reamkit 1.25.1 → 1.26.0

package/dist/esm/pdf-reader/shading.d.ts

@@ -25,3 +25,16 @@ export declare function buildShadingMap(file: PdfFile, page: PdfPage): Map<strin
  * @returns Name → fill alpha, for the states that state one below 1.
  */
 export declare function buildAlphaMap(file: PdfFile, page: PdfPage): Map<string, number>;
+/**
+ * §8.6 — how many components a colour space takes, and what they mean.
+ *
+ * `sc` / `scn` give bare numbers; only the space in force says whether `1 1 1`
+ * is white or something else, and whether a lone `1` is white (DeviceGray) or
+ * the colorant at full strength (Separation), which is usually black.
+ */
+export interface ColorSpaceInfo {
+    readonly kind: 'gray' | 'rgb' | 'cmyk' | 'tint';
+    readonly components: number;
+}
+/** The spaces a page's `/ColorSpace` resources name, by name. */
+export declare function buildColorSpaceMap(file: PdfFile, page: PdfPage): Map<string, ColorSpaceInfo>;

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

@@ -1,4 +1,4 @@
-import { PDF_NULL, PdfStream } from "../pdf/objects.js";
+import { PDF_NULL, PdfName, PdfStream } from "../pdf/objects.js";
 //#region src/pdf-reader/shading.ts
 /**
 * Resolve a page's `/Pattern` resources into gradient fills (E-PDF EP16c, ISO
@@ -179,5 +179,71 @@ function buildAlphaMap(file, page) {
 	}
 	return out;
 }
+/** The spaces a page's `/ColorSpace` resources name, by name. */
+function buildColorSpaceMap(file, page) {
+	const out = /* @__PURE__ */ new Map();
+	if (!page.resources) return out;
+	const spaces = file.get(page.resources, "ColorSpace");
+	if (!(spaces instanceof Map)) return out;
+	for (const [name, value] of spaces) {
+		const info = colorSpaceInfo(file, file.resolve(value));
+		if (info) out.set(name, info);
+	}
+	return out;
+}
+/** What a colour space object comes to, or `undefined` for one not read. */
+function colorSpaceInfo(file, cs) {
+	if (cs instanceof PdfName) return byFamily(cs.value, 0);
+	if (!Array.isArray(cs) || cs.length === 0) return void 0;
+	const head = file.resolve(cs[0]);
+	if (!(head instanceof PdfName)) return void 0;
+	if (head.value === "ICCBased") {
+		const stream = file.resolve(cs[1] ?? PDF_NULL);
+		const n = stream instanceof PdfStream ? file.get(stream.dict, "N") : void 0;
+		return byFamily("ICCBased", typeof n === "number" ? n : 3);
+	}
+	if (head.value === "Indexed") return;
+	if (head.value === "Separation" || head.value === "DeviceN") {
+		const names = file.resolve(cs[1] ?? PDF_NULL);
+		return {
+			kind: "tint",
+			components: head.value === "Separation" ? 1 : Array.isArray(names) ? names.length : 1
+		};
+	}
+	return byFamily(head.value, 0);
+}
+function byFamily(family, n) {
+	switch (family) {
+		case "DeviceGray":
+		case "CalGray":
+		case "G": return {
+			kind: "gray",
+			components: 1
+		};
+		case "DeviceRGB":
+		case "CalRGB":
+		case "Lab":
+		case "RGB": return {
+			kind: "rgb",
+			components: 3
+		};
+		case "DeviceCMYK":
+		case "CMYK": return {
+			kind: "cmyk",
+			components: 4
+		};
+		case "ICCBased": return n === 1 ? {
+			kind: "gray",
+			components: 1
+		} : n === 4 ? {
+			kind: "cmyk",
+			components: 4
+		} : {
+			kind: "rgb",
+			components: 3
+		};
+		default: return;
+	}
+}
 //#endregion
-export { buildAlphaMap, buildShadingMap };
+export { buildAlphaMap, buildColorSpaceMap, buildShadingMap };

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

@@ -3,7 +3,7 @@ import { FEATURES } from "../core/ir/features.js";
 import { IDENTITY, interpretContent, multiply } from "./content.js";
 import { collectPageAppearances } from "./annots.js";
 import { buildFonts } from "./text.js";
-import { buildAlphaMap, buildShadingMap } from "./shading.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).
