doc-codec 1.0.1 → 1.1.0

package/dist/read.d.ts

@@ -1,12 +1,29 @@
-import { Fib } from "./fib/fib.js";
-import { ContentDocument } from "document-schema.js";
+import { t as Fib } from "./fib-CtsLtF96.js";
+import { t as Prl } from "./sprm-Djg5RNiq.js";
+import { t as ParagraphProperties } from "./pap-BkFo8dZO.js";
+import { n as NumberingDefinitions } from "./numbering-DF1z1PyG.js";
+import { ContentDocument, ContentParagraph } from "document-schema.js";
 //#region src/read.d.ts
 interface DocStreams {
   readonly wordDocument: Uint8Array;
   readonly table: Uint8Array;
   readonly fib: Fib;
+  /** The raw "\x05SummaryInformation" stream bytes, or undefined when the container carries none -- a valid, spec-conformant Word Binary File need not carry document properties at all. */
+  readonly metadata: Uint8Array<ArrayBuffer> | undefined;
 }
 declare function readDocStreams(bytes: Uint8Array<ArrayBuffer>): DocStreams;
-declare function readDocContent(bytes: Uint8Array<ArrayBuffer>): ContentDocument;
+/** readDocContent's own return type: a ContentDocument (kind 'wordprocessing') plus numbering -- the list-level formatting (glyph/format, level-text template, start-at value) PlfLst/PlfLfo carry, which ContentListMembership has nowhere to hold. Mirrors ooxml.js's own DocxDocument.numbering exactly in field name and NumberingDefinitions' own shape (see list/numbering.ts's top comment for why it sits outside the shared schema rather than inside ContentListMembership); unlike DocxDocument, DocContent stays a genuine ContentDocument subtype (an intersection, not a fresh shape) since readDocContent already had one return type to widen rather than two to reconcile. */
+type DocContent = ContentDocument & {
+  readonly numbering: NumberingDefinitions;
+};
+declare function readDocContent(bytes: Uint8Array<ArrayBuffer>): DocContent;
+/** One paragraph/cell/row-ending mark, still flat -- table/read.ts's assembleBlocks is what folds a run of these into a real ContentTable. `properties` and `grpprl` are carried alongside the already-built `paragraph` because table grouping needs sprmPFInTable/sprmPFTtp/sprmPItap (properties) and, on a row's own mark, its table-defining sgc-5 sprms (grpprl) -- neither of which survives onto a plain ContentParagraph. */
+interface ParagraphEntry {
+  readonly paragraph: ContentParagraph;
+  readonly properties: ParagraphProperties;
+  readonly grpprl: readonly Prl[];
+  /** The character that terminated this paragraph in the text stream: PARAGRAPH_MARK, CELL_MARK, or SECTION_MARK. */
+  readonly terminator: number;
+}
 //#endregion
-export { DocStreams, readDocContent, readDocStreams };
+export { DocContent, DocStreams, ParagraphEntry, readDocContent, readDocStreams };

