doc-codec 2.4.3 → 2.6.0

package/dist/read.js

@@ -1,23 +1,22 @@
-import { DocFormatError, DocUnsupportedError } from "./errors.js";
+import { DocFormatError } from "./errors.js";
 import { slice } from "./bytes.js";
-import { SUMMARY_INFORMATION_STREAM, WORD_DOCUMENT_STREAM } from "./detect.js";
+import { DATA_STREAM, SUMMARY_INFORMATION_STREAM, WORD_DOCUMENT_STREAM } from "./detect.js";
 import { decryptDocStreams } from "./encryption.js";
-import { parseFib, peekFibBaseFlags } from "./fib/fib.js";
+import { PropertyBinTable } from "./prop/fkp.js";
+import { parseStsh } from "./style/stsh.js";
+import { readParagraphs } from "./text/paragraphs.js";
 import { parseClx } from "./text/piece-table.js";
 import { readTextRange } from "./text/characters.js";
-import { endsParagraph, isAnchorOnly } from "./text/special.js";
-import { readGrpprl } from "./prop/sprm.js";
-import { PropertyBinTable } from "./prop/fkp.js";
-import { applyCharacterSprms, characterIstdFromGrpprl } from "./prop/chp.js";
-import { applyParagraphSprms } from "./prop/pap.js";
-import { readSectionProperties } from "./prop/sep.js";
-import { headingLevelFromIstd, parseStsh, resolveStyleFormatting } from "./style/stsh.js";
+import { assembleBlocks } from "./table/read.js";
+import { readHeaderFooterStories } from "./headers-footers.js";
+import { parseFib, peekFibBaseFlags } from "./fib/fib.js";
+import { readAllSectionProperties } from "./prop/sep.js";
 import { parseFontTable } from "./style/fonts.js";
 import { readNumberingDefinitions } from "./list/numbering.js";
-import { assembleBlocks } from "./table/read.js";
+import { readNoteBodies } from "./notes.js";
 import { readCompoundFile, readSummaryInformation, summaryInformationToLayoutMetadata } from "archive-codec";
 //#region src/read.ts
-/** Word's own default for a new document (US Letter, one-inch margins) -- what a field this reader resolves from PlcfSed/Sepx (prop/sep.ts's readSectionProperties) falls back to when the file states nothing for it, exactly as it would fall back to Word's own implementation-dependent default for that one unstated sprm. */
+/** Word's own default for a new document (US Letter, one-inch margins) -- what a field this reader resolves from PlcfSed/Sepx (prop/sep.ts's readAllSectionProperties) falls back to when the file states nothing for it, exactly as it would fall back to Word's own implementation-dependent default for that one unstated sprm. */
 const DEFAULT_PAGE_SIZE = {
 	widthPt: 612,
 	heightPt: 792
@@ -39,177 +38,73 @@ function readDocStreams(bytes, password) {
 	let wordDocument = wordDocumentStream.bytes;
 	let table = tableStream.bytes;
 	if (flags.fEncrypted) {
-		if (flags.fObfuscated) throw new DocUnsupportedError("this document is XOR-obfuscated ([MS-DOC] 2.2.6.1); doc-codec cannot decrypt it, and reading its streams as plaintext would produce arbitrary text rather than the document's own");
-		const decrypted = decryptDocStreams(wordDocument, table, password);
+		const decrypted = decryptDocStreams(wordDocument, table, password, flags.fObfuscated);
 		wordDocument = decrypted.wordDocument;
 		table = decrypted.table;
 	}
 	const fib = parseFib(wordDocument);
 	const metadata = streams.find((stream) => stream.path === SUMMARY_INFORMATION_STREAM);
+	const data = streams.find((stream) => stream.path === DATA_STREAM);
 	return {
 		wordDocument,
 		table,
 		fib,
-		metadata: metadata?.bytes
+		metadata: metadata?.bytes,
+		data: data?.bytes
 	};
 }
 function readDocContent(bytes, password) {
-	const { wordDocument, table, fib, metadata } = readDocStreams(bytes, password);
+	const { wordDocument, table, fib, metadata, data } = readDocStreams(bytes, password);
 	const pieceTable = parseClx(slice(table, fib.fcClx, fib.lcbClx, "Clx in the Table stream"));
 	const styles = fib.lcbStshf > 0 ? parseStsh(slice(table, fib.fcStshf, fib.lcbStshf, "STSH in the Table stream")) : void 0;
-	const chpxTable = new PropertyBinTable(wordDocument, slice(table, fib.fcPlcfBteChpx, fib.lcbPlcfBteChpx, "PlcBteChpx in the Table stream"), "PlcBteChpx");
-	const papxTable = new PropertyBinTable(wordDocument, slice(table, fib.fcPlcfBtePapx, fib.lcbPlcfBtePapx, "PlcBtePapx in the Table stream"), "PlcBtePapx");
