wpd-codec 1.0.1 → 1.1.0

package/dist/index.d.ts

@@ -1,13 +1,19 @@
-import { i as WpdDiagnosticSink, n as WpdDiagnostic, r as WpdDiagnosticCodes, t as NOOP_WPD_DIAGNOSTIC_SINK } from "./diagnostics-Cp0HKphg.js";
+import { i as WpdDiagnosticSink, n as WpdDiagnostic, r as WpdDiagnosticCodes, t as NOOP_WPD_DIAGNOSTIC_SINK } from "./diagnostics-CR0pXOyb.js";
 import { ReadWpdOptions, readWpd, readWpdContent } from "./read.js";
 import { WpdBytesSchema, wpdContentCodec } from "./codec.js";
 import { WPD_FILE_ID, WPD_PREFIX_HEADER_SIZE, WpdFileHeader, hasWordPerfectFileId, readFileHeader } from "./container/header.js";
 import { PACKET_TYPE_DESIRED_FONT_DESCRIPTOR, WPD_INDEX_RECORD_SIZE, WpdPrefixPacket, packetByPrefixId, readPrefixPackets, readTypefaceName } from "./container/prefix.js";
 import { PERFECT_OFFICE_MAIN_STREAM, PERFECT_OFFICE_OBJECTS_STORAGE, WpdDocumentContainer, openWpdDocument } from "./container/container.js";
+import { PACKET_TYPE_EXTENDED_DOCUMENT_SUMMARY, readDocumentSummary } from "./container/summary.js";
 import { WpdEncryptedDocumentError, WpdFormatError, WpdNotAWordPerfectFileError, WpdUnsupportedVersionError } from "./errors.js";
 import { WPD_FILE_EXTENSION, WPD_MEDIA_TYPE } from "./format.js";
 import { FIRST_FIXED_FUNCTION, FIRST_SINGLE_BYTE_FUNCTION, FIRST_VARIABLE_FUNCTION, WpdCharacterToken, WpdFixedFunctionToken, WpdSingleByteFunctionToken, WpdToken, WpdVariableFunctionToken, tokeniseDocumentArea } from "./stream/tokenise.js";
 import { EOL_GROUP, FIRST_SINGLE_BYTE_EOL, LAST_SINGLE_BYTE_EOL, WpdEolMapping, eolMappingForSubfunction, isSingleByteEol, subfunctionForSingleByteEol } from "./stream/eol.js";
 import { ATTRIBUTE_OFF, ATTRIBUTE_ON, WpdAttribute, WpdAttributeCode, WpdRunAttributes, decodeAttributeByte, runAttributesFrom } from "./stream/attributes.js";
 import { FIRST_ASCII_CHARACTER, LAST_CHARACTER, UNMAPPED_CHARACTER, decodeSingleByteCharacter, decodeWordString, decodeWpCharacter } from "./stream/characters.js";
-export { ATTRIBUTE_OFF, ATTRIBUTE_ON, EOL_GROUP, FIRST_ASCII_CHARACTER, FIRST_FIXED_FUNCTION, FIRST_SINGLE_BYTE_EOL, FIRST_SINGLE_BYTE_FUNCTION, FIRST_VARIABLE_FUNCTION, LAST_CHARACTER, LAST_SINGLE_BYTE_EOL, NOOP_WPD_DIAGNOSTIC_SINK, PACKET_TYPE_DESIRED_FONT_DESCRIPTOR, PERFECT_OFFICE_MAIN_STREAM, PERFECT_OFFICE_OBJECTS_STORAGE, type ReadWpdOptions, UNMAPPED_CHARACTER, WPD_FILE_EXTENSION, WPD_FILE_ID, WPD_INDEX_RECORD_SIZE, WPD_MEDIA_TYPE, WPD_PREFIX_HEADER_SIZE, WpdAttribute, type WpdAttributeCode, WpdBytesSchema, type WpdCharacterToken, type WpdDiagnostic, WpdDiagnosticCodes, type WpdDiagnosticSink, type WpdDocumentContainer, WpdEncryptedDocumentError, type WpdEolMapping, type WpdFileHeader, type WpdFixedFunctionToken, WpdFormatError, WpdNotAWordPerfectFileError, type WpdPrefixPacket, type WpdRunAttributes, type WpdSingleByteFunctionToken, type WpdToken, WpdUnsupportedVersionError, type WpdVariableFunctionToken, decodeAttributeByte, decodeSingleByteCharacter, decodeWordString, decodeWpCharacter, eolMappingForSubfunction, hasWordPerfectFileId, isSingleByteEol, openWpdDocument, packetByPrefixId, readFileHeader, readPrefixPackets, readTypefaceName, readWpd, readWpdContent, runAttributesFrom, subfunctionForSingleByteEol, tokeniseDocumentArea, wpdContentCodec };
+import { POINTS_PER_INCH, WPU_PER_INCH, pointsFromWpu } from "./stream/units.js";
+import { COLUMN_GROUP, COLUMN_LEFT_MARGIN_SET, COLUMN_RIGHT_MARGIN_SET, DEFAULT_MARGIN_PT, DEFAULT_PAGE_HEIGHT_PT, DEFAULT_PAGE_WIDTH_PT, PAGE_BOTTOM_MARGIN_SET, PAGE_FORM, PAGE_GROUP, PAGE_TOP_MARGIN_SET, WpdPageForm, readMarginPt, readPageForm } from "./stream/page.js";
