From 7a2cd1ea0a63e8c44166860b7ee9c7a1aa43492b Mon Sep 17 00:00:00 2001
From: morluto <76467478+morluto@users.noreply.github.com>
Date: Tue, 13 Oct 2026 13:12:06 +0800
Subject: [PATCH] fix(domain): preserve derived byte counts and uncertain
 identity

---
 .../ManagedReconstructionService.test.ts      | 20 +++++++++++++
 src/domain/functionComparison.test.ts         | 28 +++++++++++++++++++
 src/domain/functionComparisonNormalization.ts | 21 ++++++--------
 src/domain/managed/managedReconstruction.ts   |  3 +-
 4 files changed, 59 insertions(+), 13 deletions(-)

diff --git a/src/application/managed/ManagedReconstructionService.test.ts b/src/application/managed/ManagedReconstructionService.test.ts
index 56b9d94428..793951ae9f 100644
--- a/src/application/managed/ManagedReconstructionService.test.ts
+++ b/src/application/managed/ManagedReconstructionService.test.ts
@@ -22,6 +22,26 @@ const exampleMethod = () => {
 };
 
 describe("managed decompiler reconstruction import", () => {
+  it("reports UTF-8 byte totals while preserving UTF-16 text lengths", () => {
+    const text = 'return "é"; // 😀';
+    const method = exampleMethod();
+    const result = importManagedReconstructionEvidence({
+      ...MANAGED_RECONSTRUCTION_IMPORT_EXAMPLE,
+      methods: [
+        {
+          ...method,
+          reconstruction: { ...method.reconstruction, text },
+        },
+      ],
+    });
+    if (!result.ok) throw result.error;
+
+    expect(result.value.normalized_result).toMatchObject({
+      summary: { total_text_bytes: 20 },
+      methods: [{ reconstruction: { text, text_length: 17 } }],
+    });
+  });
+
   it("rejects reconstruction text hash drift", () => {
     expect(() =>
       importManagedReconstruction({
diff --git a/src/domain/functionComparison.test.ts b/src/domain/functionComparison.test.ts
index acb505dac9..e03a277489 100644
--- a/src/domain/functionComparison.test.ts
+++ b/src/domain/functionComparison.test.ts
@@ -119,6 +119,34 @@ const observe = (
   );
 
 describe("function comparison normalized identity", () => {
+  it("keeps differing generated names ambiguous and stable names mismatched", () => {
+    const named = (base: "0x1000" | "0x2000", name: string) =>
+      functionDossierSchema.parse({
+        ...dossier("return 0;", base),
+        procedure: {
+          ...procedure(base, name),
+          signature: "int helper(void)",
+          locals: [],
+        },
+      });
+    const generated = compareFunctions(
+      observe("1", named("0x1000", "sub_1000")),
+      observe("2", named("0x2000", "sub_2000")),
+    );
+    expect(generated.function_match.status).toBe("ambiguous");
+    expect(generated.status).toBe("unknown");
+    expect(
+      generated.dimensions.find(({ dimension }) => dimension === "identity"),
+    ).toMatchObject({ status: "unknown" });
+
+    const stable = compareFunctions(
+      observe("3", named("0x1000", "firstFunction")),
+      observe("4", named("0x2000", "secondFunction")),
+    );
+    expect(stable.function_match.status).toBe("mismatched");
+    expect(stable.status).toBe("changed");
+  });
+
   it.each(["sub_deallocate", "sub_deadbeef_handler", "fcn.dispatch"])(
     "recognizes the explicit symbol %s without a generated-name prefix false positive",
     (name) => {
diff --git a/src/domain/functionComparisonNormalization.ts b/src/domain/functionComparisonNormalization.ts
index a676e1e491..5282272e14 100644
--- a/src/domain/functionComparisonNormalization.ts
+++ b/src/domain/functionComparisonNormalization.ts
@@ -16,24 +16,21 @@ export const functionMatch = (
   left: FunctionSnapshot,
   right: FunctionSnapshot,
 ): FunctionComparisonResult["function_match"] => {
-  if (
-    !isAutoName(left.dossier.procedure.name) &&
-    left.dossier.procedure.name === right.dossier.procedure.name
-  )
+  const leftName = left.dossier.procedure.name;
+  const rightName = right.dossier.procedure.name;
+  const stableNames = !isAutoName(leftName) && !isAutoName(rightName);
+  if (stableNames && leftName === rightName)
     return {
       status: "matched",
       method: "symbol",
-      left_name: left.dossier.procedure.name,
-      right_name: right.dossier.procedure.name,
+      left_name: leftName,
+      right_name: rightName,
     };
   return {
-    status:
-      left.dossier.procedure.name === right.dossier.procedure.name
-        ? "ambiguous"
-        : "mismatched",
+    status: stableNames ? "mismatched" : "ambiguous",
     method: "explicit",
-    left_name: left.dossier.procedure.name,
-    right_name: right.dossier.procedure.name,
+    left_name: leftName,
+    right_name: rightName,
   };
 };
 
diff --git a/src/domain/managed/managedReconstruction.ts b/src/domain/managed/managedReconstruction.ts
index e10be0ffd3..9600166aec 100644
--- a/src/domain/managed/managedReconstruction.ts
+++ b/src/domain/managed/managedReconstruction.ts
@@ -204,7 +204,8 @@ const buildReconstructionSummary = (
     ({ reconstruction }) => reconstruction.kind === "semantic-pseudocode",
   ).length,
   total_text_bytes: methods.reduce(
-    (total, { reconstruction }) => total + reconstruction.text_length,
+    (total, { reconstruction }) =>
+      total + Buffer.byteLength(reconstruction.text, "utf8"),
     0,
   ),
 });