-	const fonts = fib.lcbSttbfFfn > 0 ? parseFontTable(slice(table, fib.fcSttbfFfn, fib.lcbSttbfFfn, "SttbfFfn in the Table stream")) : void 0;
-	const range = readTextRange(wordDocument, pieceTable, 0, fib.ccpText);
-	const entries = readParagraphs(range.text, range.fcs, {
-		chpxTable,
-		papxTable,
+	const context = {
+		chpxTable: new PropertyBinTable(wordDocument, slice(table, fib.fcPlcfBteChpx, fib.lcbPlcfBteChpx, "PlcBteChpx in the Table stream"), "PlcBteChpx"),
+		papxTable: new PropertyBinTable(wordDocument, slice(table, fib.fcPlcfBtePapx, fib.lcbPlcfBtePapx, "PlcBtePapx in the Table stream"), "PlcBtePapx"),
 		styles,
-		fonts,
+		dataStream: data,
+		fonts: fib.lcbSttbfFfn > 0 ? parseFontTable(slice(table, fib.fcSttbfFfn, fib.lcbSttbfFfn, "SttbfFfn in the Table stream")) : void 0,
 		characterProperties: /* @__PURE__ */ new Map()
-	});
-	const blocks = assembleBlocks(entries);
+	};
+	const range = readTextRange(wordDocument, pieceTable, 0, fib.ccpText);
+	const entries = readParagraphs(range.text, range.fcs, context);
 	const numbering = readNumberingDefinitions(table, fib);
-	const sectionProperties = readSectionProperties(wordDocument, table, fib);
+	const sectionProperties = readAllSectionProperties(wordDocument, table, fib);
+	const entriesBySection = splitIntoSections(entries, sectionProperties);
+	const { footnotes, endnotes, comments } = readNoteBodies(wordDocument, table, pieceTable, context, fib);
+	const headerFooterStories = readHeaderFooterStories(wordDocument, table, pieceTable, context, fib, sectionProperties.length);
 	return {
 		kind: "wordprocessing",
 		metadata: metadata === void 0 ? {} : summaryInformationToLayoutMetadata(readSummaryInformation(metadata)),
-		sections: [{
+		sections: sectionProperties.map((properties, index) => ({
 			pageSize: {
-				widthPt: sectionProperties.pageWidthPt ?? DEFAULT_PAGE_SIZE.widthPt,
-				heightPt: sectionProperties.pageHeightPt ?? DEFAULT_PAGE_SIZE.heightPt
+				widthPt: properties.pageWidthPt ?? DEFAULT_PAGE_SIZE.widthPt,
+				heightPt: properties.pageHeightPt ?? DEFAULT_PAGE_SIZE.heightPt
 			},
 			margins: {
-				leftPt: sectionProperties.marginLeftPt ?? DEFAULT_MARGINS.leftPt,
-				rightPt: sectionProperties.marginRightPt ?? DEFAULT_MARGINS.rightPt,
-				topPt: sectionProperties.marginTopPt ?? DEFAULT_MARGINS.topPt,
-				bottomPt: sectionProperties.marginBottomPt ?? DEFAULT_MARGINS.bottomPt
+				leftPt: properties.marginLeftPt ?? DEFAULT_MARGINS.leftPt,
+				rightPt: properties.marginRightPt ?? DEFAULT_MARGINS.rightPt,
+				topPt: properties.marginTopPt ?? DEFAULT_MARGINS.topPt,
+				bottomPt: properties.marginBottomPt ?? DEFAULT_MARGINS.bottomPt
 			},
-			blocks
-		}],
-		numbering
-	};
-}
-function readParagraphs(text, fcs, context) {
-	const entries = [];
-	let start = 0;
-	for (let index = 0; index < text.length; index += 1) {
-		const code = text.charCodeAt(index);
-		if (!endsParagraph(code)) continue;
-		const markFc = fcs[index];
-		if (markFc === void 0) throw new DocFormatError(`character ${index} has no byte offset, so its paragraph's properties cannot be located`);
-		entries.push(buildParagraph(text.slice(start, index), fcs.slice(start, index), markFc, code, context));
-		start = index + 1;
-	}
-	if (start < text.length) {
-		const firstFc = fcs[start];
-		if (firstFc === void 0) throw new DocFormatError(`character ${start} has no byte offset, so the trailing paragraph's properties cannot be located`);
-		entries.push(buildParagraph(text.slice(start), fcs.slice(start), firstFc, 13, context));
-	}
-	return entries;
-}
-function buildParagraph(text, fcs, propertyFc, terminator, context) {
-	const papx = context.papxTable.papx(propertyFc);
-	const properties = {};
-	const grpprl = papx !== void 0 ? readGrpprl(papx.grpprl) : [];
-	if (papx !== void 0) {
-		properties.istd = papx.istd;
-		if (context.styles !== void 0) applyParagraphSprms(resolveStyleFormatting(context.styles, papx.istd).paragraphPrls, properties);
-		applyParagraphSprms(grpprl, properties);
-	}
-	return {
-		paragraph: {
-			kind: "paragraph",
-			runs: buildRuns(text, fcs, context, papx?.istd),
-			...paragraphAttributes(properties, context)
-		},
-		properties,
-		grpprl,
-		terminator
+			blocks: assembleBlocks(entriesBySection[index] ?? [])
