reamkit 1.1.0 → 1.3.0

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

@@ -1,4 +1,4 @@
-import { BodyElement, ShapeFill, ShapeGeometry, ShapeLine, ShapeTextBody, ShapeTransform } from '../core/document-model/index.js';
+import { BodyElement, FloatAnchor, ShapeFill, ShapeGeometry, ShapeLine, ShapeTextBody, ShapeTransform } from '../core/document-model/index.js';
 import { ColorResolver } from '../core/drawingml/colors.js';
 import { PoNode } from '../core/po-helpers.js';
 import { Pt } from '../core/ir/index.js';
@@ -18,16 +18,19 @@ export type DrawingContent = {
     readonly width: Pt;
     readonly height: Pt;
     readonly altText?: string;
+    readonly float?: FloatAnchor;
 } | {
     readonly kind: 'shape';
     readonly data: ShapeData;
     readonly altText?: string;
+    readonly float?: FloatAnchor;
 } | {
     readonly kind: 'chart';
     readonly chartRelId: string;
     readonly width: Pt;
     readonly height: Pt;
     readonly altText?: string;
+    readonly float?: FloatAnchor;
 };
 export declare function resolveMc(altContent: PoNode): ReadonlyArray<PoNode>;
 export declare function expandMcChildren(children: ReadonlyArray<PoNode>): Array<PoNode>;

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

@@ -1,10 +1,57 @@
 import { emuToPt } from "../core/ir/units.js";
-import { poAttr, poChildren, poFindDescendant, poIntAttr, poIs, poTag } from "../core/po-helpers.js";
+import { poAttr, poChildren, poFindDescendant, poIntAttr, poIs, poTag, poText } from "../core/po-helpers.js";
 import { resolveColorNode } from "../core/drawingml/colors.js";
 //#region src/word/drawing-parser.ts
 var WPS_URI = "http://schemas.microsoft.com/office/word/2010/wordprocessingShape";
 var CHART_URI = "http://schemas.openxmlformats.org/drawingml/2006/chart";
 var UNDERSTOOD_NS = new Set(["wps"]);
+function parseFloatAnchor(anchor) {
+	if (!poIs(anchor, "wp:anchor")) return void 0;
+	const behindRaw = poAttr(anchor, "behindDoc");
+	const behind = behindRaw === "1" || behindRaw === "true";
+	let wrap = "none";
+	for (const child of poChildren(anchor)) if (poIs(child, "wp:wrapSquare")) wrap = "square";
+	else if (poIs(child, "wp:wrapTight")) wrap = "tight";
+	else if (poIs(child, "wp:wrapThrough")) wrap = "through";
+	else if (poIs(child, "wp:wrapTopAndBottom")) wrap = "topAndBottom";
+	const posH = parseAnchorPos(anchor, "wp:positionH", [
+		"margin",
+		"page",
+		"column"
+	]);
+	const posV = parseAnchorPos(anchor, "wp:positionV", [
+		"margin",
+		"page",
+		"paragraph",
+		"line"
+	]);
+	return {
+		wrap,
+		...behind ? { behind: true } : {},
+		...posH ? { posH } : {},
+		...posV ? { posV } : {}
+	};
+}
+var ANCHOR_ALIGNS = new Set([
+	"left",
+	"center",
+	"right"
+]);
+function parseAnchorPos(anchor, tag, allowed) {
+	const pos = poChildren(anchor).find((c) => poIs(c, tag));
+	if (!pos) return void 0;
+	const relRaw = poAttr(pos, "relativeFrom") ?? "margin";
+	const relativeFrom = allowed.includes(relRaw) ? relRaw : allowed[0];
+	const offsetNode = poChildren(pos).find((c) => poIs(c, "wp:posOffset"));
+	const offsetRaw = offsetNode ? Number(poText(offsetNode).trim()) : NaN;
+	const alignNode = poChildren(pos).find((c) => poIs(c, "wp:align"));
+	const alignRaw = alignNode ? poText(alignNode).trim() : "";
+	return {
+		relativeFrom,
+		...Number.isFinite(offsetRaw) ? { offsetPt: emuToPt(offsetRaw) } : {},
+		...tag === "wp:positionH" && ANCHOR_ALIGNS.has(alignRaw) ? { align: alignRaw } : {}
+	};
+}
 function resolveMc(altContent) {
 	for (const choice of poChildren(altContent)) {
 		if (!poIs(choice, "mc:Choice")) continue;
@@ -30,7 +77,11 @@ function parseDrawing(drawing, resolveColor, parseBody) {
 	const descr = docPr ? poAttr(docPr, "descr") : void 0;
 	const title = docPr ? poAttr(docPr, "title") : void 0;
 	const altText = (descr ?? title)?.trim() || void 0;
-	const alt = altText ? { altText } : {};
+	const float = parseFloatAnchor(anchor);
+	const alt = {
+		...altText ? { altText } : {},
+		...float ? { float } : {}
+	};
 	const graphicData = poFindDescendant(anchor, "a:graphicData");
 	const uri = graphicData ? poAttr(graphicData, "uri") : void 0;
 	if (graphicData && uri === WPS_URI) {

