From 34bc4b9cd0b6dc51c2f33ca34d13bbc48a78e6c9 Mon Sep 17 00:00:00 2001
From: George Byers II <george@codehosted.com>
Date: Fri, 18 Sep 2026 20:35:27 -0400
Subject: [PATCH] fix: validate MPF JPEG attachments before redaction

---
 TESTING.md                       |  18 ++++++
 api/helpers.py                   |  70 ++++++++++++++++++++-
 tests/fixtures/multipicture.jpg  | Bin 0 -> 1372 bytes
 tests/test_mpf_jpeg_redaction.py | 104 +++++++++++++++++++++++++++++++
 4 files changed, 190 insertions(+), 2 deletions(-)
 create mode 100644 tests/fixtures/multipicture.jpg
 create mode 100644 tests/test_mpf_jpeg_redaction.py

diff --git a/TESTING.md b/TESTING.md
index 6b2f2862..af111974 100644
--- a/TESTING.md
+++ b/TESTING.md
@@ -36,6 +36,24 @@ npm run lint:runtime
 npx eslint --no-config-lookup -c eslint.runtime-guard.config.mjs "static/**/*.js"
 ```
 
+## Native raster redaction boundary
+
+Run `./scripts/test.sh tests/test_mpf_jpeg_redaction.py tests/test_raster_data_uri_redaction.py tests/test_security_redaction.py`.
+Native `messages[*].content[*].image_url.url` raster data may bypass text
+credential scanning only after complete container validation. MPF JPEGs require
+an APP2 MP Index with bounded TIFF entries, contiguous declared JPEG extents,
+and a complete SOF/scan/EOI sequence in each image. Gaps, overlaps, malformed
+frames, and trailing bytes fall back to text redaction. Image-shaped tool
+metadata does not acquire this exemption. This is structural validation, not
+pixel decoding or secret detection inside image metadata/pixels.
+
+`tests/fixtures/multipicture.jpg` contains two Pillow-generated 8x8 solid-color
+frames (red and blue), not private photographs; generation instructions live in
+`tests/test_mpf_jpeg_redaction.py`. The tests cover both TIFF byte orders and
+credential text appended after the final EOI. Helper timing improvements do
+not by themselves prove browser or shared-server responsiveness; verify those
+separately with a real conversation and concurrent requests.
+
 ## Python lint gate (ruff) — forward-looking, new-code-only
 
 The Python twin of the ESLint runtime guard. A curated `ruff` ruleset
diff --git a/api/helpers.py b/api/helpers.py
index 9a8e7f36..71164cce 100644
--- a/api/helpers.py
+++ b/api/helpers.py
@@ -724,12 +724,59 @@ _JPEG_SOF_MARKERS = {
 }
 
 
-def _is_complete_jpeg(raw: bytes) -> bool:
+def _mpf_image_ranges(data: bytes, base: int) -> list[tuple[int, int]] | None:
+    """Read an MP Index TIFF inside APP2; offsets are relative to its header.
+
+    Only the declared JPEG extents are trusted, not a decoder's willingness to
+    ignore trailing bytes. The caller checks contiguous coverage and each JPEG.
+    """
+    if len(data) < 8 or data[:4] not in {b"II\x2a\x00", b"MM\x00\x2a"}:
+        return None
+    order = "little" if data[:2] == b"II" else "big"
+
+    def uint(pos: int, size: int = 4) -> int:
+        return int.from_bytes(data[pos:pos + size], order)
+
+    ifd = uint(4)
+    if ifd < 8 or ifd + 2 > len(data):
+        return None
+    count = uint(ifd, 2)
+    table_end = ifd + 2 + 12 * count
+    if table_end + 4 > len(data):
+        return None
+    tags = {}
+    for pos in range(ifd + 2, table_end, 12):
+        tag = uint(pos, 2)
+        if tag in tags:
+            return None
+        tags[tag] = (uint(pos + 2, 2), uint(pos + 4), uint(pos + 8))
+    if tags.get(0xB001, ())[:2] != (4, 1):
+        return None
+    images = tags[0xB001][2]
+    entry = tags.get(0xB002)
+    if not entry or images < 2 or entry[:2] != (7, 16 * images):
+        return None
+    offset = entry[2]
+    if offset < table_end + 4 or offset + 16 * images > len(data):
+        return None
+    ranges = []
+    for index in range(images):
+        pos = offset + index * 16
+        attributes, size, relative = uint(pos), uint(pos + 4), uint(pos + 8)
+        if attributes & 0x07000000 or size < 4 or (index == 0 and relative != 0):
+            return None
+        start = 0 if index == 0 else base + relative
+        ranges.append((start, start + size))
+    return ranges
+
+
+def _is_complete_jpeg(raw: bytes, *, _allow_mpf: bool = True) -> bool:
     if len(raw) < 4 or raw[:2] != b"\xff\xd8":
         return False
     pos = 2
     saw_sof = False
     saw_scan = False
+    mpf_ranges = None
     while pos < len(raw):
         marker_start = pos
         if raw[pos] != 0xFF:
@@ -741,7 +788,20 @@ def _is_complete_jpeg(raw: bytes) -> bool:
         marker = raw[pos]
         pos += 1
         if marker == 0xD9:
-            return saw_sof and saw_scan and pos == len(raw)
+            if not saw_sof or not saw_scan:
+                return False
+            if mpf_ranges is None:
+                return pos == len(raw)
+            if mpf_ranges[0] != (0, pos):
+                return False
+            # No gaps, overlaps, undeclared pictures or post-EOI payloads.
