reamkit 1.26.0 → 1.28.0

package/dist/esm/pdf-reader/vector.js

@@ -1,35 +1,66 @@
 import { PDF_NULL, PdfName, PdfStream } from "../pdf/objects.js";
 import { FEATURES } from "../core/ir/features.js";
+import { buildAlphaMap, buildColorSpaceMap, buildShadingMap, gradientShading, shadingTypeOf } from "./shading.js";
 import { IDENTITY, interpretContent, multiply } from "./content.js";
 import { collectPageAppearances } from "./annots.js";
+import { hiddenProperties, hiddenXObject } from "./optional-content.js";
 import { buildFonts } from "./text.js";
-import { buildAlphaMap, buildColorSpaceMap, buildShadingMap } from "./shading.js";
 //#region src/pdf-reader/vector.ts
-/**
-* Every vector the page paints, its FORM XOBJECTS included (§8.8).
-*
-* A `Do` of a form is a call: its content stream draws in the caller's space
-* through the form's own `/Matrix`. Interpreting the page stream alone reads
-* only what the page drew directly, and a document that puts its artwork in
-* forms — as every CAD and drawing producer does — comes back with none of it.
-* 22060_A1_01_Plans.pdf holds eleven, and its floor plans were simply absent.
-*
-* The same walk `collectPageImages` already makes, with the same depth and
-* cycle guards, collecting paths instead of pictures.
-*/
-function paintedVectors(file, page, shadings, alphas, spaces) {
+function paintedVectors(file, page) {
 	const out = [];
+	let bareShadings = 0;
 	const visiting = /* @__PURE__ */ new Set();
+	const stateCache = /* @__PURE__ */ new Map();
+	const mapsOf = (resources) => {
+		const had = stateCache.get(resources);
+		if (had) return had;
+		const made = {
+			shadings: buildShadingMap(file, resources),
+			alphas: buildAlphaMap(file, resources),
+			spaces: buildColorSpaceMap(file, resources)
+		};
+		stateCache.set(resources, made);
+		return made;
+	};
 	const walk = (resources, content, baseCtm, depth, prefix) => {
 		if (out.length >= MAX_VECTORS) return;
 		const xobjects = resources ? file.get(resources, "XObject") : PDF_NULL;
 		const xobjDict = xobjects instanceof Map ? xobjects : void 0;
-		const result = interpretContent(content, buildFonts(file, resources), baseCtm, shadings, alphas, spaces);
+		const maps = mapsOf(resources);
+		const result = interpretContent(content, buildFonts(file, resources), baseCtm, maps.shadings, maps.alphas, maps.spaces, hiddenProperties(file, resources));
+		for (const paint of result.shadings) {
+			if (out.length >= MAX_VECTORS) break;
+			const dict = resources ? file.get(resources, "Shading") : PDF_NULL;
+			const found = dict instanceof Map ? file.resolve(dict.get(paint.name) ?? PDF_NULL) : PDF_NULL;
+			const sh = found instanceof PdfStream ? found.dict : found instanceof Map ? found : void 0;
+			if (sh && shadingTypeOf(file, sh) === 1) continue;
+			const gradient = sh ? gradientShading(file, sh) : void 0;
+			if (paint.masked) {
+				bareShadings++;
+				continue;
+			}
+			if (!sh || !gradient || !paint.clip) {
+				bareShadings++;
+				continue;
+			}
+			out.push({
+				order: paint.order,
+				segs: paint.clip.segs,
+				gradient,
+				...paint.alpha !== void 0 ? { alpha: paint.alpha } : {},
+				...paint.darkens ? { darkens: true } : {},
+				orderKey: [...prefix, paint.order]
+			});
+		}
 		const events = [
 			...result.vectors.map((vector) => ({
 				order: vector.order,
 				vector
 			})),
+			...result.outlines.map((vector) => ({
+				order: vector.order,
+				vector
+			})),
 			...result.images.map((xobject) => ({
 				order: xobject.order,
 				xobject
@@ -60,6 +91,7 @@ function paintedVectors(file, page, shadings, alphas, spaces) {
 			if (depth >= MAX_FORM_DEPTH) continue;
 			const stream = xobjDict ? file.resolve(xobjDict.get(placement.name) ?? PDF_NULL) : PDF_NULL;
 			if (!(stream instanceof PdfStream) || visiting.has(stream)) continue;
+			if (hiddenXObject(file, stream)) continue;
 			const subtype = file.get(stream.dict, "Subtype");
 			if (!(subtype instanceof PdfName) || subtype.value !== "Form") continue;
 			visiting.add(stream);
@@ -72,7 +104,10 @@ function paintedVectors(file, page, shadings, alphas, spaces) {
 	collectPageAppearances(file, page).forEach((appearance, index) => {
 		walk(appearance.resources ?? page.resources, file.streamData(appearance.stream), appearance.ctm, 1, [Number.MAX_SAFE_INTEGER, index]);
 	});
-	return out;
+	return {
+		placements: out,
+		bareShadings
+	};
 }
 /** §8.10.2 `/Matrix` — the form's own space, composed onto the placement CTM. */
 function formMatrix(file, dict) {
@@ -116,14 +151,24 @@ var MAX_FORM_DEPTH = 12;
 * clutter. Clips and the bare `sh` operator are not captured (a documented loss).
 */
 function collectPageVectors(file, page, occupied = []) {
-	const [px0, py0, px1, py1] = page.mediaBox;
+	const [px0, py0, px1, py1] = page.cropBox;
 	const pageArea = Math.max(1, Math.abs((px1 - px0) * (py1 - py0)));
-	const shadings = buildShadingMap(file, page);
-	const alphas = buildAlphaMap(file, page);
-	const spaces = buildColorSpaceMap(file, page);
 	const out = [];
 	const painted = [...occupied];
-	const raws = paintedVectors(file, page, shadings, alphas, spaces);
+	const lifted = paintedVectors(file, page);
+	const raws = lifted.placements;
+	const losses = [];
+	const bareShadings = lifted.bareShadings;
+	const asked = /* @__PURE__ */ new Set();
+	for (const raw of raws) {
+		if (raw.masked === true) asked.add("PDF soft mask (/SMask in the graphics state) is not applied; the shape is drawn at full opacity throughout");
+		if (raw.blend !== void 0) asked.add(`PDF blend mode /${raw.blend} is not performed; the shape is drawn over what it was to blend with`);
+	}
+	for (const detail of asked) losses.push({
+		severity: "degraded",
+		feature: FEATURES.images,
+		detail
+	});
 	for (const raw of raws) {
 		if (out.length >= MAX_VECTORS) break;
 		const v = clipped(raw);
@@ -139,7 +184,7 @@ function collectPageVectors(file, page, occupied = []) {
 		const short = Math.min(w, h);
 		const isBox = short >= MIN_SIDE && area >= MIN_AREA;
 		const isRule = long >= MIN_RULE_LEN && short > 0;
-		const filled = (v.gradient !== void 0 || solidFill) && (isBox || isRule) && area <= .85 * pageArea;
+		const filled = (v.gradient !== void 0 || solidFill) && (isBox || isRule || v.glyph === true);
 		const stroked = v.strokeHex !== void 0 && (v.strokeHex !== "FFFFFF" || painted.some((box) => overlaps(box, b))) && Math.max(w, h) >= MIN_STROKE_LEN && area <= .85 * pageArea;
 		if (!filled && !stroked) continue;
 		painted.push(b);
@@ -148,6 +193,7 @@ function collectPageVectors(file, page, occupied = []) {
 			segs: v.segs,
 			...filled ? v.gradient ? { gradient: v.gradient } : v.fillHex !== void 0 ? { fillHex: v.fillHex } : {} : {},
 			...filled && v.alpha !== void 0 ? { alpha: v.alpha } : {},
+			...filled && v.darkens === true ? { darkens: true } : {},
 			...stroked ? {
 				strokeHex: v.strokeHex,
 				...v.lineWidth !== void 0 ? { lineWidth: v.lineWidth } : {}
@@ -156,13 +202,19 @@ function collectPageVectors(file, page, occupied = []) {
 			...v.mcid !== void 0 ? { mcid: v.mcid } : {}
 		});
 	}
+	if (out.length >= MAX_VECTORS || raws.length >= MAX_VECTORS) losses.push({
+		severity: "dropped",
+		feature: FEATURES.shapes,
+		detail: `page carries more than ${String(MAX_VECTORS)} painted paths; the rest were not read`
+	});
+	if (bareShadings > 0) losses.push({
+		severity: "dropped",
+		feature: FEATURES.images,
+		detail: `${String(bareShadings)} bare-shading (sh) region${bareShadings === 1 ? "" : "s"} painted the clip rather than a path, and ${bareShadings === 1 ? "was" : "were"} not reconstructed`
+	});
 	return {
 		vectors: out,
-		losses: out.length >= MAX_VECTORS || raws.length >= MAX_VECTORS ? [{
-			severity: "dropped",
-			feature: FEATURES.shapes,
-			detail: `page carries more than ${String(MAX_VECTORS)} painted paths; the rest were not read`
-		}] : []
+		losses
 	};
 }
 /** Whether two boxes share any area at all. */

