@futpib/parser 1.0.3 → 1.0.6

package/src/arbitraryZig.ts

@@ -0,0 +1,380 @@
+import * as fc from 'fast-check';
+import {
+	type ZigRoot,
+	type ZigExpression,
+	type ZigTypeExpression,
+	type ZigStatement,
+	type ZigFnParam,
+	type ZigContainerMember,
+	type ZigBlockExpr,
+} from './zig.js';
+
+const zigKeywords = new Set([
+	'addrspace', 'align', 'allowzero', 'and', 'anyframe', 'anytype',
+	'asm', 'async', 'await', 'break', 'callconv', 'catch', 'comptime',
+	'const', 'continue', 'defer', 'else', 'enum', 'errdefer', 'error',
+	'export', 'extern', 'false', 'fn', 'for', 'if', 'inline',
+	'linksection', 'noalias', 'nosuspend', 'null', 'opaque', 'or',
+	'orelse', 'packed', 'pub', 'resume', 'return', 'struct',
+	'suspend', 'switch', 'test', 'threadlocal', 'true', 'try',
+	'undefined', 'union', 'unreachable', 'var', 'volatile', 'while',
+]);
+
+const arbitraryZigIdentifier = fc.stringMatching(/^[a-zA-Z_][a-zA-Z0-9_]*$/).filter(
+	id => !zigKeywords.has(id),
+);
+
+// Leaf expressions - safe to use as operands without precedence issues
+
+const arbitraryIdentifier: fc.Arbitrary<ZigExpression> = fc.record({
+	type: fc.constant('Identifier' as const),
+	name: arbitraryZigIdentifier,
+});
+
+const arbitraryIntegerLiteral: fc.Arbitrary<ZigExpression> = fc.record({
+	type: fc.constant('IntegerLiteral' as const),
+	value: fc.nat({ max: 999 }).map(n => String(n)),
+});
+
+const arbitraryStringLiteral: fc.Arbitrary<ZigExpression> = fc.record({
+	type: fc.constant('StringLiteral' as const),
+	value: fc.oneof(
+		fc.stringMatching(/^[a-zA-Z0-9 ]*$/),
+		fc.constantFrom('hello\nworld', 'tab\there', 'back\\slash', 'quote\ttab\nnewline'),
+	),
+});
+
+const arbitraryCharLiteral: fc.Arbitrary<ZigExpression> = fc.record({
+	type: fc.constant('CharLiteral' as const),
+	value: fc.stringMatching(/^[a-zA-Z0-9]$/),
+});
+
+const arbitraryEnumLiteral: fc.Arbitrary<ZigExpression> = fc.record({
+	type: fc.constant('EnumLiteral' as const),
+	name: arbitraryZigIdentifier,
+});
+
+const arbitraryBoolLiteral: fc.Arbitrary<ZigExpression> = fc.record({
+	type: fc.constant('BoolLiteral' as const),
+	value: fc.boolean(),
+});
+
+const arbitraryNullLiteral: fc.Arbitrary<ZigExpression> = fc.constant({
+	type: 'NullLiteral' as const,
+});
+
+const arbitraryUndefinedLiteral: fc.Arbitrary<ZigExpression> = fc.constant({
+	type: 'UndefinedLiteral' as const,
+});
+
+const arbitraryLeafExpression: fc.Arbitrary<ZigExpression> = fc.oneof(
+	{ weight: 3, arbitrary: arbitraryIdentifier },
+	{ weight: 2, arbitrary: arbitraryIntegerLiteral },
+	{ weight: 1, arbitrary: arbitraryStringLiteral },
+	{ weight: 1, arbitrary: arbitraryCharLiteral },
+	{ weight: 1, arbitrary: arbitraryEnumLiteral },
+	{ weight: 1, arbitrary: arbitraryBoolLiteral },
+	{ weight: 1, arbitrary: arbitraryNullLiteral },
+	{ weight: 1, arbitrary: arbitraryUndefinedLiteral },
+);
+
+// Type expressions (simple ones for use in declarations)
+
+const arbitrarySimpleTypeExpression: fc.Arbitrary<ZigTypeExpression> = arbitraryIdentifier as fc.Arbitrary<ZigTypeExpression>;
+
+const arbitraryOptionalType: fc.Arbitrary<ZigTypeExpression> = fc.record({
+	type: fc.constant('OptionalType' as const),
+	child: arbitrarySimpleTypeExpression,
+});
+
