reamkit 1.15.4 → 1.17.0

package/dist/esm/word/shading.d.ts

@@ -0,0 +1,11 @@
+/**
+ * The colour a `w:shd` actually paints: its `w:fill` background with the
+ * pattern colour blended in at the pattern's density. `clear` (and an absent
+ * `w:val`) is the fill alone; `nil` paints nothing.
+ *
+ * @param val   `w:val` — the pattern.
+ * @param color `w:color` — the pattern's own colour (`auto` ⇒ black).
+ * @param fill  `w:fill` — the background (`auto` ⇒ white).
+ * @returns The 6-hex colour to paint, or `undefined` when nothing is painted.
+ */
+export declare function shadingFillHex(val: string | undefined, color: string | undefined, fill: string | undefined): string | undefined;

package/dist/esm/word/shading.js

@@ -0,0 +1,64 @@
+//#region src/word/shading.ts
+/** §17.18.78 ST_Shd — the pattern densities, as the share of `color` over `fill`. */
+var PATTERN_DENSITY = new Map([
+	["solid", 1],
+	["pct5", .05],
+	["pct10", .1],
+	["pct12", .125],
+	["pct15", .15],
+	["pct20", .2],
+	["pct25", .25],
+	["pct30", .3],
+	["pct35", .35],
+	["pct37", .375],
+	["pct40", .4],
+	["pct45", .45],
+	["pct50", .5],
+	["pct55", .55],
+	["pct60", .6],
+	["pct62", .625],
+	["pct65", .65],
+	["pct70", .7],
+	["pct75", .75],
+	["pct80", .8],
+	["pct85", .85],
+	["pct87", .875],
+	["pct90", .9],
+	["pct95", .95],
+	["thinHorzStripe", .25],
+	["thinVertStripe", .25],
+	["thinDiagStripe", .25],
+	["thinHorzCross", .3],
+	["thinDiagCross", .3],
+	["horzStripe", .4],
+	["vertStripe", .4],
+	["diagStripe", .4],
+	["horzCross", .5],
+	["diagCross", .5]
+]);
+var HEX6 = /^[0-9A-Fa-f]{6}$/u;
+/**
+* The colour a `w:shd` actually paints: its `w:fill` background with the
+* pattern colour blended in at the pattern's density. `clear` (and an absent
+* `w:val`) is the fill alone; `nil` paints nothing.
+*
+* @param val   `w:val` — the pattern.
+* @param color `w:color` — the pattern's own colour (`auto` ⇒ black).
+* @param fill  `w:fill` — the background (`auto` ⇒ white).
+* @returns The 6-hex colour to paint, or `undefined` when nothing is painted.
+*/
+function shadingFillHex(val, color, fill) {
+	if (val === "nil") return void 0;
+	const bg = fill !== void 0 && HEX6.test(fill) ? fill.toUpperCase() : void 0;
+	const density = val !== void 0 ? PATTERN_DENSITY.get(val) : void 0;
+	if (density === void 0) return bg;
+	return blend(color !== void 0 && HEX6.test(color) ? color.toUpperCase() : "000000", bg ?? "FFFFFF", density);
+}
+/** `a` over `b` at `t` (0..1), per channel. */
+function blend(a, b, t) {
+	const ch = (hex, i) => parseInt(hex.slice(i * 2, i * 2 + 2), 16);
+	const mix = (i) => Math.round(ch(a, i) * t + ch(b, i) * (1 - t)).toString(16).padStart(2, "0");
+	return (mix(0) + mix(1) + mix(2)).toUpperCase();
+}
+//#endregion
+export { shadingFillHex };

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

