reamkit 1.31.0 → 1.32.1

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

@@ -15,6 +15,7 @@ import { OpcPackage } from "../core/opc/package.js";
 import { parseCoreProperties } from "../core/opc/core-properties.js";
 import "../core/opc/index.js";
 import { parseXml } from "../pptx/pptx-reader.js";
+import { firstTagMatch } from "../core/opc/tag-scan.js";
 import { EMPTY_NUMBERING, parseNumbering } from "./numbering-parser.js";
 import { EMPTY_SECTION, applyAuthorIds, bodyIndexForBlock, newBlockCounter, parseBackgroundColor, parseBackgroundFill, parseCommentThreads, parseDocument, parseHeaderFooter, parseNotes, parsePeople, parseSections } from "./document-parser.js";
 import { parseStyles } from "./styles-parser.js";
@@ -214,21 +215,26 @@ function makeDiagramResolver(pkg, partName, store) {
 		return resolved;
 	};
 }
+function partIndex(path) {
+	const dot = path.lastIndexOf(".");
+	if (dot < 0 || dot === path.length - 1) return "";
+	let from = dot;
+	while (from > 0 && path.charCodeAt(from - 1) >= 48 && path.charCodeAt(from - 1) <= 57) from--;
+	return path.slice(from, dot);
+}
 function siblingPart(pkg, ownerPart, dataPart, type) {
 	const own = pkg.getPartRelationships(dataPart).find((r) => r.type.endsWith(type));
 	if (own) return pkg.resolveRelatedPart(dataPart, own)?.data;
-	const index = (path) => /(\d*)\.[^.]+$/u.exec(path)?.[1] ?? "";
-	const want = index(dataPart);
+	const want = partIndex(dataPart);
 	const rels = pkg.getPartRelationships(ownerPart).filter((r) => r.type.endsWith(type));
-	const rel = rels.find((r) => index(r.target) === want) ?? (rels.length === 1 ? rels[0] : void 0);
+	const rel = rels.find((r) => partIndex(r.target) === want) ?? (rels.length === 1 ? rels[0] : void 0);
 	return rel ? pkg.resolveRelatedPart(ownerPart, rel)?.data : void 0;
 }
 function drawingFromOwner(pkg, ownerPart, dataPart) {
 	const drawings = pkg.getPartRelationships(ownerPart).filter((r) => r.type.endsWith("/diagramDrawing"));
 	if (drawings.length === 0) return void 0;
-	const index = (path) => /(\d*)\.[^.]+$/u.exec(path)?.[1] ?? "";
-	const want = index(dataPart);
-	const rel = drawings.find((r) => index(r.target) === want) ?? (drawings.length === 1 ? drawings[0] : void 0);
+	const want = partIndex(dataPart);
+	const rel = drawings.find((r) => partIndex(r.target) === want) ?? (drawings.length === 1 ? drawings[0] : void 0);
 	return rel ? pkg.resolveRelatedPart(ownerPart, rel) : void 0;
 }
 function makeImageResolver(pkg, store, partName = MAIN_DOCUMENT_PART) {
@@ -377,11 +383,10 @@ function chartOwningParts(pkg, sections, mainPart) {
 	return parts;
 }
 function detectDocxLanguage(stylesData, documentData) {
-	const re = /<w:lang\b[^>]*\bw:val="([^"]+)"/;
 	const decoder = new TextDecoder();
 	for (const data of [stylesData, documentData]) {
 		if (!data) continue;
-		const m = re.exec(decoder.decode(data));
+		const m = firstTagMatch(decoder.decode(data), "<w:lang", true, /<w:lang\b[^>]*\bw:val="([^"]+)"/y);
 		if (m?.[1]) return m[1];
 	}
 }

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

@@ -6,6 +6,7 @@ import "../pdf/index.js";
 import { parseThemeFonts } from "../core/drawingml/theme-parser.js";
 import { OpcPackage } from "../core/opc/package.js";
 import "../core/opc/index.js";
+import { tagMatches } from "../core/opc/tag-scan.js";
 import { resolveWordThemeFont } from "./theme-fonts.js";
 import { readDocx } from "./docx-reader.js";
 import { flowRenderOptions } from "../core/converter/project.js";
@@ -13,6 +14,7 @@ import { flowRenderOptions } from "../core/converter/project.js";
 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;
 /**
 * Synchronous one-shot conversion (requires `fonts`/`fontBytes`, no network).
 *
@@ -72,13 +74,11 @@ function detectDocxFamilyKeys(docx) {
 	}
 	const decoder = new TextDecoder("utf-8");
 	const themeFonts = docxThemeFonts(pkg);
-	const re = /<w:rFonts[^>]*?\bw:(ascii|asciiTheme)="([^"]+)"/g;
 	for (const part of [STYLES_PART, MAIN_DOCUMENT_PART]) {
 		const data = pkg.getPart(part);
 		if (!data) continue;
 		const xml = decoder.decode(data);
-		let m;
-		while ((m = re.exec(xml)) !== null) {
+		for (const m of tagMatches(xml, "<w:rFonts", false, ASCII_FONT)) {
 			const name = m[1] === "ascii" ? m[2] : resolveWordThemeFont(m[2], themeFonts);
 			if (name !== void 0) keys.add(resolveFamilyKey(name));
 		}