+const arbitraryPointerType: fc.Arbitrary<ZigTypeExpression> = fc.record({
+	type: fc.constant('PointerType' as const),
+	size: fc.oneof(
+		fc.constant('one' as const),
+		fc.constant('many' as const),
+		fc.constant('slice' as const),
+	),
+	isConst: fc.boolean(),
+	child: arbitrarySimpleTypeExpression,
+});
+
+const arbitraryErrorUnionType: fc.Arbitrary<ZigTypeExpression> = fc.record({
+	type: fc.constant('ErrorUnionType' as const),
+	error: arbitrarySimpleTypeExpression,
+	payload: arbitrarySimpleTypeExpression,
+});
+
+const arbitraryTypeExpression: fc.Arbitrary<ZigTypeExpression> = fc.oneof(
+	{ weight: 4, arbitrary: arbitrarySimpleTypeExpression },
+	{ weight: 1, arbitrary: arbitraryOptionalType },
+	{ weight: 1, arbitrary: arbitraryPointerType },
+	{ weight: 1, arbitrary: arbitraryErrorUnionType },
+);
+
+// Compound expressions (use leaf operands to avoid precedence issues)
+
+const arbitraryBuiltinCallExpr: fc.Arbitrary<ZigExpression> = fc.record({
+	type: fc.constant('BuiltinCallExpr' as const),
+	name: arbitraryZigIdentifier,
+	args: fc.array(arbitraryLeafExpression, { maxLength: 2 }),
+});
+
+const arbitraryFieldAccessExpr: fc.Arbitrary<ZigExpression> = fc.record({
+	type: fc.constant('FieldAccessExpr' as const),
+	operand: arbitraryIdentifier,
+	member: arbitraryZigIdentifier,
+});
+
+const arbitraryCallExpr: fc.Arbitrary<ZigExpression> = fc.record({
+	type: fc.constant('CallExpr' as const),
+	callee: arbitraryIdentifier,
+	args: fc.array(arbitraryLeafExpression, { maxLength: 2 }),
+});
+
+const arbitraryIndexExpr: fc.Arbitrary<ZigExpression> = fc.record({
+	type: fc.constant('IndexExpr' as const),
+	operand: arbitraryIdentifier,
+	index: arbitraryLeafExpression,
+});
+
+const arbitraryGroupedExpr: fc.Arbitrary<ZigExpression> = fc.record({
+	type: fc.constant('GroupedExpr' as const),
+	inner: arbitraryLeafExpression,
+});
+
+const arbitraryErrorSetExpr: fc.Arbitrary<ZigExpression> = fc.record({
+	type: fc.constant('ErrorSetExpr' as const),
+	names: fc.array(arbitraryZigIdentifier, { minLength: 1, maxLength: 3 }),
+});
+
+const arbitraryStructInitExpr: fc.Arbitrary<ZigExpression> = fc.record({
+	type: fc.constant('StructInitExpr' as const),
+	fields: fc.array(
+		fc.record({
+			type: fc.constant('StructInitField' as const),
+			name: arbitraryZigIdentifier,
+			value: arbitraryLeafExpression,
+		}),
+		{ maxLength: 3 },
+	),
+});
+
+const arbitraryArrayInitExpr: fc.Arbitrary<ZigExpression> = fc.record({
+	type: fc.constant('ArrayInitExpr' as const),
+	elements: fc.array(arbitraryLeafExpression, { maxLength: 3 }),
+});
+
+const arbitraryExpression: fc.Arbitrary<ZigExpression> = fc.oneof(
+	{ weight: 5, arbitrary: arbitraryLeafExpression },
+	{ weight: 1, arbitrary: arbitraryBuiltinCallExpr },
+	{ weight: 1, arbitrary: arbitraryFieldAccessExpr },
+	{ weight: 1, arbitrary: arbitraryCallExpr },
+	{ weight: 1, arbitrary: arbitraryIndexExpr },
+	{ weight: 1, arbitrary: arbitraryGroupedExpr },
+	{ weight: 1, arbitrary: arbitraryErrorSetExpr },
+	{ weight: 1, arbitrary: arbitraryStructInitExpr },
+	{ weight: 1, arbitrary: arbitraryArrayInitExpr },
+);
+
+// Expressions safe to use in statement positions (don't start with keywords,
