wpd-codec 0.0.0 → 1.0.1

package/dist/stream/tokenise.cjs

@@ -0,0 +1,125 @@
+Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
+const require_errors = require("../errors.cjs");
+const require_bytes_view = require("../bytes/view.cjs");
+//#region src/stream/tokenise.ts
+const FIRST_SINGLE_BYTE_FUNCTION = 128;
+const FIRST_VARIABLE_FUNCTION = 208;
+const FIRST_FIXED_FUNCTION = 240;
+const FIXED_FUNCTION_SIZES = [
+	4,
+	5,
+	3,
+	3,
+	3,
+	3,
+	4,
+	4,
+	4,
+	5,
+	5,
+	6,
+	6,
+	8,
+	8
+];
+const MIN_VARIABLE_FUNCTION_SIZE = 10;
+const VARIABLE_FUNCTION_TRAILER_SIZE = 3;
+function readVariableFunction(bytes, offset, limit) {
+	const group = require_bytes_view.byteAt(bytes, offset);
+	const subgroup = require_bytes_view.byteAt(bytes, offset + 1);
+	const size = require_bytes_view.uint16At(bytes, offset + 2);
+	const flags = require_bytes_view.byteAt(bytes, offset + 4);
+	if (size < MIN_VARIABLE_FUNCTION_SIZE) throw new require_errors.WpdFormatError(`The variable-length function 0x${group.toString(16).toUpperCase()} at offset ${offset} declares a size of ${size}, below the ${MIN_VARIABLE_FUNCTION_SIZE} bytes its own gates and fields occupy.`);
+	const end = offset + size;
+	if (end > limit) throw new require_errors.WpdFormatError(`The variable-length function 0x${group.toString(16).toUpperCase()} at offset ${offset} declares a size of ${size}, which runs past the end of the document area at offset ${limit}.`);
+	let cursor = offset + 5;
+	const prefixIds = [];
+	if ((flags & 128) !== 0) {
+		const prefixIdCount = require_bytes_view.byteAt(bytes, cursor);
+		cursor += 1;
+		for (let index = 0; index < prefixIdCount; index += 1) {
+			prefixIds.push(require_bytes_view.uint16At(bytes, cursor));
+			cursor += 2;
+		}
+	}
+	const nonDeletableSize = require_bytes_view.uint16At(bytes, cursor);
+	cursor += 2;
+	const availableForData = end - VARIABLE_FUNCTION_TRAILER_SIZE - cursor;
+	if (nonDeletableSize > availableForData) throw new require_errors.WpdFormatError(`The variable-length function 0x${group.toString(16).toUpperCase()} subgroup ${subgroup} at offset ${offset} declares ${nonDeletableSize} bytes of non-deletable data, but only ${availableForData} remain inside its own ${size}-byte extent.`);
+	const nonDeletable = require_bytes_view.sliceAt(bytes, cursor, nonDeletableSize);
+	const trailingSize = require_bytes_view.uint16At(bytes, end - 3);
+	if (trailingSize !== size) throw new require_errors.WpdFormatError(`The variable-length function 0x${group.toString(16).toUpperCase()} at offset ${offset} opens with size ${size} but closes with size ${trailingSize}.`);
+	const endGate = require_bytes_view.byteAt(bytes, end - 1);
+	if (endGate !== group) throw new require_errors.WpdFormatError(`The variable-length function at offset ${offset} opens with gate 0x${group.toString(16).toUpperCase()} but closes with 0x${endGate.toString(16).toUpperCase()}.`);
+	return {
+		token: {
+			kind: "variableFunction",
+			group,
+			subgroup,
+			size,
+			flags,
+			prefixIds,
+			nonDeletable
+		},
+		nextOffset: end
+	};
+}
+function readFixedFunction(bytes, offset, limit) {
+	const code = require_bytes_view.byteAt(bytes, offset);
+	const size = FIXED_FUNCTION_SIZES[code - 240];
+	if (size === void 0) throw new require_errors.WpdFormatError(`Function code 0x${code.toString(16).toUpperCase()} at offset ${offset} cannot appear in a document: -1 is reserved and has no assigned size.`);
+	const end = offset + size;
+	if (end > limit) throw new require_errors.WpdFormatError(`The ${size}-byte fixed-length function 0x${code.toString(16).toUpperCase()} at offset ${offset} runs past the end of the document area at offset ${limit}.`);
+	const endGate = require_bytes_view.byteAt(bytes, end - 1);
+	if (endGate !== code) throw new require_errors.WpdFormatError(`The fixed-length function at offset ${offset} opens with gate 0x${code.toString(16).toUpperCase()} but closes with 0x${endGate.toString(16).toUpperCase()}.`);
+	return {
+		token: {
+			kind: "fixedFunction",
+			code,
+			data: require_bytes_view.sliceAt(bytes, offset + 1, size - 2)
+		},
+		nextOffset: end
+	};
+}
+function tokeniseDocumentArea(bytes, offset, limit = bytes.length) {
+	const tokens = [];
+	let cursor = offset;
+	while (cursor < limit) {
+		const byte = require_bytes_view.byteAt(bytes, cursor);
+		if (byte === 0) {
+			cursor += 1;
+			continue;
+		}
+		if (byte < 128) {
+			tokens.push({
+				kind: "character",
+				byte
+			});
+			cursor += 1;
+			continue;
+		}
+		if (byte < 208) {
+			tokens.push({
+				kind: "singleByteFunction",
+				code: byte
+			});
+			cursor += 1;
+			continue;
+		}
+		if (byte < 240) {
+			const { token, nextOffset } = readVariableFunction(bytes, cursor, limit);
+			tokens.push(token);
+			cursor = nextOffset;
+			continue;
+		}
+		const { token, nextOffset } = readFixedFunction(bytes, cursor, limit);
+		tokens.push(token);
+		cursor = nextOffset;
+	}
+	return tokens;
+}
+//#endregion
+exports.FIRST_FIXED_FUNCTION = FIRST_FIXED_FUNCTION;
+exports.FIRST_SINGLE_BYTE_FUNCTION = FIRST_SINGLE_BYTE_FUNCTION;
+exports.FIRST_VARIABLE_FUNCTION = FIRST_VARIABLE_FUNCTION;
+exports.tokeniseDocumentArea = tokeniseDocumentArea;

