatal("expected error for HTTP base URL in RequestReviewer")
}
if !strings.Contains(err.Error(), "refusing HTTP request") {
t.Errorf("unexpected error message: %v", err)
}
}
func TestEscapePath_SpecialChars(t *testing.T) { func TestEscapePath_SpecialChars(t *testing.T) {
tests := []struct { tests := []struct {
input string input string
+97
View File
@@ -0,0 +1,97 @@
// Package netutil provides shared network utilities for review-bot.
// ipcheck.go implements IP-level SSRF protection by checking resolved addresses
// against known blocked CIDR ranges (RFC1918, loopback, link-local, etc.).
package netutil
import (
"fmt"
"net"
)
// blockedCIDRStrings is the canonical list of CIDR strings that should never
// be contacted by review-bot. See IsBlockedIP for the full list of covered
// address families.
//
// These are hard-coded literals: any parse failure is a programming error.
// Validity is verified by TestBlockedCIDRsValid in ipcheck_test.go.
var blockedCIDRStrings = []string{
// IPv4 loopback
"127.0.0.0/8",
// IPv4 unspecified / "this network"
"0.0.0.0/8",
// RFC1918 private ranges
"10.0.0.0/8",
"172.16.0.0/12",
"192.168.0.0/16",
// IPv4 link-local (APIPA, also used by AWS instance metadata 169.254.169.254)
"169.254.0.0/16",
// IPv4 shared address space (RFC6598, carrier-grade NAT)
"100.64.0.0/10",
// IPv4 multicast
"224.0.0.0/4",
// IPv4 reserved / broadcast
"240.0.0.0/4",
// IPv6 loopback
"::1/128",
// IPv6 unspecified
"::/128",
// IPv6 link-local
"fe80::/10",
// IPv6 unique local (ULA) — RFC4193
"fc00::/7",
// IPv6 multicast
"ff00::/8",
}
// blockedCIDRs is the parsed form of blockedCIDRStrings.
// Any entry that fails to parse is recorded in blockedCIDRParseErrors instead
// of panicking; tests verify this slice is always empty via TestBlockedCIDRsValid.
var (
blockedCIDRs []*net.IPNet
blockedCIDRParseErrors []string
)
func init() {
blockedCIDRs = make([]*net.IPNet, 0, len(blockedCIDRStrings))
for _, r := range blockedCIDRStrings {
_, cidr, err := net.ParseCIDR(r)
if err != nil {
// Record the error rather than panicking; TestBlockedCIDRsValid
// will catch this during tests, and the CI build will fail.
blockedCIDRParseErrors = append(blockedCIDRParseErrors,
fmt.Sprintf("ipcheck: invalid built-in CIDR %q: %v", r, err))
continue
}
blockedCIDRs = append(blockedCIDRs, cidr)
}
}
// BlockedCIDRParseErrors returns any errors encountered parsing the built-in
// CIDR list. In correct code this will always be empty; tests assert it is.
func BlockedCIDRParseErrors() []string {
return blockedCIDRParseErrors
}
// IsBlockedIP reports whether ip is in a blocked address range.
// It is exported for use by the gitea package's safe dialer, the validate-url
// subcommand, and tests outside this package.
//
// IPv6-mapped IPv4 addresses (e.g. ::ffff:192.168.1.1) are normalized to their
// IPv4 form before checking so that IPv4 CIDRs catch them.
//
// Based on:
// - RFC1918 private ranges
// - RFC5735 / RFC4193 special-use IPv4/IPv6 ranges
// - RFC4291 IPv6 link-local / loopback
func IsBlockedIP(ip net.IP) bool {
// Normalize IPv6-mapped IPv4 addresses (::ffff:x.x.x.x) to plain IPv4.
