reamkit 1.33.2 → 1.34.1

package/dist/esm/word/doc/doc-text.js

@@ -7,6 +7,12 @@ var OFF_CCPFTN = 80;
 var OFF_CCPHDD = 84;
 var OFF_FCPLCFHDD = 242;
 var OFF_LCBPLCFHDD = 246;
+var OFF_FCDOP = 402;
+var OFF_LCBDOP = 406;
+var OFF_FCSTSHF = 162;
+var OFF_LCBSTSHF = 166;
+var OFF_FCSTTBFFFN = 274;
+var OFF_LCBSTTBFFFN = 278;
 var OFF_FCPLCFBTECHPX = 250;
 var OFF_LCBPLCFBTECHPX = 254;
 var OFF_FCPLCFBTEPAPX = 258;
@@ -31,6 +37,11 @@ var SPRM_C_KUL = 10814;
 var SPRM_C_HPS = 19011;
 var SPRM_C_FSPEC = 2133;
 var SPRM_C_PIC_LOCATION = 27139;
+var SPRM_C_ISTD = 18992;
+var SPRM_C_RG_FTC0 = 19023;
+var SPRM_C_RG_FTC1 = 19024;
+var SPRM_C_RG_FTC2 = 19025;
+var SPRM_C_FTC_BI = 19038;
 var PIC_CHAR = 1;
 var SPRM_P_JC = 9219;
 var SPRM_P_DXA_RIGHT = 33806;
@@ -43,13 +54,26 @@ var SPRM_P_F_TTP = 9239;
 var SPRM_P_F_WIDOW_CONTROL = 9265;
 var SPRM_P_F_KEEP = 9221;
 var SPRM_P_F_KEEP_FOLLOW = 9222;
+var SPRM_P_F_PAGE_BREAK_BEFORE = 9223;
 var SPRM_T_DEF_TABLE = 54792;
 var SPRM_T_DEF_TABLE_SHD = 54802;
 var MAX_ROW_CELLS = 256;
 var SPRM_P_ILVL = 9738;
 var SPRM_P_ILFO = 17931;
+var SPRM_S_BKC = 12297;
+var SPRM_S_F_TITLE_PAGE = 12298;
+var SPRM_S_NFC_PGN = 12302;
+var SPRM_S_F_PGN_RESTART = 12305;
+var SPRM_S_PGN_START_97 = 20508;
 var SPRM_S_XA_PAGE = 45087;
 var SPRM_S_YA_PAGE = 45088;
+var SPRM_S_DXA_LEFT = 45089;
+var SPRM_S_DXA_RIGHT = 45090;
+var SPRM_S_DYA_TOP = 36899;
+var SPRM_S_DYA_BOTTOM = 36900;
+var SPRM_S_DYA_HDR_TOP = 45079;
+var SPRM_S_DYA_HDR_BOTTOM = 45080;
+var SPRM_S_DZA_GUTTER = 45093;
 var MAX_PAGE_TWIPS = 31680;
 var SPRA_LEN = [
 	1,
@@ -63,35 +87,88 @@ var SPRA_LEN = [
 ];
 var SPRM_LONG_OPERAND = /* @__PURE__ */ new Set([54792, 50701]);
 var CELL_MARK = 7;
+var PAGE_BREAK = 12;
+var COLUMN_BREAK = 14;
+var LINE_BREAK = 11;
 var FIELD_BEGIN = 19;
 var FIELD_SEPARATOR = 20;
 var FIELD_END = 21;
+var MAX_FIELD_INSTR = 64;
 var MAX_TEXT = 1 << 24;
+var STK_PARAGRAPH = 1;
+var STK_CHARACTER = 2;
+var ISTD_NIL = 4095;
+var MAX_STYLE_DEPTH = 16;
 var EMPTY = {
 	paragraphs: [],
 	encrypted: false
 };
-function readSectionPageSize(wd, table) {
+function readSections(wd, table) {
+	const sections = [];
+	const ends = /* @__PURE__ */ new Map();
 	const fc = u32(wd, OFF_FCPLCFSED);
 	const lcb = u32(wd, OFF_LCBPLCFSED);
-	if (lcb < 20 || fc + lcb > table.length) return void 0;
-	const fcSepx = u32(table.subarray(fc, fc + lcb), (Math.floor((lcb - 4) / 16) + 1) * 4 + 2);
-	if (fcSepx === 4294967295 || fcSepx + 2 > wd.length) return void 0;
-	const cb = u16(wd, fcSepx);
-	const grpprl = wd.subarray(fcSepx + 2, Math.min(wd.length, fcSepx + 2 + cb));
-	let xa = 0;
-	let ya = 0;
-	for (const { sprm, op } of sprms(grpprl)) if (sprm === SPRM_S_XA_PAGE) xa = u16(grpprl, op);
-	else if (sprm === SPRM_S_YA_PAGE) ya = u16(grpprl, op);
-	if (xa < 1 || ya < 1 || xa > MAX_PAGE_TWIPS || ya > MAX_PAGE_TWIPS) return void 0;