+            for start, end in mpf_ranges[1:]:
+                if start != pos or end > len(raw):
+                    return False
+                if not _is_complete_jpeg(raw[start:end], _allow_mpf=False):
+                    return False
+                pos = end
+            return pos == len(raw)
         if marker in {0x00, 0x01, 0xD8} or 0xD0 <= marker <= 0xD7:
             return False
         if pos + 2 > len(raw):
@@ -752,6 +812,12 @@ def _is_complete_jpeg(raw: bytes) -> bool:
         segment_end = pos + segment_length
         if segment_end > len(raw):
             return False
+        if _allow_mpf and marker == 0xE2 and raw[pos + 2:pos + 6] == b"MPF\x00":
+            if mpf_ranges is not None:
+                return False
+            mpf_ranges = _mpf_image_ranges(raw[pos + 6:segment_end], pos + 6)
+            if mpf_ranges is None:
+                return False
         if marker in _JPEG_SOF_MARKERS:
             if segment_length < 8:
                 return False
diff --git a/tests/fixtures/multipicture.jpg b/tests/fixtures/multipicture.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..16b2e30aba86f3a9bac57e302ed67e723197a310
GIT binary patch
literal 1372
zcmex=<NpH&0WUXCHwH#V1_nkTWcdGxA;UMojlt7Xi-CiIfq|KUVFNn@3y@`CXkfs&
zfd!<9feD0x;tD|82#6tonc*==0tr+C`NyGR3TWW}Z3brs4mLJ+HdYRHc6Lrq4lW)M
zULJ049!Vi#ei2zIc{y1r85so?J#_^o9c39A4RcK$149!N6M1z@TMHu_J!2CikRgnm
zoSZz|JQBRT5=M$Libf=Z{|6WZIe@<9U}jWeU=n0x7G(T?gh3t{BCL#H0P_(e6Eh1d
z8#@Ol7dKGBRsjYkMrLLv7G_pf7GMZ5)&k`jSOi&x6b&8OgaZ@Vl?p|S8YeE~P<GmQ
zP&DY`2NmO_q9#r*F>wh=DOELf4NWZ*Q!{f5ODks=S2uSLPp{yR(6I1`$f)F$)U@=B
z%&g*)(z5c3%Btp;*0%PJ&aO$5r%atTea6gLixw|gx@`H1m8&*w-m-Pu_8mKS9XfpE
z=&|D`PM*4S`O4L6*Kgds_3+W-Cr_U}fAR9w$4{TXeEs(Q$Io9Ne=#yJL%anfAwEO%
zmmttzOe`$SEbJhEF*20{F|!~GtD+&BkYgZwVxh2-Q6q<l)5L`v4{|CS2YnDtD!Rxe
zreg9?^&`k@V4o4^u_m&72KN}kU$+=|m>GeQ$1KQT&+zEQqOPBlzYG7r`5#*nA5D0p
U36C}j@8c<rMf(eF4*$Oi0E0KH+5i9m

literal 0
HcmV?d00001

diff --git a/tests/test_mpf_jpeg_redaction.py b/tests/test_mpf_jpeg_redaction.py
new file mode 100644
index 00000000..072df1a5
--- /dev/null
+++ b/tests/test_mpf_jpeg_redaction.py
@@ -0,0 +1,104 @@
+"""MPF fixture: Pillow MPO, two generated 8x8 red/blue frames.
+
+Generate with Image.new('RGB', (8, 8), color) for red, blue;
+first.save('multipicture.jpg', format='MPO', save_all=True,
+           append_images=remaining). No private photographs are used.
