reamkit 1.33.1 → 1.34.0

package/dist/esm/word/document-parser.js

@@ -734,22 +734,24 @@ function collectRuns(container, out, ctx, href, anchor, anchored, sdtRunProps) {
 		if (poIs(child, "w:r")) {
 			const parsed = parseRun(child, ctx, anchored);
 			const ranges = ctx.openCommentRanges && ctx.openCommentRanges.size > 0 ? [...ctx.openCommentRanges] : void 0;
-			const run = href !== void 0 || anchor !== void 0 || ranges !== void 0 || sdtRunProps ? {
-				...parsed.run,
-				...sdtRunProps ? { properties: {
-					...sdtRunProps,
-					...parsed.run.properties
-				} } : {},
-				...href !== void 0 ? { href } : {},
-				...anchor !== void 0 ? { anchor } : {},
-				...ranges !== void 0 ? { commentRangeRefs: ranges } : {}
-			} : parsed.run;
-			out.push({
-				run,
-				...parsed.fldChar ? { fldChar: parsed.fldChar } : {},
-				...parsed.instrText !== void 0 ? { instrText: parsed.instrText } : {},
-				...parsed.checkBox !== void 0 ? { checkBox: parsed.checkBox } : {}
-			});
+			for (const own of parsed.runs) {
+				const run = href !== void 0 || anchor !== void 0 || ranges !== void 0 || sdtRunProps ? {
+					...own,
+					...sdtRunProps ? { properties: {
+						...sdtRunProps,
+						...own.properties
+					} } : {},
+					...href !== void 0 ? { href } : {},
+					...anchor !== void 0 ? { anchor } : {},
+					...ranges !== void 0 ? { commentRangeRefs: ranges } : {}
+				} : own;
+				out.push({
+					run,
+					...parsed.fldChar ? { fldChar: parsed.fldChar } : {},
+					...parsed.instrText !== void 0 ? { instrText: parsed.instrText } : {},
+					...parsed.checkBox !== void 0 ? { checkBox: parsed.checkBox } : {}
+				});
+			}
 			continue;
 		}
 		if (poIs(child, "m:oMath")) {
@@ -890,6 +892,9 @@ function symbolChar(font, char) {
 function parseRun(r, ctx, anchored) {
 	const rPr = poChildren(r).find((c) => poIs(c, "w:rPr"));
 	const properties = parseRunProperties(rPr ? poElementToFlat(rPr) : void 0, ctx.themeFonts);
+	const children = expandMcChildren(poChildren(r));
+	const cuts = !children.some((c) => poIs(c, "w:fldChar") || poIs(c, "w:instrText"));
+	const runs = [];
 	let text = "";
 	let pageBreak = false;
 	let columnBreak = false;
@@ -901,7 +906,28 @@ function parseRun(r, ctx, anchored) {
 	let endnoteRef;
 	let commentRef;
 	let noteNumber = false;
-	for (const child of expandMcChildren(poChildren(r))) {