+	if (lcb < 20 || fc + lcb > table.length) return {
+		sections,
+		ends
+	};
+	const plc = table.subarray(fc, fc + lcb);
+	const n = Math.floor((lcb - 4) / 16);
+	for (let i = 0; i < n; i++) {
+		if (i + 1 < n) ends.set(u32(plc, (i + 1) * 4), i);
+		const fcSepx = u32(plc, (n + 1) * 4 + i * 12 + 2);
+		if (fcSepx === 4294967295 || fcSepx + 2 > wd.length) {
+			sections.push({});
+			continue;
+		}
+		const cb = u16(wd, fcSepx);
+		const grpprl = wd.subarray(fcSepx + 2, Math.min(wd.length, fcSepx + 2 + cb));
+		let bkc;
+		let titlePage;
+		let nfcPgn;
+		let restart = false;
+		let pgnStart = 1;
+		let xa = 0;
+		let ya = 0;
+		const margins = {};
+		const margin = (key, twips) => {
+			if (twips <= MAX_PAGE_TWIPS) margins[key] = twips;
+		};
+		for (const { sprm, op } of sprms(grpprl)) if (sprm === SPRM_S_BKC) bkc = grpprl[op];
+		else if (sprm === SPRM_S_F_TITLE_PAGE) titlePage = grpprl[op] !== 0;
+		else if (sprm === SPRM_S_NFC_PGN) nfcPgn = grpprl[op];
+		else if (sprm === SPRM_S_F_PGN_RESTART) restart = grpprl[op] !== 0;
+		else if (sprm === SPRM_S_PGN_START_97) pgnStart = u16(grpprl, op);
+		else if (sprm === SPRM_S_XA_PAGE) xa = u16(grpprl, op);
+		else if (sprm === SPRM_S_YA_PAGE) ya = u16(grpprl, op);
+		else if (sprm === SPRM_S_DXA_LEFT) margin("leftTwips", u16(grpprl, op));
+		else if (sprm === SPRM_S_DXA_RIGHT) margin("rightTwips", u16(grpprl, op));
+		else if (sprm === SPRM_S_DYA_TOP) margin("topTwips", Math.abs(i16(grpprl, op)));
+		else if (sprm === SPRM_S_DYA_BOTTOM) margin("bottomTwips", Math.abs(i16(grpprl, op)));
+		else if (sprm === SPRM_S_DYA_HDR_TOP) margin("headerTwips", u16(grpprl, op));
+		else if (sprm === SPRM_S_DYA_HDR_BOTTOM) margin("footerTwips", u16(grpprl, op));
+		else if (sprm === SPRM_S_DZA_GUTTER) margin("gutterTwips", u16(grpprl, op));
+		const sane = xa >= 1 && ya >= 1 && xa <= MAX_PAGE_TWIPS && ya <= MAX_PAGE_TWIPS;
+		sections.push({
+			...bkc !== void 0 ? { bkc } : {},
+			...sane ? { pageSize: {
+				widthPt: xa / 20,
+				heightPt: ya / 20
+			} } : {},
+			...titlePage ? { titlePage } : {},
+			...Object.keys(margins).length > 0 ? { margins } : {},
+			...restart ? { pageNumberStart: pgnStart } : {},
+			...nfcPgn ? { pageNumberNfc: nfcPgn } : {}
+		});
+	}
 	return {
-		widthPt: xa / 20,
-		heightPt: ya / 20
+		sections,
+		ends
 	};
 }
 /**
 * Extract a `.doc`'s `WordDocument` stream into its main-document paragraphs plus
-* the section's header/footer stories, list tables and first-section page size.
+* the section's header/footer stories, list tables and sections.
 * Joins the piece table (CLX → CPs/FCs), CHPX run formatting and PAPX paragraph
 * formatting, splitting at the CR / cell marks. Every offset is bounds-checked, so
 * a structurally odd file yields no text rather than throwing.