@@ -1,13 +1,14 @@
 import { eighthPtToPt, twipsToPt } from "../core/ir/units.js";
 import { EMPTY_STYLE_SHEET } from "../core/style-cascade/resolver.js";
 import "../core/style-cascade/index.js";
-import { asArray, asElement, getAttr, getVal } from "./xml-helpers.js";
+import { asArray, asElement, getAttr, getVal, parseIntAttr } from "./xml-helpers.js";
 import { parseRunProperties } from "./run-properties.js";
 import { parseParagraphProperties } from "./paragraph-properties.js";
 import { XMLParser } from "fast-xml-parser";
 //#region src/word/styles-parser.ts
 var decoder = new TextDecoder("utf-8");
 var parser = new XMLParser({
+	htmlEntities: true,
 	ignoreAttributes: false,
 	attributeNamePrefix: "@_",
 	textNodeName: "#text",
@@ -118,12 +119,17 @@ function parseTableStyleLayer(el) {
 	const tcPr = asElement(el["w:tcPr"]);
 	const borders = parseFlatBorders(tblPr ? asElement(tblPr["w:tblBorders"]) : void 0) ?? parseFlatBorders(tcPr ? asElement(tcPr["w:tcBorders"]) : void 0);
 	const cellMargins = parseFlatCellMargins(tblPr ? asElement(tblPr["w:tblCellMar"]) : void 0);
+	const tblInd = tblPr ? asElement(tblPr["w:tblInd"]) : void 0;
+	const indW = tblInd ? parseIntAttr(tblInd, "w") : void 0;
+	const indType = tblInd ? getAttr(tblInd, "type") : void 0;
+	const indentPt = indW !== void 0 && (indType === void 0 || indType === "dxa") ? twipsToPt(indW) : void 0;
 	const shading = parseFlatShading(tcPr ? tcPr["w:shd"] : void 0);
 	const runProperties = parseRunProperties(el["w:rPr"]);
 	const paragraphProperties = parseParagraphProperties(el["w:pPr"]);
 	const layer = {
 		...borders ? { borders } : {},
 		...cellMargins ? { cellMargins } : {},
+		...indentPt !== void 0 ? { indentPt } : {},
 		...shading ? { shading } : {},
 		...Object.keys(runProperties).length > 0 ? { runProperties } : {},
 		...Object.keys(paragraphProperties).length > 0 ? { paragraphProperties } : {}
@@ -136,7 +142,8 @@ var FLAT_BORDER_STYLES = new Set([
 	"double",
 	"thick",
 	"dotted",
-	"dashed"
+	"dashed",
+	"dashSmallGap"
 ]);
 function parseFlatBorder(node) {
 	const el = asElement(node);

package/dist/esm/word/table-parser.d.ts

@@ -1,4 +1,4 @@
-import { Table } from '../core/document-model/index.js';
+import { CellBorders, Table } from '../core/document-model/index.js';
 import { PoNode } from '../core/po-helpers.js';
 import { ParseContext } from './document-parser.js';
 /**
@@ -13,3 +13,11 @@ import { ParseContext } from './document-parser.js';
  * @returns The parsed table.
  */
 export declare function parseTable(tbl: PoNode, ctx?: ParseContext): Table;
+/**
+ * A `w:tcBorders` / `w:tblBorders` / §17.6.10 `w:pgBorders` element: the rule on
+ * each edge. Every one of them is spelled the same way.
+ *
+ * @param node The borders element, or `undefined`.
+ * @returns The edges that name a rule, or `undefined` when none do.
