reamkit 1.26.0 → 1.27.0

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

@@ -616,8 +616,8 @@ function geomXml(g) {
 				case "close": return "<a:close/>";
 			}
 		}).join("");
-		const w = c1(g.custom.pathWidth);
-		const h = c1(g.custom.pathHeight);
+		const w = Math.max(1, c1(g.custom.pathWidth));
+		const h = Math.max(1, c1(g.custom.pathHeight));
 		return `<a:custGeom><a:avLst/><a:gdLst/><a:rect l="0" t="0" r="${w}" b="${h}"/><a:pathLst><a:path w="${w}" h="${h}">${cmds}</a:path></a:pathLst></a:custGeom>`;
 	}
 	const gds = g.adjust && g.adjust.size > 0 ? [...g.adjust].map(([n, v]) => `<a:gd name="${escapeAttr(n)}" fmla="val ${v}"/>`).join("") : "";
@@ -826,19 +826,27 @@ function runXml(run, state, scope) {
 }
 function rPrXml(r) {
 	const out = [];
+	const fonts = rFontsXml(r.fontFamily);
+	if (fonts) out.push(fonts);
 	if (r.bold !== DEFAULT_RUN.bold) out.push(toggle("w:b", r.bold));
 	if (r.italic !== DEFAULT_RUN.italic) out.push(toggle("w:i", r.italic));
 	if (r.strike !== DEFAULT_RUN.strike) out.push(toggle("w:strike", r.strike));
-	if (r.underline !== DEFAULT_RUN.underline) out.push(`<w:u w:val="${r.underline}"/>`);
-	const fonts = rFontsXml(r.fontFamily);
-	if (fonts) out.push(fonts);
-	if (r.fontSizePt !== DEFAULT_RUN.fontSizePt) out.push(`<w:sz w:val="${Math.round(r.fontSizePt * 2)}"/>`);
 	if (r.colorHex !== DEFAULT_RUN.colorHex) out.push(`<w:color w:val="${r.colorHex}"/>`);
+	if (r.fontSizePt !== DEFAULT_RUN.fontSizePt) out.push(`<w:sz w:val="${Math.round(r.fontSizePt * 2)}"/>`);
+	if (r.underline !== DEFAULT_RUN.underline) out.push(underlineXml(r));
+	if (r.shadingColorHex !== void 0) out.push(runShdXml(r.shadingColorHex));
 	if (r.verticalAlign !== DEFAULT_RUN.verticalAlign) out.push(`<w:vertAlign w:val="${r.verticalAlign}"/>`);
 	if (r.rtl !== DEFAULT_RUN.rtl) out.push(toggle("w:rtl", r.rtl));
 	if (r.lang !== void 0) out.push(`<w:lang w:val="${escapeAttr(r.lang)}"/>`);
 	return out.length > 0 ? `<w:rPr>${out.join("")}</w:rPr>` : "";
 }
+function underlineXml(r) {
+	const color = r.underlineColorHex !== void 0 ? ` w:color="${r.underlineColorHex}"` : "";
+	return `<w:u w:val="${r.underline}"${color}/>`;
+}
+function runShdXml(fill) {
+	return `<w:shd w:val="clear" w:color="auto" w:fill="${fill}"/>`;
+}
 function pPrBody(p) {
 	const out = [];
 	if (p.numbering) out.push(`<w:numPr><w:ilvl w:val="${p.numbering.ilvl}"/><w:numId w:val="${escapeAttr(p.numbering.numId)}"/></w:numPr>`);
@@ -942,14 +950,18 @@ function rawParaPrXml(p) {
 }
 function rawRunPrXml(r) {
 	const out = [];
+	const fonts = r.fontFamily ? rawRFontsXml(r.fontFamily) : "";
+	if (fonts) out.push(fonts);
 	if (r.bold !== void 0) out.push(toggle("w:b", r.bold));
 	if (r.italic !== void 0) out.push(toggle("w:i", r.italic));
 	if (r.strike !== void 0) out.push(toggle("w:strike", r.strike));
-	if (r.underline !== void 0) out.push(`<w:u w:val="${r.underline}"/>`);
-	const fonts = r.fontFamily ? rawRFontsXml(r.fontFamily) : "";
-	if (fonts) out.push(fonts);
-	if (r.fontSizePt !== void 0) out.push(`<w:sz w:val="${Math.round(r.fontSizePt * 2)}"/>`);
 	if (r.colorHex !== void 0) out.push(`<w:color w:val="${r.colorHex}"/>`);
+	if (r.fontSizePt !== void 0) out.push(`<w:sz w:val="${Math.round(r.fontSizePt * 2)}"/>`);