@@ -17,14 +17,14 @@ import { buildAlphaMap, buildShadingMap } from "./shading.js";
 * The same walk `collectPageImages` already makes, with the same depth and
 * cycle guards, collecting paths instead of pictures.
 */
-function paintedVectors(file, page, shadings, alphas) {
+function paintedVectors(file, page, shadings, alphas, spaces) {
 	const out = [];
 	const visiting = /* @__PURE__ */ new Set();
 	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);
+		const result = interpretContent(content, buildFonts(file, resources), baseCtm, shadings, alphas, spaces);
 		const events = [
 			...result.vectors.map((vector) => ({
 				order: vector.order,
@@ -120,9 +120,10 @@ function collectPageVectors(file, page, occupied = []) {
 	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);
+	const raws = paintedVectors(file, page, shadings, alphas, spaces);
 	for (const raw of raws) {
 		if (out.length >= MAX_VECTORS) break;
 		const v = clipped(raw);

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

@@ -43,10 +43,6 @@ var RASTER_MEDIA = {
 		ext: "jpeg",
 		contentType: "image/jpeg"
 	},
-	jpeg2000: {
-		ext: "jp2",
-		contentType: "image/jp2"
-	},
 	gif: {
 		ext: "gif",
 		contentType: "image/gif"
@@ -56,6 +52,18 @@ var RASTER_MEDIA = {
 		contentType: "image/bmp"
 	}
 };
+/**
+* Why an image was not written, in terms of the image — for the loss report,
+* which used to say "image bytes missing" whatever the reason, including for
+* bytes that were all there and in a format the format does not admit.
+*/
+function imageRefusal(resource, state) {
+	const bytes = resource === void 0 ? void 0 : state.resources.get(resource);
+	if (!bytes) return "image bytes missing";
+	const format = detectImageFormat(bytes);
+	if (format === "jpeg2000") return "JPEG 2000 picture dropped — §15.2.14 admits no such image part, and no reader of a .docx shows one";
+	return `picture dropped — ${format ?? "an unrecognised"} is not an image part a .docx may carry`;
+}
 function mediaInfo(bytes) {
 	const raster = detectImageFormat(bytes);
 	if (raster) return RASTER_MEDIA[raster];
@@ -137,9 +145,7 @@ function writeDocx(flow) {
 		if (i === sections.length - 1) finalSectPr = sp;
 		else sectPrByClosingIndex.set(sec.endIndex - 1, sp);
 	});
-	flow.body.forEach((el, idx) => {
-		emitBlock(body, el, losses, state, docScope, sectPrByClosingIndex.get(idx));
-	});
+	emitBody(body, flow.body, losses, state, docScope, sectPrByClosingIndex);
 	if (finalSectPr) body.push(finalSectPr);
 	const documentXml = `<?xml version="1.0" encoding="UTF-8" standalone="yes"?><w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main" xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships"><w:body>${body.join("")}</w:body></w:document>`;
 	const numberingPart = flow.body.some((el) => el.kind === "paragraph" && el.paragraph.properties.numbering !== void 0) && flow.numbering ? {
@@ -365,6 +371,59 @@ var docxWriter = {
 	supports: new Set([FEATURES.text]),
 	write: (doc) => writeDocx(doc)
 };
+/**
+* The document body, with consecutive floating drawings gathered into one
+* carrier paragraph.
+*
+* §20.4.2.3 — a floating drawing is placed by its anchor and not by where its
+* paragraph lands, so the paragraph is only somewhere for the anchor to hang
+* from. Given one paragraph EACH, a page of placed artwork is that many
+* paragraphs of flow: 160F-2019.pdf's one-page form is 355 drawings, and Word
+* ran it to sixteen pages of empty lines with the artwork anchored to them.
+* Gathered, the page costs one paragraph however much is placed on it.
+*
+* A block that closes a section keeps its own paragraph — the section break
+* rides that paragraph's properties, and it must be the last one in the
+* section.
+*/
+function emitBody(out, blocks, losses, state, scope, sectPrByClosingIndex) {
+	let carried = [];
+	const flush = () => {
+		if (carried.length > 0) out.push(`<w:p>${carried.join("")}</w:p>`);
+		carried = [];
+	};
+	blocks.forEach((el, idx) => {
+		const closing = sectPrByClosingIndex.get(idx);
+		const anchored = closing === void 0 ? floatingDrawingRun(el, losses, state, scope) : void 0;
+		if (anchored !== void 0) {
+			carried.push(anchored);
+			return;
+		}
+		flush();
+		emitBlock(out, el, losses, state, scope, closing);
+	});
+	flush();
+}
+/**
+* The run holding one block's drawing, when that block states where on the page
+* it goes — `undefined` for everything that flows, which the ordinary block
+* emitter handles.