+		})),
+		numbering,
+		footnotes,
+		endnotes,
+		comments,
+		headerFooterStories
 	};
 }
-function paragraphAttributes(properties, context) {
-	const attributes = {};
-	const istd = properties.istd;
-	if (istd !== void 0) {
-		const style = context.styles?.styles[istd];
-		if (style !== void 0 && style.name !== "") attributes.styleId = style.name;
-		const headingLevel = headingLevelFromIstd(istd);
-		if (headingLevel !== void 0) attributes.headingLevel = headingLevel;
-	}
-	if (attributes.headingLevel === void 0 && properties.outlineLevel !== void 0) attributes.headingLevel = properties.outlineLevel + 1;
-	if (properties.alignment !== void 0) attributes.alignment = properties.alignment;
-	if (properties.spacingBeforePt !== void 0) attributes.spacingBeforePt = properties.spacingBeforePt;
-	if (properties.spacingAfterPt !== void 0) attributes.spacingAfterPt = properties.spacingAfterPt;
-	if (properties.lineSpacing !== void 0) attributes.lineSpacing = properties.lineSpacing;
-	if (properties.indentLeftPt !== void 0) attributes.indentLeftPt = properties.indentLeftPt;
-	if (properties.indentRightPt !== void 0) attributes.indentRightPt = properties.indentRightPt;
-	if (properties.indentFirstLinePt !== void 0) attributes.indentFirstLinePt = properties.indentFirstLinePt;
-	if (properties.pageBreakBefore === true) attributes.pageBreakBefore = true;
-	if (properties.listId !== void 0) attributes.list = {
-		numId: String(properties.listId),
-		level: properties.listLevel ?? 0
-	};
-	return attributes;
-}
-function buildRuns(text, fcs, context, paragraphIstd) {
-	const runs = [];
-	let currentKey;
-	let currentText = "";
-	let currentProperties = {};
-	const paragraphStyleCharacterPrls = context.styles !== void 0 && paragraphIstd !== void 0 ? resolveStyleFormatting(context.styles, paragraphIstd).characterPrls : [];
-	const enclosingInstruction = [];
-	let inInstruction = false;
-	const flush = () => {
-		if (currentText !== "") runs.push({
-			text: currentText,
-			...currentProperties
-		});
-		currentText = "";
-	};
-	for (let index = 0; index < text.length; index += 1) {
-		const code = text.charCodeAt(index);
-		if (code === 19) {
-			flush();
-			enclosingInstruction.push(inInstruction);
-			inInstruction = true;
-			continue;
-		}
-		if (code === 20) {
-			inInstruction = false;
-			continue;
-		}
-		if (code === 21) {
-			inInstruction = enclosingInstruction.pop() ?? inInstruction;
-			continue;
-		}
-		if (inInstruction || isAnchorOnly(code)) continue;
-		const fc = fcs[index];
-		if (fc === void 0) throw new DocFormatError(`character ${index} of a paragraph has no byte offset, so its formatting cannot be located`);
-		const grpprl = context.chpxTable.chpxGrpprl(fc);
-		const chpxKey = grpprl === void 0 ? "none" : `${grpprl.byteOffset}:${grpprl.byteLength}`;
-		const key = `${paragraphIstd ?? "none"}:${chpxKey}`;
-		if (key !== currentKey) {
-			flush();
-			currentKey = key;
-			let properties = context.characterProperties.get(key);
-			if (properties === void 0) {
-				properties = {};
-				applyCharacterSprms(paragraphStyleCharacterPrls, properties, context.fonts);
-				if (grpprl !== void 0) {
-					const runPrls = readGrpprl(grpprl);
-					const characterIstd = characterIstdFromGrpprl(runPrls);
-					if (characterIstd !== void 0 && context.styles !== void 0) applyCharacterSprms(resolveStyleFormatting(context.styles, characterIstd).characterPrls, properties, context.fonts);
-					applyCharacterSprms(runPrls, properties, context.fonts);
-				}
-				context.characterProperties.set(key, properties);
-			}
-			currentProperties = properties;
-		}
-		currentText += String.fromCharCode(code === 11 ? 10 : code);
+function splitIntoSections(entries, sections) {
+	const groups = sections.map(() => []);
+	let sectionIndex = 0;
+	for (const entry of entries) {
+		const group = groups[sectionIndex];
+		if (group === void 0) throw new DocFormatError(`internal defect: section index ${sectionIndex} has no group despite ${sections.length} sections`);
+		group.push(entry);
+		if (entry.endCp === sections[sectionIndex + 1]?.startCp) sectionIndex += 1;
 	}
-	flush();
-	return runs;
+	return groups;
 }
 //#endregion
 export { readDocContent, readDocStreams };