+import { TAB_GROUP, WpdTabEffect, tabEffectFor } from "./stream/tab.js";
+import { DISPLAY_NUMBER_GROUP, STYLE_GROUP, WpdStyleSemantics, isParagraphNumberDisplayOff, isParagraphNumberDisplayOn, isStyleScopeCloser, isStyleScopeOpener, readDisplayNumberLevel, readSystemStyleNumber, styleSemanticsFor } from "./stream/style.js";
+import { CELL_FILL_COLORS_SUBFUNCTION, CELL_INFORMATION_SUBFUNCTION, CELL_SPANNING_SUBFUNCTION, CHARACTER_DEFINE_TABLE_END, CHARACTER_TABLE_COLUMN, CHARACTER_TABLE_DEFINITION, ROW_INFORMATION_SUBFUNCTION, WpdCellFill, WpdCellInformation, WpdCellSpanning, WpdEmbeddedSubfunction, WpdEmbeddedSubfunctions, WpdRowInformation, findEmbeddedSubfunction, readCellFill, readCellInformation, readCellSpanning, readEmbeddedSubfunctions, readRowInformation, readTableColumnWidthPt } from "./stream/table.js";
+export { ATTRIBUTE_OFF, ATTRIBUTE_ON, CELL_FILL_COLORS_SUBFUNCTION, CELL_INFORMATION_SUBFUNCTION, CELL_SPANNING_SUBFUNCTION, CHARACTER_DEFINE_TABLE_END, CHARACTER_TABLE_COLUMN, CHARACTER_TABLE_DEFINITION, COLUMN_GROUP, COLUMN_LEFT_MARGIN_SET, COLUMN_RIGHT_MARGIN_SET, DEFAULT_MARGIN_PT, DEFAULT_PAGE_HEIGHT_PT, DEFAULT_PAGE_WIDTH_PT, DISPLAY_NUMBER_GROUP, EOL_GROUP, FIRST_ASCII_CHARACTER, FIRST_FIXED_FUNCTION, FIRST_SINGLE_BYTE_EOL, FIRST_SINGLE_BYTE_FUNCTION, FIRST_VARIABLE_FUNCTION, LAST_CHARACTER, LAST_SINGLE_BYTE_EOL, NOOP_WPD_DIAGNOSTIC_SINK, PACKET_TYPE_DESIRED_FONT_DESCRIPTOR, PACKET_TYPE_EXTENDED_DOCUMENT_SUMMARY, PAGE_BOTTOM_MARGIN_SET, PAGE_FORM, PAGE_GROUP, PAGE_TOP_MARGIN_SET, PERFECT_OFFICE_MAIN_STREAM, PERFECT_OFFICE_OBJECTS_STORAGE, POINTS_PER_INCH, ROW_INFORMATION_SUBFUNCTION, type ReadWpdOptions, STYLE_GROUP, TAB_GROUP, UNMAPPED_CHARACTER, WPD_FILE_EXTENSION, WPD_FILE_ID, WPD_INDEX_RECORD_SIZE, WPD_MEDIA_TYPE, WPD_PREFIX_HEADER_SIZE, WPU_PER_INCH, WpdAttribute, type WpdAttributeCode, WpdBytesSchema, type WpdCellFill, type WpdCellInformation, type WpdCellSpanning, type WpdCharacterToken, type WpdDiagnostic, WpdDiagnosticCodes, type WpdDiagnosticSink, type WpdDocumentContainer, type WpdEmbeddedSubfunction, type WpdEmbeddedSubfunctions, WpdEncryptedDocumentError, type WpdEolMapping, type WpdFileHeader, type WpdFixedFunctionToken, WpdFormatError, WpdNotAWordPerfectFileError, type WpdPageForm, type WpdPrefixPacket, type WpdRowInformation, type WpdRunAttributes, type WpdSingleByteFunctionToken, type WpdStyleSemantics, type WpdTabEffect, type WpdToken, WpdUnsupportedVersionError, type WpdVariableFunctionToken, decodeAttributeByte, decodeSingleByteCharacter, decodeWordString, decodeWpCharacter, eolMappingForSubfunction, findEmbeddedSubfunction, hasWordPerfectFileId, isParagraphNumberDisplayOff, isParagraphNumberDisplayOn, isSingleByteEol, isStyleScopeCloser, isStyleScopeOpener, openWpdDocument, packetByPrefixId, pointsFromWpu, readCellFill, readCellInformation, readCellSpanning, readDisplayNumberLevel, readDocumentSummary, readEmbeddedSubfunctions, readFileHeader, readMarginPt, readPageForm, readPrefixPackets, readRowInformation, readSystemStyleNumber, readTableColumnWidthPt, readTypefaceName, readWpd, readWpdContent, runAttributesFrom, styleSemanticsFor, subfunctionForSingleByteEol, tabEffectFor, tokeniseDocumentArea, wpdContentCodec };

package/dist/index.js

@@ -3,11 +3,17 @@ import { WPD_FILE_ID, WPD_PREFIX_HEADER_SIZE, hasWordPerfectFileId, readFileHead
 import { FIRST_ASCII_CHARACTER, LAST_CHARACTER, UNMAPPED_CHARACTER, decodeSingleByteCharacter, decodeWordString, decodeWpCharacter } from "./stream/characters.js";
 import { PACKET_TYPE_DESIRED_FONT_DESCRIPTOR, WPD_INDEX_RECORD_SIZE, packetByPrefixId, readPrefixPackets, readTypefaceName } from "./container/prefix.js";
