fix(python): validate shared pyc header facts

src/application/python/PythonPycService.test.ts

@@ -66,6 +66,28 @@ const report: PythonPyc = {
   ],
 };
 
+it.each([
+  {
+    flags: 1,
+    hash_based: true,
+    source_hash_hex: "",
+    source_mtime: null,
+    source_size: null,
+  },
+  { flags: 1 },
+])("refuses inconsistent provider header facts %j", async (header) => {
+  const result = await new PythonPycService({
+    close: async () => {},
+    identity,
+    inspect: async () =>
+      ok({ ...report, header: { ...report.header, ...header } }),
+  }).inspect({ path: report.artifact.path });
+  expect(result).toMatchObject({
+    ok: false,
+    error: { _tag: "AnalysisOutputError" },
+  });
+});
+
 it("rejects a relative path and extra arguments before reading", async () => {
   let calls = 0;
   const service = new PythonPycService({

src/captures/TsharkCaptureProvider.ts

@@ -369,10 +369,7 @@ export class TsharkCaptureProvider implements PacketCapturePort {
         resourcePath = "path" in input ? input.path : command;
         // TShark resolves `-e` aliases before emitting canonical field names;
         // the command arguments, not output names, own field selection.
-        const matchingPackets = parsePacketOutput(
-          output.stdout.text,
-          input.packet_limit,
-        );
+        const matchingPackets = parsePacketOutput(output.stdout.text);
         const start = Math.min(input.match_offset, matchingPackets.length);
         const end =
           input.match_limit === null

src/captures/parsePacketOutput.ts

@@ -6,7 +6,7 @@ import { packetFieldsSchema } from "../domain/captures/packetCapture.js";
 const producerSchema = z.array(
   z.object({ _source: z.object({ layers: packetFieldsSchema }) }),
 );
-export const parsePacketOutput = (text: string, packetLimit: number | null) => {
+export const parsePacketOutput = (text: string) => {
   const decoded: unknown = JSON.parse(text);
   const packets = producerSchema
     .parse(decoded)
@@ -19,11 +19,6 @@ export const parsePacketOutput = (text: string, packetLimit: number | null) => {
         "inspect_packet_capture",
         "TShark returned unordered frames.",
       );
-    if (packetLimit !== null && frame > BigInt(packetLimit))
-      throw new AnalysisOutputError(
-        "inspect_packet_capture",
-        "TShark returned a frame beyond the selected input packet limit.",
-      );
     previous = frame;
   }
   return packets;

src/domain/python/pythonPyc.test.ts

@@ -3,9 +3,49 @@ import {
   inspectPythonPycInputSchema,
   pythonPycMarshalProfile,
   PYTHON_PYC_MARSHAL_PROFILES,
+  pythonPycHeaderSchema,
 } from "./pythonPyc.js";
 import { pythonBytecodeIdentity } from "./pythonBytecodeIdentity.js";
 
+const hashHeader = {
+  magic: 3531,
+  python_version: "3.12",
+  code_layout: "py311",
+  flags: 3,
+  hash_based: true,
+  check_source: true,
+  source_hash_hex: "0123456789abcdef",
+  source_mtime: null,
+  source_size: null,
+};
+
+it.each(["", "ab", "g".repeat(16)])(
+  "rejects an invalid source hash %j",
+  (source_hash_hex) => {
+    expect(
+      pythonPycHeaderSchema.safeParse({ ...hashHeader, source_hash_hex })
+        .success,
+    ).toBe(false);
+  },
+);
+
+it.each([
+  { hash_based: false },
+  { check_source: false },
+  { source_hash_hex: null },
+  { source_mtime: 0 },
+  { source_size: 0 },
+])("keeps header facts consistent with the recorded flags: %j", (changed) => {
+  expect(
+    pythonPycHeaderSchema.safeParse({ ...hashHeader, ...changed }).success,
+  ).toBe(false);
+});
+
+it("preserves unknown raw header flags and valid hash bytes", () => {
+  const header = { ...hashHeader, flags: 0x80000003 };
+  expect(pythonPycHeaderSchema.parse(header)).toEqual(header);
+});
+
 it("keeps pinned CPython identity separate from standalone marshal coverage", () => {
   expect(pythonPycMarshalProfile(3394)).toEqual({
     code_layout: "py37",

src/domain/python/pythonPyc.ts

@@ -205,17 +205,53 @@ const pythonCodeObjectSchema = z.strictObject({
 });
 export type PythonPycCodeObject = z.infer<typeof pythonCodeObjectSchema>;
 