package/dist/stream/tokenise.d.cts

@@ -0,0 +1,30 @@
+//#region src/stream/tokenise.d.ts
+interface WpdCharacterToken {
+  readonly kind: "character";
+  readonly byte: number;
+}
+interface WpdSingleByteFunctionToken {
+  readonly kind: "singleByteFunction";
+  readonly code: number;
+}
+interface WpdVariableFunctionToken {
+  readonly kind: "variableFunction";
+  readonly group: number;
+  readonly subgroup: number;
+  readonly size: number;
+  readonly flags: number;
+  readonly prefixIds: readonly number[];
+  readonly nonDeletable: Uint8Array;
+}
+interface WpdFixedFunctionToken {
+  readonly kind: "fixedFunction";
+  readonly code: number;
+  readonly data: Uint8Array;
+}
+type WpdToken = WpdCharacterToken | WpdSingleByteFunctionToken | WpdVariableFunctionToken | WpdFixedFunctionToken;
+declare const FIRST_SINGLE_BYTE_FUNCTION = 128;
+declare const FIRST_VARIABLE_FUNCTION = 208;
+declare const FIRST_FIXED_FUNCTION = 240;
+declare function tokeniseDocumentArea(bytes: Uint8Array, offset: number, limit?: number): WpdToken[];
+//#endregion
+export { FIRST_FIXED_FUNCTION, FIRST_SINGLE_BYTE_FUNCTION, FIRST_VARIABLE_FUNCTION, WpdCharacterToken, WpdFixedFunctionToken, WpdSingleByteFunctionToken, WpdToken, WpdVariableFunctionToken, tokeniseDocumentArea };

package/dist/stream/tokenise.d.ts

@@ -0,0 +1,30 @@
+//#region src/stream/tokenise.d.ts
+interface WpdCharacterToken {
+  readonly kind: "character";
+  readonly byte: number;
+}
+interface WpdSingleByteFunctionToken {
+  readonly kind: "singleByteFunction";
+  readonly code: number;
+}
+interface WpdVariableFunctionToken {
+  readonly kind: "variableFunction";
+  readonly group: number;
+  readonly subgroup: number;
+  readonly size: number;
+  readonly flags: number;
+  readonly prefixIds: readonly number[];
+  readonly nonDeletable: Uint8Array;
+}
+interface WpdFixedFunctionToken {
+  readonly kind: "fixedFunction";
+  readonly code: number;
+  readonly data: Uint8Array;
+}
+type WpdToken = WpdCharacterToken | WpdSingleByteFunctionToken | WpdVariableFunctionToken | WpdFixedFunctionToken;
+declare const FIRST_SINGLE_BYTE_FUNCTION = 128;
+declare const FIRST_VARIABLE_FUNCTION = 208;
+declare const FIRST_FIXED_FUNCTION = 240;
+declare function tokeniseDocumentArea(bytes: Uint8Array, offset: number, limit?: number): WpdToken[];
+//#endregion
+export { FIRST_FIXED_FUNCTION, FIRST_SINGLE_BYTE_FUNCTION, FIRST_VARIABLE_FUNCTION, WpdCharacterToken, WpdFixedFunctionToken, WpdSingleByteFunctionToken, WpdToken, WpdVariableFunctionToken, tokeniseDocumentArea };