package/dist/read.js

@@ -1,6 +1,6 @@
 import { DocFormatError } from "./errors.js";
 import { slice } from "./bytes.js";
-import { WORD_DOCUMENT_STREAM } from "./detect.js";
+import { SUMMARY_INFORMATION_STREAM, WORD_DOCUMENT_STREAM } from "./detect.js";
 import { parseFib, tableStreamName } from "./fib/fib.js";
 import { parseClx } from "./text/piece-table.js";
 import { readTextRange } from "./text/characters.js";
@@ -9,10 +9,14 @@ import { readGrpprl } from "./prop/sprm.js";
 import { PropertyBinTable } from "./prop/fkp.js";
 import { applyCharacterSprms } from "./prop/chp.js";
 import { applyParagraphSprms } from "./prop/pap.js";
+import { readSectionProperties } from "./prop/sep.js";
 import { headingLevelFromIstd, parseStsh } from "./style/stsh.js";
-import { readCompoundFile } from "archive-codec";
+import { parseFontTable } from "./style/fonts.js";
+import { readNumberingDefinitions } from "./list/numbering.js";
+import { assembleBlocks } from "./table/read.js";
+import { readCompoundFile, readSummaryInformation, summaryInformationToLayoutMetadata } from "archive-codec";
 //#region src/read.ts
-/** The page geometry every section is given, because this reader does not yet read a document's own. US Letter with one-inch margins is Word's own default for a new document; a document that states otherwise is not yet consulted, so this is a placeholder the schema requires rather than a fact read from the file. */
+/** 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. */
 const DEFAULT_PAGE_SIZE = {
 	widthPt: 612,
 	heightPt: 792
@@ -31,64 +35,86 @@ function readDocStreams(bytes) {
 	const wanted = tableStreamName(fib);
 	const table = streams.find((stream) => stream.path === wanted);
 	if (table === void 0) throw new DocFormatError(`FibBase.fWhichTblStm selects the "${wanted}" stream, which this compound file does not contain`);
+	const metadata = streams.find((stream) => stream.path === SUMMARY_INFORMATION_STREAM);
 	return {
 		wordDocument: wordDocument.bytes,
 		table: table.bytes,
-		fib
+		fib,
+		metadata: metadata?.bytes
 	};
 }
 function readDocContent(bytes) {
-	const { wordDocument, table, fib } = readDocStreams(bytes);
+	const { wordDocument, table, fib, metadata } = readDocStreams(bytes);
 	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 blocks = readParagraphs(range.text, range.fcs, {
+	const entries = readParagraphs(range.text, range.fcs, {
 		chpxTable,
 		papxTable,
 		styles,
+		fonts,
 		characterProperties: /* @__PURE__ */ new Map()
 	});
+	const blocks = assembleBlocks(entries);
+	const numbering = readNumberingDefinitions(table, fib);
+	const sectionProperties = readSectionProperties(wordDocument, table, fib);
 	return {
 		kind: "wordprocessing",
-		metadata: {},
+		metadata: metadata === void 0 ? {} : summaryInformationToLayoutMetadata(readSummaryInformation(metadata)),
 		sections: [{
-			pageSize: DEFAULT_PAGE_SIZE,
-			margins: DEFAULT_MARGINS,
+			pageSize: {
+				widthPt: sectionProperties.pageWidthPt ?? DEFAULT_PAGE_SIZE.widthPt,
+				heightPt: sectionProperties.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
+			},
 			blocks
-		}]
+		}],
+		numbering
 	};
 }
 function readParagraphs(text, fcs, context) {
-	const blocks = [];
+	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`);
-		blocks.push(buildParagraph(text.slice(start, index), fcs.slice(start, index), markFc, context));
+		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`);
-		blocks.push(buildParagraph(text.slice(start), fcs.slice(start), firstFc, context));
+		entries.push(buildParagraph(text.slice(start), fcs.slice(start), firstFc, 13, context));
 	}
-	return blocks;
+	return entries;
 }
-function buildParagraph(text, fcs, propertyFc, context) {
+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;
-		applyParagraphSprms(readGrpprl(papx.grpprl), properties);
+		applyParagraphSprms(grpprl, properties);
 	}
 	return {
-		kind: "paragraph",
-		runs: buildRuns(text, fcs, context),
-		...paragraphAttributes(properties, context)
+		paragraph: {
+			kind: "paragraph",
+			runs: buildRuns(text, fcs, context),
+			...paragraphAttributes(properties, context)
+		},
+		properties,
+		grpprl,
+		terminator
 	};
 }
 function paragraphAttributes(properties, context) {
@@ -155,7 +181,7 @@ function buildRuns(text, fcs, context) {
 			let properties = context.characterProperties.get(key);
 			if (properties === void 0) {
 				properties = {};
-				if (grpprl !== void 0) applyCharacterSprms(readGrpprl(grpprl), properties);
+				if (grpprl !== void 0) applyCharacterSprms(readGrpprl(grpprl), properties, context.fonts);
 				context.characterProperties.set(key, properties);
 			}
 			currentProperties = properties;

