Files
remark42/backend/app/store/image/fs_store.go
T
Dmitry VerkhoturovandGitHub c9ba8520c7 fix(auth): close OAuth open-redirect by wiring AllowedRedirectHosts (#2049)
* fix(auth): close OAuth open-redirect by wiring AllowedRedirectHosts

Bump go-pkgz/auth/v2 to master (v2.1.2-0.20260421203319-686683f19cf7)
which carries the `from` redirect validator from go-pkgz/auth#275.

The library default with a nil AllowedRedirectHosts is permissive
(preserves legacy behavior for existing consumers on a dep bump), so
just bumping the dep leaves remark42 vulnerable — a crafted
/auth/<provider>/login?from=https://evil.example.com/... still issues
the 307 to the attacker host after the user completes legitimate
OAuth. Verified end-to-end against a local dev-auth instance before
and after this commit.

Wire Opts.AllowedRedirectHosts in getAuthenticator to the operator's
existing --allowed-hosts config, stripping the CSP "self" sentinel
which is not a real hostname. RemarkURL's own host is always implicit
per the library contract, so a default single-site deployment gains
the protection with no config change. Multi-host embeds work as soon
as their embedding hosts are added to AllowedHosts (they already need
to be there for CSP frame-ancestors).

Refreshed vendor tree to match the new module version.

* chore(lint): suppress G703 false positives on image Save

CI's newer gosec flags os.MkdirAll/os.WriteFile in FileSystem.Save with
G703 because id flows in from the caller. id is validated at the HTTP
layer (safePictureSegment in rest_public.go) and dst is derived via
f.location — not a real traversal. Targeted //nolint with reason.

* fix(auth): normalise AllowedRedirectHosts entries + add unit test

Address Copilot review on PR #2049. The previous closure passed raw
s.AllowedHosts entries straight to the auth library, but --allowed-hosts
holds CSP frame-ancestors source expressions: scheme-prefixed values
(https://blog.example.com), entries with ports, and wildcards
(*.cdn.example.com) are all valid there but the auth library compares
against u.Hostname() and would silently drop them — breaking legitimate
redirects on multi-host deployments.

Extract getAllowedRedirectHosts that:
* trims whitespace, drops empty / 'self' / "self" / wildcard entries
* prepends https:// if scheme missing then url.Parse to extract Hostname
* logs a warning on parse failure rather than poisoning the allowlist

Wire the closure in getAuthenticator to call the helper.

Test_getAllowedRedirectHosts covers all the edge cases Copilot flagged
(scheme stripping, port handling, self spellings, wildcards, empty,
mixed real-world).

* fix(auth): preserve explicit port in AllowedRedirectHosts + clarify fs_store nolint

Address Copilot follow-up on PR #2049:

* getAllowedRedirectHosts stripped explicit ports via u.Hostname(), which
  broadened the allowlist. The auth validator checks both Hostname() and
  Host, so an entry like admin.example.com:8443 can and should be kept
  host:port — allowing only that port, not any. Emit u.Host when
  u.Port() != "", u.Hostname() otherwise. Updated tests.

* fs_store Save nolint rationale said "id validated at HTTP layer", but
  Save is reached via image.Service.Save and SaveWithID (cache), neither
  of which is HTTP validation. id is actually a server-generated hash in
  both paths. Updated the comment.
2026-04-21 19:09:26 -05:00