+// which would cause the parser's while/for/block label detector to throw
+// a non-recoverable error from zigIdentifierParser).
+const arbitraryNonKeywordLeafExpression: fc.Arbitrary<ZigExpression> = fc.oneof(
+	{ weight: 3, arbitrary: arbitraryIdentifier },
+	{ weight: 2, arbitrary: arbitraryIntegerLiteral },
+	{ weight: 1, arbitrary: arbitraryStringLiteral },
+	{ weight: 1, arbitrary: arbitraryCharLiteral },
+	{ weight: 1, arbitrary: arbitraryEnumLiteral },
+);
+
+const arbitraryStatementSafeExpression: fc.Arbitrary<ZigExpression> = fc.oneof(
+	{ weight: 5, arbitrary: arbitraryNonKeywordLeafExpression },
+	{ weight: 1, arbitrary: arbitraryBuiltinCallExpr },
+	{ weight: 1, arbitrary: arbitraryFieldAccessExpr },
+	{ weight: 1, arbitrary: arbitraryCallExpr },
+	{ weight: 1, arbitrary: arbitraryIndexExpr },
+	{ weight: 1, arbitrary: arbitraryGroupedExpr },
+);
+
+// Statements
+
+// Expression statements are bare expressions (no wrapper)
+const arbitraryExprStmt: fc.Arbitrary<ZigStatement> = arbitraryStatementSafeExpression as fc.Arbitrary<ZigStatement>;
+
+const arbitraryReturnStmt: fc.Arbitrary<ZigStatement> = fc.oneof(
+	fc.constant({ type: 'ReturnStmt' as const }),
+	fc.record({
+		type: fc.constant('ReturnStmt' as const),
+		value: arbitraryLeafExpression,
+	}),
+);
+
+const arbitraryBreakStmt: fc.Arbitrary<ZigStatement> = fc.constant({
+	type: 'BreakStmt' as const,
+});
+
+const arbitraryContinueStmt: fc.Arbitrary<ZigStatement> = fc.constant({
+	type: 'ContinueStmt' as const,
+});
+
+const arbitraryVarDeclStmt: fc.Arbitrary<ZigStatement> = fc.record({
+	type: fc.constant('VarDecl' as const),
+	isConst: fc.boolean(),
+	isPub: fc.constant(false),
+	isExtern: fc.constant(false),
+	isComptime: fc.constant(false),
+	isThreadlocal: fc.constant(false),
+	name: arbitraryZigIdentifier,
+}).chain(base =>
+	fc.oneof(
+		// With type and init
+		fc.record({
+			typeExpr: arbitraryTypeExpression,
+			initExpr: arbitraryLeafExpression,
+		}).map(extra => ({ ...base, ...extra })),
+		// With init only
+		fc.record({
+			initExpr: arbitraryLeafExpression,
+		}).map(extra => ({ ...base, ...extra })),
+		// With type only (extern style)
+		fc.record({
+			typeExpr: arbitraryTypeExpression,
+		}).map(extra => ({ ...base, isExtern: true, ...extra })),
+	),
+);
+
+const arbitraryAssignStmt: fc.Arbitrary<ZigStatement> = fc.record({
+	type: fc.constant('AssignStmt' as const),
+	target: arbitraryIdentifier as fc.Arbitrary<ZigExpression>,
+	operator: fc.oneof(
+		fc.constant('=' as const),
+		fc.constant('+=' as const),
+		fc.constant('-=' as const),
+		fc.constant('*=' as const),
+	),
+	value: arbitraryLeafExpression,
+});
+
+const arbitraryBlockStmt: fc.Arbitrary<ZigStatement> = fc.record({
+	type: fc.constant('BlockExpr' as const),
+	statements: fc.constant([] as ZigStatement[]),
+}) as fc.Arbitrary<ZigStatement>;
+
+const arbitraryDeferStmt: fc.Arbitrary<ZigStatement> = fc.record({
+	type: fc.constant('DeferStmt' as const),
+	isErrdefer: fc.boolean(),
+	body: arbitraryStatementSafeExpression as fc.Arbitrary<ZigStatement>,
+});
+
+const arbitraryStatement: fc.Arbitrary<ZigStatement> = fc.oneof(
+	{ weight: 2, arbitrary: arbitraryExprStmt },
+	{ weight: 2, arbitrary: arbitraryVarDeclStmt },
