Reject malformed PDF media uploads

CHANGELOG.md

@@ -2,6 +2,15 @@
 
 # Changelog
 
+## v0.1.0-preview.14 — 2026-09-03
+
+- Reject header-only, truncated, and structurally invalid PDF uploads in the
+  bounded media preparer. Accepted attachments now require a supported PDF
+  version, terminal EOF marker, numeric in-range `startxref`, and either a
+  traditional xref/trailer or xref-stream object at the declared offset.
+- Keep PDF handling storage-neutral and non-rendering: applications still own
+  authorization, reference tracking, attachment disposition, and lifecycle.
+
 ## v0.1.0-preview.13 — 2026-09-03
 
 - Complete the password-plus-recovery-code flow with a short-lived restricted

media/media.go

@@ -20,6 +20,7 @@ import (
 	"mime"
 	"net/http"
 	"path/filepath"
+	"strconv"
 	"strings"
 	"time"
 	"unicode/utf8"
@@ -121,7 +122,7 @@ func Prepare(reader io.Reader, originalName string, limits Limits) (Prepared, er
 	}
 
 	detected := http.DetectContentType(data)
-	if detected == "application/pdf" && bytes.HasPrefix(data, []byte("%PDF-")) {
+	if detected == "application/pdf" && validPDF(data) {
 		result := Prepared{Data: append([]byte(nil), data...), MediaType: "application/pdf", Kind: KindAttachment, OriginalName: name}
 		result.Digest = sha256.Sum256(result.Data)
 		return result, nil
@@ -156,6 +157,61 @@ func Prepare(reader io.Reader, originalName string, limits Limits) (Prepared, er
 	return result, nil
 }
 
+// validPDF performs a deliberately bounded structural check without trying to
+// render or interpret document content. It rejects header-only spoofing and
+// truncated uploads by requiring a supported header, terminal EOF marker, a
+// numeric startxref offset, and either a traditional xref table with trailer
+// or an xref-stream object at that offset.
+func validPDF(data []byte) bool {
+	if len(data) < 32 || !bytes.HasPrefix(data, []byte("%PDF-")) {
+		return false
+	}
+	headerEnd := bytes.IndexAny(data, "\r\n")
+	if headerEnd < 8 || headerEnd > 32 {
+		return false
+	}
+	header := string(bytes.TrimSpace(data[:headerEnd]))
+	if header != "%PDF-1.0" && header != "%PDF-1.1" && header != "%PDF-1.2" && header != "%PDF-1.3" && header != "%PDF-1.4" && header != "%PDF-1.5" && header != "%PDF-1.6" && header != "%PDF-1.7" && header != "%PDF-2.0" {
+		return false
+	}
+	trimmed := bytes.TrimRight(data, "\x00\t\n\f\r ")
+	if !bytes.HasSuffix(trimmed, []byte("%%EOF")) {
+		return false
+	}
+	eof := len(trimmed) - len("%%EOF")
+	start := bytes.LastIndex(trimmed[:eof], []byte("startxref"))
+	if start < headerEnd {
+		return false
+	}
+	cursor := start + len("startxref")
+	for cursor < eof && (trimmed[cursor] == ' ' || trimmed[cursor] == '\t' || trimmed[cursor] == '\r' || trimmed[cursor] == '\n' || trimmed[cursor] == '\f') {
+		cursor++
+	}
+	digits := cursor
+	for cursor < eof && trimmed[cursor] >= '0' && trimmed[cursor] <= '9' && cursor-digits < 20 {
+		cursor++
+	}
+	if cursor == digits {
+		return false
+	}
+	if len(bytes.TrimSpace(trimmed[cursor:eof])) != 0 {
+		return false
+	}
+	offset, err := strconv.ParseInt(string(trimmed[digits:cursor]), 10, 64)
+	if err != nil || offset < int64(headerEnd+1) || offset >= int64(start) {
+		return false
+	}
+	target := trimmed[int(offset):start]
+	if bytes.HasPrefix(target, []byte("xref")) {
+		return bytes.Contains(target, []byte("trailer"))
+	}
+	lineEnd := bytes.IndexByte(target, '\n')
+	if lineEnd < 5 || lineEnd > 80 || !bytes.Contains(target[:lineEnd], []byte(" obj")) {
+		return false
+	}
+	return bytes.Contains(target, []byte("/Type /XRef")) || bytes.Contains(target, []byte("/Type/XRef"))
+}
+
 func Extension(mediaType string) string {
 	switch mediaType {
 	case "image/jpeg":