+*/
+function floatingDrawingRun(el, losses, state, scope) {
+	if (el.kind === "image" && el.image.float) {
+		const drawing = drawingXml(el.image.resource, el.image.width, el.image.height, el.image.altText, state, scope, el.image.float);
+		if (drawing === "") {
+			losses.push({
+				severity: "dropped",
+				feature: FEATURES.images,
+				detail: imageRefusal(el.image.resource, state)
+			});
+			return "";
+		}
+		return `<w:r>${drawing}</w:r>`;
+	}
+	if (el.kind === "shape" && el.shape.float) return `<w:r>${shapeDrawingXml(el.shape, losses, state, scope)}</w:r>`;
+}
 function emitBlock(out, el, losses, state, scope, closingSectPr) {
 	if (el.kind === "paragraph") {
 		out.push(paragraphXml(el.paragraph, state, scope, closingSectPr));
@@ -376,13 +435,13 @@ function emitBlock(out, el, losses, state, scope, closingSectPr) {
 		return;
 	}
 	if (el.kind === "image") {
-		const drawing = drawingXml(el.image.resource, el.image.width, el.image.height, el.image.altText, state, scope);
+		const drawing = drawingXml(el.image.resource, el.image.width, el.image.height, el.image.altText, state, scope, el.image.float);
 		if (drawing) out.push(`<w:p>${pPrWithSect(el.image.paragraphProperties, closingSectPr)}<w:r>${drawing}</w:r></w:p>`);
 		else {
 			losses.push({
 				severity: "dropped",
 				feature: FEATURES.images,
-				detail: "image bytes missing"
+				detail: imageRefusal(el.image.resource, state)
 			});
 			if (closingSectPr) out.push(`<w:p><w:pPr>${closingSectPr}</w:pPr></w:p>`);
 		}
@@ -425,14 +484,82 @@ function chartBlockXml(chart, state, scope, losses) {
 	const id = ++state.drawingSeq;
 	return `<w:drawing><wp:inline xmlns:wp="http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing"><wp:extent cx="${cx}" cy="${cy}"/><wp:docPr id="${id}" name="Chart ${id}"${chart.altText ? ` descr="${escapeAttr(chart.altText)}"` : ""}/><a:graphic xmlns:a="http://schemas.openxmlformats.org/drawingml/2006/main"><a:graphicData uri="http://schemas.openxmlformats.org/drawingml/2006/chart"><c:chart xmlns:c="http://schemas.openxmlformats.org/drawingml/2006/chart" xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships" r:id="${relId}"/></a:graphicData></a:graphic></wp:inline></w:drawing>`;
 }
-function drawingXml(resource, widthPt, heightPt, altText, state, scope) {
+/**
+* §20.4.2.3 — the frame a drawing is placed by: `wp:anchor` where the block
+* states where on the page it belongs, `wp:inline` where it belongs in the run
+* of text.
+*
+* Everything used to be inline. The document model has carried the anchor since
+* the reader learned to read one, and the PDF reconstruction fills it in for
+* every line and rule it lifts off a page — and all of it was thrown away here,
+* so a converted page came out as a column of drawings stacked in document
+* order. S2.pdf's two colour wheels marched down the left margin as eighteen
+* separate wedges; 160F-2019.pdf's one-page form ran to sixteen pages, because
+* each of its 355 drawings took a paragraph of its own in the flow.
+*
+* The attribute list is not decoration: CT_Anchor requires `simplePos`,
+* `relativeHeight`, `behindDoc`, `locked`, `layoutInCell` and `allowOverlap`,
+* and the children are a SEQUENCE — position, extent, wrap, docPr, graphic —
+* that Word refuses out of order.
+*
+* @param float The placement, or `undefined` for a drawing that flows.
+* @param cx    Width in EMU.
+* @param cy    Height in EMU.
+* @param head  The `wp:docPr` (and anything else) that precedes the graphic.
+* @param graphic The `a:graphic` element the frame carries.