+	{ weight: 2, arbitrary: arbitraryReturnStmt },
+	{ weight: 1, arbitrary: arbitraryAssignStmt },
+	{ weight: 1, arbitrary: arbitraryBreakStmt },
+	{ weight: 1, arbitrary: arbitraryContinueStmt },
+	{ weight: 1, arbitrary: arbitraryBlockStmt },
+	{ weight: 1, arbitrary: arbitraryDeferStmt },
+);
+
+// Function parameters
+
+const arbitraryFnParam: fc.Arbitrary<ZigFnParam> = fc.oneof(
+	// Named parameter
+	fc.record({
+		type: fc.constant('FnParam' as const),
+		name: arbitraryZigIdentifier,
+		isComptime: fc.constant(false),
+		isNoalias: fc.constant(false),
+		typeExpr: arbitraryTypeExpression,
+	}),
+	// Unnamed parameter
+	fc.record({
+		type: fc.constant('FnParam' as const),
+		isComptime: fc.constant(false),
+		isNoalias: fc.constant(false),
+		typeExpr: arbitraryTypeExpression,
+	}),
+);
+
+// Block expression
+
+const arbitraryBlockExpr: fc.Arbitrary<ZigBlockExpr> = fc.record({
+	type: fc.constant('BlockExpr' as const),
+	statements: fc.array(arbitraryStatement, { maxLength: 3 }),
+});
+
+// Top-level declarations
+
+const arbitraryFnDecl: fc.Arbitrary<ZigContainerMember> = fc.oneof(
+	// Function with body
+	fc.record({
+		type: fc.constant('FnDecl' as const),
+		isPub: fc.boolean(),
+		isExtern: fc.constant(false),
+		isExport: fc.constant(false),
+		isInline: fc.constant(false),
+		isComptime: fc.constant(false),
+		name: arbitraryZigIdentifier,
+		params: fc.array(arbitraryFnParam, { maxLength: 2 }),
+		returnType: arbitraryTypeExpression,
+		body: arbitraryBlockExpr,
+	}),
+	// Extern function (no body)
+	fc.record({
+		type: fc.constant('FnDecl' as const),
+		isPub: fc.boolean(),
+		isExtern: fc.constant(true),
+		isExport: fc.constant(false),
+		isInline: fc.constant(false),
+		isComptime: fc.constant(false),
+		name: arbitraryZigIdentifier,
+		params: fc.array(arbitraryFnParam, { maxLength: 2 }),
+		returnType: arbitraryTypeExpression,
+	}),
+);
+
+const arbitraryVarDecl: fc.Arbitrary<ZigContainerMember> = fc.record({
+	type: fc.constant('VarDecl' as const),
+	isConst: fc.boolean(),
+	isPub: fc.boolean(),
+	isExtern: fc.constant(false),
+	isComptime: fc.constant(false),
+	isThreadlocal: fc.constant(false),
+	name: arbitraryZigIdentifier,
+}).chain(base =>
+	fc.oneof(
+		// With type and init
+		fc.record({
+			typeExpr: arbitraryTypeExpression,
+			initExpr: arbitraryLeafExpression,
+		}).map(extra => ({ ...base, ...extra })),
+		// With init only
+		fc.record({
+			initExpr: arbitraryLeafExpression,
+		}).map(extra => ({ ...base, ...extra })),
+	),
+);
+
+const arbitraryTestDecl: fc.Arbitrary<ZigContainerMember> = fc.oneof(
+	fc.record({
+		type: fc.constant('TestDecl' as const),
+		name: fc.stringMatching(/^[a-zA-Z0-9 ]+$/),
+		body: arbitraryBlockExpr,
+	}),
+	fc.record({
+		type: fc.constant('TestDecl' as const),
+		body: arbitraryBlockExpr,
+	}),
+);
+
+const arbitraryContainerMember: fc.Arbitrary<ZigContainerMember> = fc.oneof(
+	{ weight: 3, arbitrary: arbitraryFnDecl },
+	{ weight: 3, arbitrary: arbitraryVarDecl },
+	{ weight: 1, arbitrary: arbitraryTestDecl },
+);
+
+export const arbitraryZigRoot: fc.Arbitrary<ZigRoot> = fc.record({
+	type: fc.constant('Root' as const),
+	members: fc.array(arbitraryContainerMember, { minLength: 1, maxLength: 4 }),
+});