+ */
+export declare function parseBorders(node: PoNode | undefined): CellBorders | undefined;

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

@@ -1,6 +1,6 @@
-import { eighthPtToPt, twipsToPt } from "../core/ir/units.js";
-import { poAttr, poChildren, poChildrenWith, poFirstChild, poIntAttr, poToggle, poVal } from "../core/po-helpers.js";
-import { DEFAULT_PARSE_CONTEXT, parseBodyElements } from "./document-parser.js";
+import { eighthPtToPt, pt, twipsToPt } from "../core/ir/units.js";
+import { poAttr, poChildren, poChildrenWith, poFirstChild, poIntAttr, poIs, poToggle, poVal } from "../core/po-helpers.js";
+import { DEFAULT_PARSE_CONTEXT, parseBodyElements, sdtRunProperties } from "./document-parser.js";
 //#region src/word/table-parser.ts
 var BORDER_STYLES = new Set([
 	"none",
@@ -8,7 +8,23 @@ var BORDER_STYLES = new Set([
 	"double",
 	"thick",
 	"dotted",
-	"dashed"
+	"dashed",
+	"dashSmallGap"
+]);
+var MULTI_LINE_BORDERS = new Set([
+	"triple",
+	"doubleWave",
+	"thinThickSmallGap",
+	"thickThinSmallGap",
+	"thinThickThinSmallGap",
+	"thinThickMediumGap",
+	"thickThinMediumGap",
+	"thinThickThinMediumGap",
+	"thinThickLargeGap",
+	"thickThinLargeGap",
+	"thinThickThinLargeGap",
+	"outset",
+	"inset"
 ]);
 var WIDTH_TYPES = new Set([
 	"auto",
@@ -21,6 +37,34 @@ var HEIGHT_RULES = new Set([
 	"atLeast",
 	"exact"
 ]);
+function parseTablePosition(tblpPr) {
+	const horz = poAttr(tblpPr, "horzAnchor");
+	const vert = poAttr(tblpPr, "vertAnchor");
+	const hRel = horz === "page" ? "page" : horz === "margin" ? "margin" : "column";
+	const vRel = vert === "page" ? "page" : vert === "margin" ? "margin" : "paragraph";
+	const xSpec = poAttr(tblpPr, "tblpXSpec");
+	const align = xSpec === "center" ? "center" : xSpec === "right" ? "right" : xSpec === "left" ? "left" : void 0;
+	const x = poIntAttr(tblpPr, "tblpX");
+	const y = poIntAttr(tblpPr, "tblpY");
+	const dist = (name) => twipsToPt(poIntAttr(tblpPr, name) ?? 0);
+	return {
+		wrap: "square",
+		posH: {
+			relativeFrom: hRel,
+			...align ? { align } : { offsetPt: twipsToPt(x ?? 0) }
+		},
+		posV: {
+			relativeFrom: vRel,
+			offsetPt: twipsToPt(y ?? 0)
+		},
+		wrapDist: {
+			topPt: dist("topFromText"),
+			bottomPt: dist("bottomFromText"),
+			leftPt: dist("leftFromText"),
+			rightPt: dist("rightFromText")
+		}
+	};
+}
 /**
 * Parse a `w:tbl` (ECMA-376 Part 1 §17.4) in preserveOrder shape into the typed
 * {@link Table} model: its properties (style ref, look, width, layout, borders,
@@ -36,26 +80,53 @@ function parseTable(tbl, ctx = DEFAULT_PARSE_CONTEXT) {
 	const properties = parseTableProperties(poFirstChild(tbl, "w:tblPr"));
 	const grid = parseTableGrid(poFirstChild(tbl, "w:tblGrid"));
 	const draftRows = [];
-	for (const tr of poChildrenWith(tbl, "w:tr")) draftRows.push(parseTableRow(tr, ctx));
+	for (const tr of poChildrenThroughSdt(tbl, "w:tr")) {
+		if (isDeletedRow(tr)) continue;
+		draftRows.push(parseTableRow(tr, ctx));
+	}
 	const roles = resolveMergeRoles(draftRows.map((r) => r.cells));
+	const rows = draftRows.map((draft, rowIdx) => ({
+		properties: draft.properties,
+		cells: draft.cells.map((d, cellIdx) => {
+			const merge = roles[rowIdx][cellIdx];
+			return merge ? {