 import { PERFECT_OFFICE_MAIN_STREAM, PERFECT_OFFICE_OBJECTS_STORAGE, openWpdDocument } from "./container/container.js";
+import { PACKET_TYPE_EXTENDED_DOCUMENT_SUMMARY, readDocumentSummary } from "./container/summary.js";
 import { NOOP_WPD_DIAGNOSTIC_SINK, WpdDiagnosticCodes } from "./diagnostics.js";
 import { ATTRIBUTE_OFF, ATTRIBUTE_ON, WpdAttribute, decodeAttributeByte, runAttributesFrom } from "./stream/attributes.js";
 import { EOL_GROUP, FIRST_SINGLE_BYTE_EOL, LAST_SINGLE_BYTE_EOL, eolMappingForSubfunction, isSingleByteEol, subfunctionForSingleByteEol } from "./stream/eol.js";
+import { POINTS_PER_INCH, WPU_PER_INCH, pointsFromWpu } from "./stream/units.js";
+import { COLUMN_GROUP, COLUMN_LEFT_MARGIN_SET, COLUMN_RIGHT_MARGIN_SET, DEFAULT_MARGIN_PT, DEFAULT_PAGE_HEIGHT_PT, DEFAULT_PAGE_WIDTH_PT, PAGE_BOTTOM_MARGIN_SET, PAGE_FORM, PAGE_GROUP, PAGE_TOP_MARGIN_SET, readMarginPt, readPageForm } from "./stream/page.js";
+import { DISPLAY_NUMBER_GROUP, STYLE_GROUP, isParagraphNumberDisplayOff, isParagraphNumberDisplayOn, isStyleScopeCloser, isStyleScopeOpener, readDisplayNumberLevel, readSystemStyleNumber, styleSemanticsFor } from "./stream/style.js";
+import { CELL_FILL_COLORS_SUBFUNCTION, CELL_INFORMATION_SUBFUNCTION, CELL_SPANNING_SUBFUNCTION, CHARACTER_DEFINE_TABLE_END, CHARACTER_TABLE_COLUMN, CHARACTER_TABLE_DEFINITION, ROW_INFORMATION_SUBFUNCTION, findEmbeddedSubfunction, readCellFill, readCellInformation, readCellSpanning, readEmbeddedSubfunctions, readRowInformation, readTableColumnWidthPt } from "./stream/table.js";
+import { TAB_GROUP, tabEffectFor } from "./stream/tab.js";
 import { FIRST_FIXED_FUNCTION, FIRST_SINGLE_BYTE_FUNCTION, FIRST_VARIABLE_FUNCTION, tokeniseDocumentArea } from "./stream/tokenise.js";
 import { readWpd, readWpdContent } from "./read.js";
 import { WpdBytesSchema, wpdContentCodec } from "./codec.js";
 import { WPD_FILE_EXTENSION, WPD_MEDIA_TYPE } from "./format.js";
-export { ATTRIBUTE_OFF, ATTRIBUTE_ON, EOL_GROUP, FIRST_ASCII_CHARACTER, FIRST_FIXED_FUNCTION, FIRST_SINGLE_BYTE_EOL, FIRST_SINGLE_BYTE_FUNCTION, FIRST_VARIABLE_FUNCTION, LAST_CHARACTER, LAST_SINGLE_BYTE_EOL, NOOP_WPD_DIAGNOSTIC_SINK, PACKET_TYPE_DESIRED_FONT_DESCRIPTOR, PERFECT_OFFICE_MAIN_STREAM, PERFECT_OFFICE_OBJECTS_STORAGE, UNMAPPED_CHARACTER, WPD_FILE_EXTENSION, WPD_FILE_ID, WPD_INDEX_RECORD_SIZE, WPD_MEDIA_TYPE, WPD_PREFIX_HEADER_SIZE, WpdAttribute, WpdBytesSchema, WpdDiagnosticCodes, WpdEncryptedDocumentError, WpdFormatError, WpdNotAWordPerfectFileError, WpdUnsupportedVersionError, decodeAttributeByte, decodeSingleByteCharacter, decodeWordString, decodeWpCharacter, eolMappingForSubfunction, hasWordPerfectFileId, isSingleByteEol, openWpdDocument, packetByPrefixId, readFileHeader, readPrefixPackets, readTypefaceName, readWpd, readWpdContent, runAttributesFrom, subfunctionForSingleByteEol, tokeniseDocumentArea, wpdContentCodec };
+export { ATTRIBUTE_OFF, ATTRIBUTE_ON, CELL_FILL_COLORS_SUBFUNCTION, CELL_INFORMATION_SUBFUNCTION, CELL_SPANNING_SUBFUNCTION, CHARACTER_DEFINE_TABLE_END, CHARACTER_TABLE_COLUMN, CHARACTER_TABLE_DEFINITION, COLUMN_GROUP, COLUMN_LEFT_MARGIN_SET, COLUMN_RIGHT_MARGIN_SET, DEFAULT_MARGIN_PT, DEFAULT_PAGE_HEIGHT_PT, DEFAULT_PAGE_WIDTH_PT, DISPLAY_NUMBER_GROUP, EOL_GROUP, FIRST_ASCII_CHARACTER, FIRST_FIXED_FUNCTION, FIRST_SINGLE_BYTE_EOL, FIRST_SINGLE_BYTE_FUNCTION, FIRST_VARIABLE_FUNCTION, LAST_CHARACTER, LAST_SINGLE_BYTE_EOL, NOOP_WPD_DIAGNOSTIC_SINK, PACKET_TYPE_DESIRED_FONT_DESCRIPTOR, PACKET_TYPE_EXTENDED_DOCUMENT_SUMMARY, PAGE_BOTTOM_MARGIN_SET