package/dist/stream/tokenise.js

@@ -0,0 +1,121 @@
+import { WpdFormatError } from "../errors.js";
+import { byteAt, sliceAt, uint16At } from "../bytes/view.js";
+//#region src/stream/tokenise.ts
+const FIRST_SINGLE_BYTE_FUNCTION = 128;
+const FIRST_VARIABLE_FUNCTION = 208;
+const FIRST_FIXED_FUNCTION = 240;
+const FIXED_FUNCTION_SIZES = [
+	4,
+	5,
+	3,
+	3,
+	3,
+	3,
+	4,
+	4,
+	4,
+	5,
+	5,
+	6,
+	6,
+	8,
+	8
+];
+const MIN_VARIABLE_FUNCTION_SIZE = 10;
+const VARIABLE_FUNCTION_TRAILER_SIZE = 3;
+function readVariableFunction(bytes, offset, limit) {
+	const group = byteAt(bytes, offset);
+	const subgroup = byteAt(bytes, offset + 1);
+	const size = uint16At(bytes, offset + 2);
+	const flags = byteAt(bytes, offset + 4);
+	if (size < MIN_VARIABLE_FUNCTION_SIZE) throw new WpdFormatError(`The variable-length function 0x${group.toString(16).toUpperCase()} at offset ${offset} declares a size of ${size}, below the ${MIN_VARIABLE_FUNCTION_SIZE} bytes its own gates and fields occupy.`);
+	const end = offset + size;
+	if (end > limit) throw new WpdFormatError(`The variable-length function 0x${group.toString(16).toUpperCase()} at offset ${offset} declares a size of ${size}, which runs past the end of the document area at offset ${limit}.`);
+	let cursor = offset + 5;
+	const prefixIds = [];
+	if ((flags & 128) !== 0) {
+		const prefixIdCount = byteAt(bytes, cursor);
+		cursor += 1;
+		for (let index = 0; index < prefixIdCount; index += 1) {
+			prefixIds.push(uint16At(bytes, cursor));
+			cursor += 2;
+		}
+	}
+	const nonDeletableSize = uint16At(bytes, cursor);
+	cursor += 2;
+	const availableForData = end - VARIABLE_FUNCTION_TRAILER_SIZE - cursor;
+	if (nonDeletableSize > availableForData) throw new WpdFormatError(`The variable-length function 0x${group.toString(16).toUpperCase()} subgroup ${subgroup} at offset ${offset} declares ${nonDeletableSize} bytes of non-deletable data, but only ${availableForData} remain inside its own ${size}-byte extent.`);
+	const nonDeletable = sliceAt(bytes, cursor, nonDeletableSize);
+	const trailingSize = uint16At(bytes, end - 3);
+	if (trailingSize !== size) throw new WpdFormatError(`The variable-length function 0x${group.toString(16).toUpperCase()} at offset ${offset} opens with size ${size} but closes with size ${trailingSize}.`);
+	const endGate = byteAt(bytes, end - 1);
+	if (endGate !== group) throw new WpdFormatError(`The variable-length function at offset ${offset} opens with gate 0x${group.toString(16).toUpperCase()} but closes with 0x${endGate.toString(16).toUpperCase()}.`);
+	return {
+		token: {
+			kind: "variableFunction",
+			group,
+			subgroup,
+			size,
+			flags,
+			prefixIds,
+			nonDeletable
+		},
+		nextOffset: end
+	};
+}
+function readFixedFunction(bytes, offset, limit) {
+	const code = byteAt(bytes, offset);
+	const size = FIXED_FUNCTION_SIZES[code - 240];
+	if (size === void 0) throw new WpdFormatError(`Function code 0x${code.toString(16).toUpperCase()} at offset ${offset} cannot appear in a document: -1 is reserved and has no assigned size.`);
+	const end = offset + size;
+	if (end > limit) throw new WpdFormatError(`The ${size}-byte fixed-length function 0x${code.toString(16).toUpperCase()} at offset ${offset} runs past the end of the document area at offset ${limit}.`);
+	const endGate = byteAt(bytes, end - 1);
+	if (endGate !== code) throw new WpdFormatError(`The fixed-length function at offset ${offset} opens with gate 0x${code.toString(16).toUpperCase()} but closes with 0x${endGate.toString(16).toUpperCase()}.`);
+	return {
+		token: {
+			kind: "fixedFunction",
+			code,
+			data: sliceAt(bytes, offset + 1, size - 2)
+		},
+		nextOffset: end
+	};
+}
+function tokeniseDocumentArea(bytes, offset, limit = bytes.length) {
+	const tokens = [];
+	let cursor = offset;
+	while (cursor < limit) {
+		const byte = byteAt(bytes, cursor);
+		if (byte === 0) {
+			cursor += 1;
+			continue;
+		}
+		if (byte < 128) {
+			tokens.push({
+				kind: "character",
+				byte
+			});
+			cursor += 1;
+			continue;
+		}
+		if (byte < 208) {
+			tokens.push({
+				kind: "singleByteFunction",
+				code: byte
+			});
+			cursor += 1;
+			continue;
+		}
+		if (byte < 240) {
+			const { token, nextOffset } = readVariableFunction(bytes, cursor, limit);
+			tokens.push(token);
+			cursor = nextOffset;
+			continue;
+		}
+		const { token, nextOffset } = readFixedFunction(bytes, cursor, limit);
+		tokens.push(token);
+		cursor = nextOffset;
+	}
+	return tokens;
+}
+//#endregion
+export { FIRST_FIXED_FUNCTION, FIRST_SINGLE_BYTE_FUNCTION, FIRST_VARIABLE_FUNCTION, tokeniseDocumentArea };