+	const cut = (last) => {
+		if (text !== "" || inlineImage !== void 0 || pageBreak || columnBreak || footnoteRef !== void 0 || endnoteRef !== void 0 || commentRef !== void 0 || noteNumber || last && runs.length === 0) runs.push({
+			text,
+			properties,
+			...inlineImage ? { inlineImage } : {},
+			...pageBreak ? { pageBreak: true } : {},
+			...columnBreak ? { columnBreak: true } : {},
+			...footnoteRef !== void 0 ? { footnoteRef } : {},
+			...endnoteRef !== void 0 ? { endnoteRef } : {},
+			...commentRef !== void 0 ? { commentRef } : {},
+			...noteNumber ? { noteNumber: true } : {}
+		});
+		text = "";
+		pageBreak = false;
+		columnBreak = false;
+		inlineImage = void 0;
+		footnoteRef = void 0;
+		endnoteRef = void 0;
+		commentRef = void 0;
+		noteNumber = false;
+	};
+	for (const child of children) {
 		if (poIs(child, "w:rPr")) continue;
 		if (poIs(child, "w:fldChar")) {
 			const t = poAttr(child, "fldCharType");
@@ -937,11 +963,14 @@ function parseRun(r, ctx, anchored) {
 		else if (poIs(child, "w:ptab")) text += "	";
 		else if (poIs(child, "w:br")) {
 			const breakType = poAttr(child, "type");
+			const forced = cuts && (breakType === "page" || breakType === "column");
+			if (forced) cut(false);
 			if (breakType === "page") pageBreak = true;
 			else {
 				if (breakType === "column") columnBreak = true;
 				text += "\n";
 			}
+			if (forced) cut(false);
 		} else if (poIs(child, "w:sym")) text += symbolChar(poAttr(child, "font"), poAttr(child, "char"));
 		else if (poIs(child, "w:noBreakHyphen")) text += "‑";
 		else if (poIs(child, "w:softHyphen")) text += "­";
@@ -981,18 +1010,9 @@ function parseRun(r, ctx, anchored) {
 			} else if (content && anchored) anchored.push(...blocksForDrawing(content, {}, ctx) ?? []);
 		}
 	}
+	cut(true);
 	return {
-		run: {
-			text,
-			properties,
-			...inlineImage ? { inlineImage } : {},
-			...pageBreak ? { pageBreak: true } : {},
-			...columnBreak ? { columnBreak: true } : {},
-			...footnoteRef !== void 0 ? { footnoteRef } : {},
-			...endnoteRef !== void 0 ? { endnoteRef } : {},
-			...commentRef !== void 0 ? { commentRef } : {},
-			...noteNumber ? { noteNumber: true } : {}
-		},
+		runs,
 		...fldChar ? { fldChar } : {},
 		...instrText !== void 0 ? { instrText } : {},
 		...checkBox !== void 0 ? { checkBox } : {}

