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/domain/objcClassSymbol.ts

@@ -8,7 +8,8 @@ export const objcClassSymbol = (
       readonly metaclass: boolean;
     }
   | undefined => {
-  const leaf = objcSymbolLeaf(name);
+  const namespace = name.lastIndexOf("::");
+  const leaf = namespace < 0 ? name : name.slice(namespace + 2);
   const match = /^_?(OBJC_(META)?CLASS_\$_([^\s]+))$/u.exec(leaf);
   if (match?.[1] === undefined || match[3] === undefined) return undefined;
   return {
@@ -17,9 +18,3 @@ export const objcClassSymbol = (
     metaclass: match[2] !== undefined,
   };
 };
-
-/** Preserve the exact runtime symbol leaf after a provider namespace. */
-export const objcSymbolLeaf = (name: string): string => {
-  const namespace = name.lastIndexOf("::");
-  return namespace < 0 ? name : name.slice(namespace + 2);
-};

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",

src/domain/symbolAnalysis.test.ts

@@ -282,23 +282,3 @@ it("distinguishes protocol identity from protocol member contexts", () => {
     unclassified: [],
   });
 });
-
-it("selects exact Objective-C protocol symbol leaves", () => {
-  const declarations = [
-    "OBJC_PROTOCOL_$_Runnable",
-    "_OBJC_PROTOCOL_$_Runnable",
-    "__OBJC_PROTOCOL_$_Runnable",
-    "Framework::__OBJC_PROTOCOL_$_Runnable",
-  ].map((name, index) => ({ address: `0x${index}`, name }));
-  const otherNames = [
-    "s__OBJC_PROTOCOL_$_Runnable_1000",
-    "PTR__OBJC_PROTOCOL_$_Runnable_1000",
-    "-[Example useOBJC_PROTOCOL_$_Runnable]",
-    "__OBJC_PROTOCOL_REFERENCE_$_Runnable",
-    "__OBJC_PROTOCOL_$_",
-  ].map((name, index) => ({ address: `0x${index + 4}`, name }));
-  expect(discoverObjcProtocols([...otherNames, ...declarations])).toEqual({
-    count: declarations.length,
-    protocols: declarations,
-  });
-});

src/domain/symbolAnalysis.ts

@@ -1,4 +1,4 @@
-import { objcClassSymbol, objcSymbolLeaf } from "./objcClassSymbol.js";
+import { objcClassSymbol } from "./objcClassSymbol.js";
 import type { JsonValue } from "./jsonValue.js";
 import type { AddressedName } from "./native/nativeInventory.js";
 import type {
@@ -14,8 +14,11 @@ import {
 /** Local class records emitted by Apple's fragile Objective-C ABI. */
 const LEGACY_OBJC_CLASS_PREFIX = "l_OBJC_CLASS_" as const;
 
-/** Objective-C protocol declaration labels with their object-file prefixes. */
-const OBJC_PROTOCOL_SYMBOL = /^_{0,2}OBJC_PROTOCOL_\$_[^\s]+$/u;
+/**
+ * Objective-C protocol declaration marker. `PROTOCOL_REFERENCE` slots
+ * share the substring but are uses, not definitions.
+ */
+const OBJC_PROTOCOL_MARKER = "OBJC_PROTOCOL_$_" as const;
 
 export interface ObjcClassImportBindingScan {
   readonly bindings: readonly MachoImportBinding[];
@@ -63,7 +66,7 @@ export const discoverObjcProtocols = (
   const protocols = uniqueByName(
     names.filter(
       ({ name }) =>
-        OBJC_PROTOCOL_SYMBOL.test(objcSymbolLeaf(name)) ||
+        name.includes(OBJC_PROTOCOL_MARKER) ||
         classifySwiftSymbol(name)?.protocolIdentity === true,
     ),
   );