package/package.json

@@ -1,5 +1,99 @@
 {
   "name": "wpd-codec",
-  "version": "0.0.0",
-  "private": false
+  "version": "1.0.1",
+  "description": "Hand-written read-only WordPerfect 6.x-X6 (.wpd) reader over document-schema.js's ContentDocument",
+  "type": "module",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/ExaDev/documents.js.git",
+    "directory": "packages/wpd-codec"
+  },
+  "homepage": "https://github.com/ExaDev/documents.js/tree/main/packages/wpd-codec",
+  "bugs": {
+    "url": "https://github.com/ExaDev/documents.js/issues"
+  },
+  "exports": {
+    ".": {
+      "types": {
+        "import": "./dist/index.d.ts",
+        "require": "./dist/index.d.cts"
+      },
+      "import": "./dist/index.js",
+      "require": "./dist/index.cjs"
+    },
+    "./*": {
+      "types": {
+        "import": "./dist/*.d.ts",
+        "require": "./dist/*.d.cts"
+      },
+      "import": "./dist/*.js",
+      "require": "./dist/*.cjs"
+    }
+  },
+  "main": "./dist/index.cjs",
+  "module": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "files": [
+    "dist"
+  ],
+  "publishConfig": {
+    "access": "public",
+    "provenance": true,
+    "registry": "https://registry.npmjs.org/"
+  },
+  "sideEffects": false,
+  "engines": {
+    "node": ">=20"
+  },
+  "scripts": {
+    "build": "turbo run _build",
+    "_build": "tsdown",
+    "prepublishOnly": "pnpm run lint && pnpm run typecheck && tsdown && publint && attw --pack",
+    "lint": "turbo run _lint",
+    "_lint": "eslint . --fix --cache --max-warnings 0",
+    "typecheck": "turbo run _typecheck _typecheck:node _typecheck:attw",
+    "_typecheck": "tsc -p tsconfig.json",
+    "_typecheck:node": "tsc -p tsconfig.node.json",
+    "_typecheck:attw": "attw --pack",
+    "test": "turbo run _test",
+    "_test": "vitest run --project unit",
+    "test:workers": "turbo run _test:workers",
+    "_test:workers": "vitest run --config vitest.workers.config.ts",
+    "test:watch": "vitest --project unit",
+    "test:coverage": "turbo run _test:coverage",
+    "_test:coverage": "vitest run --project unit --coverage",
+    "test:smoke": "turbo run _test:smoke",
+    "_test:smoke": "vitest run --project smoke",
+    "prepare": "husky",
+    "release": "semantic-release"
+  },
+  "keywords": [
+    "wordperfect",
+    "wpd",
+    "codec",
+    "legacy",
+    "document",
+    "zod"
+  ],
+  "license": "MIT",
+  "packageManager": "pnpm@11.6.0",
+  "dependencies": {
+    "archive-codec": "^1.3.0",
+    "document-schema.js": "^5.4.0",
+    "zod": "^4.4.3"
+  },
+  "devDependencies": {
+    "@arethetypeswrong/cli": "^0.18.5",
+    "@cloudflare/vitest-pool-workers": "^0.21.2",
+    "@types/node": "^26.1.1",
+    "@vitest/coverage-v8": "^4.1.10",
+    "eslint": "^10.8.0",
+    "husky": "^9.1.7",
+    "publint": "^0.3.21",
+    "semantic-release": "^25.0.8",
+    "tsdown": "^0.22.13",
+    "turbo": "^2.10.8",
+    "typescript": "^6.0.3",
+    "vitest": "^4.1.10"
+  }
 }