package/dist/esm/word/docx-to-pdf.d.ts

@@ -38,11 +38,11 @@ export interface ConvertDocxOptions extends Omit<StyledRenderOptions, 'registry'
  */
 export declare function convertDocxToPdfSync(docx: Uint8Array, options: ConvertDocxOptions): Uint8Array;
 /**
- * The distinct curated font families the document references — the `w:ascii` of
- * `styles.xml` + `document.xml`, and the theme slots they point at instead
- * (§17.3.2.26) — always including the sans default `'arimo'` as a fallback for
- * unstyled runs / math / charts. The async path fetches one substitute set per
- * family so each run renders in the right one.
+ * The distinct curated font families the document references — the `w:ascii`
+ * and `w:hAnsi` of `styles.xml` + `document.xml`, and the theme slots they point
+ * at instead (§17.3.2.26) — always including the sans default `'arimo'` as a
+ * fallback for unstyled runs / math / charts. The async path fetches one
+ * substitute set per family so each run renders in the right one.
  *
  * @param docx The `.docx` bytes.
  * @returns The resolved {@link FamilyKey} set; just the default when the package

package/dist/esm/word/docx-to-pdf.js

@@ -1,5 +1,6 @@
 import { fetchFontSet, resolveFamilyKey } from "../core/fonts/remote-fonts.js";
 import { FontRegistry } from "../core/font/font-registry.js";
+import "../core/font/index.js";
 import { signPdf } from "../pdf/signature.js";
 import { renderStyledPdf, renderStyledPdfEncrypted } from "../pdf/styled-page-renderer.js";
 import "../pdf/index.js";
@@ -15,6 +16,7 @@ var STYLES_PART = "word/styles.xml";
 var THEME_PART = "word/theme/theme1.xml";
 var MAIN_DOCUMENT_PART = "word/document.xml";
 var ASCII_FONT = /<w:rFonts[^>]*?\bw:(ascii|asciiTheme)="([^"]+)"/y;
+var HANSI_FONT = /<w:rFonts[^>]*?\bw:(hAnsi|hAnsiTheme)="([^"]+)"/y;
 /**
 * Synchronous one-shot conversion (requires `fonts`/`fontBytes`, no network).
 *
@@ -54,11 +56,11 @@ function prepareDocxStyledRender(docx, options) {
 	};
 }
 /**
-* The distinct curated font families the document references — the `w:ascii` of
-* `styles.xml` + `document.xml`, and the theme slots they point at instead
-* (§17.3.2.26) — always including the sans default `'arimo'` as a fallback for
-* unstyled runs / math / charts. The async path fetches one substitute set per
-* family so each run renders in the right one.
+* The distinct curated font families the document references — the `w:ascii`
+* and `w:hAnsi` of `styles.xml` + `document.xml`, and the theme slots they point
+* at instead (§17.3.2.26) — always including the sans default `'arimo'` as a
+* fallback for unstyled runs / math / charts. The async path fetches one
+* substitute set per family so each run renders in the right one.
 *
 * @param docx The `.docx` bytes.
 * @returns The resolved {@link FamilyKey} set; just the default when the package
@@ -78,8 +80,8 @@ function detectDocxFamilyKeys(docx) {
 		const data = pkg.getPart(part);
 		if (!data) continue;
 		const xml = decoder.decode(data);
-		for (const m of tagMatches(xml, "<w:rFonts", false, ASCII_FONT)) {
-			const name = m[1] === "ascii" ? m[2] : resolveWordThemeFont(m[2], themeFonts);
+		for (const slot of [ASCII_FONT, HANSI_FONT]) for (const m of tagMatches(xml, "<w:rFonts", false, slot)) {
+			const name = m[1] === "ascii" || m[1] === "hAnsi" ? m[2] : resolveWordThemeFont(m[2], themeFonts);
 			if (name !== void 0) keys.add(resolveFamilyKey(name));
 		}
 	}

package/dist/esm/word/docx-writer.js

@@ -1235,11 +1235,9 @@ function numberingXml(numbering) {
 	return out.join("");
 }
 function levelXml(level) {
-	const inner = [
-		`<w:start w:val="${level.start}"/>`,
-		`<w:numFmt w:val="${level.format}"/>`,
-		`<w:lvlText w:val="${escapeAttr(level.lvlText)}"/>`
-	];
+	const inner = [`<w:start w:val="${level.start}"/>`, `<w:numFmt w:val="${level.format}"/>`];
+	if (level.suffix !== void 0) inner.push(`<w:suff w:val="${level.suffix}"/>`);
+	inner.push(`<w:lvlText w:val="${escapeAttr(level.lvlText)}"/>`);
 	const pPr = rawParaPrXml(level.paragraphProperties);
 	if (pPr) inner.push(pPr);
 	const rPr = rawRunPrXml(level.runProperties);

package/dist/esm/word/font-embed.js

@@ -1,4 +1,5 @@
 import { buildTrueType, editableEmbedding } from "../core/font/ttf-build.js";
+import "../core/font/index.js";
 import { FEATURES } from "../core/ir/features.js";
 import { obfuscateEmbeddedFont } from "./font-table.js";
 import { md5 } from "../core/crypto/primitives.js";

package/dist/esm/word/font-table.js

@@ -1,4 +1,5 @@
 import { FontRegistry } from "../core/font/font-registry.js";
+import "../core/font/index.js";
 import { isWordChar, tagMatches } from "../core/opc/tag-scan.js";
 //#region src/word/font-table.ts
 var FONT_TABLE_PART = "word/fontTable.xml";