@@ -121,22 +198,160 @@ function extractDocContent(bytes) {
 	const plcPcd = findPlcPcd(table.subarray(fcClx, fcClx + lcbClx));
 	if (!plcPcd) return EMPTY;
 	const pieces = parsePieces(plcPcd);
-	const chpx = parsePlcBte(wd, table, OFF_FCPLCFBTECHPX, OFF_LCBPLCFBTECHPX, parseChpxFkp);
+	const fonts = readFonts(wd, table);
+	const chpx = parsePlcBte(wd, table, OFF_FCPLCFBTECHPX, OFF_LCBPLCFBTECHPX, (page, out) => parseChpxFkp(page, out, fonts));
 	const papx = parsePlcBte(wd, table, OFF_FCPLCFBTEPAPX, OFF_LCBPLCFBTEPAPX, parsePapxFkp);
 	const ccpText = u32(wd, OFF_CCPTEXT);
 	const data = cfb.readStream("Data");
-	const main = buildParagraphs(wd, pieces, chpx, papx, 0, ccpText, data);
-	const headerFooters = extractHeaderFooters(wd, table, pieces, chpx, papx, data, ccpText);
+	const { sections, ends } = readSections(wd, table);
+	const styles = parseStyles(wd, table, fonts);
+	const main = buildParagraphs(wd, pieces, chpx, papx, 0, ccpText, data, ends, styles, fonts);
+	const headerFooters = extractHeaderFooters(wd, table, pieces, chpx, papx, data, ccpText, Math.max(1, sections.length), styles, fonts);
 	const listTables = parseListTables(wd, table);
-	const pageSize = readSectionPageSize(wd, table);
+	const fcDop = u32(wd, OFF_FCDOP);
+	const facingPages = u32(wd, OFF_LCBDOP) > 0 && fcDop < table.length && (table[fcDop] & 1) !== 0;
 	return {
 		paragraphs: main,
 		...headerFooters ? { headerFooters } : {},
+		...facingPages ? { facingPages } : {},
 		...listTables ? { listTables } : {},
-		...pageSize ? { pageSize } : {},
+		...sections.length > 0 ? { sections } : {},
 		encrypted: false
 	};
 }
+function readFonts(wd, table) {
+	const fonts = [];
+	const fc = u32(wd, OFF_FCSTTBFFFN);
+	const lcb = u32(wd, OFF_LCBSTTBFFFN);
+	if (lcb < 4 || fc + lcb > table.length) return fonts;
+	const sttb = table.subarray(fc, fc + lcb);
+	const count = u16(sttb, 0);
+	let p = 4;
+	for (let i = 0; i < count && p < sttb.length; i++) {
+		const end = Math.min(sttb.length, p + 1 + sttb[p]);
+		let name = "";
+		for (let q = p + 40; q + 1 < end; q += 2) {
+			const c = u16(sttb, q);
+			if (c === 0) break;
+			name += String.fromCharCode(c);
+		}
+		fonts.push(name);
+		p = end;
+	}
+	return fonts;
+}
+function parseStyles(wd, table, fonts) {
+	const fc = u32(wd, OFF_FCSTSHF);
+	const lcb = u32(wd, OFF_LCBSTSHF);
+	if (lcb < 6 || fc + lcb > table.length) return void 0;
+	const stsh = table.subarray(fc, fc + lcb);
+	const cbStshi = u16(stsh, 0);
+	if (cbStshi < 4 || 2 + cbStshi > stsh.length) return void 0;
+	const cstd = u16(stsh, 2);
+	const cbStdBase = u16(stsh, 4);
+	const font = (key, off) => {
+		const name = cbStshi >= off ? fonts[u16(stsh, off)] : void 0;
+		return name ? { [key]: name } : {};
+	};
+	const defaults = {
+		...font("fontAscii", 14),
+		...font("fontEastAsia", 16),
+		...font("fontOther", 18),
+		...font("fontBi", 20)
+	};
+	const stds = [];
+	let p = 2 + cbStshi;
+	for (let istd = 0; istd < cstd && p + 2 <= stsh.length; istd++) {
+		const cbStd = u16(stsh, p);
+		p += 2;
+		stds.push(cbStd > 0 && p + cbStd <= stsh.length ? parseStd(stsh.subarray(p, p + cbStd), cbStdBase) : void 0);
+		p += cbStd;
+	}
+	const none = {
+		para: {},
+		chars: defaults
+	};
+	const memo = /* @__PURE__ */ new Map();
+	const paragraph = (istd, depth = 0) => {