package/dist/sprm-Djg5RNiq.d.cts

@@ -0,0 +1,26 @@
+//#region src/prop/sprm.d.ts
+interface Sprm {
+  /** The raw 16-bit opcode, which is what the property tables in [MS-DOC] 2.6.1-2.6.5 are keyed on. */
+  readonly value: number;
+  readonly ispmd: number;
+  readonly fSpec: 0 | 1;
+  readonly sgc: number;
+  readonly spra: number;
+}
+interface Prl {
+  readonly sprm: Sprm;
+  readonly operand: Uint8Array;
+}
+/** Sprm.sgc, [MS-DOC] 2.6.1 -- which property family the opcode belongs to. */
+declare const SGC: {
+  readonly paragraph: 1;
+  readonly character: 2;
+  readonly picture: 3;
+  readonly section: 4;
+  readonly table: 5;
+};
+declare function decodeSprm(value: number): Sprm;
+declare function operandSize(sprm: Sprm, bytes: Uint8Array, offset: number): number;
+declare function readGrpprl(bytes: Uint8Array): Prl[];
+//#endregion
+export { operandSize as a, decodeSprm as i, SGC as n, readGrpprl as o, Sprm as r, Prl as t };

package/dist/sprm-Djg5RNiq.d.ts

@@ -0,0 +1,26 @@
+//#region src/prop/sprm.d.ts
+interface Sprm {
+  /** The raw 16-bit opcode, which is what the property tables in [MS-DOC] 2.6.1-2.6.5 are keyed on. */
+  readonly value: number;
+  readonly ispmd: number;
+  readonly fSpec: 0 | 1;
+  readonly sgc: number;
+  readonly spra: number;
+}
+interface Prl {
+  readonly sprm: Sprm;
+  readonly operand: Uint8Array;
+}
+/** Sprm.sgc, [MS-DOC] 2.6.1 -- which property family the opcode belongs to. */
+declare const SGC: {
+  readonly paragraph: 1;
+  readonly character: 2;
+  readonly picture: 3;
+  readonly section: 4;
+  readonly table: 5;
+};
+declare function decodeSprm(value: number): Sprm;
+declare function operandSize(sprm: Sprm, bytes: Uint8Array, offset: number): number;
+declare function readGrpprl(bytes: Uint8Array): Prl[];
+//#endregion
+export { operandSize as a, decodeSprm as i, SGC as n, readGrpprl as o, Sprm as r, Prl as t };

package/dist/style/fonts.cjs

@@ -0,0 +1,68 @@
+Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
+const require_errors = require("../errors.cjs");
+const require_bytes = require("../bytes.cjs");
+//#region src/style/fonts.ts
+/** FFN's fixed head before xszFfn: ffid(1) + wWeight(2) + chs(1) + ixchSzAlt(1) + panose(10) + fs(24). */
+const FFN_FIXED_SIZE = 39;
+/** chs's ANSI_CHARSET, [MS-DOC] 2.9.87's own value 0 -- the character set every font this package writes is declared to use, since the document's own text is written as Unicode regardless (see text/piece-table-write.ts). */
+const FFN_CHARSET_ANSI = 0;
+/** A non-extended STTB's own per-entry length prefix is one byte, so one FFN record (fixed head plus a null-terminated name, each name character and terminator being 2 bytes) cannot exceed this. */
+const MAX_ENTRY_BYTES = 255;
+function parseFontTable(sttbfFfn) {
+	const cData = require_bytes.readUint16LE(sttbfFfn, 0);
+	const cbExtra = require_bytes.readUint16LE(sttbfFfn, 2);
+	if (cbExtra !== 0) throw new require_errors.DocFormatError(`SttbfFfn.cbExtra is ${cbExtra}, but [MS-DOC] 2.9.253 requires it to be 0`);
+	const names = [];
+	let cursor = 4;
+	for (let index = 0; index < cData; index += 1) {
+		const cch = require_bytes.readUint8(sttbfFfn, cursor);
+		const record = require_bytes.slice(sttbfFfn, cursor + 1, cch, `SttbfFfn entry ${index}`);
+		names.push(readFfnName(record, index));
+		cursor += 1 + cch;
+	}
+	return names;
+}
+function readFfnName(record, index) {
+	if (record.length < 41) throw new require_errors.DocFormatError(`FFN record ${index} is ${record.length} bytes, shorter than the fixed ${FFN_FIXED_SIZE}-byte head plus xszFfn's own null terminator`);
+	let name = "";
+	for (let offset = FFN_FIXED_SIZE; offset + 2 <= record.length; offset += 2) {
+		const unit = require_bytes.readUint16LE(record, offset);
+		if (unit === 0) return name;
+		name += String.fromCharCode(unit);
+	}
+	throw new require_errors.DocFormatError(`FFN record ${index}'s xszFfn runs to the end of the record with no null terminator`);
+}
+function buildFontTable(names) {
+	const bytes = [];
+	const push16 = (value) => {
+		bytes.push(value & 255, value >> 8 & 255);
+	};
+	push16(names.length);
+	push16(0);
+	for (const name of names) {
+		const record = buildFfnRecord(name);
+		if (record.length > MAX_ENTRY_BYTES) throw new require_errors.DocFormatError(`font name ${JSON.stringify(name)} produces a ${record.length}-byte FFN record, past the ${MAX_ENTRY_BYTES}-byte limit a non-extended STTB's one-byte cch can address`);
+		bytes.push(record.length, ...record);
+	}
+	return new Uint8Array(bytes);
+}
+function buildFfnRecord(name) {
+	const record = [
+		0,
+		144,
+		1,
+		FFN_CHARSET_ANSI,
+		0,
+		...new Array(10).fill(0),
+		...new Array(24).fill(0)
+	];
+	for (const character of name) {
+		const code = character.charCodeAt(0);
+		record.push(code & 255, code >> 8 & 255);
+	}
+	record.push(0, 0);
+	return record;
+}
+//#endregion
+exports.buildFontTable = buildFontTable;
+exports.parseFontTable = parseFontTable;