if v4 := ip.To4(); v4 != nil {
ip = v4
}
for _, cidr := range blockedCIDRs {
if cidr.Contains(ip) {
return true
}
}
return false
}
+142
View File
@@ -0,0 +1,142 @@
package netutil
import (
"net"
"testing"
)
func TestIsBlockedIP(t *testing.T) {
blocked := []struct {
name string
ip string
}{
// IPv4 loopback
{"loopback 127.0.0.1", "127.0.0.1"},
{"loopback 127.0.0.2", "127.0.0.2"},
{"loopback 127.255.255.255", "127.255.255.255"},
// IPv4 unspecified
{"unspecified 0.0.0.0", "0.0.0.0"},
{"unspecified 0.1.2.3", "0.1.2.3"},
// RFC1918
{"RFC1918 10.0.0.1", "10.0.0.1"},
{"RFC1918 10.255.255.255", "10.255.255.255"},
{"RFC1918 172.16.0.1", "172.16.0.1"},
{"RFC1918 172.31.255.255", "172.31.255.255"},
{"RFC1918 192.168.0.1", "192.168.0.1"},
{"RFC1918 192.168.255.255", "192.168.255.255"},
// Link-local (APIPA / AWS metadata)
{"link-local 169.254.0.1", "169.254.0.1"},
{"link-local 169.254.169.254", "169.254.169.254"},
// Shared address space (carrier-grade NAT)
{"CGN 100.64.0.1", "100.64.0.1"},
{"CGN 100.127.255.255", "100.127.255.255"},
// Multicast
{"multicast 224.0.0.1", "224.0.0.1"},
{"multicast 239.255.255.255", "239.255.255.255"},
// Reserved
{"reserved 240.0.0.1", "240.0.0.1"},
{"broadcast 255.255.255.255", "255.255.255.255"},
// IPv6 loopback
{"IPv6 loopback ::1", "::1"},
// IPv6 unspecified
{"IPv6 unspecified ::", "::"},
// IPv6 link-local
{"IPv6 link-local fe80::1", "fe80::1"},
{"IPv6 link-local fe80::dead:beef", "fe80::dead:beef"},
// IPv6 ULA
{"IPv6 ULA fc00::1", "fc00::1"},
{"IPv6 ULA fd00::1", "fd00::1"},
// IPv6 multicast
{"IPv6 multicast ff02::1", "ff02::1"},
}
for _, tc := range blocked {
t.Run(tc.name, func(t *testing.T) {
ip := net.ParseIP(tc.ip)
if ip == nil {
t.Fatalf("failed to parse IP %q", tc.ip)
}
if !IsBlockedIP(ip) {
t.Errorf("IsBlockedIP(%q) = false, want true", tc.ip)
}
})
}
allowed := []struct {
name string
ip string
}{
{"public 8.8.8.8", "8.8.8.8"},
{"public 1.1.1.1", "1.1.1.1"},
{"public 198.51.100.1", "198.51.100.1"}, // RFC5737 TEST-NET-2 — a documentation-only range;
// not assigned to any real host, but intentionally left unblocked here because
// it has no special routing treatment (unlike RFC1918/loopback/link-local) and
// blocking it would require tracking every RFC5737 range without meaningful
// security benefit (no server should ever listen on a TEST-NET address).
{"public 151.101.1.1", "151.101.1.1"}, // Fastly
{"public IPv6 2001:4860:4860::8888", "2001:4860:4860::8888"}, // Google DNS
{"public IPv6 2606:4700:4700::1111", "2606:4700:4700::1111"}, // Cloudflare DNS
}
for _, tc := range allowed {
t.Run(tc.name, func(t *testing.T) {
ip := net.ParseIP(tc.ip)
if ip == nil {
t.Fatalf("failed to parse IP %q", tc.ip)
}
if IsBlockedIP(ip) {
t.Errorf("IsBlockedIP(%q) = true, want false", tc.ip)
}
})
}
}
func TestIsBlockedIPv6MappedIPv4(t *testing.T) {
// ::ffff:192.168.1.1 is an IPv6-mapped IPv4 address — should be blocked as RFC1918.
// Construct it manually as a 16-byte IP.
mapped := net.IP{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0xff, 0xff, 192, 168, 1, 1}
if !IsBlockedIP(mapped) {
t.Errorf("IsBlockedIP(::ffff:192.168.1.1) = false, want true (IPv6-mapped IPv4 must be normalized)")
}
// ::ffff:8.8.8.8 — IPv6-mapped public IP — should be allowed.
mappedPublic := net.IP{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0xff, 0xff, 8, 8, 8, 8}
if IsBlockedIP(mappedPublic) {
t.Errorf("IsBlockedIP(::ffff:8.8.8.8) = true, want false")
}
}
func TestIsBlockedIPEdgeCases(t *testing.T) {
// The boundary between RFC1918 and public ranges.
// 172.15.255.255 is NOT private (just below 172.16.0.0/12).
notPrivate := net.ParseIP("172.15.255.255")
if IsBlockedIP(notPrivate) {
t.Errorf("IsBlockedIP(172.15.255.255) = true, want false (outside 172.16.0.0/12)")
}
// 172.32.0.0 is NOT private (just above 172.31.255.255).
notPrivate2 := net.ParseIP("172.32.0.0")
if IsBlockedIP(notPrivate2) {
t.Errorf("IsBlockedIP(172.32.0.0) = true, want false (outside 172.16.0.0/12)")
}
// CGN: 100.63.255.255 is NOT in 100.64.0.0/10.
notCGN := net.ParseIP("100.63.255.255")
if IsBlockedIP(notCGN) {
t.Errorf("IsBlockedIP(100.63.255.255) = true, want false (outside 100.64.0.0/10)")
}
// CGN: 100.128.0.0 is NOT in 100.64.0.0/10.
notCGN2 := net.ParseIP("100.128.0.0")
if IsBlockedIP(notCGN2) {
t.Errorf("IsBlockedIP(100.128.0.0) = true, want false (outside 100.64.0.0/10)")
}
}
// TestBlockedCIDRsValid verifies that all entries in blockedCIDRStrings parse
// successfully. This catches programming errors in the CIDR list without
// requiring a startup panic. The init() function records parse failures in
// blockedCIDRParseErrors rather than panicking; this test makes those failures
// visible as test failures during CI.
func TestBlockedCIDRsValid(t *testing.T) {
for _, msg := range BlockedCIDRParseErrors() {
t.Errorf("CIDR parse error: %s", msg)
}
}