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

@@ -156,7 +156,8 @@ function writeDocx(flow) {
 		familiesUsed: /* @__PURE__ */ new Map(),
 		...flow.faceOutlines ? { faceOutlines: flow.faceOutlines } : {},
 		facesUsed: /* @__PURE__ */ new Set(),
-		zRanks: zRanksOf(flow)
+		zRanks: zRanksOf(flow),
+		overlaid: /* @__PURE__ */ new Set()
 	};
 	const docScope = newScope();
 	const extraParts = [];
@@ -355,7 +356,10 @@ function zRanksOf(flow) {
 		for (const el of blocks) {
 			const z = el.kind === "image" ? el.image.float?.zOrder : el.kind === "shape" ? el.shape.float?.zOrder : el.kind === "chart" ? el.chart.float?.zOrder : void 0;
 			if (z !== void 0 && Number.isFinite(z)) seen.add(z);
-			if (el.kind === "table") for (const row of el.table.rows) for (const cell of row.cells) visit(cell.content);
+			if (el.kind === "table") {
+				for (const row of el.table.rows) for (const cell of row.cells) visit(cell.content);
+				visit(el.table.overlay?.drawings ?? []);
+			}
 		}
 	};
 	visit(flow.body);
@@ -593,6 +597,11 @@ function emitBlock(out, el, losses, state, scope, closingSectPr) {
 	}
 	if (el.kind === "table") {
 		out.push(tableXml(el.table, losses, state, scope));
+		for (const drawing of el.table.overlay?.drawings ?? []) {
+			if (state.overlaid.has(drawing)) continue;
+			state.overlaid.add(drawing);
+			emitBlock(out, drawing, losses, state, scope);
+		}
 		if (closingSectPr) out.push(`<w:p><w:pPr>${closingSectPr}</w:pPr></w:p>`);
 		return;
 	}