, PAGE_FORM, PAGE_GROUP, PAGE_TOP_MARGIN_SET, PERFECT_OFFICE_MAIN_STREAM, PERFECT_OFFICE_OBJECTS_STORAGE, POINTS_PER_INCH, ROW_INFORMATION_SUBFUNCTION, STYLE_GROUP, TAB_GROUP, UNMAPPED_CHARACTER, WPD_FILE_EXTENSION, WPD_FILE_ID, WPD_INDEX_RECORD_SIZE, WPD_MEDIA_TYPE, WPD_PREFIX_HEADER_SIZE, WPU_PER_INCH, WpdAttribute, WpdBytesSchema, WpdDiagnosticCodes, WpdEncryptedDocumentError, WpdFormatError, WpdNotAWordPerfectFileError, WpdUnsupportedVersionError, decodeAttributeByte, decodeSingleByteCharacter, decodeWordString, decodeWpCharacter, eolMappingForSubfunction, findEmbeddedSubfunction, hasWordPerfectFileId, isParagraphNumberDisplayOff, isParagraphNumberDisplayOn, isSingleByteEol, isStyleScopeCloser, isStyleScopeOpener, openWpdDocument, packetByPrefixId, pointsFromWpu, readCellFill, readCellInformation, readCellSpanning, readDisplayNumberLevel, readDocumentSummary, readEmbeddedSubfunctions, readFileHeader, readMarginPt, readPageForm, readPrefixPackets, readRowInformation, readSystemStyleNumber, readTableColumnWidthPt, readTypefaceName, readWpd, readWpdContent, runAttributesFrom, styleSemanticsFor, subfunctionForSingleByteEol, tabEffectFor, tokeniseDocumentArea, wpdContentCodec };

package/dist/read.cjs

@@ -3,9 +3,14 @@ const require_bytes_view = require("./bytes/view.cjs");
 const require_stream_characters = require("./stream/characters.cjs");
 const require_container_prefix = require("./container/prefix.cjs");
 const require_container_container = require("./container/container.cjs");
+const require_container_summary = require("./container/summary.cjs");
 const require_diagnostics = require("./diagnostics.cjs");
 const require_stream_attributes = require("./stream/attributes.cjs");
 const require_stream_eol = require("./stream/eol.cjs");
+const require_stream_page = require("./stream/page.cjs");
+const require_stream_style = require("./stream/style.cjs");
+const require_stream_table = require("./stream/table.cjs");
+const require_stream_tab = require("./stream/tab.cjs");
 const require_stream_tokenise = require("./stream/tokenise.cjs");
 let document_schema_js = require("document-schema.js");
 //#region src/read.ts
@@ -21,6 +26,11 @@ const SOFT_END_OF_CENTER_ALIGN = 136;
 const HARD_END_OF_CENTER_ALIGN = 137;
 const START_OF_TEXT_TO_SKIP = 141;
 const END_OF_TEXT_TO_SKIP = 142;
+const CROSS_REFERENCE_GROUP = 213;
+const HEADER_FOOTER_GROUP = 214;
+const FOOTNOTE_ENDNOTE_GROUP = 215;
+const MERGE_GROUP = 222;
+const BOX_GROUP = 223;
 const PARAGRAPH_GROUP = 211;
 const PARAGRAPH_SET_JUSTIFICATION = 5;
 const CHARACTER_GROUP = 212;
@@ -38,10 +48,18 @@ const JUSTIFICATION = [
 ];
 const THREE_THOUSAND_SIX_HUNDREDTHS_PER_POINT = 50;
 const COLOR_COMPONENT_MAX = 255;
-const DEFAULT_MARGIN_PT = 72;
+const NO_SPAN = 1;
 function sameAttributes(a, b) {
 	return a.bold === b.bold && a.italic === b.italic && a.underline === b.underline && a.strike === b.strike;
 }
+function reportOnce(state, sink, code, message) {
+	if (state.reported.has(code)) return;
+	state.reported.add(code);
+	sink({
+		code,
+		message
+	});
+}
 function buildRun(state) {
 	return {
 		text: state.text,
@@ -59,20 +77,104 @@ function flushRun(state) {
 	state.runs.push(buildRun(state));
 	state.text = "";
 }
+function targetBlocks(state) {
+	return state.table === void 0 ? state.blocks : state.table.cellBlocks;
+}
 function flushParagraph(state) {
 	flushRun(state);
+	const alignment = state.pendingAlignment ?? state.alignment;
 	const paragraph = {
 		kind: "paragraph",
 		runs: state.runs.splice(0, state.runs.length),
-		...state.alignment === void 0 ? {} : { alignment: state.alignment }