+	if (r.underline !== void 0) out.push(underlineXml({
+		...r,
+		underline: r.underline
+	}));
+	if (r.shadingColorHex !== void 0) out.push(runShdXml(r.shadingColorHex));
 	if (r.verticalAlign !== void 0) out.push(`<w:vertAlign w:val="${r.verticalAlign}"/>`);
 	return out.length > 0 ? `<w:rPr>${out.join("")}</w:rPr>` : "";
 }
@@ -987,8 +999,58 @@ function colsXml(cols) {
 	return `<w:cols w:num="${cols.count}" w:space="${twips(cols.spacePt)}"/>`;
 }
 var twips = (pt) => Math.round(pt * 20);
+/**
+* XML 1.0 §2.2 — the characters a document may contain at all.
+*
+* Everything below U+0020 except tab, newline and return is FORBIDDEN, and no
+* escape exists for them: `&#2;` is as ill-formed as the byte. A package with
+* one in it is not a document — LibreOffice says "source file could not be
+* loaded" and Word says nothing useful either.
+*
+* They reach here from the PDF reader. A subset font that states neither a
+* `/ToUnicode` nor an `/Encoding /Differences` says nothing about what its
+* codes mean, and the codes of a subset start at 1, 2, 3 — so the last-resort
+* Latin-1 reading turns a page of text into control characters. Reading it
+* better is `font.ts`'s business; what this must guarantee is that nothing the
+* reader believes can produce a package that will not open.
+*
+* A lone surrogate is the same case: a half of a pair is not a character.
+*/
+/**
+* XML 1.0 §2.2 — whether a code point may appear in a document at all.
+*
+* Everything below U+0020 except tab, newline and return is forbidden, and no
+* escape exists for them: `&#2;` is as ill-formed as the byte itself. So is a
+* lone surrogate — half a pair is not a character — and so are U+FFFE/U+FFFF.
+*/
+function xmlAllows(cp) {
+	if (cp === 9 || cp === 10 || cp === 13) return true;
+	if (cp < 32) return false;
+	if (cp >= 55296 && cp <= 57343) return false;
+	if (cp === 65534 || cp === 65535) return false;
+	return cp <= 1114111;
+}
+/**
+* Escape text for XML, and drop what XML cannot carry.
+*
+* The dropping is not fussiness: a package with one control character in it is
+* not a document. LibreOffice answers "source file could not be loaded" and
+* Word refuses it too, so the whole conversion is lost over one byte.
+*
+* They reach here from the PDF reader. A subset font that states neither a
+* `/ToUnicode` nor an `/Encoding /Differences` says nothing about what its
+* codes mean, and a subset's codes start at 1, 2, 3 — so the last-resort
+* Latin-1 reading turns a page of prose into control characters. Reading such
+* a font better is `font.ts`'s business; what this guarantees is that nothing
+* the reader believes can produce a package that will not open.
+*/
 function escapeXml(s) {
-	return s.replace(/&/g, "&amp;").replace(/</g, "&lt;").replace(/>/g, "&gt;");
+	let out = "";
+	for (const ch of s) {
+		if (!xmlAllows(ch.codePointAt(0) ?? 0)) continue;
+		out += ch === "&" ? "&amp;" : ch === "<" ? "&lt;" : ch === ">" ? "&gt;" : ch;
+	}
+	return out;
 }
 function escapeAttr(s) {
 	return escapeXml(s).replace(/"/g, "&quot;");

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "reamkit",
-  "version": "1.26.0",
+  "version": "1.27.0",
   "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>",