+"""
+import base64
+from pathlib import Path
+import struct
+
+import pytest
+
+from api import helpers
+
+
+RAW = (Path(__file__).parent / 'fixtures' / 'multipicture.jpg').read_bytes()
+BASE = RAW.index(b'MPF\0') + 4
+IFD = BASE + int.from_bytes(RAW[BASE + 4:BASE + 8], 'little')
+TAGS = {int.from_bytes(RAW[p:p + 2], 'little'): p
+        for p in range(IFD + 2, IFD + 2 + 12 * int.from_bytes(RAW[IFD:IFD + 2], 'little'), 12)}
+ENTRIES = BASE + int.from_bytes(RAW[TAGS[0xB002] + 8:TAGS[0xB002] + 12], 'little')
+
+
+def uri(raw):
+    return 'data:image/jpeg;base64,' + base64.b64encode(raw).decode('ascii')
+
+
+def session(value):
+    return {'messages': [{'role': 'user', 'content': [
+        {'type': 'image_url', 'image_url': {'url': value}}]}]}
+
+
+@pytest.mark.parametrize('byte_order', ['little', 'big'])
+def test_mpf_preserves_native_image_without_text_redaction(monkeypatch, byte_order):
+    raw = bytearray(RAW)
+    if byte_order == 'big':
+        raw[BASE:BASE + 4] = b'MM\x00\x2a'
+        struct.pack_into('>I', raw, BASE + 4, IFD - BASE)
+        struct.pack_into('>H', raw, IFD, len(TAGS))
+        for tag, pos in TAGS.items():
+            kind, count, value = struct.unpack_from('<HII', RAW, pos + 2)
+            struct.pack_into('>HHI', raw, pos, tag, kind, count)
+            if tag != 0xB000:  # Version is four raw ASCII bytes, not an integer.
+                struct.pack_into('>I', raw, pos + 8, value)
+        for pos in range(ENTRIES, ENTRIES + 32, 16):
+            struct.pack_into('>IIIHH', raw, pos, *struct.unpack_from('<IIIHH', RAW, pos))
+    value = uri(raw)
+    calls = []
+    monkeypatch.setattr(helpers, '_redact_text', lambda text, **kw: calls.append(text) or text)
+    assert helpers._is_native_raster_data_uri(value)
+    assert helpers.redact_session_data(session(value)) == session(value)
+    assert value not in calls
+
+
+@pytest.mark.parametrize('attack', ['trailing_secret', 'gap', 'overlap', 'size',
+                                   'count', 'table_offset', 'truncated', 'bad_frame',
+                                   'first_offset', 'undeclared_image'])
+def test_mpf_rejects_unaccounted_or_invalid_bytes(attack):
+    raw = bytearray(RAW)
+    if attack == 'trailing_secret':
+        raw += b'API_KEY=private-trailing-secret'
+    elif attack == 'undeclared_image':
+        raw += RAW
+    elif attack == 'truncated':
+        del raw[-1:]
+    elif attack == 'count':
+        struct.pack_into('<I', raw, TAGS[0xB001] + 8, 999999)
+    elif attack == 'table_offset':
+        struct.pack_into('<I', raw, TAGS[0xB002] + 8, len(raw))
+    elif attack == 'first_offset':
+        struct.pack_into('<I', raw, ENTRIES + 8, 1)
+    elif attack == 'bad_frame':
+        raw[-2:] = b'xx'
+    else:
+        pos = ENTRIES + 16 + (4 if attack == 'size' else 8)
+        value = int.from_bytes(raw[pos:pos + 4], 'little')
+        struct.pack_into('<I', raw, pos, value + (-1 if attack == 'overlap' else 1))
+    assert not helpers._is_native_raster_data_uri(uri(raw))
+
+
+def test_mpf_appended_base64_secret_reaches_real_redactor():
+    # Pad inside a declared JPEG COM segment, keeping all MP offsets valid,
+    # so the appended decoded credential aligns to a base64 quantum.
+    raw = bytearray(RAW)
+    padding = b'\xff\xfe\x00\x04xx' if len(raw) % 3 == 0 else (
+        b'\xff\xfe\x00\x02' if len(raw) % 3 == 2 else b'\xff\xfe\x00\x03x')
+    raw[-2:-2] = padding
+    last_size = ENTRIES + 16 + 4
+    struct.pack_into('<I', raw, last_size, int.from_bytes(RAW[last_size:last_size + 4], 'little') + len(padding))
+    assert len(raw) % 3 == 0
+    secret = 'AKIA' + 'Z' * 16
+    value = uri(raw + base64.b64decode(secret))
+    assert value.endswith(secret)
+    assert not helpers._is_native_raster_data_uri(value)
+    result = helpers.redact_session_data(session(value))
+    assert secret not in result['messages'][0]['content'][0]['image_url']['url']
+
+
+def test_mpf_outside_authoritative_image_boundary_still_redacts(monkeypatch):
+    value = uri(RAW)
+    calls = []
+    monkeypatch.setattr(helpers, '_redact_text', lambda text, **kw: calls.append(text) or text)
+    helpers.redact_session_data({'tool_calls': [{'args': session(value)}]})
+    assert value in calls