+		...alignment === void 0 ? {} : { alignment },
+		...state.pendingHeadingLevel === void 0 ? {} : { headingLevel: state.pendingHeadingLevel },
+		...state.pendingListLevel === void 0 ? {} : { list: { level: state.pendingListLevel } }
 	};
-	state.blocks.push(paragraph);
+	state.pendingHeadingLevel = void 0;
+	state.pendingListLevel = void 0;
+	state.pendingAlignment = void 0;
+	targetBlocks(state).push(paragraph);
+}
+function flushParagraphIfContent(state) {
+	if (state.text.length === 0 && state.runs.length === 0) return;
+	flushParagraph(state);
+}
+function effectiveStyle(state) {
+	for (let index = state.styleScopes.length - 1; index >= 0; index -= 1) {
+		const scope = state.styleScopes[index];
+		if (scope !== void 0) return scope;
+	}
 }
 function appendText(state, text) {
-	if (state.skipDepth > 0) return;
+	if (state.skipDepth > 0 || state.numberDisplayDepth > 0) return;
+	if (state.pendingHeadingLevel === void 0 && state.pendingListLevel === void 0) {
+		const style = effectiveStyle(state);
+		if (style !== void 0) {
+			state.pendingHeadingLevel = style.headingLevel;
+			state.pendingListLevel = style.listLevel;
+		}
+	}
 	state.text += text;
 }
-function applyEolMapping(state, mapping, sink) {
+function readCellAttributes(state, nonDeletable, sink) {
+	const { subfunctions, truncated } = require_stream_table.readEmbeddedSubfunctions(nonDeletable);
+	if (truncated) reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.TableAttributesTruncated, "A cell's embedded attribute list held a record of undocumented length, so the attributes after it were not read.");
+	const information = require_stream_table.findEmbeddedSubfunction(subfunctions, 132);
+	const spanning = require_stream_table.findEmbeddedSubfunction(subfunctions, 133);
+	const fill = require_stream_table.findEmbeddedSubfunction(subfunctions, 134);
+	const row = require_stream_table.findEmbeddedSubfunction(subfunctions, 128);
+	const cellSpanning = spanning === void 0 ? void 0 : require_stream_table.readCellSpanning(spanning);
+	const cellFill = fill === void 0 ? void 0 : require_stream_table.readCellFill(fill);
+	if (cellFill?.blended === true) reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.CellFillBlended, "A cell is filled with a shaded blend of two colours; its background colour is used and the blend is not reproduced.");
+	return {
+		attributes: {
+			alignment: information === void 0 ? void 0 : require_stream_table.readCellInformation(information)?.alignment,
+			background: cellFill?.background,
+			columnSpan: cellSpanning?.columnSpan ?? NO_SPAN,
+			rowSpan: cellSpanning?.rowSpan ?? NO_SPAN,
+			covered: cellSpanning?.coveredFromLeft === true || cellSpanning?.coveredFromAbove === true
+		},
+		rowHeightPt: row === void 0 ? void 0 : require_stream_table.readRowInformation(row)?.heightPt
+	};
+}
+function closeCell(state, table, attributes) {
+	flushParagraphIfContent(state);
+	const blocks = table.cellBlocks;
+	table.cellBlocks = [];
+	if (attributes.covered) return;
+	if (attributes.alignment !== void 0) {
+		for (const block of blocks) if (block.kind === "paragraph") block.alignment = attributes.alignment;
+	}
+	const cell = {
+		blocks,
+		...attributes.columnSpan > NO_SPAN ? { colSpan: attributes.columnSpan } : {},
+		...attributes.rowSpan > NO_SPAN ? { rowSpan: attributes.rowSpan } : {},
+		...attributes.background === void 0 ? {} : { background: attributes.background }
+	};
+	table.cells.push(cell);
+}
+function closeRow(table) {
+	if (table.cells.length === 0) return;
+	const row = {
+		cells: table.cells,