+				...d.cell,
+				properties: {
+					...d.cell.properties,
+					merge
+				}
+			} : d.cell;
+		})
+	}));
+	if (poFirstChild(poFirstChild(tbl, "w:tblPr"), "w:bidiVisual")) return {
+		properties,
+		grid: [...grid].reverse(),
+		rows: rows.map((row) => ({
+			...row,
+			cells: [...row.cells].reverse()
+		}))
+	};
 	return {
 		properties,
 		grid,
-		rows: draftRows.map((draft, rowIdx) => ({
-			properties: draft.properties,
-			cells: draft.cells.map((d, cellIdx) => {
-				const merge = roles[rowIdx][cellIdx];
-				return merge ? {
-					...d.cell,
-					properties: {
-						...d.cell.properties,
-						merge
-					}
-				} : d.cell;
-			})
-		}))
+		rows
 	};
 }
+function poChildrenThroughSdt(node, tag) {
+	return childrenThroughSdt(node, tag).map((c) => c.node);
+}
+function childrenThroughSdt(node, tag) {
+	const out = [];
+	for (const child of poChildren(node)) if (poIs(child, tag)) out.push({ node: child });
+	else if (poIs(child, "w:sdt")) {
+		const props = sdtRunProperties(child);
+		for (const inner of poChildrenWith(poFirstChild(child, "w:sdtContent"), tag)) out.push({
+			node: inner,
+			...props ? { sdtRunProps: props } : {}
+		});
+	}
+	return out;
+}
 function resolveMergeRoles(rows) {
 	const out = rows.map((r) => new Array(r.length).fill(void 0));
 	const colSlots = /* @__PURE__ */ new Map();
@@ -109,11 +180,23 @@ function parseTableProperties(tblPr) {
 		else if (w !== void 0) out.widthPt = twipsToPt(w);
 		if (type && WIDTH_TYPES.has(type)) out.widthType = type;
 	}
+	const tblInd = poFirstChild(tblPr, "w:tblInd");
+	if (tblInd) {
+		const w = poIntAttr(tblInd, "w");
+		const type = poAttr(tblInd, "type");
+		if (w !== void 0 && (type === void 0 || type === "dxa")) out.indentPt = twipsToPt(w);
+	}
+	const tblpPr = poFirstChild(tblPr, "w:tblpPr");
+	if (tblpPr) out.float = parseTablePosition(tblpPr);
 	const tblLayout = poFirstChild(tblPr, "w:tblLayout");
 	if (tblLayout) {
 		const t = poAttr(tblLayout, "type");
 		if (t === "fixed" || t === "auto") out.layout = t;
 	}
+	const tblJc = poVal(poFirstChild(tblPr, "w:jc"));
+	if (tblJc === "center") out.alignment = "center";
+	else if (tblJc === "right" || tblJc === "end") out.alignment = "right";
+	else if (tblJc === "left" || tblJc === "start") out.alignment = "left";
 	const borders = parseBorders(poFirstChild(tblPr, "w:tblBorders"));
 	if (borders) out.borders = borders;
 	const margins = parseCellMargins(poFirstChild(tblPr, "w:tblCellMar"));
@@ -154,15 +237,61 @@ function parseTableGrid(tblGrid) {
 	}
 	return cols;
 }
+/**
+* §17.13.5.15 — whether a row was deleted, under the reader's accept-all
+* reading of tracked changes (the same one that drops `w:del` runs).
+*
+* Two spellings. `w:trPr/w:del` says it outright. The other is how a producer
+* writes a row whose CONTENT was taken away: every paragraph mark in it carries
+* `w:del` or `w:moveFrom`, which deletes the mark, and a row whose every mark
+* is gone is a row that is gone. TC-table-DnD-move.docx moves a table that way,
+* and accepting only the runs left a ghost of empty bordered rows behind where
+* Word and LibreOffice leave nothing.
+*
+* @param tr The `w:tr` element.
+* @returns Whether to drop the row.