+		const hit = memo.get(istd);
+		if (hit) return hit;
+		const std = stds[istd];
+		let out;
+		if (depth >= MAX_STYLE_DEPTH) out = none;
+		else if (std?.stk !== STK_PARAGRAPH) out = istd !== 0 ? paragraph(0, depth + 1) : none;
+		else {
+			const base = std.base !== ISTD_NIL && std.base !== istd ? paragraph(std.base, depth + 1) : none;
+			out = {
+				para: std.papx ? {
+					...base.para,
+					...decodePapxGrpprl(std.papx)
+				} : base.para,
+				chars: std.chpx ? {
+					...base.chars,
+					...decodeChpxGrpprl(std.chpx, base.chars, fonts)
+				} : base.chars
+			};
+		}
+		memo.set(istd, out);
+		return out;
+	};
+	const character = (istd, under, depth = 0) => {
+		const std = stds[istd];
+		if (std?.stk !== STK_CHARACTER || depth >= MAX_STYLE_DEPTH) return under;
+		const base = std.base !== ISTD_NIL && std.base !== istd ? character(std.base, under, depth + 1) : under;
+		return std.chpx ? {
+			...base,
+			...decodeChpxGrpprl(std.chpx, base, fonts)
+		} : base;
+	};
+	return {
+		paragraph: (istd) => paragraph(istd),
+		character: (istd, base) => character(istd, base)
+	};
+}
+function parseStd(std, cbStdBase) {
+	if (std.length < 6) return void 0;
+	const stk = u16(std, 2) & 15;
+	const base = u16(std, 2) >> 4;
+	const cupx = u16(std, 4) & 15;
+	let q = cbStdBase;
+	if (q + 2 > std.length) return {
+		stk,
+		base
+	};
+	q += 2 + u16(std, q) * 2 + 2;
+	const upx = [];
+	for (let k = 0; k < cupx && q + 2 <= std.length; k++) {
+		const cb = u16(std, q);
+		q += 2;
+		if (q + cb > std.length) break;
+		upx.push(std.subarray(q, q + cb));
+		q += cb + (cb & 1);
+	}
+	if (stk === STK_PARAGRAPH) {
+		const papx = upx[0] && upx[0].length >= 2 ? upx[0].subarray(2) : void 0;
+		return {
+			stk,
+			base,
+			...papx ? { papx } : {},
+			...upx[1] ? { chpx: upx[1] } : {}
+		};
+	}
+	if (stk === STK_CHARACTER) return {
+		stk,
+		base,
+		...upx[0] ? { chpx: upx[0] } : {}
+	};
+	return {
+		stk,
+		base
+	};
+}
+function charStyleOf(grpprl) {
+	let istd;
+	for (const s of sprms(grpprl)) if (s.sprm === SPRM_C_ISTD) istd = u16(grpprl, s.op);
+	return istd;
+}
 function parseListTables(wd, table) {
 	const lstFc = u32(wd, OFF_FCPLFLST);
 	const lstLcb = u32(wd, OFF_LCBPLFLST);
@@ -184,6 +399,7 @@ function readLvl(d, off) {
 	if (off + 28 > d.length) return void 0;
 	const iStartAt = u32(d, off) | 0;
 	const nfc = d[off + 4];
+	const follow = d[off + 15];
 	const cbChpx = d[off + 24];
 	const cbPapx = d[off + 25];
 	let p = off + 28 + cbPapx + cbChpx;
@@ -199,7 +415,8 @@ function readLvl(d, off) {
 		info: {
 			nfc,
 			iStartAt,
-			xst
+			xst,
+			follow
 		},
 		next: p
 	};
@@ -259,7 +476,7 @@ function colorHex(r, g, b) {
 		b
 	].map((c) => c.toString(16).padStart(2, "0")).join("").toUpperCase();
 }
-function extractHeaderFooters(wd, table, pieces, chpx, papx, data, ccpText) {
+function extractHeaderFooters(wd, table, pieces, chpx, papx, data, ccpText, sectionCount, styles, fonts) {
 	if (u32(wd, OFF_CCPHDD) === 0) return void 0;
 	const fc = u32(wd, OFF_FCPLCFHDD);
 	const lcb = u32(wd, OFF_LCBPLCFHDD);
@@ -268,28 +485,36 @@ function extractHeaderFooters(wd, table, pieces, chpx, papx, data, ccpText) {
 	const n = Math.floor(lcb / 4);
 	const hddStart = ccpText + u32(wd, OFF_CCPFTN);
 	const story = (i) => {
-		if (i + 1 >= n) return void 0;