217 lines
6.4 KiB
Go

package image
import (
"context"
"fmt"
"hash/crc64"
"io"
"math"
"os"
"path"
"path/filepath"
"strconv"
"strings"
"sync"
"time"
log "github.com/go-pkgz/lgr"
)
// FileSystem provides image Store for local files. Saves and loads files from Location, restricts max size.
type FileSystem struct {
Location string
Staging string
Partitions int
moveLock sync.Mutex // needed only for deleting images or moving them from staging to permanent storage
crc struct {
*crc64.Table
sync.Once
mask string
divider uint64
}
}
// Save saves image with given id to local FS, staging directory.
// Files partitioned across multiple subdirectories, and the final path includes part, i.e. /location/user1/03/123-4567
func (f *FileSystem) Save(id string, img []byte) error {
dst := f.location(f.Staging, id)
if err := os.MkdirAll(path.Dir(dst), 0o700); err != nil { //nolint:gosec // id is server-generated hash via image.Service (Save / SaveWithID); dst computed via f.location
return fmt.Errorf("can't make image directory: %w", err)
}
if err := os.WriteFile(dst, img, 0o600); err != nil { //nolint:gosec // same as MkdirAll above
return fmt.Errorf("can't write image file with id %s: %w", id, err)
}
log.Printf("[DEBUG] file %s saved for image %s, size=%d", dst, id, len(img))
return nil
}
// Commit file stored in staging location by moving it to permanent location
func (f *FileSystem) Commit(id string) error {
f.moveLock.Lock()
defer f.moveLock.Unlock()
log.Printf("[DEBUG] Commit image %s", id)
stagingImage, permImage := f.location(f.Staging, id), f.location(f.Location, id)
if err := os.MkdirAll(path.Dir(permImage), 0o700); err != nil {
return fmt.Errorf("can't make image directory: %w", err)
}
if err := os.Rename(stagingImage, permImage); err != nil {
return fmt.Errorf("failed to commit image %s: %w", id, err)
}
return nil
}
// ResetCleanupTimer resets cleanup timer for the image
func (f *FileSystem) ResetCleanupTimer(id string) error {
file := f.location(f.Staging, id)
_, err := os.Stat(file)
if err != nil {
return fmt.Errorf("can't get image stats for %s: %w", id, err)
}
// we don't need to update access time (second arg),
// but reading it is platform-dependent and looks different on darwin and linux,
// so it's easier to update it as well
if err = os.Chtimes(file, time.Now(), time.Now()); err != nil {
return fmt.Errorf("problem updating %s modification time: %w", file, err)
}
return nil
}
// Load image from FS. Uses id to get partition subdirectory.
func (f *FileSystem) Load(id string) ([]byte, error) {
// get image file by id. first try permanent location and if not found - staging
img := func(id string) (file string, err error) {
file = f.location(f.Location, id)
_, err = os.Stat(file)
if err != nil {
file = f.location(f.Staging, id)
_, err = os.Stat(file)
}
if err != nil {
return file, fmt.Errorf("can't get image stats for %s: %w", id, err)
}
return file, nil
}
imgFile, err := img(id)
if err != nil {
return nil, fmt.Errorf("can't get image file for %s: %w", id, err)
}
fh, err := os.Open(imgFile) //nolint:gosec // we open file from known location
if err != nil {
return nil, fmt.Errorf("can't load image %s: %w", id, err)
}
return io.ReadAll(fh)
}
// Delete image from storage
func (f *FileSystem) Delete(id string) error {
f.moveLock.Lock()
defer f.moveLock.Unlock()
staging := f.location(f.Staging, id)
// file doesn't exist on staging, delete from permanent location
if _, err := os.Stat(staging); os.IsNotExist(err) {
file := f.location(f.Location, id)
e := os.Remove(file)
_ = os.Remove(path.Dir(file)) // try to remove directory
return e
}
// delete file from staging
err := os.Remove(staging)
_ = os.Remove(path.Dir(staging)) // try to remove directory
return err
}
// Cleanup runs scan of staging and removes old files based on ttl
func (f *FileSystem) Cleanup(_ context.Context, ttl time.Duration) error {
if _, err := os.Stat(f.Staging); os.IsNotExist(err) {
return nil
}
f.moveLock.Lock()
defer f.moveLock.Unlock()
// we can ignore context as on local FS remove is relatively fast operation
err := filepath.Walk(f.Staging, func(fpath string, info os.FileInfo, err error) error {
if err != nil {
return err
}
if info.IsDir() {
return nil
}
age := time.Since(info.ModTime())
if age > (ttl + 100*time.Millisecond) { // delay cleanup triggering to allow commit
log.Printf("[INFO] remove staging image %s, age %v", fpath, age)
rmErr := os.Remove(fpath) //nolint:gosec // staging dir is server-only, no untrusted symlinks land here
_ = os.Remove(path.Dir(fpath)) //nolint:gosec // same staging dir
return rmErr
}
return nil
})
if err != nil {
return fmt.Errorf("failed to cleanup images: %w", err)
}
return nil
}
// Info returns meta information about storage
func (f *FileSystem) Info() (StoreInfo, error) {
if _, err := os.Stat(f.Staging); os.IsNotExist(err) {
return StoreInfo{}, nil
}
var ts time.Time
err := filepath.Walk(f.Staging, func(_ string, info os.FileInfo, err error) error {
if err != nil {
return err
}
if info.IsDir() {
return nil
}
created := info.ModTime()
if ts.IsZero() || created.Before(ts) {
ts = created
}
return nil
})
if err != nil {
return StoreInfo{}, fmt.Errorf("problem retrieving first timestamp from staging images on fs: %w", err)
}
return StoreInfo{FirstStagingImageTS: ts}, nil
}
// location gets full path for id by adding partition to the final path in order to keep files in different subdirectories
// and avoid too many files in a single place.
// the end result is a full path like this - /tmp/images/user1/92/xxx-yyy.png.
// Number of partitions defined by FileSystem.Partitions
func (f *FileSystem) location(base, id string) string {
partition := func(id string) string {
f.crc.Do(func() {
f.crc.Table = crc64.MakeTable(crc64.ECMA)
p := int(math.Round(math.Log10(float64(f.Partitions))))
f.crc.mask = "%0" + strconv.Itoa(p) + "d"
f.crc.divider = uint64(math.Pow(10, float64(p)))
})
checksum64 := crc64.Checksum([]byte(id), f.crc.Table)
partition := checksum64 % f.crc.divider
return fmt.Sprintf(f.crc.mask, partition)
}
user, file := "unknown", id // default if no user in id
if elems := strings.Split(id, "/"); len(elems) == 2 {
user, file = elems[0], elems[1] // user in id
}
if f.Partitions == 0 {
return path.Join(base, user, file) // avoid partition directory if 0 Partitions
}
return path.Join(base, user, partition(id), file)
}