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

@@ -1,6 +1,6 @@
-export type { ImageResolver, ParseContext } from './document-parser.js';
+export type { HyperlinkResolver, ImageResolver, ParseContext } from './document-parser.js';
 export { extractParagraphs } from './text-extractor.js';
-export { parseDocument, parseBodyElements, parseSection, parseSections, parseHeaderFooter, EMPTY_SECTION, } from './document-parser.js';
+export { parseDocument, parseBodyElements, parseSection, parseSections, parseHeaderFooter, parseNotes, EMPTY_SECTION, } from './document-parser.js';
 export { parseRunProperties } from './run-properties.js';
 export { parseParagraphProperties } from './paragraph-properties.js';
 export { parseStyles } from './styles-parser.js';

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

@@ -1,3 +1,4 @@
+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";
@@ -50,7 +51,8 @@ function parseStyles(stylesXml) {
 			...basedOnVal ? { basedOn: basedOnVal } : {},
 			isDefault,
 			runProperties: parseRunProperties(el["w:rPr"]),
-			paragraphProperties: parseParagraphProperties(el["w:pPr"])
+			paragraphProperties: parseParagraphProperties(el["w:pPr"]),
+			...type === "table" ? parseTableStyleParts(el) : {}
 		};
 		styles.set(id, style);
 	}
@@ -60,5 +62,136 @@ function parseStyles(stylesXml) {
 		styles
 	};
 }
+var CONDITION_TYPES = new Set([
+	"wholeTable",
+	"band1Vert",
+	"band2Vert",
+	"band1Horz",
+	"band2Horz",
+	"firstCol",
+	"lastCol",
+	"firstRow",
+	"lastRow",
+	"nwCell",
+	"neCell",
+	"swCell",
+	"seCell"
+]);
+function parseTableStyleParts(el) {
+	const out = {};
+	const base = parseTableStyleLayer(el);
+	if (base) out.tableLayer = base;
+	const tblPr = asElement(el["w:tblPr"]);
+	if (tblPr) {
+		const rowBand = intVal(tblPr["w:tblStyleRowBandSize"]);
+		const colBand = intVal(tblPr["w:tblStyleColBandSize"]);
+		if (rowBand !== void 0 && rowBand > 0) out.rowBandSize = rowBand;
+		if (colBand !== void 0 && colBand > 0) out.colBandSize = colBand;
+	}
+	const conditions = [];
+	for (const c of asArray(el["w:tblStylePr"])) {
+		const cond = asElement(c);
+		if (!cond) continue;
+		const typeAttr = getAttr(cond, "type");
+		if (!typeAttr || !CONDITION_TYPES.has(typeAttr)) continue;
+		const layer = parseTableStyleLayer(cond);
+		if (layer) conditions.push({
+			type: typeAttr,
+			layer
+		});
+	}
+	if (conditions.length > 0) out.tableConditions = conditions;
+	return out;
+}
+function parseTableStyleLayer(el) {
+	const tblPr = asElement(el["w:tblPr"]);
+	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 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 } : {},
+		...shading ? { shading } : {},
+		...Object.keys(runProperties).length > 0 ? { runProperties } : {},
+		...Object.keys(paragraphProperties).length > 0 ? { paragraphProperties } : {}
+	};
+	return Object.keys(layer).length > 0 ? layer : void 0;
+}
+var FLAT_BORDER_STYLES = new Set([
+	"none",
+	"single",
+	"double",
+	"thick",
+	"dotted",
+	"dashed"
+]);
+function parseFlatBorder(node) {
+	const el = asElement(node);
+	if (!el) return void 0;
+	const val = getAttr(el, "val");
+	if (!val || !FLAT_BORDER_STYLES.has(val)) return void 0;
+	const szRaw = getAttr(el, "sz");
+	const sz = szRaw !== void 0 ? Number(szRaw) : NaN;
+	const color = getAttr(el, "color");
+	return {
+		style: val,
+		...Number.isFinite(sz) ? { width: eighthPtToPt(sz) } : {},
+		...color && color !== "auto" && /^[0-9A-Fa-f]{6}$/.test(color) ? { colorHex: color.toUpperCase() } : {}
+	};
+}
+function parseFlatBorders(el) {
+	if (!el) return void 0;
+	const top = parseFlatBorder(el["w:top"]);
+	const right = parseFlatBorder(el["w:right"] ?? el["w:end"]);
+	const bottom = parseFlatBorder(el["w:bottom"]);
+	const left = parseFlatBorder(el["w:left"] ?? el["w:start"]);
+	const insideH = parseFlatBorder(el["w:insideH"]);
+	const insideV = parseFlatBorder(el["w:insideV"]);
+	const out = {
+		...top ? { top } : {},
+		...right ? { right } : {},
+		...bottom ? { bottom } : {},
+		...left ? { left } : {},
+		...insideH ? { insideH } : {},
+		...insideV ? { insideV } : {}
+	};
+	return Object.keys(out).length > 0 ? out : void 0;
+}
+function parseFlatCellMargins(el) {
+	if (!el) return void 0;
+	const read = (node) => {
+		const m = asElement(node);
+		if (!m) return void 0;
+		const w = getAttr(m, "w");
+		const n = w !== void 0 ? Number(w) : NaN;
+		return Number.isFinite(n) ? n : void 0;
+	};
+	const top = read(el["w:top"]);
+	const bottom = read(el["w:bottom"]);
+	const left = read(el["w:left"] ?? el["w:start"]);
+	const right = read(el["w:right"] ?? el["w:end"]);
+	const out = {
+		...top !== void 0 ? { top: twipsToPt(top) } : {},
+		...bottom !== void 0 ? { bottom: twipsToPt(bottom) } : {},
+		...left !== void 0 ? { left: twipsToPt(left) } : {},
+		...right !== void 0 ? { right: twipsToPt(right) } : {}
+	};
+	return Object.keys(out).length > 0 ? out : void 0;
+}
+function parseFlatShading(node) {
+	const el = asElement(node);
+	if (!el) return void 0;
+	const fill = getAttr(el, "fill");
+	if (fill && fill !== "auto" && /^[0-9A-Fa-f]{6}$/.test(fill)) return { colorHex: fill.toUpperCase() };
+}
+function intVal(node) {
+	const v = getVal(node);
+	if (v === void 0) return void 0;
+	const n = Number(v);
+	return Number.isFinite(n) ? n : void 0;
+}
 //#endregion
 export { parseStyles };