+		if (i + 1 >= n) return "inherit";
 		const a = hddStart + u32(plc, i * 4);
 		const b = hddStart + u32(plc, (i + 1) * 4);
-		if (b <= a) return void 0;
-		const paras = buildParagraphs(wd, pieces, chpx, papx, a, b, data);
+		if (b <= a) return "inherit";
+		const paras = buildParagraphs(wd, pieces, chpx, papx, a, b, data, void 0, styles, fonts);
+		const last = paras[paras.length - 1];
+		if (last && last.runs.length === 0) paras.pop();
 		return paras.some((p) => p.runs.some((r) => r.text.length > 0 || r.picture)) ? paras : void 0;
 	};
-	const evenHeader = story(6);
-	const defaultHeader = story(7);
-	const evenFooter = story(8);
-	const defaultFooter = story(9);
-	const firstHeader = story(10);
-	const firstFooter = story(11);
-	const hf = {
-		...defaultHeader ? { defaultHeader } : {},
-		...defaultFooter ? { defaultFooter } : {},
-		...firstHeader ? { firstHeader } : {},
-		...firstFooter ? { firstFooter } : {},
-		...evenHeader ? { evenHeader } : {},
-		...evenFooter ? { evenFooter } : {}
-	};
-	return Object.keys(hf).length > 0 ? hf : void 0;
+	const SLOTS = [
+		"evenHeader",
+		"defaultHeader",
+		"evenFooter",
+		"defaultFooter",
+		"firstHeader",
+		"firstFooter"
+	];
+	const out = [];
+	let prev = {};
+	for (let sec = 0; sec < sectionCount; sec++) {
+		const hf = {};
+		SLOTS.forEach((slot, k) => {
+			const got = story(6 + sec * 6 + k);
+			const paras = got === "inherit" ? prev[slot] : got;
+			if (paras) hf[slot] = paras;
+		});
+		out.push(hf);
+		prev = hf;
+	}
+	return out.some((hf) => Object.keys(hf).length > 0) ? out : void 0;
 }
 function findPlcPcd(clx) {
 	let p = 0;
@@ -342,24 +567,32 @@ function parsePlcBte(wd, table, fcOff, lcbOff, parsePage) {
 	out.sort((a, b) => a.fcStart - b.fcStart);
 	return out;
 }
-function parseChpxFkp(page, out) {
+function parseChpxFkp(page, out, fonts) {
 	const crun = page[511];
 	if (crun === 0 || (crun + 1) * 4 + crun > 511) return;
 	for (let i = 0; i < crun; i++) {
 		const fcStart = u32(page, i * 4);
 		const fcEnd = u32(page, (i + 1) * 4);
 		if (fcEnd <= fcStart) continue;
-		let props = {};
 		const wordOff = page[(crun + 1) * 4 + i];
 		if (wordOff !== 0) {
 			const chpxOff = wordOff * 2;
 			const cb = chpxOff < FKP_SIZE ? page[chpxOff] : 0;
-			if (cb > 0 && chpxOff + 1 + cb <= FKP_SIZE) props = decodeChpxGrpprl(page.subarray(chpxOff + 1, chpxOff + 1 + cb));
+			if (cb > 0 && chpxOff + 1 + cb <= FKP_SIZE) {
+				const grpprl = page.subarray(chpxOff + 1, chpxOff + 1 + cb);
+				out.push({
+					fcStart,
+					fcEnd,
+					props: decodeChpxGrpprl(grpprl, {}, fonts),
+					grpprl
+				});
+				continue;
+			}
 		}
 		out.push({
 			fcStart,
 			fcEnd,
-			props
+			props: {}
 		});
 	}
 }
@@ -371,34 +604,49 @@ function parsePapxFkp(page, out) {
 		const fcStart = u32(page, i * 4);
 		const fcEnd = u32(page, (i + 1) * 4);
 		if (fcEnd <= fcStart) continue;
-		let props = {};
 		const papxOff = page[bxBase + i * 13] * 2;
 		if (papxOff !== 0 && papxOff < FKP_SIZE) {
 			const cb = page[papxOff];
 			const size = cb !== 0 ? 2 * cb - 1 : 2 * (page[papxOff + 1] ?? 0);
 			const start = cb !== 0 ? papxOff + 1 : papxOff + 2;
-			if (size >= 2 && start + size <= FKP_SIZE) props = decodePapxGrpprl(page.subarray(start + 2, start + size));
+			if (size >= 2 && start + size <= FKP_SIZE) {