package/dist/stsh-Bj0YSEOp.d.ts

@@ -0,0 +1,40 @@
+import { t as Prl } from "./sprm-DEyvdOo0.js";
+//#region src/style/stsh.d.ts
+/** StdfBase.sti's "0x0FFE for user-defined styles"; anything else is an application-defined style whose sti identifies it. */
+declare const STI_USER_DEFINED = 4094;
+/** StdfBase.stk, [MS-DOC] 2.9.269 -- which of ECMA-376's ST_StyleType values this style is. */
+declare const STK: {
+  readonly paragraph: 1;
+  readonly character: 2;
+  readonly table: 3;
+  readonly numbering: 4;
+};
+interface Style {
+  /** The style's index in STSH.rglpstd, which is what sprmPIstd and sprmCIstd name. */
+  readonly istd: number;
+  /** The invariant application-defined style identifier, or STI_USER_DEFINED for a style the document itself defines. */
+  readonly sti: number;
+  readonly stk: number;
+  /** The istd this style inherits from, or undefined when StdfBase.istdBase is 0x0FFF ("this style does not inherit from any other style"). */
+  readonly istdBase: number | undefined;
+  /** The style's primary name, from the STD's own Xstz. */
+  readonly name: string;
+  /** The style's own paragraph formatting -- StkParaGRLPUPX's lpUpxPapx.PAPX.grpprlPapx -- present only for a paragraph style (STK.paragraph). Undefined for every other stk: table and numbering styles carry their own formatting shapes this package does not read (see the README's own scope note). */
+  readonly grpprlPapx: readonly Prl[] | undefined;
+  /** The style's own character formatting -- a paragraph style's StkParaGRLPUPX.lpUpxChpx.CHPX.grpprlChpx (the run-level defaults a heading's own text falls back to) or a character style's StkCharGRLPUPX.lpUpxChpx.CHPX.grpprlChpx -- present for STK.paragraph and STK.character, undefined otherwise. */
+  readonly grpprlChpx: readonly Prl[] | undefined;
+}
+interface StyleSheet {
+  /** Indexed by istd; a hole is a slot [MS-DOC] permits to be empty ("A style definition can be empty, in which case cbStd MUST be 0"). */
+  readonly styles: readonly (Style | undefined)[];
+}
+declare function parseStsh(stsh: Uint8Array): StyleSheet;
+interface ResolvedStyleFormatting {
+  readonly paragraphPrls: readonly Prl[];
+  readonly characterPrls: readonly Prl[];
+}
+declare function resolveStyleFormatting(styleSheet: StyleSheet, istd: number): ResolvedStyleFormatting;
+declare function headingLevelFromIstd(istd: number): number | undefined;
+declare function buildStshForStyles(names: ReadonlyMap<number, string>): Uint8Array;
+//#endregion
+export { StyleSheet as a, parseStsh as c, Style as i, resolveStyleFormatting as l, STI_USER_DEFINED as n, buildStshForStyles as o, STK as r, headingLevelFromIstd as s, ResolvedStyleFormatting as t };

package/dist/stsh-CK59x6XG.d.cts

@@ -0,0 +1,40 @@
+import { t as Prl } from "./sprm-DEyvdOo0.cjs";
+//#region src/style/stsh.d.ts
+/** StdfBase.sti's "0x0FFE for user-defined styles"; anything else is an application-defined style whose sti identifies it. */
+declare const STI_USER_DEFINED = 4094;
+/** StdfBase.stk, [MS-DOC] 2.9.269 -- which of ECMA-376's ST_StyleType values this style is. */
+declare const STK: {
+  readonly paragraph: 1;
+  readonly character: 2;
+  readonly table: 3;
+  readonly numbering: 4;
+};
+interface Style {
+  /** The style's index in STSH.rglpstd, which is what sprmPIstd and sprmCIstd name. */
+  readonly istd: number;
+  /** The invariant application-defined style identifier, or STI_USER_DEFINED for a style the document itself defines. */
+  readonly sti: number;
+  readonly stk: number;
+  /** The istd this style inherits from, or undefined when StdfBase.istdBase is 0x0FFF ("this style does not inherit from any other style"). */
+  readonly istdBase: number | undefined;
+  /** The style's primary name, from the STD's own Xstz. */
+  readonly name: string;
+  /** The style's own paragraph formatting -- StkParaGRLPUPX's lpUpxPapx.PAPX.grpprlPapx -- present only for a paragraph style (STK.paragraph). Undefined for every other stk: table and numbering styles carry their own formatting shapes this package does not read (see the README's own scope note). */
+  readonly grpprlPapx: readonly Prl[] | undefined;
+  /** The style's own character formatting -- a paragraph style's StkParaGRLPUPX.lpUpxChpx.CHPX.grpprlChpx (the run-level defaults a heading's own text falls back to) or a character style's StkCharGRLPUPX.lpUpxChpx.CHPX.grpprlChpx -- present for STK.paragraph and STK.character, undefined otherwise. */
+  readonly grpprlChpx: readonly Prl[] | undefined;
+}
+interface StyleSheet {
+  /** Indexed by istd; a hole is a slot [MS-DOC] permits to be empty ("A style definition can be empty, in which case cbStd MUST be 0"). */
+  readonly styles: readonly (Style | undefined)[];
+}
+declare function parseStsh(stsh: Uint8Array): StyleSheet;
+interface ResolvedStyleFormatting {
+  readonly paragraphPrls: readonly Prl[];
+  readonly characterPrls: readonly Prl[];
+}
+declare function resolveStyleFormatting(styleSheet: StyleSheet, istd: number): ResolvedStyleFormatting;
+declare function headingLevelFromIstd(istd: number): number | undefined;
+declare function buildStshForStyles(names: ReadonlyMap<number, string>): Uint8Array;
+//#endregion
+export { StyleSheet as a, parseStsh as c, Style as i, resolveStyleFormatting as l, STI_USER_DEFINED as n, buildStshForStyles as o, STK as r, headingLevelFromIstd as s, ResolvedStyleFormatting as t };