media/media_test.go

@@ -5,6 +5,7 @@ package media
 import (
 	"bytes"
 	"errors"
+	"fmt"
 	"image"
 	"image/color"
 	"image/jpeg"
@@ -33,7 +34,8 @@ func TestPrepareReencodesRasterAndStripsTrailingData(t *testing.T) {
 }
 
 func TestPreparePDFIsAttachment(t *testing.T) {
-	prepared, err := Prepare(strings.NewReader("%PDF-1.7\nsmall fixture"), "guide.pdf", Limits{})
+	pdf := minimalPDF()
+	prepared, err := Prepare(bytes.NewReader(pdf), "guide.pdf", Limits{})
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -42,6 +44,18 @@ func TestPreparePDFIsAttachment(t *testing.T) {
 	}
 }
 
+func TestPrepareRejectsMalformedPDF(t *testing.T) {
+	for _, source := range []string{
+		"%PDF-1.7\nsmall fixture",
+		"%PDF-9.9\nxref\ntrailer\nstartxref\n9\n%%EOF",
+		"%PDF-1.7\nxref\ntrailer\nstartxref\n999999\n%%EOF",
+	} {
+		if _, err := Prepare(strings.NewReader(source), "broken.pdf", Limits{}); !errors.Is(err, ErrInvalidMedia) {
+			t.Fatalf("malformed PDF error=%v source=%q", err, source)
+		}
+	}
+}
+
 func TestPrepareRejectsActiveAndOversizedInput(t *testing.T) {
 	if _, err := Prepare(strings.NewReader("<svg><script/></svg>"), "bad.svg", Limits{}); !errors.Is(err, ErrInvalidMedia) {
 		t.Fatalf("svg err=%v", err)
@@ -50,3 +64,9 @@ func TestPrepareRejectsActiveAndOversizedInput(t *testing.T) {
 		t.Fatalf("large err=%v", err)
 	}
 }
+
+func minimalPDF() []byte {
+	prefix := []byte("%PDF-1.7\n1 0 obj\n<< /Type /Catalog >>\nendobj\n")
+	offset := len(prefix)
+	return append(prefix, []byte(fmt.Sprintf("xref\n0 2\n0000000000 65535 f \n0000000009 00000 n \ntrailer\n<< /Size 2 /Root 1 0 R >>\nstartxref\n%d\n%%%%EOF\n", offset))...)
+}

medialocal/store_test.go

@@ -5,6 +5,7 @@ package medialocal
 import (
 	"bytes"
 	"errors"
+	"fmt"
 	"io"
 	"os"
 	"path/filepath"
@@ -20,7 +21,9 @@ func TestStoreRoundTripAndIdempotentPut(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	prepared, err := media.Prepare(bytes.NewReader([]byte("%PDF-1.7\nfixture")), "fixture.pdf", media.Limits{})
+	prefix := []byte("%PDF-1.7\n1 0 obj\n<< /Type /Catalog >>\nendobj\n")
+	pdf := append(prefix, []byte(fmt.Sprintf("xref\n0 2\n0000000000 65535 f \n0000000009 00000 n \ntrailer\n<< /Size 2 /Root 1 0 R >>\nstartxref\n%d\n%%%%EOF\n", len(prefix)))...)
+	prepared, err := media.Prepare(bytes.NewReader(pdf), "fixture.pdf", media.Limits{})
 	if err != nil {
 		t.Fatal(err)
 	}