+				const grpprl = page.subarray(start + 2, start + size);
+				const istd = u16(page, start);
+				out.push({
+					fcStart,
+					fcEnd,
+					props: decodePapxGrpprl(grpprl),
+					grpprl,
+					istd
+				});
+				continue;
+			}
 		}
 		out.push({
 			fcStart,
 			fcEnd,
-			props
+			props: {}
 		});
 	}
 }
-function decodeChpxGrpprl(d) {
+function decodeChpxGrpprl(d, style = {}, fonts = []) {
 	let bold;
 	let italic;
 	let underlineKul;
 	let sizeHalfPts;
 	let fSpec;
 	let picOffset;
+	const named = {};
+	const font = (key, ftc) => {
+		const name = fonts[ftc];
+		if (name) named[key] = name;
+	};
 	for (const s of sprms(d)) switch (s.sprm) {
 		case SPRM_C_FBOLD:
-			bold = toggleBool(d[s.op]);
+			bold = toggleBool(d[s.op], style.bold);
 			break;
 		case SPRM_C_FITALIC:
-			italic = toggleBool(d[s.op]);
+			italic = toggleBool(d[s.op], style.italic);
 			break;
 		case SPRM_C_KUL:
 			underlineKul = d[s.op];
@@ -406,6 +654,18 @@ function decodeChpxGrpprl(d) {
 		case SPRM_C_HPS:
 			sizeHalfPts = u16(d, s.op);
 			break;
+		case SPRM_C_RG_FTC0:
+			font("fontAscii", u16(d, s.op));
+			break;
+		case SPRM_C_RG_FTC1:
+			font("fontEastAsia", u16(d, s.op));
+			break;
+		case SPRM_C_RG_FTC2:
+			font("fontOther", u16(d, s.op));
+			break;
+		case SPRM_C_FTC_BI:
+			font("fontBi", u16(d, s.op));
+			break;
 		case SPRM_C_FSPEC:
 			fSpec = d[s.op] !== 0;
 			break;
@@ -416,6 +676,7 @@ function decodeChpxGrpprl(d) {
 		...italic !== void 0 ? { italic } : {},
 		...underlineKul !== void 0 ? { underlineKul } : {},
 		...sizeHalfPts !== void 0 ? { sizeHalfPts } : {},
+		...named,
 		...fSpec ? { fSpec } : {},
 		...picOffset !== void 0 ? { picOffset } : {}
 	};
@@ -437,6 +698,7 @@ function decodePapxGrpprl(d) {
 	let widowControl;
 	let keepNext;
 	let keepLines;
+	let pageBreakBefore;
 	for (const s of sprms(d)) switch (s.sprm) {
 		case SPRM_P_JC:
 			jc = d[s.op];
@@ -471,6 +733,9 @@ function decodePapxGrpprl(d) {
 		case SPRM_P_F_KEEP:
 			keepLines = d[s.op] !== 0;
 			break;
+		case SPRM_P_F_PAGE_BREAK_BEFORE:
+			pageBreakBefore = d[s.op] !== 0;
+			break;
 		case SPRM_T_DEF_TABLE: {
 			const cols = d[s.op] ?? 0;
 			const edges = [];
@@ -483,7 +748,7 @@ function decodePapxGrpprl(d) {
 			break;
 		}
 		case SPRM_P_ILVL:
-			listIlvl = d[s.op];
+			listIlvl = d[s.op] <= 8 ? d[s.op] : 0;
 			break;
 		case SPRM_T_DEF_TABLE_SHD: {
 			const shd = [];
@@ -512,7 +777,8 @@ function decodePapxGrpprl(d) {
 		...listIlvl !== void 0 ? { listIlvl } : {},
 		...widowControl !== void 0 ? { widowControl } : {},
 		...keepNext !== void 0 ? { keepNext } : {},
-		...keepLines !== void 0 ? { keepLines } : {}
+		...keepLines !== void 0 ? { keepLines } : {},
+		...pageBreakBefore !== void 0 ? { pageBreakBefore } : {}
 	};
 }
 function* sprms(d) {
@@ -544,9 +810,11 @@ function* sprms(d) {
 		p = op + len;
 	}
 }
-function toggleBool(v) {
+function toggleBool(v, style) {
 	if (v === 0) return false;
-	if (v === 1 || v === 129) return true;
+	if (v === 1) return true;
+	if (v === 128) return style;
+	if (v === 129) return style !== true;
 }
 function extractPicture(data, fcPic) {
 	if (fcPic < 0 || fcPic + 36 > data.length) return void 0;
@@ -605,32 +873,71 @@ function magicAt(d, off, magic) {
 	for (let i = 0; i < magic.length; i++) if (d[off + i] !== magic[i]) return false;