package/dist/esm/word/numbering-parser.js

@@ -126,12 +126,15 @@ function parseLevel(node, picBullets, themeFonts) {
 	const picId = getValVal(lvlEl["w:lvlPicBulletId"]);
 	const picBullet = picId !== void 0 ? picBullets.get(picId) : void 0;
 	const isLegal = "w:isLgl" in lvlEl && getValVal(lvlEl["w:isLgl"]) !== "0";
+	const suff = getValVal(lvlEl["w:suff"]);
+	const suffix = suff === "space" || suff === "nothing" ? suff : void 0;
 	return {
 		ilvl,
 		start: Number.isFinite(start) ? start : 1,
 		format,
 		lvlText,
 		...isLegal ? { isLegal: true } : {},
+		...suffix ? { suffix } : {},
 		...picBullet ? { picBullet } : {},
 		paragraphProperties: parseParagraphProperties(lvlEl["w:pPr"], 0, themeFonts),
 		runProperties: parseRunProperties(lvlEl["w:rPr"], themeFonts)

package/dist/esm/word/paragraph-properties.js

@@ -1,5 +1,5 @@
 import { eighthPtToPt, pt, twipsToPt } from "../core/ir/units.js";
-import { asArray, asElement, getAttr, getVal, parseIntAttr, parseToggle } from "./xml-helpers.js";
+import { asArray, asElement, foldRepeatedProperties, getAttr, getVal, parseIntAttr, parseToggle } from "./xml-helpers.js";
 import { parseRunProperties } from "./run-properties.js";
 //#region src/word/paragraph-properties.ts
 var ALIGNMENTS = /* @__PURE__ */ new Set([
@@ -105,7 +105,7 @@ function parseFramePr(node) {
 * @returns The extracted properties; an empty object when `pPr` is absent.
 */
 function parseParagraphProperties(pPr, autoSpacingPt = 0, themeFonts) {
-	const el = asElement(pPr);
+	const el = foldRepeatedProperties(pPr);
 	if (!el) return {};
 	const out = {};
 	if ("w:pStyle" in el) {

package/dist/esm/word/run-properties.js

@@ -1,6 +1,6 @@
 import { halfPtToPt, twipsToPt } from "../core/ir/units.js";
 import { resolveThemeFont } from "../core/drawingml/theme-parser.js";
-import { asElement, getAttr, getVal, parseHalfPointAttr, parseToggle } from "./xml-helpers.js";
+import { asElement, foldRepeatedProperties, getAttr, getVal, parseHalfPointAttr, parseToggle } from "./xml-helpers.js";
 import { resolveWordThemeFont } from "./theme-fonts.js";
 import { shadingFillHex } from "./shading.js";
 //#region src/word/run-properties.ts
@@ -69,7 +69,7 @@ var HIGHLIGHT_COLORS = /* @__PURE__ */ new Map([
 * @returns The parsed properties; an empty object when `rPr` is absent.
 */
 function parseRunProperties(rPr, themeFonts) {
-	const el = asElement(rPr);
+	const el = foldRepeatedProperties(rPr);
 	if (!el) return {};
 	const out = {};
 	if ("w:rStyle" in el) {

package/dist/esm/word/xml-helpers.d.ts

@@ -58,6 +58,24 @@ export declare function parseIntAttr(node: unknown, name: string): number | unde
  * @returns The toggle state, or `undefined` when the property is absent.
  */
 export declare function parseToggle(node: unknown): boolean | undefined;
+/**
+ * Narrow a `w:rPr` or `w:pPr` to an element, with every property written more
+ * than once in it folded into the one Word applies.
+ *
+ * The schema allows each property once, but producers repeat them: LibreOffice
+ * writes `<w:b w:val="false"/>` twice, and the flat tree hands the pair over as
+ * an array, which every reader took for nothing and {@link parseToggle} for
+ * "on", so the run came out bold. Word applies the occurrences in order, each
+ * over the ones before it (Word for Mac, measured): what a later one states
+ * wins, what it leaves unsaid stands. Two `w:ind` keep the left indent of one
+ * and the right of the other, two `w:pBdr` draw the rules of both, and two
+ * `w:rFonts` name the faces of both.
+ *
+ * @param node The parsed `w:rPr` / `w:pPr`.
+ * @returns The element with its repeats folded (the same object when it has
+ *          none), or `undefined` when `node` is not a single element.
+ */
+export declare function foldRepeatedProperties(node: unknown): XmlElement | undefined;
 /**
  * Read the `#text` content of a node. A bare string/number is returned directly;
  * an element returns its `#text` child (coerced to string), else the empty string.

