Return bounded Range errors for hosted and raw reads

src/msg/extensions/hosting.py

@@ -330,7 +330,7 @@ def hosted_range(request, size, etag):
     requested = request.headers.get('range')
     if not requested or request.headers.get('if-range') not in {None, etag}:
         return (0, size), 200, {}
-    match = re.fullmatch(r'bytes=(\d*)-(\d*)', requested)
+    match = re.fullmatch(r'bytes=(\d*)-(\d*)', requested) if len(requested) <= 128 else None
     require(match is not None and any(match.groups()), 'range_not_satisfiable')
     left, right = match.groups()
     if left:

src/msg/transports/http_routes.py

@@ -2431,7 +2431,11 @@ async def dispatch(request: Request):
                 status = 200
                 requested_range = request.headers.get('range')
                 if requested_range and (request.headers.get('if-range') in {None, etag}):
-                    match = re.fullmatch(r'bytes=(\d*)-(\d*)', requested_range)
+                    match = (
+                        re.fullmatch(r'bytes=(\d*)-(\d*)', requested_range)
+                        if len(requested_range) <= 128
+                        else None
+                    )
                     require(match is not None and any(match.groups()), 'range_not_satisfiable')
                     left, right = match.groups()
                     if left:

tests/test_hosting_same_origin.py

@@ -91,6 +91,15 @@ async def test_same_domain_hosted_html_head_304_range_raw_and_no_write_route(ins
         partial = await http.get('/@web-owner/web/index.html', headers={'Range': 'bytes=0-14'})
         assert partial.status_code == 206 and partial.content == html[:15]
         isolated(partial)
+        for malformed in (
+            'bytes=' + '9' * 5000 + '-',
+            'bytes=0-' + '9' * 5000,
+            'bytes=-' + '9' * 5000,
+        ):
+            refused = await http.get('/@web-owner/web/index.html', headers={'Range': malformed})
+            assert refused.status_code == 416
+            assert refused.json()['error'] == 'range_not_satisfiable'
+            isolated(refused)
         raw = await http.get(f'/_read/{published}/raw')
         assert raw.status_code == 200
         assert 'attachment' in raw.headers['content-disposition']

tests/test_http_range_bounds.py

@@ -0,0 +1,70 @@
+"""Malformed Range fields have bounded parsing and stable client errors."""
+
+import pytest
+from starlette.requests import Request
+
+from msg.core.errors import Failure
+from msg.extensions.hosting import hosted_range
+
+
+@pytest.mark.parametrize(
+    'value',
+    ['bytes=' + '9' * 5000 + '-', 'bytes=0-' + '9' * 5000, 'bytes=-' + '9' * 5000],
+    ids=['start', 'end', 'suffix'],
+)
+def test_hosted_range_rejects_oversized_numbers_with_contract_error(value):
+    request = Request({'type': 'http', 'headers': [(b'range', value.encode())]})
+    with pytest.raises(Failure, match='range_not_satisfiable'):
+        hosted_range(request, 12, '"etag"')
+
+
+@pytest.mark.parametrize(
+    'value,expected',
+    [('bytes=0-4', (0, 5)), ('bytes=5-', (5, 12)), ('bytes=-5', (7, 12)), ('bytes=-99', (0, 12))],
+)
+def test_hosted_range_keeps_normal_prefix_open_and_suffix_ranges(value, expected):
+    request = Request({'type': 'http', 'headers': [(b'range', value.encode())]})
+    interval, status, _ = hosted_range(request, 12, '"etag"')
+    assert interval == expected and status == 206
+
+
+def test_stale_if_range_ignores_range_before_parsing():
+    request = Request({
+        'type': 'http',
+        'headers': [(b'range', b'bytes=' + b'9' * 5000 + b'-'), (b'if-range', b'"stale"')],
+    })
+    assert hosted_range(request, 12, '"etag"') == ((0, 12), 200, {})
+
+
+@pytest.mark.asyncio
+async def test_raw_http_rejects_oversized_range_without_read_effects(installed, monkeypatch):
+    import httpx
+    from read_only_evidence import readonly_evidence
+    from test_service import call, register
+
+    from msg.transports.http import create_app
+
+    app, _ = installed
+    key, subject, _ = await register(app, 'range-reader')
+    created = await call(
+        app,
+        'content.post_create',
+        {'parent': '/main', 'body': 'range content'},
+        key=key,
+        subject=subject,
+    )
+    assert created.status == 'ok'
+    async with httpx.AsyncClient(
+        transport=httpx.ASGITransport(app=create_app(app)), base_url=app.settings.service_url
+    ) as http:
+        async with readonly_evidence(app, monkeypatch):
+            for value in (
+                'bytes=' + '9' * 5000 + '-',
+                'bytes=0-' + '9' * 5000,
+                'bytes=-' + '9' * 5000,
+            ):
+                result = await http.get(
+                    '/_id/' + created.resources[0].id + '/raw', headers={'Range': value}
+                )
+                assert result.status_code == 416
+                assert result.json()['error']['code'] == 'range_not_satisfiable'