+		...table.rowHeightPt === void 0 ? {} : { heightPt: table.rowHeightPt }
+	};
+	table.cells = [];
+	table.rowHeightPt = void 0;
+	table.rows.push(row);
+}
+function closeTable(state) {
+	const table = state.table;
+	if (table === void 0) return;
+	state.table = void 0;
+	if (table.rows.length === 0) return;
+	targetBlocks(state).push({
+		kind: "table",
+		rows: table.rows,
+		columnWidthsPt: table.columnWidthsPt
+	});
+}
+function applyEolMapping(state, mapping, nonDeletable, sink) {
 	switch (mapping) {
 		case "ignore": return;
 		case "space":
@@ -82,35 +184,38 @@ function applyEolMapping(state, mapping, sink) {
 			flushParagraph(state);
 			return;
 		case "hardEndOfColumn":
-			sink({
-				code: require_diagnostics.WpdDiagnosticCodes.ColumnBreakFlattened,
-				message: "A column break became a paragraph break."
-			});
+			reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.ColumnBreakFlattened, "A column break became a paragraph break.");
 			flushParagraph(state);
 			return;
 		case "hardEndOfPage":
 			flushParagraph(state);
-			state.blocks.push({ kind: "pageBreak" });
+			targetBlocks(state).push({ kind: "pageBreak" });
 			return;
 		case "tableCell":
 		case "tableRow":
 		case "hardTableRow":
-		case "tableOff":
-			if (!state.tableReported) {
-				state.tableReported = true;
-				sink({
-					code: require_diagnostics.WpdDiagnosticCodes.TableFlattened,
-					message: "This document contains a table; its cells and rows became paragraphs, and the table structure was not reconstructed."
-				});
+		case "tableOff": {
+			const table = state.table;
+			if (table === void 0) {
+				reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.TableFlattened, "A table cell or row boundary appeared with no table definition open; its text became a paragraph.");
+				flushParagraph(state);
+				return;
 			}
-			flushParagraph(state);
+			const { attributes, rowHeightPt } = readCellAttributes(state, nonDeletable, sink);
+			if (rowHeightPt !== void 0) table.rowHeightPt = rowHeightPt;
+			const cellIsOpen = state.text.length > 0 || state.runs.length > 0 || table.cellBlocks.length > 0;
+			if (mapping !== "tableOff" || cellIsOpen) closeCell(state, table, attributes);
+			if (mapping === "tableCell") return;
+			closeRow(table);
+			if (mapping === "tableOff") closeTable(state);
 			return;
+		}
 	}
 }
 function applySingleByteFunction(state, code, sink) {
 	if (require_stream_eol.isSingleByteEol(code)) {
 		const mapping = require_stream_eol.eolMappingForSubfunction(require_stream_eol.subfunctionForSingleByteEol(code));
-		if (mapping !== void 0) applyEolMapping(state, mapping, sink);
+		if (mapping !== void 0) applyEolMapping(state, mapping, /* @__PURE__ */ new Uint8Array(0), sink);
 		return;
 	}
 	switch (code) {
@@ -162,19 +267,82 @@ function applyFontFaceChange(state, prefixIds, container, sink) {
 	flushRun(state);
 	state.fontFamily = typeface;
 }
-function applyVariableFunction(state, token, container, sink) {
-	if (token.group === 208) {
-		const mapping = require_stream_eol.eolMappingForSubfunction(token.subgroup);
-		if (mapping !== void 0) applyEolMapping(state, mapping, sink);
+function setPageDimension(state, field, value, sink) {
+	const current = state.page[field];
+	if (current === void 0) {
+		state.page[field] = value;
+		return;
+	}
+	if (current !== value && !state.page.changeReported) {
+		state.page.changeReported = true;
+		sink({
+			code: require_diagnostics.WpdDiagnosticCodes.PageGeometryChanged,
+			message: "This document changes its page size or margins partway through; the section carries the geometry the document opens with."