package/dist/style/stsh.d.cts

@@ -1,40 +1,2 @@
-import { t as Prl } from "../sprm-DEyvdOo0.cjs";
-//#region src/style/stsh.d.ts
-/** StdfBase.sti's "0x0FFE for user-defined styles"; anything else is an application-defined style whose sti identifies it. */
-declare const STI_USER_DEFINED = 4094;
-/** StdfBase.stk, [MS-DOC] 2.9.269 -- which of ECMA-376's ST_StyleType values this style is. */
-declare const STK: {
-  readonly paragraph: 1;
-  readonly character: 2;
-  readonly table: 3;
-  readonly numbering: 4;
-};
-interface Style {
-  /** The style's index in STSH.rglpstd, which is what sprmPIstd and sprmCIstd name. */
-  readonly istd: number;
-  /** The invariant application-defined style identifier, or STI_USER_DEFINED for a style the document itself defines. */
-  readonly sti: number;
-  readonly stk: number;
-  /** The istd this style inherits from, or undefined when StdfBase.istdBase is 0x0FFF ("this style does not inherit from any other style"). */
-  readonly istdBase: number | undefined;
-  /** The style's primary name, from the STD's own Xstz. */
-  readonly name: string;
-  /** The style's own paragraph formatting -- StkParaGRLPUPX's lpUpxPapx.PAPX.grpprlPapx -- present only for a paragraph style (STK.paragraph). Undefined for every other stk: table and numbering styles carry their own formatting shapes this package does not read (see the README's own scope note). */
-  readonly grpprlPapx: readonly Prl[] | undefined;
-  /** The style's own character formatting -- a paragraph style's StkParaGRLPUPX.lpUpxChpx.CHPX.grpprlChpx (the run-level defaults a heading's own text falls back to) or a character style's StkCharGRLPUPX.lpUpxChpx.CHPX.grpprlChpx -- present for STK.paragraph and STK.character, undefined otherwise. */
-  readonly grpprlChpx: readonly Prl[] | undefined;
-}
-interface StyleSheet {
-  /** Indexed by istd; a hole is a slot [MS-DOC] permits to be empty ("A style definition can be empty, in which case cbStd MUST be 0"). */
-  readonly styles: readonly (Style | undefined)[];
-}
-declare function parseStsh(stsh: Uint8Array): StyleSheet;
-interface ResolvedStyleFormatting {
-  readonly paragraphPrls: readonly Prl[];
-  readonly characterPrls: readonly Prl[];
-}
-declare function resolveStyleFormatting(styleSheet: StyleSheet, istd: number): ResolvedStyleFormatting;
-declare function headingLevelFromIstd(istd: number): number | undefined;
-declare function buildStshForStyles(names: ReadonlyMap<number, string>): Uint8Array;
-//#endregion
+import { a as StyleSheet, c as parseStsh, i as Style, l as resolveStyleFormatting, n as STI_USER_DEFINED, o as buildStshForStyles, r as STK, s as headingLevelFromIstd, t as ResolvedStyleFormatting } from "../stsh-CK59x6XG.cjs";
 export { ResolvedStyleFormatting, STI_USER_DEFINED, STK, Style, StyleSheet, buildStshForStyles, headingLevelFromIstd, parseStsh, resolveStyleFormatting };