+*/
+function isDeletedRow(tr) {
+	const trPr = poFirstChild(tr, "w:trPr");
+	if (trPr && (poFirstChild(trPr, "w:del") ?? poFirstChild(trPr, "w:moveFrom"))) return true;
+	const cells = poChildrenThroughSdt(tr, "w:tc");
+	if (cells.length === 0) return false;
+	let marks = 0;
+	for (const tc of cells) {
+		for (const p of poChildrenWith(tc, "w:p")) {
+			const rPr = poFirstChild(poFirstChild(p, "w:pPr"), "w:rPr");
+			if (!rPr || !(poFirstChild(rPr, "w:del") ?? poFirstChild(rPr, "w:moveFrom"))) return false;
+			marks++;
+		}
+		if (poChildrenWith(tc, "w:tbl").length > 0) return false;
+	}
+	return marks > 0;
+}
 function parseTableRow(tr, ctx) {
 	const properties = parseRowProperties(poFirstChild(tr, "w:trPr"));
 	const cells = [];
-	for (const tc of poChildrenWith(tr, "w:tc")) cells.push(parseTableCell(tc, ctx));
+	for (const { node, sdtRunProps } of childrenThroughSdt(tr, "w:tc")) cells.push(parseTableCell(node, ctx, sdtRunProps));
 	return {
 		properties,
 		cells
 	};
 }
+function parseRowConditional(cnf) {
+	if (!cnf) return void 0;
+	const bits = poAttr(cnf, "val");
+	const bit = (i) => bits?.[i] === "1";
+	const flag = (name) => {
+		const v = poAttr(cnf, name);
+		return v === void 0 ? void 0 : v === "1" || v === "true" || v === "on";
+	};
+	const firstRow = flag("firstRow") ?? bit(0);
+	const lastRow = flag("lastRow") ?? bit(1);
+	if (!firstRow && !lastRow) return void 0;
+	return {
+		...firstRow ? { firstRow } : {},
+		...lastRow ? { lastRow } : {}
+	};
+}
 function parseRowProperties(trPr) {
 	if (!trPr) return {};
 	const out = {};
@@ -172,17 +301,22 @@ function parseRowProperties(trPr) {
 		const rule = poAttr(trHeight, "hRule");
 		if (val !== void 0) out.height = twipsToPt(val);
 		if (rule && HEIGHT_RULES.has(rule)) out.heightRule = rule;
+		else if (val !== void 0) out.heightRule = "atLeast";
 	}
 	if (poFirstChild(trPr, "w:cantSplit")) out.cantSplit = poToggle(poFirstChild(trPr, "w:cantSplit")) ?? true;
 	if (poFirstChild(trPr, "w:tblHeader")) out.isHeader = poToggle(poFirstChild(trPr, "w:tblHeader")) ?? true;
+	const before = poIntAttr(poFirstChild(trPr, "w:gridBefore"), "val");
+	if (before !== void 0 && before > 0) out.gridBefore = before;
+	const cnf = parseRowConditional(poFirstChild(trPr, "w:cnfStyle"));
+	if (cnf) out.conditional = cnf;
 	return out;
 }
-function parseTableCell(tc, ctx) {
+function parseTableCell(tc, ctx, sdtRunProps) {
 	const { properties, vMerge } = parseCellProperties(poFirstChild(tc, "w:tcPr"));
 	return {
 		cell: {
 			properties,
-			content: parseBodyElements(poChildren(tc), ctx)
+			content: parseBodyElements(poChildren(tc), ctx, void 0, sdtRunProps)
 		},
 		...vMerge ? { vMerge } : {}
 	};
@@ -194,7 +328,12 @@ function parseCellProperties(tcPr) {
 	const tcW = poFirstChild(tcPr, "w:tcW");
 	if (tcW) {
 		const w = poIntAttr(tcW, "w");
-		if (w !== void 0) out.width = twipsToPt(w);
+		const type = poAttr(tcW, "type");
+		if (type === "pct" || type === "dxa" || type === "auto" || type === "nil") out.widthType = type;
+		if (w !== void 0) {
+			if (type === "pct") out.widthFraction = w / 5e3;
+			else if (type !== "auto" && type !== "nil") out.width = twipsToPt(w);
+		}
 	}
 	const gridSpan = poFirstChild(tcPr, "w:gridSpan");
 	if (gridSpan) {
@@ -212,11 +351,21 @@ function parseCellProperties(tcPr) {
 		const fill = poAttr(shd, "fill");
 		if (fill && fill !== "auto" && /^[0-9A-Fa-f]{6}$/.test(fill)) out.shading = { colorHex: fill.toUpperCase() };
 	}
+	const vAlign = poVal(poFirstChild(tcPr, "w:vAlign"));
+	if (vAlign === "center" || vAlign === "bottom" || vAlign === "top") out.verticalAlign = vAlign;
+	if (poFirstChild(tcPr, "w:hideMark")) out.hideMark = true;
 	return {
 		properties: out,
 		...rawVMerge ? { vMerge: rawVMerge } : {}
 	};
 }
+/**
+* A `w:tcBorders` / `w:tblBorders` / §17.6.10 `w:pgBorders` element: the rule on
+* each edge. Every one of them is spelled the same way.
+*
+* @param node The borders element, or `undefined`.
+* @returns The edges that name a rule, or `undefined` when none do.