package/src/arrayParser.ts

@@ -8,7 +8,7 @@ export const createArrayParser = <ElementOutput, Sequence>(
 		const elements: ElementOutput[] = [];
 
 		while (true) {
-			const elementParserContext = parserContext.lookahead();
+			using elementParserContext = parserContext.lookahead();
 			const initialPosition = elementParserContext.position;
 			try {
 				const element = await elementParser(elementParserContext);
@@ -24,8 +24,6 @@ export const createArrayParser = <ElementOutput, Sequence>(
 				}
 
 				throw error;
-			} finally {
-				elementParserContext.dispose();
 			}
 		}
 	};

package/src/backsmali.ts

@@ -7,19 +7,22 @@ import { smaliClass } from './smali.js';
 export async function baksmaliClass(
 	dexStream: Uint8Array | AsyncIterable<Uint8Array>,
 	smaliFilePath: string,
+	dexEntry?: string,
 ): Promise<string> {
 	const inputFilePath = temporaryFile();
 	const outputDirectoryPath = temporaryDirectory();
 
 	await fs.writeFile(inputFilePath, dexStream);
 
+	const inputPath = dexEntry ? `${inputFilePath}/${dexEntry}` : inputFilePath;
+
 	await execa('baksmali', [
 		'disassemble',
 		'--classes',
 		'L' + smaliFilePath + ';',
 		'--output',
 		outputDirectoryPath,
-		inputFilePath,
+		inputPath,
 	]);
 
 	await fs.unlink(inputFilePath);
@@ -38,20 +41,23 @@ export async function baksmaliClass(
 export async function backsmaliSmaliIsolateClass(
 	dexStream: Uint8Array | AsyncIterable<Uint8Array>,
 	smaliFilePath: string,
+	dexEntry?: string,
 ): Promise<Uint8Array> {
-	const smali = await baksmaliClass(dexStream, smaliFilePath);
+	const smali = await baksmaliClass(dexStream, smaliFilePath, dexEntry);
 	return smaliClass(smali);
 }
 
-export async function baksmaliListClasses(dexStream: Uint8Array | AsyncIterable<Uint8Array>): Promise<string[]> {
+export async function baksmaliListClasses(dexStream: Uint8Array | AsyncIterable<Uint8Array>, dexEntry?: string): Promise<string[]> {
 	const inputFilePath = temporaryFile();
 
 	await fs.writeFile(inputFilePath, dexStream);
 