+		});
+	}
+}
+function applyPageGroup(state, token, sink) {
+	if (token.subgroup === 17) {
+		const form = require_stream_page.readPageForm(token.nonDeletable);
+		if (form === void 0) return;
+		if (form.landscape) reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.LandscapeOrientationUnmapped, "The document's form declares a landscape orientation; the form's own stated width and length are used as written, since a page size carries no orientation.");
+		setPageDimension(state, "widthPt", form.widthPt, sink);
+		setPageDimension(state, "heightPt", form.heightPt, sink);
 		return;
 	}
-	if (token.group === PARAGRAPH_GROUP && token.subgroup === PARAGRAPH_SET_JUSTIFICATION) {
-		const mode = token.nonDeletable[0];
-		if (mode !== void 0) state.alignment = JUSTIFICATION[mode];
+	const margin = require_stream_page.readMarginPt(token.nonDeletable);
+	if (margin === void 0) return;
+	if (token.subgroup === 0) setPageDimension(state, "topPt", margin, sink);
+	else if (token.subgroup === 1) setPageDimension(state, "bottomPt", margin, sink);
+}
+function applyColumnGroup(state, token, sink) {
+	const margin = require_stream_page.readMarginPt(token.nonDeletable);
+	if (margin === void 0) return;
+	if (token.subgroup === 0) setPageDimension(state, "leftPt", margin, sink);
+	else if (token.subgroup === 1) setPageDimension(state, "rightPt", margin, sink);
+}
+function applyStyleGroup(state, token) {
+	if (require_stream_style.isStyleScopeOpener(token.subgroup)) {
+		const systemStyleNumber = require_stream_style.readSystemStyleNumber(token.nonDeletable);
+		state.styleScopes.push(systemStyleNumber === void 0 ? void 0 : require_stream_style.styleSemanticsFor(systemStyleNumber));
 		return;
 	}
-	if (token.group !== CHARACTER_GROUP) return;
+	if (require_stream_style.isStyleScopeCloser(token.subgroup)) state.styleScopes.pop();
+}
+function applyDisplayNumberGroup(state, token, sink) {
+	if (require_stream_style.isParagraphNumberDisplayOn(token.subgroup)) {
+		const level = require_stream_style.readDisplayNumberLevel(token.nonDeletable);
+		if (level !== void 0 && state.pendingListLevel === void 0) state.pendingListLevel = level;
+		state.numberDisplayDepth += 1;
+		reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.OutlineNumberRegenerated, "An outline number's rendered digits were replaced by the list membership that regenerates them.");
+		return;
+	}
+	if (require_stream_style.isParagraphNumberDisplayOff(token.subgroup)) state.numberDisplayDepth = Math.max(0, state.numberDisplayDepth - 1);
+}
+function applyCharacterGroup(state, token, container, sink) {
 	switch (token.subgroup) {
+		case 42:
+			closeTable(state);
+			flushParagraphIfContent(state);
+			state.table = {
+				columnWidthsPt: [],
+				rows: [],
+				cells: [],
+				cellBlocks: [],
+				definingColumns: true,
+				rowHeightPt: void 0
+			};
+			return;
+		case 44: {
+			const table = state.table;
+			if (!table?.definingColumns) return;
+			const widthPt = require_stream_table.readTableColumnWidthPt(token.nonDeletable);
+			if (widthPt !== void 0) table.columnWidthsPt.push(widthPt);
+			return;
+		}
+		case 43:
+			if (state.table !== void 0) state.table.definingColumns = false;
+			return;
 		case CHARACTER_FONT_FACE_CHANGE:
 			applyFontFaceChange(state, token.prefixIds, container, sink);
 			return;
@@ -200,6 +368,59 @@ function applyVariableFunction(state, token, container, sink) {
 		default: return;
 	}
 }
+function applyVariableFunction(state, token, container, sink) {
+	switch (token.group) {
+		case 208: {
+			const mapping = require_stream_eol.eolMappingForSubfunction(token.subgroup);
+			if (mapping !== void 0) applyEolMapping(state, mapping, token.nonDeletable, sink);
+			return;
+		}
+		case 209:
+			applyPageGroup(state, token, sink);
+			return;
+		case 210:
+			applyColumnGroup(state, token, sink);
+			return;
+		case PARAGRAPH_GROUP: {
+			if (token.subgroup !== PARAGRAPH_SET_JUSTIFICATION) return;
+			const mode = token.nonDeletable[0];
+			if (mode !== void 0) state.alignment = JUSTIFICATION[mode];
+			return;
+		}
+		case CHARACTER_GROUP:
+			applyCharacterGroup(state, token, container, sink);
+			return;
+		case 224: {
+			const effect = require_stream_tab.tabEffectFor(token.subgroup);
+			if (effect === void 0) return;
+			if (effect.kind === "tab") appendText(state, "	");