package/dist/style/stsh.d.ts

@@ -1,40 +1,2 @@
-import { t as Prl } from "../sprm-DEyvdOo0.js";
-//#region src/style/stsh.d.ts
-/** StdfBase.sti's "0x0FFE for user-defined styles"; anything else is an application-defined style whose sti identifies it. */
-declare const STI_USER_DEFINED = 4094;
-/** StdfBase.stk, [MS-DOC] 2.9.269 -- which of ECMA-376's ST_StyleType values this style is. */
-declare const STK: {
-  readonly paragraph: 1;
-  readonly character: 2;
-  readonly table: 3;
-  readonly numbering: 4;
-};
-interface Style {
-  /** The style's index in STSH.rglpstd, which is what sprmPIstd and sprmCIstd name. */
-  readonly istd: number;
-  /** The invariant application-defined style identifier, or STI_USER_DEFINED for a style the document itself defines. */
-  readonly sti: number;
-  readonly stk: number;
-  /** The istd this style inherits from, or undefined when StdfBase.istdBase is 0x0FFF ("this style does not inherit from any other style"). */
-  readonly istdBase: number | undefined;
-  /** The style's primary name, from the STD's own Xstz. */
-  readonly name: string;
-  /** The style's own paragraph formatting -- StkParaGRLPUPX's lpUpxPapx.PAPX.grpprlPapx -- present only for a paragraph style (STK.paragraph). Undefined for every other stk: table and numbering styles carry their own formatting shapes this package does not read (see the README's own scope note). */
-  readonly grpprlPapx: readonly Prl[] | undefined;
-  /** The style's own character formatting -- a paragraph style's StkParaGRLPUPX.lpUpxChpx.CHPX.grpprlChpx (the run-level defaults a heading's own text falls back to) or a character style's StkCharGRLPUPX.lpUpxChpx.CHPX.grpprlChpx -- present for STK.paragraph and STK.character, undefined otherwise. */
-  readonly grpprlChpx: readonly Prl[] | undefined;
-}
-interface StyleSheet {
-  /** Indexed by istd; a hole is a slot [MS-DOC] permits to be empty ("A style definition can be empty, in which case cbStd MUST be 0"). */
-  readonly styles: readonly (Style | undefined)[];
-}
-declare function parseStsh(stsh: Uint8Array): StyleSheet;
-interface ResolvedStyleFormatting {
-  readonly paragraphPrls: readonly Prl[];
-  readonly characterPrls: readonly Prl[];
-}
-declare function resolveStyleFormatting(styleSheet: StyleSheet, istd: number): ResolvedStyleFormatting;
-declare function headingLevelFromIstd(istd: number): number | undefined;
-declare function buildStshForStyles(names: ReadonlyMap<number, string>): Uint8Array;
-//#endregion
+import { a as StyleSheet, c as parseStsh, i as Style, l as resolveStyleFormatting, n as STI_USER_DEFINED, o as buildStshForStyles, r as STK, s as headingLevelFromIstd, t as ResolvedStyleFormatting } from "../stsh-Bj0YSEOp.js";
 export { ResolvedStyleFormatting, STI_USER_DEFINED, STK, Style, StyleSheet, buildStshForStyles, headingLevelFromIstd, parseStsh, resolveStyleFormatting };

package/dist/subdocument.cjs

@@ -0,0 +1,20 @@
+Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
+const require_bytes = require("./bytes.cjs");
+const require_plc = require("./plc.cjs");
+const require_text_paragraphs = require("./text/paragraphs.cjs");
+const require_text_characters = require("./text/characters.cjs");
+//#region src/subdocument.ts
+function readSubdocumentStories(wordDocument, table, pieceTable, context, subdocStartCp, subdocLength, boundaryFc, boundaryLcb, what) {
+	if (subdocLength <= 0 || boundaryLcb <= 0) return [];
+	const range = require_text_characters.readTextRange(wordDocument, pieceTable, subdocStartCp, subdocStartCp + subdocLength);
+	const entries = require_text_paragraphs.readParagraphs(range.text, range.fcs, context);
+	const plc = require_plc.parsePlc(require_bytes.slice(table, boundaryFc, boundaryLcb, `${what} in the Table stream`), 0, what);
+	return require_text_paragraphs.splitEntriesByBoundaries(entries, plc.keys).slice(0, -1).map((story) => story.slice(0, -1));
+}
+/** Joins a story's own paragraphs into plain text -- one line per paragraph, matching how ooxml.js's own Footnote/Comment reading concatenates a footnote or comment body's `w:t` runs rather than preserving paragraph/run structure. A footnote, endnote, or comment in this family is carried as text alone; only a header or footer keeps real block flow (headers-footers.ts). Non-paragraph blocks a story's own entries produced (an inline picture split out of one of its paragraphs) contribute no text of their own here, the same way an image has no plain-text spelling in ooxml.js's own footnote/comment reading either. */
+function storyText(entries) {
+	return entries.map((entry) => entry.blocks.flatMap((block) => block.kind === "paragraph" ? block.runs : []).map((run) => run.text).join("")).join("\n");
+}
+//#endregion
+exports.readSubdocumentStories = readSubdocumentStories;
+exports.storyText = storyText;