+	const inputPath = dexEntry ? `${inputFilePath}/${dexEntry}` : inputFilePath;
+
 	const result = await execa('baksmali', [
 		'list',
 		'classes',
-		inputFilePath,
+		inputPath,
 	]);
 
 	await fs.unlink(inputFilePath);
@@ -72,3 +78,28 @@ export async function baksmaliListClasses(dexStream: Uint8Array | AsyncIterable<
 
 	return classes;
 }
+
+export async function baksmaliListDexFiles(apkStream: Uint8Array | AsyncIterable<Uint8Array>): Promise<string[]> {
+	const inputFilePath = temporaryFile();
+
+	await fs.writeFile(inputFilePath, apkStream);
+
+	const result = await execa('baksmali', [
+		'list',
+		'dex',
+		inputFilePath,
+	]);
+
+	await fs.unlink(inputFilePath);
+
+	if (result.stderr) {
+		throw new Error(`baksmali error: ${result.stderr}`);
+	}
+
+	const dexFiles = result.stdout
+		.split('\n')
+		.filter(line => line.trim())
+		.map(line => line.trim());
+
+	return dexFiles;
+}

package/src/bash.ts

@@ -0,0 +1,127 @@
+// Word: a single argument/token (may contain expansions)
+export type BashWord = {
+	parts: BashWordPart[];
+};
+
+export type BashWordPart =
+	| BashWordPartLiteral
+	| BashWordPartSingleQuoted
+	| BashWordPartDoubleQuoted
+	| BashWordPartVariable
+	| BashWordPartVariableBraced
+	| BashWordPartCommandSubstitution
+	| BashWordPartBacktickSubstitution
+	| BashWordPartArithmeticExpansion
+	| BashWordPartProcessSubstitution;
+
+export type BashWordPartLiteral = {
+	type: 'literal';
+	value: string;
+};
+
+export type BashWordPartSingleQuoted = {
+	type: 'singleQuoted';
+	value: string;
+};
+
+export type BashWordPartDoubleQuoted = {
+	type: 'doubleQuoted';
+	parts: BashWordPart[];
+};
+
+export type BashWordPartVariable = {
+	type: 'variable';
+	name: string;
+};
+
+export type BashWordPartVariableBraced = {
+	type: 'variableBraced';
+	name: string;
+	operator?: string;
+	operand?: BashWord;
+};
+
+export type BashWordPartCommandSubstitution = {
+	type: 'commandSubstitution';
+	command: BashCommand;
+};
+
+export type BashWordPartBacktickSubstitution = {
+	type: 'backtickSubstitution';
+	command: BashCommand;
+};
+
+export type BashWordPartArithmeticExpansion = {
+	type: 'arithmeticExpansion';
+	expression: string;
+};
+
+export type BashWordPartProcessSubstitution = {
+	type: 'processSubstitution';
+	direction: '<' | '>';
+	command: BashCommand;
+};
+
+// Redirect: file descriptor operations
+export type BashRedirect = {
+	fd?: number;
+	operator: '>' | '>>' | '<' | '<<' | '<<<' | '>&' | '<&' | '>|';
+	target: BashWord | BashHereDoc;
+};
+
+export type BashHereDoc = {
+	type: 'hereDoc';
+	delimiter: string;
+	content: string;
+	quoted: boolean;
+};
+
+// Assignment
+export type BashAssignment = {
+	name: string;
+	value?: BashWord;
+};
+
+// Simple command: name + args + redirects
+export type BashSimpleCommand = {
+	type: 'simple';
+	name?: BashWord;
+	args: BashWord[];
+	redirects: BashRedirect[];
+	assignments: BashAssignment[];
+};
+
+// Compound commands (structural syntax only)
+export type BashSubshell = {
+	type: 'subshell';
+	body: BashCommand;
+};
+
+export type BashBraceGroup = {
+	type: 'braceGroup';
+	body: BashCommand;
+};
+
+export type BashCommandUnit = BashSimpleCommand | BashSubshell | BashBraceGroup;
+
+// Pipeline: cmd1 | cmd2 | cmd3
+export type BashPipeline = {
+	type: 'pipeline';
+	negated: boolean;
+	commands: BashCommandUnit[];
+};
+
+export type BashCommandListSeparator = '&&' | '||' | ';' | '&' | '\n';
+
+// Command list: pipelines connected by && || ; &
+export type BashCommandList = {
+	type: 'list';
+	entries: {
+		pipeline: BashPipeline;
+		separator?: BashCommandListSeparator;
+	}[];
+};
+
+// Top-level
+export type BashCommand = BashCommandList;
+export type BashScript = BashCommand;