diff --git a/src/domain/apple/binaryPlistMaterializer.test.ts b/src/domain/apple/binaryPlistMaterializer.test.ts index 924001ef9f..6ac08def4b 100644 --- a/src/domain/apple/binaryPlistMaterializer.test.ts +++ b/src/domain/apple/binaryPlistMaterializer.test.ts @@ -250,6 +250,44 @@ it("honors a byte order mark in a data object", () => { ).toBe(false); }); +it.each([ + ["UTF-8 string", Buffer.from([0x72, 0x61, 0xff])], + ["UTF-16BE string", Buffer.from([0x62, 0, 0x61, 0xd8, 0])], + ["UTF-16 data", Buffer.from([0x45, 0xff, 0xfe, 0x61, 0, 0x62])], +] as const)("reports partial text beside malformed %s", (_name, malformed) => { + const bytes = source([ + dictionary(1, 2), + text("key"), + text("kept"), + malformed, + ]); + expect(decodeBinaryPlistText(bytes, budget)).toEqual({ + text: "key\nkept\na\ufffd", + complete: false, + }); +}); + +it.each([ + ["valid marked UTF-8", [0xef, 0xbb, 0xbf, 0xc3, 0xa9], "\ufeffé", true], + [ + "lossy marked UTF-8", + [0xef, 0xbb, 0xbf, 0x61, 0xff], + "\ufeffa\ufffd", + false, + ], + ["unmarked raw data", [0x61, 0xff], "a\ufffd", true], +] as const)( + "tracks stored data text coverage for %s", + (_name, payload, value, complete) => { + const data = Buffer.from([0x40 | payload.length, ...payload]); + const bytes = source([dictionary(1, 2), text("key"), data]); + expect(decodeBinaryPlistText(bytes, budget)).toEqual({ + text: `key\n${value}`, + complete, + }); + }, +); + it("keeps readable text beside an unreferenced string it cannot read", () => { const bytes = source([ dictionary(1, 2), diff --git a/src/domain/apple/binaryPlistMaterializer.ts b/src/domain/apple/binaryPlistMaterializer.ts index fca6fa4538..0b36ae7891 100644 --- a/src/domain/apple/binaryPlistMaterializer.ts +++ b/src/domain/apple/binaryPlistMaterializer.ts @@ -6,6 +6,7 @@ import { createPlistDuplicateKeyCollector, type PlistDuplicateKeys, } from "./plistDuplicateKeys.js"; +import { isStrictUtf8 } from "../strictUtf8.js"; import { byteOrderMarkEncoding } from "../textByteOrderMark.js"; import type { ParsedPropertyList } from "../propertyListKeys.js"; import { @@ -191,27 +192,59 @@ const storedPlistPayload = ( return { type, payload: bytes.subarray(start, end) }; }; -/** The text a data payload holds, honoring a byte order mark; null when no reader scans it. */ -const storedDataText = (payload: Buffer): string | null => { - const encoding = byteOrderMarkEncoding(payload); - if (encoding === "utf-32") return null; - return encoding === null - ? payload.toString("utf8") - : new TextDecoder(encoding).decode(payload); +/** Decode declared wide text while preserving readable portions on a malformed input. */ +const storedWideText = ( + payload: Buffer, + encoding: "utf-16le" | "utf-16be", + ignoreBOM: boolean, +): BinaryPlistText => { + try { + const text = new TextDecoder(encoding, { fatal: true, ignoreBOM }).decode( + payload, + ); + return { text, complete: true }; + } catch (cause: unknown) { + if (!(cause instanceof TypeError)) throw cause; + return { + text: new TextDecoder(encoding, { ignoreBOM }).decode(payload), + complete: false, + }; + } }; -/** - * Decode a payload `storedPlistPayload` found; call only once it fits the - * budget. Null when its encoding cannot be scanned. - */ -const storedPlistText = (type: number, payload: Buffer): string | null => - type === 5 - ? payload.toString("latin1") - : type === 6 - ? Buffer.from(payload).swap16().toString("utf16le") - : type === 4 - ? storedDataText(payload) - : payload.toString("utf8"); +/** Scan data bytes as before unless a byte order mark names a specific encoding. */ +const storedDataText = (payload: Buffer): BinaryPlistText | null => { + const encoding = byteOrderMarkEncoding(payload); + if (encoding === "utf-32") return null; + if (encoding !== null) return storedWideText(payload, encoding, false); + const text = payload.toString("utf8"); + // This decode preserves a UTF-8 BOM. It declares text just as a wide + // mark does; unmarked data remains eligible for best-effort raw scanning. + return { + text, + complete: !text.startsWith("\ufeff") || isStrictUtf8(payload), + }; +}; + +/** Decode a stored payload after its byte budget is admitted. */ +const storedPlistText = ( + type: number, + payload: Buffer, +): BinaryPlistText | null => { + switch (type) { + case 4: + return storedDataText(payload); + case 5: + return { text: payload.toString("latin1"), complete: true }; + case 6: + return storedWideText(payload, "utf-16be", true); + default: + return { + text: payload.toString("utf8"), + complete: isStrictUtf8(payload), + }; + } +}; /** The text a binary plist stores; `complete` is false when part of it went unread. */ export interface BinaryPlistText { @@ -256,8 +289,8 @@ export const decodeBinaryPlistText = ( } decoded += stored.payload.length; const text = storedPlistText(stored.type, stored.payload); - if (text === null) complete = false; - else texts.push(text); + if (text?.complete !== true) complete = false; + if (text !== null) texts.push(text.text); } return { text: texts.join("\n"), complete }; };