package/dist/style/fonts.d.cts

@@ -0,0 +1,5 @@
+//#region src/style/fonts.d.ts
+declare function parseFontTable(sttbfFfn: Uint8Array): string[];
+declare function buildFontTable(names: readonly string[]): Uint8Array;
+//#endregion
+export { buildFontTable, parseFontTable };

package/dist/style/fonts.d.ts

@@ -0,0 +1,5 @@
+//#region src/style/fonts.d.ts
+declare function parseFontTable(sttbfFfn: Uint8Array): string[];
+declare function buildFontTable(names: readonly string[]): Uint8Array;
+//#endregion
+export { buildFontTable, parseFontTable };

package/dist/style/fonts.js

@@ -0,0 +1,66 @@
+import { DocFormatError } from "../errors.js";
+import { readUint16LE, readUint8, slice } from "../bytes.js";
+//#region src/style/fonts.ts
+/** FFN's fixed head before xszFfn: ffid(1) + wWeight(2) + chs(1) + ixchSzAlt(1) + panose(10) + fs(24). */
+const FFN_FIXED_SIZE = 39;
+/** chs's ANSI_CHARSET, [MS-DOC] 2.9.87's own value 0 -- the character set every font this package writes is declared to use, since the document's own text is written as Unicode regardless (see text/piece-table-write.ts). */
+const FFN_CHARSET_ANSI = 0;
+/** A non-extended STTB's own per-entry length prefix is one byte, so one FFN record (fixed head plus a null-terminated name, each name character and terminator being 2 bytes) cannot exceed this. */
+const MAX_ENTRY_BYTES = 255;
+function parseFontTable(sttbfFfn) {
+	const cData = readUint16LE(sttbfFfn, 0);
+	const cbExtra = readUint16LE(sttbfFfn, 2);
+	if (cbExtra !== 0) throw new DocFormatError(`SttbfFfn.cbExtra is ${cbExtra}, but [MS-DOC] 2.9.253 requires it to be 0`);
+	const names = [];
+	let cursor = 4;
+	for (let index = 0; index < cData; index += 1) {
+		const cch = readUint8(sttbfFfn, cursor);
+		const record = slice(sttbfFfn, cursor + 1, cch, `SttbfFfn entry ${index}`);
+		names.push(readFfnName(record, index));
+		cursor += 1 + cch;
+	}
+	return names;
+}
+function readFfnName(record, index) {
+	if (record.length < 41) throw new DocFormatError(`FFN record ${index} is ${record.length} bytes, shorter than the fixed ${FFN_FIXED_SIZE}-byte head plus xszFfn's own null terminator`);
+	let name = "";
+	for (let offset = FFN_FIXED_SIZE; offset + 2 <= record.length; offset += 2) {
+		const unit = readUint16LE(record, offset);
+		if (unit === 0) return name;
+		name += String.fromCharCode(unit);
+	}
+	throw new DocFormatError(`FFN record ${index}'s xszFfn runs to the end of the record with no null terminator`);
+}
+function buildFontTable(names) {
+	const bytes = [];
+	const push16 = (value) => {
+		bytes.push(value & 255, value >> 8 & 255);
+	};
+	push16(names.length);
+	push16(0);
+	for (const name of names) {
+		const record = buildFfnRecord(name);
+		if (record.length > MAX_ENTRY_BYTES) throw new DocFormatError(`font name ${JSON.stringify(name)} produces a ${record.length}-byte FFN record, past the ${MAX_ENTRY_BYTES}-byte limit a non-extended STTB's one-byte cch can address`);
+		bytes.push(record.length, ...record);
+	}
+	return new Uint8Array(bytes);
+}
+function buildFfnRecord(name) {
+	const record = [
+		0,
+		144,
+		1,
+		FFN_CHARSET_ANSI,
+		0,
+		...new Array(10).fill(0),
+		...new Array(24).fill(0)
+	];
+	for (const character of name) {
+		const code = character.charCodeAt(0);
+		record.push(code & 255, code >> 8 & 255);
+	}
+	record.push(0, 0);
+	return record;
+}
+//#endregion
+export { buildFontTable, parseFontTable };