 	return true;
 }
-function buildParagraphs(wd, pieces, chpx, papx, cpStart, cpEnd, data) {
+function buildParagraphs(wd, pieces, chpx, papx, cpStart, cpEnd, data, sectionEnds = /* @__PURE__ */ new Map(), styles, fonts = []) {
 	const end = Math.min(cpEnd, cpStart + MAX_TEXT);
 	const paragraphs = [];
 	let runs = [];
+	let chpxOf = [];
 	let cur = "";
-	let curProps = {};
+	let curChpx;
 	let lastFc = 0;
 	let fieldDepth = 0;
 	let codeDepth = 0;
+	let instr = "";
+	let page;
+	const push = (run, from) => {
+		runs.push(run);
+		chpxOf.push(from);
+	};
 	const flushRun = () => {
 		if (cur.length > 0) {
-			runs.push({
+			push({
 				text: cur,
-				props: curProps
-			});
+				props: curChpx?.props ?? {}
+			}, curChpx);
 			cur = "";
 		}
 	};
-	const endParagraph = (fc, cellMark) => {
+	const runProps = (from, base) => {
+		if (!from?.grpprl) return base;
+		const cistd = charStyleOf(from.grpprl);
+		const under = cistd !== void 0 && styles ? styles.character(cistd, base) : base;
+		return {
+			...under,
+			...decodeChpxGrpprl(from.grpprl, under, fonts)
+		};
+	};
+	const endParagraph = (fc, cellMark, sectionEnd) => {
 		flushRun();
+		const pap = lookupRun(papx, fc);
+		const style = styles?.paragraph(pap?.istd ?? 0);
+		const settled = [];
+		runs.forEach((run, i) => {
+			const props = style && !run.picture ? runProps(chpxOf[i], style.chars) : run.props;
+			const prev = settled[settled.length - 1];
+			if (prev && plainText(prev) && plainText(run) && sameProps(prev.props, props)) settled[settled.length - 1] = {
+				text: prev.text + run.text,
+				props: prev.props
+			};
+			else settled.push({
+				...run,
+				props
+			});
+		});
+		const markChpx = lookupRun(chpx, fc);
+		const mark = style ? runProps(markChpx, style.chars) : markChpx?.props;
 		paragraphs.push({
-			runs,
-			props: lookup(papx, fc) ?? {},
-			...cellMark ? { cellMark: true } : {}
+			runs: settled,
+			props: style ? {
+				...style.para,
+				...pap?.props ?? {}
+			} : pap?.props ?? {},
+			...cellMark ? { cellMark: true } : {},
+			...sectionEnd !== void 0 ? { sectionEnd } : {},
+			...mark && Object.keys(mark).length > 0 ? { mark } : {}
 		});
 		runs = [];
+		chpxOf = [];
 	};
 	for (const piece of pieces) {
 		if (piece.cpStart >= end) break;
@@ -643,45 +950,91 @@ function buildParagraphs(wd, pieces, chpx, papx, cpStart, cpEnd, data) {
 			lastFc = fc;
 			const code = piece.compressed ? byteCode(wd, fc) : unitCode(wd, fc);
 			if (code < 0) continue;
-			if (code === CR || code === CELL_MARK) {
-				endParagraph(fc, code === CELL_MARK);
+			const sectionEnd = code === PAGE_BREAK ? sectionEnds.get(piece.cpStart + k + 1) : void 0;
+			if (code === CR || code === CELL_MARK || sectionEnd !== void 0) {
+				endParagraph(fc, code === CELL_MARK, sectionEnd);
 				continue;
 			}
 			if (code === FIELD_BEGIN) {
+				if (codeDepth === 0) instr = "";
 				fieldDepth++;
 				codeDepth++;
 				continue;
 			}
 			if (code === FIELD_SEPARATOR) {
-				if (codeDepth > 0) codeDepth--;
+				const ended = codeDepth > 0;
+				if (ended) codeDepth--;
+				const field = ended && codeDepth === 0 && !page ? pageField(instr) : void 0;
+				if (field) {
+					flushRun();
+					const sep = lookupRun(chpx, fc);
+					page = {
+						field,
+						depth: fieldDepth,