+*/
 function parseBorders(node) {
 	if (!node) return void 0;
 	const out = {};
@@ -236,12 +385,15 @@ function parseBorders(node) {
 }
 function parseBorder(node) {
 	if (!node) return void 0;
-	const val = poVal(node);
-	if (!val || !BORDER_STYLES.has(val)) return void 0;
+	const raw = poVal(node);
+	if (!raw) return void 0;
+	const val = raw === "nil" ? "none" : BORDER_STYLES.has(raw) ? raw : MULTI_LINE_BORDERS.has(raw) ? "double" : "single";
 	const sz = poIntAttr(node, "sz");
 	const color = poAttr(node, "color");
 	const out = { style: val };
-	if (sz !== void 0) out.width = eighthPtToPt(sz);
+	if (sz !== void 0) out.width = eighthPtToPt(MULTI_LINE_BORDERS.has(raw) ? sz * 3 : sz);
+	const space = poIntAttr(node, "space");
+	if (space !== void 0) out.spacePt = pt(space);
 	if (color && color !== "auto" && /^[0-9A-Fa-f]{6}$/.test(color)) out.colorHex = color.toUpperCase();
 	return out;
 }
@@ -252,11 +404,11 @@ function parseCellMargins(node) {
 	const bottom = poIntAttr(poFirstChild(node, "w:bottom"), "w");
 	const left = poIntAttr(poFirstChild(node, "w:left") ?? poFirstChild(node, "w:start"), "w");
 	const right = poIntAttr(poFirstChild(node, "w:right") ?? poFirstChild(node, "w:end"), "w");
-	if (top !== void 0) out.top = twipsToPt(top);
-	if (bottom !== void 0) out.bottom = twipsToPt(bottom);
-	if (left !== void 0) out.left = twipsToPt(left);
-	if (right !== void 0) out.right = twipsToPt(right);
+	if (top !== void 0) out.top = twipsToPt(Math.max(0, top));
+	if (bottom !== void 0) out.bottom = twipsToPt(Math.max(0, bottom));
+	if (left !== void 0) out.left = twipsToPt(Math.max(0, left));
+	if (right !== void 0) out.right = twipsToPt(Math.max(0, right));
 	return Object.keys(out).length > 0 ? out : void 0;
 }
 //#endregion
-export { parseTable };
+export { parseBorders, parseTable };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "reamkit",
-  "version": "1.15.4",
+  "version": "1.17.0",
   "description": "Ream — convert DOCX, XLSX, PPTX and PDF to PDF, SVG, HTML, DOCX and XLSX, built from scratch on the ECMA-376 and ISO 32000 specifications. Parse once, convert anywhere.",
   "license": "MIT",
   "author": "Alex Krassavin <info@reamkit.dev>",
@@ -60,7 +60,7 @@
     "test:types": "tsc -p tsconfig.json --noEmit",
     "test:eslint": "eslint .",
     "test:build": "publint --strict && attw --pack . --profile esm-only",
-    "typecheck": "tsc -p tsconfig.json --noEmit",
+    "typecheck": "tsc -p tsconfig.eslint.json --noEmit",
     "lint": "eslint .",
     "lint:fix": "eslint . --fix",
     "format": "prettier --write \"src/**/*.ts\" \"tests/**/*.ts\" \"scripts/**/*.ts\"",
@@ -73,7 +73,11 @@
     "corpus:roundtrip:xlsx": "tsx scripts/corpus/xlsx-roundtrip.ts",
     "corpus:xlsx:invariants": "tsx scripts/corpus/xlsx-invariants.ts",
     "corpus:fixtures": "tsx scripts/corpus/sync-real-fixtures.ts",
-    "corpus:golden": "tsx scripts/corpus/make-golden.ts"
+    "corpus:golden": "tsx scripts/corpus/make-golden.ts",
+    "corpus:visual": "tsx scripts/corpus/visual-diff.ts",
+    "corpus:scout": "tsx scripts/corpus/scout.ts",
+    "corpus:pixel": "tsx scripts/corpus/pixel-scout.ts",
+    "typecheck:dist": "tsc -p tsconfig.json --noEmit"
   },
   "dependencies": {
     "fast-xml-parser": "5.7.0",