+*/
+function drawingFrame(float, cx, cy, head, graphic, state, rotation60k) {
+	const WP_NS = " xmlns:wp=\"http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing\"";
+	if (!float) return `<w:drawing><wp:inline${WP_NS}><wp:extent cx="${cx}" cy="${cy}"/>${head}${graphic}</wp:inline></w:drawing>`;
+	const emu = (pt) => Number.isFinite(pt) ? Math.round((pt ?? 0) * EMU_PER_PT) : 0;
+	const d = float.wrapDist;
+	const dist = ` distT="${String(emu(d?.topPt))}" distB="${String(emu(d?.bottomPt))}" distL="${String(emu(d?.leftPt))}" distR="${String(emu(d?.rightPt))}"`;
+	const z = float.zOrder ?? ++state.drawingSeq;
+	const attrs = `${dist} simplePos="0" relativeHeight="${String(z)}" behindDoc="${float.behind ? "1" : "0"}" locked="0" layoutInCell="${float.inCell === false ? "0" : "1"}" allowOverlap="1"`;
+	const pos = (tag, p, fallback) => {
+		const from = p?.relativeFrom ?? fallback;
+		const body = p?.align !== void 0 ? `<wp:align>${p.align}</wp:align>` : `<wp:posOffset>${String(emu(p?.offsetPt))}</wp:posOffset>`;
+		return `<${tag} relativeFrom="${escapeAttr(from)}">${body}</${tag}>`;
+	};
+	const side = ` wrapText="${float.wrapSide ?? "bothSides"}"`;
+	const wrap = float.wrap === "square" ? `<wp:wrapSquare${side}/>` : float.wrap === "tight" ? `<wp:wrapTight${side}/>` : float.wrap === "through" ? `<wp:wrapThrough${side}/>` : float.wrap === "topAndBottom" || float.wrap === "notBeside" ? "<wp:wrapTopAndBottom/>" : "<wp:wrapNone/>";
+	return `<w:drawing><wp:anchor${WP_NS}${attrs}><wp:simplePos x="0" y="0"/>` + pos("wp:positionH", float.posH, "column") + pos("wp:positionV", float.posV, "paragraph") + `<wp:extent cx="${cx}" cy="${cy}"/>` + effectExtentXml(cx, cy, rotation60k) + wrap + head + graphic + "</wp:anchor></w:drawing>";
+}
+/**
+* §20.4.2.3 `wp:effectExtent` — how far the drawing reaches PAST the extent.
+*
+* `wp:extent` is the shape's own size, unrotated: fdo75722-dml.docx states
+* 1964690 × 1240790 there and repeats it in `a:ext`, then carries the turn's
+* overhang here (l 198120, t 433705, r 199390, b 430530). Written as zeroes —
+* which they always were — a renderer reserves the flat box for a shape that
+* draws across the turned one, and 160F-2019.pdf's "Nature", set on its side
+* down the middle of a column, came back lying flat across it.
+*
+* A box `w × h` turned by θ about its centre spans `|w·cosθ| + |h·sinθ|`
+* across and `|w·sinθ| + |h·cosθ|` down; half of each growth hangs off each
+* side. A turn that makes the box no wider hangs off by nothing.
+*/
+function effectExtentXml(cx, cy, rotation60k) {
+	const none = "<wp:effectExtent l=\"0\" t=\"0\" r=\"0\" b=\"0\"/>";
+	if (rotation60k === void 0 || !Number.isFinite(rotation60k) || rotation60k % 216e5 === 0) return none;
+	const rad = rotation60k / 6e4 * Math.PI / 180;
+	const c = Math.abs(Math.cos(rad));
+	const s = Math.abs(Math.sin(rad));
+	const overhang = (span, turned) => Math.max(0, Math.round((turned - span) / 2));
+	const h = overhang(cx, cx * c + cy * s);
+	const v = overhang(cy, cx * s + cy * c);
+	return h === 0 && v === 0 ? none : `<wp:effectExtent l="${String(h)}" t="${String(v)}" r="${String(h)}" b="${String(v)}"/>`;
+}
+function drawingXml(resource, widthPt, heightPt, altText, state, scope, float) {
 	if (resource === void 0) return "";
 	const relId = mediaRelId(resource, state, scope);
 	if (relId === void 0) return "";
 	const cx = Math.round(widthPt * EMU_PER_PT);
 	const cy = Math.round(heightPt * EMU_PER_PT);
 	const id = ++state.drawingSeq;
