reamkit 1.26.0 → 1.28.0

package/dist/esm/pdf-reader/flow-build.js

@@ -27,11 +27,11 @@ function paragraphBlock(text, outlineLevel) {
 * survives as its own run) and squashing whitespace. With no hrefs this
 * collapses to a single run — the same shape {@link paragraphBlock} produces.
 */
-function paragraphFromRuns(spans, outlineLevel) {
+function paragraphFromRuns(spans, outlineLevel, placement) {
 	const merged = [];
 	for (const s of spans) {
 		const last = merged[merged.length - 1];
-		if (last && last.href === s.href && last.sizePt === s.sizePt && last.colorHex === s.colorHex && last.fontName === s.fontName && last.outline?.colorHex === s.outline?.colorHex && last.outline?.widthPt === s.outline?.widthPt && last.bold === s.bold && last.italic === s.italic) last.text += s.text;
+		if (last && last.href === s.href && last.sizePt === s.sizePt && last.colorHex === s.colorHex && last.fontName === s.fontName && last.outline?.colorHex === s.outline?.colorHex && last.outline?.widthPt === s.outline?.widthPt && last.bold === s.bold && last.italic === s.italic && last.script === s.script && sameMarkup(last.markup, s.markup)) last.text += s.text;
 		else merged.push({
 			text: s.text,
 			...s.href !== void 0 ? { href: s.href } : {},
@@ -40,12 +40,14 @@ function paragraphFromRuns(spans, outlineLevel) {
 			...s.fontName !== void 0 ? { fontName: s.fontName } : {},
 			...s.outline !== void 0 ? { outline: s.outline } : {},
 			...s.bold !== void 0 ? { bold: s.bold } : {},
-			...s.italic !== void 0 ? { italic: s.italic } : {}
+			...s.italic !== void 0 ? { italic: s.italic } : {},
+			...s.script !== void 0 ? { script: s.script } : {},
+			...s.markup !== void 0 ? { markup: s.markup } : {}
 		});
 	}
 	const runs = merged.map((m) => ({
 		...m,
-		text: m.text.replace(/\s+/g, " ")
+		text: m.text.replace(/[^\S\t]+/gu, " ")
 	})).filter((m) => m.text.length > 0);
 	if (runs.length > 0) {
 		runs[0].text = runs[0].text.replace(/^ /, "");
@@ -54,7 +56,10 @@ function paragraphFromRuns(spans, outlineLevel) {
 	return {
 		kind: "paragraph",
 		paragraph: {
-			properties: outlineLevel !== void 0 ? { outlineLevel } : {},
+			properties: {
+				...outlineLevel !== void 0 ? { outlineLevel } : {},
+				...placement
+			},
 			runs: runs.filter((r) => r.text.length > 0).map((r) => ({
 				text: r.text,
 				properties: {
@@ -63,22 +68,32 @@ function paragraphFromRuns(spans, outlineLevel) {
 					...r.fontName !== void 0 ? { fontFamily: { ascii: r.fontName } } : {},
 					...r.outline !== void 0 ? { textOutline: r.outline } : {},
 					...r.bold ? { bold: true } : {},
-					...r.italic ? { italic: true } : {}
+					...r.italic ? { italic: true } : {},
+					...r.script !== void 0 ? { verticalAlign: r.script } : {},
+					...r.markup?.highlightHex !== void 0 ? { shadingColorHex: r.markup.highlightHex } : {},
+					...r.markup?.underline !== void 0 ? { underline: r.markup.underline } : {},
+					...r.markup?.underlineHex !== void 0 ? { underlineColorHex: r.markup.underlineHex } : {},
+					...r.markup?.strike === true ? { strike: true } : {}
 				},
 				...r.href ? { href: r.href } : {}
 			}))
 		}
 	};
 }
+/** Whether two runs are marked the same way, so they may join into one. */
+function sameMarkup(a, b) {
+	return a?.highlightHex === b?.highlightHex && a?.underline === b?.underline && a?.underlineHex === b?.underlineHex && a?.strike === b?.strike;
+}
 /**
 * Store a {@link PdfImage}'s bytes (content-addressed dedup) and build the image
 * {@link BodyElement} that references them, sized in points from the placement
 * CTM. `alt` becomes the block's alt text when given.
 */
-function imageBlock(image, resources, alt, frame, zOrder) {
+function imageBlock(image, resources, alt, frame, zOrder, behind = false) {
 	const resource = resources.put(image.bytes);
 	const float = frame !== void 0 ? {
 		wrap: "none",
+		...behind ? { behind: true } : {},
 		...zOrder !== void 0 ? { zOrder } : {},
 		posH: {
 			relativeFrom: "page",
@@ -96,6 +111,9 @@ function imageBlock(image, resources, alt, frame, zOrder) {
 			resource,
 			width: pt(image.widthPt),
 			height: pt(image.heightPt),
+			...image.rotationDeg !== void 0 ? { rotation60k: Math.round(-image.rotationDeg * 6e4) } : {},
+			...image.crop ? { crop: image.crop } : {},
+			...image.alpha !== void 0 ? { alpha: image.alpha } : {},
 			paragraphProperties: {},
 			...alt ? { altText: alt } : {}
 		}
@@ -172,7 +190,7 @@ function dedupeLosses(losses) {
 * a paragraph: 22060_A1_01_Plans.pdf is one A3 sheet of vectors, and stacking
 * its forty-nine paths one under another spilled it onto a second page.