package/dist/esm/word/xml-helpers.js

@@ -90,5 +90,52 @@ function parseToggle(node) {
 	if (v === void 0) return true;
 	return !(v === "false" || v === "0" || v === "off");
 }
+/**
+* Narrow a `w:rPr` or `w:pPr` to an element, with every property written more
+* than once in it folded into the one Word applies.
+*
+* The schema allows each property once, but producers repeat them: LibreOffice
+* writes `<w:b w:val="false"/>` twice, and the flat tree hands the pair over as
+* an array, which every reader took for nothing and {@link parseToggle} for
+* "on", so the run came out bold. Word applies the occurrences in order, each
+* over the ones before it (Word for Mac, measured): what a later one states
+* wins, what it leaves unsaid stands. Two `w:ind` keep the left indent of one
+* and the right of the other, two `w:pBdr` draw the rules of both, and two
+* `w:rFonts` name the faces of both.
+*
+* @param node The parsed `w:rPr` / `w:pPr`.
+* @returns The element with its repeats folded (the same object when it has
+*          none), or `undefined` when `node` is not a single element.
+*/
+function foldRepeatedProperties(node) {
+	const el = asElement(node);
+	if (!el) return void 0;
+	let out;
+	for (const [tag, v] of Object.entries(el)) {
+		if (!Array.isArray(v)) continue;
+		out ??= { ...el };
+		out[tag] = fold(v);
+	}
+	return out ?? el;
+}
+function fold(v) {
+	return Array.isArray(v) ? v.reduce(applyOver) : v;
+}
+/**
+* One occurrence of a property element applied over an earlier one: its
+* attributes win, its child elements apply over theirs in turn, and its tab
+* stops join the ones already set (Word keeps the stops of both `w:tabs`).
+*/
+function applyOver(earlier, later) {
+	const a = asElement(earlier) ?? {};
+	const b = asElement(later) ?? {};
+	if (valOnly(a) && valOnly(b)) return later;
+	const out = { ...a };
+	for (const [k, v] of Object.entries(b)) out[k] = !Object.hasOwn(out, k) || k.startsWith("@_") ? v : k === "w:tab" ? [...asArray(out[k]), ...asArray(v)] : applyOver(fold(out[k]), fold(v));
+	return out;
+}
+function valOnly(el) {
+	return Object.keys(el).every((k) => k === "@_w:val" || k === "@_val");
+}
 //#endregion
-export { asArray, asElement, getAttr, getVal, parseHalfPointAttr, parseIntAttr, parseToggle };
+export { asArray, asElement, foldRepeatedProperties, getAttr, getVal, parseHalfPointAttr, parseIntAttr, parseToggle };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "reamkit",
-  "version": "1.33.1",
+  "version": "1.34.0",
   "description": "Ream — convert DOCX, XLSX, PPTX, PDF and legacy DOC, XLS and PPT to PDF, SVG, HTML, Markdown, DOCX and XLSX, built from scratch on the ECMA-376 and ISO 32000 specifications. Parse once, convert anywhere.",
   "license": "MIT",
   "author": "Alexandr Krassavin <info@reamkit.dev>",
@@ -102,8 +102,8 @@
     "publint": "^0.3.21",
     "tsx": "^4.19.0",
     "typescript": "^5.7.0",
-    "vite": "8.3.2",
-    "vitest": "^5.0.3"
+    "vite": "8.3.5",
+    "vitest": "5.0.5"
   },
   "overrides": {
     "@microsoft/api-extractor": {