package/dist/style/stsh.cjs

@@ -58,8 +58,31 @@ function readXstz(bytes, offset, what) {
 function headingLevelFromIstd(istd) {
 	return istd >= 1 && istd <= 9 ? istd : void 0;
 }
+function buildEmptyStsh() {
+	const stshi = [];
+	const push16 = (value) => {
+		stshi.push(value & 255, value >> 8 & 255);
+	};
+	push16(0);
+	push16(STDF_SIZE_WITHOUT_POST_2000);
+	push16(1);
+	push16(0);
+	push16(15);
+	push16(0);
+	push16(0);
+	push16(0);
+	push16(0);
+	push16(0);
+	push16(4);
+	return new Uint8Array([
+		stshi.length & 255,
+		stshi.length >> 8 & 255,
+		...stshi
+	]);
+}
 //#endregion
 exports.STI_USER_DEFINED = STI_USER_DEFINED;
 exports.STK = STK;
+exports.buildEmptyStsh = buildEmptyStsh;
 exports.headingLevelFromIstd = headingLevelFromIstd;
 exports.parseStsh = parseStsh;

package/dist/style/stsh.d.cts

@@ -25,5 +25,6 @@ interface StyleSheet {
 }
 declare function parseStsh(stsh: Uint8Array): StyleSheet;
 declare function headingLevelFromIstd(istd: number): number | undefined;
+declare function buildEmptyStsh(): Uint8Array;
 //#endregion
-export { STI_USER_DEFINED, STK, Style, StyleSheet, headingLevelFromIstd, parseStsh };
+export { STI_USER_DEFINED, STK, Style, StyleSheet, buildEmptyStsh, headingLevelFromIstd, parseStsh };

package/dist/style/stsh.d.ts

@@ -25,5 +25,6 @@ interface StyleSheet {
 }
 declare function parseStsh(stsh: Uint8Array): StyleSheet;
 declare function headingLevelFromIstd(istd: number): number | undefined;
+declare function buildEmptyStsh(): Uint8Array;
 //#endregion
-export { STI_USER_DEFINED, STK, Style, StyleSheet, headingLevelFromIstd, parseStsh };
+export { STI_USER_DEFINED, STK, Style, StyleSheet, buildEmptyStsh, headingLevelFromIstd, parseStsh };

package/dist/style/stsh.js

@@ -57,5 +57,27 @@ function readXstz(bytes, offset, what) {
 function headingLevelFromIstd(istd) {
 	return istd >= 1 && istd <= 9 ? istd : void 0;
 }
+function buildEmptyStsh() {
+	const stshi = [];
+	const push16 = (value) => {
+		stshi.push(value & 255, value >> 8 & 255);
+	};
+	push16(0);
+	push16(STDF_SIZE_WITHOUT_POST_2000);
+	push16(1);
+	push16(0);
+	push16(15);
+	push16(0);
+	push16(0);
+	push16(0);
+	push16(0);
+	push16(0);
+	push16(4);
+	return new Uint8Array([
+		stshi.length & 255,
+		stshi.length >> 8 & 255,
+		...stshi
+	]);
+}
 //#endregion
-export { STI_USER_DEFINED, STK, headingLevelFromIstd, parseStsh };
+export { STI_USER_DEFINED, STK, buildEmptyStsh, headingLevelFromIstd, parseStsh };