package/dist/subdocument.d.cts

@@ -0,0 +1,8 @@
+import { n as ReadContext, t as ParagraphEntry } from "./paragraphs-DQUp8_HB.cjs";
+import { n as PieceTable } from "./piece-table-CA18mhjv.cjs";
+//#region src/subdocument.d.ts
+declare function readSubdocumentStories(wordDocument: Uint8Array, table: Uint8Array, pieceTable: PieceTable, context: ReadContext, subdocStartCp: number, subdocLength: number, boundaryFc: number, boundaryLcb: number, what: string): readonly ParagraphEntry[][];
+/** Joins a story's own paragraphs into plain text -- one line per paragraph, matching how ooxml.js's own Footnote/Comment reading concatenates a footnote or comment body's `w:t` runs rather than preserving paragraph/run structure. A footnote, endnote, or comment in this family is carried as text alone; only a header or footer keeps real block flow (headers-footers.ts). Non-paragraph blocks a story's own entries produced (an inline picture split out of one of its paragraphs) contribute no text of their own here, the same way an image has no plain-text spelling in ooxml.js's own footnote/comment reading either. */
+declare function storyText(entries: readonly ParagraphEntry[]): string;
+//#endregion
+export { readSubdocumentStories, storyText };

package/dist/subdocument.d.ts

@@ -0,0 +1,8 @@
+import { n as ReadContext, t as ParagraphEntry } from "./paragraphs-s2lJ6w2I.js";
+import { n as PieceTable } from "./piece-table-CA18mhjv.js";
+//#region src/subdocument.d.ts
+declare function readSubdocumentStories(wordDocument: Uint8Array, table: Uint8Array, pieceTable: PieceTable, context: ReadContext, subdocStartCp: number, subdocLength: number, boundaryFc: number, boundaryLcb: number, what: string): readonly ParagraphEntry[][];
+/** Joins a story's own paragraphs into plain text -- one line per paragraph, matching how ooxml.js's own Footnote/Comment reading concatenates a footnote or comment body's `w:t` runs rather than preserving paragraph/run structure. A footnote, endnote, or comment in this family is carried as text alone; only a header or footer keeps real block flow (headers-footers.ts). Non-paragraph blocks a story's own entries produced (an inline picture split out of one of its paragraphs) contribute no text of their own here, the same way an image has no plain-text spelling in ooxml.js's own footnote/comment reading either. */
+declare function storyText(entries: readonly ParagraphEntry[]): string;
+//#endregion
+export { readSubdocumentStories, storyText };

package/dist/subdocument.js

@@ -0,0 +1,18 @@
+import { slice } from "./bytes.js";
+import { parsePlc } from "./plc.js";
+import { readParagraphs, splitEntriesByBoundaries } from "./text/paragraphs.js";
+import { readTextRange } from "./text/characters.js";
+//#region src/subdocument.ts
+function readSubdocumentStories(wordDocument, table, pieceTable, context, subdocStartCp, subdocLength, boundaryFc, boundaryLcb, what) {
+	if (subdocLength <= 0 || boundaryLcb <= 0) return [];
+	const range = readTextRange(wordDocument, pieceTable, subdocStartCp, subdocStartCp + subdocLength);
+	const entries = readParagraphs(range.text, range.fcs, context);
+	const plc = parsePlc(slice(table, boundaryFc, boundaryLcb, `${what} in the Table stream`), 0, what);
+	return splitEntriesByBoundaries(entries, plc.keys).slice(0, -1).map((story) => story.slice(0, -1));
+}
+/** Joins a story's own paragraphs into plain text -- one line per paragraph, matching how ooxml.js's own Footnote/Comment reading concatenates a footnote or comment body's `w:t` runs rather than preserving paragraph/run structure. A footnote, endnote, or comment in this family is carried as text alone; only a header or footer keeps real block flow (headers-footers.ts). Non-paragraph blocks a story's own entries produced (an inline picture split out of one of its paragraphs) contribute no text of their own here, the same way an image has no plain-text spelling in ooxml.js's own footnote/comment reading either. */
+function storyText(entries) {
+	return entries.map((entry) => entry.blocks.flatMap((block) => block.kind === "paragraph" ? block.runs : []).map((run) => run.text).join("")).join("\n");
+}
+//#endregion
+export { readSubdocumentStories, storyText };