+			else state.pendingAlignment = effect.alignment;
+			return;
+		}
+		case 221:
+			applyStyleGroup(state, token);
+			return;
+		case 218:
+			applyDisplayNumberGroup(state, token, sink);
+			return;
+		case CROSS_REFERENCE_GROUP:
+			reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.CrossReferenceFlattened, "This document contains a cross-reference; its displayed text survives as ordinary text, and the reference's own target binding does not.");
+			return;
+		case HEADER_FOOTER_GROUP:
+			reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.HeaderFooterDropped, "This document declares a header, footer, or watermark, which the flat content model has no page-furniture position for.");
+			return;
+		case FOOTNOTE_ENDNOTE_GROUP:
+			reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.NoteDropped, "This document contains a footnote or endnote; its text lives in a prefix packet the flat content model has nowhere to put.");
+			return;
+		case MERGE_GROUP:
+			reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.MergeCodeDropped, "This document contains merge codes, which are a form-letter template's placeholders rather than text.");
+			return;
+		case BOX_GROUP:
+			reportOnce(state, sink, require_diagnostics.WpdDiagnosticCodes.BoxDropped, "This document contains a box -- a figure, text box, equation, or graphic -- whose contents were not read.");
+			return;
+		default: return;
+	}
+}
 function applyFixedFunction(state, token, sink) {
 	if (token.code === EXTENDED_CHARACTER) {
 		const characterNumber = token.data[0];
@@ -241,7 +462,22 @@ function foldTokens(tokens, container, sink) {
 		color: void 0,
 		alignment: void 0,
 		skipDepth: 0,
-		tableReported: false
+		styleScopes: [],
+		pendingHeadingLevel: void 0,
+		pendingListLevel: void 0,
+		pendingAlignment: void 0,
+		numberDisplayDepth: 0,
+		page: {
+			widthPt: void 0,
+			heightPt: void 0,
+			topPt: void 0,
+			rightPt: void 0,
+			bottomPt: void 0,
+			leftPt: void 0,
+			changeReported: false
+		},
+		table: void 0,
+		reported: /* @__PURE__ */ new Set()
 	};
 	for (const token of tokens) switch (token.kind) {
 		case "character": {
@@ -267,22 +503,36 @@ function foldTokens(tokens, container, sink) {
 	}
 	flushRun(state);
 	if (state.runs.length > 0) flushParagraph(state);
-	return state.blocks;
+	if (state.table !== void 0) {
+		closeRow(state.table);
+		closeTable(state);
+	}
+	return {
+		blocks: state.blocks,
+		page: state.page
+	};
+}
+function readMetadata(container) {
+	const packet = container.packets.find((candidate) => candidate.packetType === 18);
+	return packet === void 0 ? {} : require_container_summary.readDocumentSummary(packet.bytes);
 }
 function readWpdContent(bytes, options = {}) {
 	const sink = options.sink ?? require_diagnostics.NOOP_WPD_DIAGNOSTIC_SINK;
 	const container = require_container_container.openWpdDocument(bytes);
-	const blocks = foldTokens(require_stream_tokenise.tokeniseDocumentArea(container.bytes, container.documentAreaOffset, container.documentAreaEnd), container, sink);
+	const { blocks, page } = foldTokens(require_stream_tokenise.tokeniseDocumentArea(container.bytes, container.documentAreaOffset, container.documentAreaEnd), container, sink);
 	return {
 		kind: "wordprocessing",
-		metadata: {},
+		metadata: readMetadata(container),
 		sections: [{
-			pageSize: document_schema_js.PAGE_SIZE_LETTER,
+			pageSize: {
+				widthPt: page.widthPt ?? 612,
+				heightPt: page.heightPt ?? 792
+			},
 			margins: {
-				topPt: DEFAULT_MARGIN_PT,
-				rightPt: DEFAULT_MARGIN_PT,
-				bottomPt: DEFAULT_MARGIN_PT,
-				leftPt: DEFAULT_MARGIN_PT
+				topPt: page.topPt ?? 72,
+				rightPt: page.rightPt ?? 72,
+				bottomPt: page.bottomPt ?? 72,
+				leftPt: page.leftPt ?? 72
 			},
 			blocks
 		}]

package/dist/read.d.cts

@@ -1,4 +1,4 @@
-import { i as WpdDiagnosticSink } from "./diagnostics-Cp0HKphg.cjs";
+import { i as WpdDiagnosticSink } from "./diagnostics-CR0pXOyb.cjs";
 import { ContentDocument, DocumentTree } from "document-schema.js";
 //#region src/read.d.ts
 interface ReadWpdOptions {

package/dist/read.d.ts

@@ -1,4 +1,4 @@
-import { i as WpdDiagnosticSink } from "./diagnostics-Cp0HKphg.js";
+import { i as WpdDiagnosticSink } from "./diagnostics-CR0pXOyb.js";
 import { ContentDocument, DocumentTree } from "document-schema.js";
 //#region src/read.d.ts
 interface ReadWpdOptions {