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

@@ -262,7 +262,10 @@ function parseSectPrNode(sectPr) {
 		const type = poAttr(child, "type") ?? "default";
 		const pitch = poIntAttr(child, "linePitch");
 		if ((type === "lines" || type === "linesAndChars") && pitch !== void 0 && pitch > 0) gridLinePitchPt = twipsToPt(pitch);
-	} else if (poIs(child, "w:type")) sectionStart = poAttr(child, "val") === "continuous" ? "continuous" : "nextPage";
+	} else if (poIs(child, "w:type")) {
+		const val = poAttr(child, "val");
+		sectionStart = val === "continuous" || val === "oddPage" || val === "evenPage" ? val : "nextPage";
+	}
 	return {
 		pageSize,
 		...margins ? { margins } : {},

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>` : "";
 }
@@ -964,6 +976,7 @@ function sectPrXml(s, hf) {
 	const parts = [];
 	for (const h of hf.headers) parts.push(`<w:headerReference w:type="${h.type}" r:id="${h.relId}"/>`);
 	for (const f of hf.footers) parts.push(`<w:footerReference w:type="${f.type}" r:id="${f.relId}"/>`);
+	if (s.sectionStart !== void 0 && s.sectionStart !== "nextPage") parts.push(`<w:type w:val="${s.sectionStart}"/>`);
 	if (s.pageSize) {
 		const orient = s.pageSize.orientation === "landscape" ? " w:orient=\"landscape\"" : "";
 		parts.push(`<w:pgSz w:w="${twips(s.pageSize.width)}" w:h="${twips(s.pageSize.height)}"${orient}/>`);
@@ -987,8 +1000,58 @@ 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": "1.28.0",
   "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>",