-export const pythonPycHeaderSchema = z.strictObject({
-  magic: z.number().int().min(0).max(0xffff),
-  python_version: z.string().min(1),
-  code_layout: z.enum(PYTHON_PYC_CODE_LAYOUTS),
-  flags: uint,
-  hash_based: z.boolean(),
-  check_source: z.boolean(),
-  source_hash_hex: z.string().nullable(),
-  source_mtime: uint.nullable(),
-  source_size: uint.nullable(),
-});
+export const pythonPycHeaderSchema = z
+  .strictObject({
+    magic: z.number().int().min(0).max(0xffff),
+    python_version: z.string().min(1),
+    code_layout: z.enum(PYTHON_PYC_CODE_LAYOUTS),
+    flags: uint,
+    hash_based: z.boolean(),
+    check_source: z.boolean(),
+    source_hash_hex: z
+      .string()
+      .regex(/^[0-9a-f]{16}$/u)
+      .nullable(),
+    source_mtime: uint.nullable(),
+    source_size: uint.nullable(),
+  })
+  .superRefine((header, context) => {
+    for (const [field, flag] of [
+      ["hash_based", PYTHON_PYC_HASH_BASED],
+      ["check_source", PYTHON_PYC_CHECK_SOURCE],
+    ] as const)
+      if (header[field] !== ((header.flags & flag) !== 0))
+        context.addIssue({
+          code: "custom",
+          path: [field],
+          message: `${field} disagrees with its recorded header flag.`,
+        });
+    if (header.hash_based === (header.source_hash_hex === null))
+      context.addIssue({
+        code: "custom",
+        path: ["source_hash_hex"],
+        message:
+          "Hash-based pycs require an 8-byte source hash; others must omit it.",
+      });
+    if (header.hash_based === (header.source_mtime !== null))
+      context.addIssue({
+        code: "custom",
+        path: ["source_mtime"],
+        message: "Timestamp pycs require mtime; hash-based pycs must omit it.",
+      });
+    if (header.hash_based === (header.source_size !== null))
+      context.addIssue({
+        code: "custom",
+        path: ["source_size"],
+        message:
+          "Timestamp pycs require source_size; hash-based pycs must omit it.",
+      });
+  });
 
 /** Bounded CPython 3.7–3.13 pyc observation; marshal is walked, bytecode is not decompiled. */
 export const pythonPycSchema = z
@@ -253,29 +289,6 @@ export const pythonPycSchema = z
         message: "Marshal byte count disagrees with the header-stripped file.",
       });
     }
-    if (report.header.hash_based === (report.header.source_hash_hex === null)) {
-      context.addIssue({
-        code: "custom",
-        path: ["header", "source_hash_hex"],
-        message:
-          "Hash-based pycs require an 8-byte source hash; others must omit it.",
-      });
-    }
-    if (report.header.hash_based === (report.header.source_mtime !== null)) {
-      context.addIssue({
-        code: "custom",
-        path: ["header", "source_mtime"],
-        message: "Timestamp pycs require mtime; hash-based pycs must omit it.",
-      });
-    }
-    if (report.header.hash_based === (report.header.source_size !== null)) {
-      context.addIssue({
-        code: "custom",
-        path: ["header", "source_size"],
-        message:
-          "Timestamp pycs require source_size; hash-based pycs must omit it.",
-      });
-    }
     if (report.coverage.code_objects !== report.code_objects.length) {
       context.addIssue({
         code: "custom",

tests/boundary/captures/captureProvider.test.ts

@@ -247,26 +247,6 @@ it("preserves TShark canonical output for a caller-selected field alias", async
   expect(report.tool.arguments).toContain("bootp.type");
 });
 
-it("refuses producer frames outside the selected input read scope before paging", async () => {
-  const f = await fixture("many");
-  const request = { path: f.path, fields: ["ip.addr"], match_limit: 1 };
-  expect(
-    await f.service.inspect({ ...request, packet_limit: 10 }),
-  ).toMatchObject({
-    ok: false,
-    error: { _tag: "AnalysisOutputError", operation: "inspect_packet_capture" },
-  });
-  const admitted = await f.service.inspect({ ...request, packet_limit: 90 });
-  if (!admitted.ok) throw admitted.error;
-  const report = packetCaptureSchema.parse(admitted.value.normalized_result);
-  expect(report.output_window).toEqual({
-    matching_packets: 90,
-    returned_packets: 1,
-    has_more: true,
-  });
-  expect(report.packets[0]?.["frame.number"]).toEqual(["1"]);
-});
-
 it("returns canonical alias output through the public packet-capture CLI", async () => {
   const f = await fixture("alias-canonical");
   const cli = Cli.create("capture-alias-test", { sync: false });

tests/boundary/captures/packetOutput.test.ts

@@ -17,7 +17,7 @@ it("preserves repeated field values and missing fields while discarding export i
     },
     { _source: { layers: { "frame.number": ["7"] } } },
   ]);
-  expect(parsePacketOutput(output, 7)).toEqual([
+  expect(parsePacketOutput(output)).toEqual([
     { "frame.number": ["2"], "ip.addr": ["192.0.2.1", "192.0.2.2"] },
     { "frame.number": ["7"] },
   ]);
@@ -49,7 +49,7 @@ it.each([
 ])("rejects malformed producer fields: $layers", ({ layers, issue }) => {
   let failure: unknown;
   try {
-    parsePacketOutput(JSON.stringify([{ _source: { layers } }]), null);
+    parsePacketOutput(JSON.stringify([{ _source: { layers } }]));
   } catch (error) {
     failure = error;
   }
@@ -72,7 +72,6 @@ it.each([
           _source: { layers: { "frame.number": frame } },
         })),
       ),
-      null,
     );
   } catch (error) {
     failure = error;
@@ -89,18 +88,7 @@ it("preserves canonical field names emitted for a selected alias", () => {
       },
     },
   ]);
-  expect(parsePacketOutput(output, null)).toEqual([
+  expect(parsePacketOutput(output)).toEqual([
     { "frame.number": ["1"], "dhcp.type": ["1"] },
   ]);
 });
-
-it("validates every frame against the input read limit while allowing filtered gaps", () => {
-  const output = JSON.stringify(
-    [2, 7].map((frame) => ({
-      _source: { layers: { "frame.number": [String(frame)] } },
-    })),
-  );
-  expect(parsePacketOutput(output, 7)).toHaveLength(2);
-  expect(() => parsePacketOutput(output, 6)).toThrow(AnalysisOutputError);
-  expect(parsePacketOutput(output, null)).toHaveLength(2);
-});