 */
-function shapeBlock(v, frame, zOrder) {
+function shapeBlock(v, frame, zOrder, behind = false) {
 	const w = v.maxX - v.minX;
 	const h = v.maxY - v.minY;
 	const fx = (x) => x - v.minX;
@@ -222,6 +240,7 @@ function shapeBlock(v, frame, zOrder) {
 	} : void 0;
 	const float = frame !== void 0 ? {
 		wrap: "none",
+		...behind || v.darkens === true ? { behind: true } : {},
 		...zOrder !== void 0 ? { zOrder } : {},
 		posH: {
 			relativeFrom: "page",
@@ -287,17 +306,98 @@ function sectionFromPdfPages(pages) {
 	};
 }
 /**
+* The margins the SOURCE used, measured off where its words actually sit.
+*
+* A PDF states none — text is placed anywhere on the MediaBox — so the reader
+* used to leave them at zero rather than invent an inch. But the words
+* themselves say where the margin was: the leftmost glyph on the page is the
+* left margin, and reflowing inside it keeps the measure the author set instead
+* of running the text from edge to edge.
+*
+* Measured on the MEDIAN page rather than the extreme one, so a single full-
+* bleed rule or a page number in the corner does not collapse the margin for
+* the whole document, and clamped so a strange page cannot leave no text area
+* at all.
+*
+* @param section  The section the page box gave, or `undefined`.
+* @param shown    Each page as it is shown, for its own width and height.
+* @param pageRuns Each page's runs, already placed on the shown page.
+* @returns The section with measured margins, or `section` when nothing is
+*          measurable.
+*/
+/** What the measure gives back, so the widest line still fits when re-set. */
+var SLACK = .01;
+/** How much of the sheet a margin down the page may take. */
+var DEEPEST_MARGIN = .5;
+/** How far a face's ascender stands above its baseline, as a fraction of the size. */
+var ASCENDER = .8;
+/** And its descender below — the two together are a little over one em. */
+var DESCENDER = .22;
+function withMeasuredMargins(section, shown, pageRuns) {
+	if (!section?.pageSize) return section;
+	const width = section.pageSize.width;
+	const height = section.pageSize.height;
+	const lefts = [];
+	const rights = [];
+	const tops = [];
+	const bottoms = [];
+	pageRuns.forEach((runs, i) => {
+		const page = shown[i];
+		if (!page || runs.length === 0) return;
+		let minX = Infinity;
+		let maxX = -Infinity;
+		let minY = Infinity;
+		let maxY = -Infinity;
+		let topSize = 0;
+		let bottomSize = 0;
+		for (const r of runs) {
+			if (!Number.isFinite(r.x) || !Number.isFinite(r.y)) continue;
+			minX = Math.min(minX, r.x);
+			maxX = Math.max(maxX, r.endX);
+			if (r.y < minY) {
+				minY = r.y;
+				bottomSize = r.fontSizePt;
+			}
+			if (r.y > maxY) {
+				maxY = r.y;
+				topSize = r.fontSizePt;
+			}
+		}
+		if (!Number.isFinite(minX) || !Number.isFinite(minY)) return;
+		lefts.push(minX);
+		rights.push(page.width - maxX);
+		tops.push(page.height - maxY - topSize * ASCENDER);
+		bottoms.push(minY - bottomSize * DESCENDER);
+	});
+	if (lefts.length === 0) return section;
+	const median = (xs) => {
+		const s = [...xs].sort((a, b) => a - b);
+		return s[Math.floor(s.length / 2)] ?? 0;
+	};
+	const clamp = (v, span, most = 1 / 3) => pt(Math.max(0, Math.min(v, span * most)));
+	return {
+		...section,
+		margins: {
+			left: clamp(median(lefts), width),
+			right: clamp(median(rights) - width * SLACK, width),
+			top: clamp(Math.min(...tops), height, DEEPEST_MARGIN),
+			bottom: clamp(Math.min(...bottoms) - height * SLACK, height, DEEPEST_MARGIN)
+		}
+	};
+}
+/**
 * Assemble the final {@link FlowDoc} for a reconstruction: the body elements
 * with their styles resolved against the empty style sheet, the lifted-image
 * resource store, and the optional page {@link SectionProperties}. Shared by
 * both reconstruction paths (the tagged fast-path EP3 and the heuristic layout
 * path EP4).
 */
-function buildFlowDoc(body, resources = new ResourceStore(), section, embeddedFonts) {
+function buildFlowDoc(body, resources = new ResourceStore(), section, embeddedFonts, sections = [], headersFooters) {
 	return {
 		kind: "flow",
 		body: resolveBodyStyles([...body], EMPTY_STYLE_SHEET),
-		sections: [],
+		sections,
+		...headersFooters && headersFooters.size > 0 ? { headersFooters } : {},
 		...section ? { section } : {},
 		...embeddedFonts && embeddedFonts.size > 0 ? { embeddedFonts } : {},
 		styles: EMPTY_STYLE_SHEET,
@@ -305,4 +405,4 @@ function buildFlowDoc(body, resources = new ResourceStore(), section, embeddedFo
 	};
 }
 //#endregion
-export { buildFlowDoc, dedupeLosses, imageBlock, paragraphBlock, paragraphFromRuns, positionedText, sectionFromPdfPages, shapeBlock };
+export { buildFlowDoc, dedupeLosses, imageBlock, paragraphBlock, paragraphFromRuns, positionedText, sectionFromPdfPages, shapeBlock, withMeasuredMargins };