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

@@ -1,4 +1,5 @@
 import { FontRegistry } from "../core/font/font-registry.js";
+import { isWordChar, tagMatches } from "../core/opc/tag-scan.js";
 //#region src/word/font-table.ts
 var FONT_TABLE_PART = "word/fontTable.xml";
 var REL_FONT = "http://schemas.openxmlformats.org/officeDocument/2006/relationships/font";
@@ -48,19 +49,15 @@ function obfuscateEmbeddedFont(data, fontKey) {
 function parseFontTable(data) {
 	const xml = new TextDecoder("utf-8").decode(data);
 	const out = [];
-	const fontRe = /<w:font\b[^>]*\bw:name="([^"]+)"[^>]*>([\s\S]*?)<\/w:font>/g;
-	let fm;
-	while ((fm = fontRe.exec(xml)) !== null) {
+	const closed = xml.slice(0, xml.lastIndexOf("</w:font>") + 9);
+	for (const fm of tagMatches(closed, "<w:font", true, FONT)) {
 		const name = fm[1];
-		const inner = fm[2];
 		const embeds = {};
-		const embedRe = /<w:embed(Regular|Bold|Italic|BoldItalic)\b[^>]*\br:id="([^"]+)"[^>]*\bw:fontKey="([^"]+)"/g;
-		let em;
-		while ((em = embedRe.exec(inner)) !== null) {
-			const variant = VARIANT_BY_TAG[em[1]];
+		for (const em of embedRefs(fm[2])) {
+			const variant = VARIANT_BY_TAG[em.tag];
 			if (variant) embeds[variant] = {
-				rId: em[2],
-				fontKey: em[3]
+				rId: em.rId,
+				fontKey: em.fontKey
 			};
 		}
 		if (Object.keys(embeds).length > 0) out.push({
@@ -70,6 +67,88 @@ function parseFontTable(data) {
 	}
 	return out;
 }
+/** A named `w:font` and everything up to its close. Sticky, for tagMatches. */
+var FONT = /<w:font\b[^>]*\bw:name="([^"]+)"[^>]*>([\s\S]*?)<\/w:font>/y;
+var EMBED_HEAD = /<w:embed(Regular|Bold|Italic|BoldItalic)\b/y;
+/**
+* The faces a font's element embeds — what
+* `/<w:embed(Regular|Bold|Italic|BoldItalic)\b[^>]*\br:id="([^"]+)"[^>]*\bw:fontKey="([^"]+)"/g`
+* finds in it, in one pass. The expression gave the tag back one `r:id` at a
+* time and read the rest of it again for a `w:fontKey` after each, quadratic
+* in a tag with many; and searching, it read each tag again from every
+* `<w:embed` inside it. Where a start fails, every later one before the tag's
+* `>` does too, so the search moves past it.
+*/
+function embedRefs(inner) {
+	const out = [];
+	let at = inner.indexOf("<w:embed");
+	while (at >= 0) {
+		EMBED_HEAD.lastIndex = at;
+		const head = EMBED_HEAD.exec(inner);
+		if (head === null) {
+			at = inner.indexOf("<w:embed", at + 1);
+			continue;
+		}
+		const from = EMBED_HEAD.lastIndex;
+		const tagEnd = endOfStretch(inner, from);
+		const found = embedRef(inner, from, tagEnd);
+		if (found) {
+			out.push({
+				tag: head[1],
+				rId: found.rId,
+				fontKey: found.fontKey
+			});
+			at = inner.indexOf("<w:embed", found.end);
+		} else if (tagEnd < inner.length) at = inner.indexOf("<w:embed", tagEnd + 1);
+		else break;
+	}
+	return out;
+}
+/** The first `>` at or after `from`, or the end of the text. */
+function endOfStretch(s, from) {
+	const gt = s.indexOf(">", from);
+	return gt < 0 ? s.length : gt;
+}
+/**
+* The reference one `<w:embed…` makes, its attributes read from `from` up to
+* the tag's `>` at `tagEnd`. The first `[^>]*` gives back the last `r:id`
+* first; the second takes the last `w:fontKey` before the `>` that follows the
+* `r:id`'s value — and that one is the same for every `r:id` whose value ends
+* in one stretch, so it is looked for once per stretch.
+*/
+function embedRef(s, from, tagEnd) {
+	const keys = /* @__PURE__ */ new Map();
+	const lastKey = (stretchEnd) => {
+		if (keys.has(stretchEnd)) return keys.get(stretchEnd);
+		const start = s.lastIndexOf(">", stretchEnd - 1) + 1;
+		let key;
+		for (let f = s.lastIndexOf("w:fontKey=\"", stretchEnd - 11); f >= start; f = s.lastIndexOf("w:fontKey=\"", f - 1)) {
+			const close = s.indexOf("\"", f + 11);
+			if (!isWordChar(s.charCodeAt(f - 1)) && close > f + 11) {
+				key = {
+					at: f,
+					value: s.slice(f + 11, close),
+					end: close + 1
+				};
+				break;
+			}
+			if (f === 0) break;
+		}
+		keys.set(stretchEnd, key);
+		return key;
+	};
+	for (let p = s.lastIndexOf("r:id=\"", tagEnd - 6); p >= from; p = s.lastIndexOf("r:id=\"", p - 1)) {
+		if (isWordChar(s.charCodeAt(p - 1))) continue;
+		const close = s.indexOf("\"", p + 6);
+		if (close <= p + 6) continue;
+		const key = lastKey(close < tagEnd ? tagEnd : endOfStretch(s, close + 1));
+		if (key !== void 0 && key.at > close) return {
+			rId: s.slice(p + 6, close),
+			fontKey: key.value,
+			end: key.end
+		};
+	}
+}
 /**
 * De-obfuscate and load every embedded font in the package, building one
 * {@link FontRegistry} per font keyed by its normalized (trimmed, lower-cased)

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

@@ -171,7 +171,7 @@ var CSS_UNITS = new Map([
 	["px", .75]
 ]);
 function cssLengthPt(raw) {
-	const m = raw !== void 0 ? /^\s*(-?[\d.]+)\s*([a-z%]*)\s*$/u.exec(raw) : null;
+	const m = raw !== void 0 ? /^(-?[\d.]+)\s*([a-z%]*)$/u.exec(raw.trim()) : null;
 	if (!m) return void 0;
 	const n = Number(m[1]);
 	if (!Number.isFinite(n)) return void 0;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "reamkit",
-  "version": "1.31.0",
+  "version": "1.32.1",
   "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": "Alex Krassavin <info@reamkit.dev>",
@@ -84,10 +84,14 @@
     "corpus:pixel": "tsx scripts/corpus/pixel-scout.ts",
     "stand": "tsx stand/serve.ts",
     "stand:build": "tsx stand/build.ts",
+    "stand:html": "tsx stand/serve.ts html",
+    "stand:html:build": "tsx stand/html-build.ts",
+    "stand:svg": "tsx stand/serve.ts svg",
+    "stand:svg:build": "tsx stand/svg-build.ts",
     "typecheck:dist": "tsc -p tsconfig.json --noEmit"
   },
   "dependencies": {
-    "fast-xml-parser": "5.7.0",
+    "fast-xml-parser": "5.7.3",
     "fflate": "^0.8.2"
   },
   "devDependencies": {
@@ -99,7 +103,7 @@
     "tsx": "^4.19.0",
     "typescript": "^5.7.0",
     "vite": "8.0.5",
-    "vitest": "^2.1.0"
+    "vitest": "^5.0.3"
   },
   "engines": {
     "node": ">=20"