+						paragraph: paragraphs.length,
+						at: runs.length,
+						...sep ? { sep } : {}
+					};
+				}
 				continue;
 			}
 			if (code === FIELD_END) {
+				if (page?.depth === fieldDepth) {
+					if (page.paragraph === paragraphs.length) {
+						flushRun();
+						const result = runs.splice(page.at);
+						const from = chpxOf.splice(page.at)[0] ?? page.sep;
+						push({
+							text: result.map((r) => r.text).join(""),
+							props: from?.props ?? {},
+							field: page.field
+						}, from);
+					}
+					page = void 0;
+				}
 				if (fieldDepth > 0) fieldDepth--;
 				if (codeDepth > fieldDepth) codeDepth = fieldDepth;
 				continue;
 			}
-			if (codeDepth > 0) continue;
+			if (codeDepth > 0) {
+				if (codeDepth === 1 && instr.length < MAX_FIELD_INSTR) instr += String.fromCharCode(code);
+				continue;
+			}
 			if (code === PIC_CHAR) {
 				const props = lookup(chpx, fc) ?? {};
 				const picture = data && props.fSpec && props.picOffset !== void 0 ? extractPicture(data, props.picOffset) : void 0;
 				if (picture) {
 					flushRun();
-					runs.push({
+					push({
 						text: "",
 						props: {},
 						picture
-					});
+					}, void 0);
 				}
 				continue;
 			}
-			const ch = code === 9 || code === 11 ? " " : code < 32 ? "" : String.fromCharCode(code);
+			if (code === PAGE_BREAK || code === COLUMN_BREAK) {
+				flushRun();
+				const from = lookupRun(chpx, fc);
+				const props = from?.props ?? {};
+				push(code === PAGE_BREAK ? {
+					text: "",
+					props,
+					pageBreak: true
+				} : {
+					text: "\n",
+					props,
+					columnBreak: true
+				}, from);
+				continue;
+			}
+			const ch = code === LINE_BREAK ? "\n" : code === 9 ? " " : code < 32 ? "" : String.fromCharCode(code);
 			if (ch.length === 0) continue;
-			const props = lookup(chpx, fc) ?? {};
-			if (cur.length === 0) curProps = props;
-			else if (!sameProps(props, curProps)) {
+			const from = lookupRun(chpx, fc);
+			if (cur.length === 0) curChpx = from;
+			else if (from !== curChpx) {
 				flushRun();
-				curProps = props;
+				curChpx = from;
 			}
 			cur += ch;
 		}
@@ -690,6 +1043,9 @@ function buildParagraphs(wd, pieces, chpx, papx, cpStart, cpEnd, data) {
 	return paragraphs;
 }
 function lookup(runs, fc) {
+	return lookupRun(runs, fc)?.props;
+}
+function lookupRun(runs, fc) {
 	let lo = 0;
 	let hi = runs.length - 1;
 	while (lo <= hi) {
@@ -697,11 +1053,18 @@ function lookup(runs, fc) {
 		const r = runs[mid];
 		if (fc < r.fcStart) hi = mid - 1;
 		else if (fc >= r.fcEnd) lo = mid + 1;
-		else return r.props;
+		else return r;
 	}
 }
+function plainText(run) {
+	return !run.picture && !run.pageBreak && !run.columnBreak && !run.field;
+}
+function pageField(code) {
+	const keyword = /^\s*([A-Za-z]+)/.exec(code)?.[1]?.toUpperCase();
+	return keyword === "PAGE" || keyword === "NUMPAGES" ? keyword : void 0;
+}
 function sameProps(a, b) {
-	return a.bold === b.bold && a.italic === b.italic && a.underlineKul === b.underlineKul && a.sizeHalfPts === b.sizeHalfPts;
+	return a.bold === b.bold && a.italic === b.italic && a.underlineKul === b.underlineKul && a.sizeHalfPts === b.sizeHalfPts && a.fontAscii === b.fontAscii && a.fontEastAsia === b.fontEastAsia && a.fontOther === b.fontOther && a.fontBi === b.fontBi;
 }
 var CP1252_HIGH = [
 	8364,