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

@@ -80,6 +80,16 @@ function resolveMergeRoles(rows) {
 function parseTableProperties(tblPr) {
 	if (!tblPr) return {};
 	const out = {};
+	const tblStyle = poFirstChild(tblPr, "w:tblStyle");
+	if (tblStyle) {
+		const id = poVal(tblStyle);
+		if (id) out.styleId = id;
+	}
+	const tblLook = poFirstChild(tblPr, "w:tblLook");
+	if (tblLook) {
+		const look = parseTableLook(tblLook);
+		if (look) out.look = look;
+	}
 	const tblW = poFirstChild(tblPr, "w:tblW");
 	if (tblW) {
 		const w = poIntAttr(tblW, "w");
@@ -99,6 +109,31 @@ function parseTableProperties(tblPr) {
 	if (margins) out.defaultCellMargins = margins;
 	return out;
 }
+function parseTableLook(node) {
+	const flag = (name, bit) => {
+		const attr = poAttr(node, name);
+		if (attr !== void 0) return attr === "1" || attr === "true";
+		const valRaw = poAttr(node, "val");
+		if (valRaw === void 0) return void 0;
+		const mask = parseInt(valRaw, 16);
+		return Number.isFinite(mask) ? (mask & bit) !== 0 : void 0;
+	};
+	const firstRow = flag("firstRow", 32);
+	const lastRow = flag("lastRow", 64);
+	const firstColumn = flag("firstColumn", 128);
+	const lastColumn = flag("lastColumn", 256);
+	const noHBand = flag("noHBand", 512);
+	const noVBand = flag("noVBand", 1024);
+	const out = {
+		...firstRow !== void 0 ? { firstRow } : {},
+		...lastRow !== void 0 ? { lastRow } : {},
+		...firstColumn !== void 0 ? { firstColumn } : {},
+		...lastColumn !== void 0 ? { lastColumn } : {},
+		...noHBand !== void 0 ? { noHBand } : {},
+		...noVBand !== void 0 ? { noVBand } : {}
+	};
+	return Object.keys(out).length > 0 ? out : void 0;
+}
 function parseTableGrid(tblGrid) {
 	if (!tblGrid) return [];
 	const cols = [];

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "reamkit",
-  "version": "1.1.0",
+  "version": "1.3.0",
   "description": "Ream — DOCX/XLSX to PDF (and SVG), built from scratch on the ECMA-376 and ISO 32000 specifications. Parse once, convert anywhere.",
   "license": "MIT",
   "author": "Alex Krassavin <info@reamkit.dev>",