-	return `<w:drawing><wp:inline xmlns:wp="http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing"><wp:extent cx="${cx}" cy="${cy}"/><wp:docPr id="${id}" name="Image ${id}"${altText ? ` descr="${escapeAttr(altText)}"` : ""}/><a:graphic xmlns:a="http://schemas.openxmlformats.org/drawingml/2006/main"><a:graphicData uri="http://schemas.openxmlformats.org/drawingml/2006/picture"><pic:pic xmlns:pic="http://schemas.openxmlformats.org/drawingml/2006/picture"><pic:nvPicPr><pic:cNvPr id="${id}" name="Image ${id}"/><pic:cNvPicPr/></pic:nvPicPr><pic:blipFill><a:blip r:embed="${relId}"/><a:stretch><a:fillRect/></a:stretch></pic:blipFill><pic:spPr><a:xfrm><a:off x="0" y="0"/><a:ext cx="${cx}" cy="${cy}"/></a:xfrm><a:prstGeom prst="rect"><a:avLst/></a:prstGeom></pic:spPr></pic:pic></a:graphicData></a:graphic></wp:inline></w:drawing>`;
+	const descr = altText ? ` descr="${escapeAttr(altText)}"` : "";
+	const graphic = `<a:graphic xmlns:a="http://schemas.openxmlformats.org/drawingml/2006/main"><a:graphicData uri="http://schemas.openxmlformats.org/drawingml/2006/picture"><pic:pic xmlns:pic="http://schemas.openxmlformats.org/drawingml/2006/picture"><pic:nvPicPr><pic:cNvPr id="${id}" name="Image ${id}"/><pic:cNvPicPr/></pic:nvPicPr><pic:blipFill><a:blip r:embed="${relId}"/><a:stretch><a:fillRect/></a:stretch></pic:blipFill><pic:spPr><a:xfrm><a:off x="0" y="0"/><a:ext cx="${cx}" cy="${cy}"/></a:xfrm><a:prstGeom prst="rect"><a:avLst/></a:prstGeom></pic:spPr></pic:pic></a:graphicData></a:graphic>`;
+	return drawingFrame(float, cx, cy, `<wp:docPr id="${id}" name="Image ${id}"${descr}/>`, graphic, state);
 }
 var WPS_URI = "http://schemas.microsoft.com/office/word/2010/wordprocessingShape";
 function shapeDrawingXml(shape, losses, state, scope) {
@@ -442,7 +569,8 @@ function shapeDrawingXml(shape, losses, state, scope) {
 	const descr = shape.altText ? ` descr="${escapeAttr(shape.altText)}"` : "";
 	const spPr = `<wps:spPr>${xfrmXml(shape.transform, cx, cy)}${geomXml(shape.geometry)}${fillXml(shape.fill)}${shape.line ? lineXml(shape.line) : ""}</wps:spPr>`;
 	const txbx = shape.text ? txbxXml(shape.text, losses, state, scope) : "";
-	return `<w:drawing><wp:inline xmlns:wp="http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing"><wp:extent cx="${cx}" cy="${cy}"/><wp:docPr id="${id}" name="Shape ${id}"${descr}/><a:graphic xmlns:a="http://schemas.openxmlformats.org/drawingml/2006/main"><a:graphicData uri="${WPS_URI}"><wps:wsp xmlns:wps="http://schemas.microsoft.com/office/word/2010/wordprocessingShape"><wps:cNvSpPr/>` + spPr + txbx + bodyPrXml(shape.text) + "</wps:wsp></a:graphicData></a:graphic></wp:inline></w:drawing>";
+	const graphic = `<a:graphic xmlns:a="http://schemas.openxmlformats.org/drawingml/2006/main"><a:graphicData uri="${WPS_URI}"><wps:wsp xmlns:wps="http://schemas.microsoft.com/office/word/2010/wordprocessingShape"><wps:cNvSpPr/>` + spPr + txbx + bodyPrXml(shape.text) + "</wps:wsp></a:graphicData></a:graphic>";
+	return drawingFrame(shape.float, cx, cy, `<wp:docPr id="${id}" name="Shape ${id}"${descr}/>`, graphic, state, shape.transform?.rotation60k);
 }
 function xfrmXml(t, cx, cy) {
 	return `<a:xfrm${t?.rotation60k !== void 0 ? ` rot="${t.rotation60k}"` : ""}${t?.flipH ? " flipH=\"1\"" : ""}${t?.flipV ? " flipV=\"1\"" : ""}><a:off x="0" y="0"/><a:ext cx="${cx}" cy="${cy}"/></a:xfrm>`;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "reamkit",
-  "version": "1.25.1",
+  "version": "1.26.0",
   "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>",