package/dist/table/read.cjs

@@ -23,30 +23,44 @@ function effectiveColumnBoundaryTolerance(definitions) {
 	return narrowestRealGapTwips === void 0 ? TWIPS_PER_POINT : Math.min(TWIPS_PER_POINT, narrowestRealGapTwips - 1);
 }
 function assembleBlocks(entries) {
+	return walkBlocksAtDepth(entries, 0);
+}
+function effectiveTableDepth(properties) {
+	if (properties.tableDepth !== void 0) return properties.tableDepth;
+	return properties.inTable === true ? 1 : 0;
+}
+function isCellBoundary(entry, tableDepth) {
+	if (tableDepth === 1) return entry.terminator === 7 && entry.properties.tableRowEnd !== true;
+	return entry.terminator === 13 && entry.properties.innerTableCellMark === true;
+}
+function isRowBoundary(entry, tableDepth) {
+	return tableDepth === 1 ? entry.properties.tableRowEnd === true : entry.properties.innerTtpMark === true;
+}
+function walkBlocksAtDepth(entries, depth) {
 	const blocks = [];
 	let index = 0;
 	while (index < entries.length) {
 		const entry = entries[index];
 		if (entry === void 0) break;
-		if (entry.properties.inTable !== true) {
-			blocks.push(entry.paragraph);
+		if (effectiveTableDepth(entry.properties) <= depth) {
+			blocks.push(...entry.blocks);
 			index += 1;
 			continue;
 		}
-		const { runEntries, nextIndex } = collectTableRun(entries, index);
-		const table = tryAssembleTable(runEntries);
-		blocks.push(...table !== void 0 ? [table] : runEntries.map((run) => run.paragraph));
+		const tableDepth = depth + 1;
+		const { runEntries, nextIndex } = collectTableRun(entries, index, tableDepth);
+		const table = tryAssembleTable(runEntries, tableDepth);
+		blocks.push(...table !== void 0 ? [table] : runEntries.flatMap((run) => run.blocks));
 		index = nextIndex;
 	}
 	return blocks;
 }
-function collectTableRun(entries, start) {
+function collectTableRun(entries, start, tableDepth) {
 	const runEntries = [];
 	let index = start;
 	while (index < entries.length) {
 		const entry = entries[index];
-		if (entry?.properties.inTable !== true) break;
-		if (entry.properties.tableDepth !== void 0 && entry.properties.tableDepth > 1 || entry.properties.nestedTableMark === true) throw new require_errors.DocUnsupportedError("doc-codec does not support a table nested inside a table cell (table depth greater than 1)");
+		if (entry === void 0 || effectiveTableDepth(entry.properties) < tableDepth) break;
 		runEntries.push(entry);
 		index += 1;
 	}
@@ -55,15 +69,15 @@ function collectTableRun(entries, start) {
 		nextIndex: index
 	};
 }
-function tryAssembleTable(runEntries) {
+function tryAssembleTable(runEntries, tableDepth) {
 	const rawRows = [];
 	const rowDefinitions = [];
 	const rowHeights = [];
-	let cellParagraphs = [];
+	let cellEntries = [];
 	let rowCells = [];
 	for (const entry of runEntries) {
-		cellParagraphs.push(entry.paragraph);
-		if (entry.properties.tableRowEnd === true) {
+		cellEntries.push(entry);
+		if (isRowBoundary(entry, tableDepth)) {
 			const rowProperties = require_table_tap.applyTableSprms(entry.grpprl, {});
 			const definition = rowProperties.definition;
 			if (definition === void 0) return void 0;
@@ -83,15 +97,15 @@ function tryAssembleTable(runEntries) {
 			rowDefinitions.push(definition);
 			rowHeights.push(rowProperties.heightPt);
 			rowCells = [];
-			cellParagraphs = [];
+			cellEntries = [];
 			continue;
 		}
-		if (entry.terminator === 7) {
-			rowCells.push({ blocks: cellParagraphs });
-			cellParagraphs = [];
+		if (isCellBoundary(entry, tableDepth)) {
+			rowCells.push({ blocks: walkBlocksAtDepth(cellEntries, tableDepth) });
+			cellEntries = [];
 		}
 	}
-	if (cellParagraphs.length > 0 || rowCells.length > 0) throw new require_errors.DocFormatError("a table's paragraphs end without a row-ending mark to close the row's last cell");
+	if (cellEntries.length > 0 || rowCells.length > 0) throw new require_errors.DocFormatError("a table's paragraphs end without a row-ending mark to close the row's last cell");
 	const toleranceTwips = effectiveColumnBoundaryTolerance(rowDefinitions);
 	const columnBoundariesTwips = canonicalColumnBoundariesTwips(rowDefinitions, toleranceTwips);
 	return {

package/dist/table/read.d.cts

@@ -1,4 +1,4 @@
-import { ParagraphEntry } from "../read.cjs";
+import { t as ParagraphEntry } from "../paragraphs-DQUp8_HB.cjs";
 import { ContentBlock } from "document-schema.js";
 //#region src/table/read.d.ts
 declare function assembleBlocks(entries: readonly ParagraphEntry[]): ContentBlock[];

package/dist/table/read.d.ts

@@ -1,4 +1,4 @@
-import { ParagraphEntry } from "../read.js";
+import { t as ParagraphEntry } from "../paragraphs-s2lJ6w2I.js";
 import { ContentBlock } from "document-schema.js";
 //#region src/table/read.d.ts
 declare function assembleBlocks(entries: readonly ParagraphEntry[]): ContentBlock[];