pdf-codec 3.1.3 → 3.2.0

package/dist/interpret.js

@@ -1,6 +1,7 @@
 import { asArray, asName, asNumber, dictGet } from "./objects.js";
-import { decodeStream } from "./filters.js";
+import { decodePdfString } from "./pdf-text.js";
 import { BEZIER_KAPPA, IDENTITY_MATRIX, applyMatrix, multiplyMatrices, translationMatrix } from "./matrix.js";
+import { decodeStream } from "./filters.js";
 import { readContentStream } from "./content-read.js";
 import { COLOR_BLACK } from "document-schema.js";
 //#region src/interpret.ts
@@ -34,6 +35,22 @@ function computeTrm(ctm, ts) {
 function numAt(operands, index) {
 	return asNumber(operands[index]) ?? 0;
 }
+function markedContentProperties(operand, resources, resolver) {
+	if (operand?.kind === "dict") return operand;
+	if (operand?.kind === "name") {
+		const properties = resolver.resolveDict(dictGet(resources, "Properties"));
+		return properties === void 0 ? void 0 : resolver.resolveDict(dictGet(properties, operand.name));
+	}
+}
+function markedContentStrings(props) {
+	if (props === void 0) return {};
+	const actualTextObj = dictGet(props, "ActualText");
+	const altObj = dictGet(props, "Alt");
+	return {
+		...actualTextObj?.kind === "string" ? { actualText: decodePdfString(actualTextObj.bytes) } : {},
+		...altObj?.kind === "string" ? { alt: decodePdfString(altObj.bytes) } : {}
+	};
+}
 function grayColor(value) {
 	return {
 		r: value,
@@ -252,7 +269,7 @@ function interpretContentStream(bytes, resources, context) {
 		fillColor: COLOR_BLACK,
 		strokeColor: COLOR_BLACK,
 		lineWidth: DEFAULT_LINE_WIDTH_PT
-	}, context, items, 0);
+	}, context, items, 0, []);
 	return items;
 }
 function lastPointOf(subpath) {
@@ -265,13 +282,38 @@ function lastPointOf(subpath) {
 		y: last.yPt
 	};
 }
-function runContentStream(bytes, resources, initialState, context, items, depth) {
+function runContentStream(bytes, resources, initialState, context, items, depth, markedContent) {
 	const operations = readContentStream(bytes, context.sink);
 	const gsStack = [];
 	let gs = initialState;
 	let ts = defaultTextState();
 	let pathSubpaths = [];
 	let currentSubpath;
+	const inScope = (key) => {
+		for (let i = markedContent.length - 1; i >= 0; i--) {
+			const value = markedContent[i]?.[key];
+			if (value !== void 0) return value;
+		}
+	};
+	const pushItem = (item) => {
+		if (item.kind === "text") {
+			const layer = item.layerName ?? inScope("layerName");
+			const actualText = inScope("actualText");
+			const alt = inScope("alt");
+			items.push({
+				...item,
+				...layer !== void 0 ? { layerName: layer } : {},
+				...actualText !== void 0 ? { actualText } : {},
+				...alt !== void 0 ? { alt } : {}
+			});
+			return;
+		}
+		const layer = item.layerName ?? inScope("layerName");
+		items.push(layer === void 0 ? item : {
+			...item,
+			layerName: layer
+		});
+	};
 	const finalizeCurrentSubpath = () => {
 		if (currentSubpath !== void 0) {
 			pathSubpaths.push(currentSubpath);
@@ -345,7 +387,7 @@ function runContentStream(bytes, resources, initialState, context, items, depth)
 					widthPt: gs.lineWidth
 				} : void 0
 			};
-			items.push(classifyShape(pathSubpaths, paint) ?? {
+			pushItem(classifyShape(pathSubpaths, paint) ?? {
 				kind: "path",
 				subpaths: pathSubpaths,
 				fillRule: paintFillRuleFor(operator),
@@ -393,7 +435,7 @@ function runContentStream(bytes, resources, initialState, context, items, depth)
 			at += chunk.length;
 		}
 		const endMatrix = computeTrm(gs.ctm, ts);
-		items.push({
+		pushItem({
 			kind: "text",
 			codes: combined,
 			fontResourceName: ts.fontResourceName,
@@ -428,12 +470,14 @@ function runContentStream(bytes, resources, initialState, context, items, depth)
 			return;
 		}
 		const subtype = asName(dictGet(xobj.dict, "Subtype"));
+		const ownLayer = context.layerNameOf?.(dictGet(xobj.dict, "OC"));
 		if (subtype === "Image") {
-			items.push({
+			pushItem({
 				kind: "image",
 				resourceName: name,
 				resources,
-				matrix: gs.ctm
+				matrix: gs.ctm,
+				...ownLayer !== void 0 ? { layerName: ownLayer } : {}
 			});
 			return;
 		}
@@ -454,12 +498,13 @@ function runContentStream(bytes, resources, initialState, context, items, depth)
 				...gs,
 				ctm: multiplyMatrices(formMatrix, gs.ctm)
 			};
-			runContentStream(decoded.bytes, formResources, formState, context, items, depth + 1);
+			const formLayer = ownLayer ?? inScope("layerName");
+			runContentStream(decoded.bytes, formResources, formState, context, items, depth + 1, [formLayer !== void 0 ? { layerName: formLayer } : {}]);
 		}
 	};
 	for (const token of operations) {
 		if (token.kind === "inlineImage") {
-			items.push({
+			pushItem({
 				kind: "inlineImage",
 				dict: token.image.dict,
 				data: token.image.data,
@@ -713,7 +758,22 @@ function runContentStream(bytes, resources, initialState, context, items, depth)
 				if (array !== void 0) showTextArray(array);
 				break;
 			}
-			case "Do": handleDo(asName(operands[0]));
+			case "Do":
+				handleDo(asName(operands[0]));
+				break;
+			case "BDC": {
+				const props = markedContentProperties(operands[1], resources, context.resolver);
+				const layerName = context.layerNameOf?.(props === void 0 ? void 0 : dictGet(props, "OC"));
+				markedContent.push({
+					...layerName !== void 0 ? { layerName } : {},
+					...markedContentStrings(props)
+				});
+				break;
+			}
+			case "BMC":
+				markedContent.push({});
+				break;
+			case "EMC": markedContent.pop();
 		}
 	}
 }

package/dist/layout.cjs

@@ -14,6 +14,9 @@ const LayoutTextSchema = zod.z.object({
 	widthPt: zod.z.number().nonnegative().optional(),
 	rotationDeg: zod.z.number().optional(),
 	underline: zod.z.boolean().optional(),
+	layer: zod.z.string().optional(),
+	actualText: zod.z.string().optional(),
+	alt: zod.z.string().optional(),
 	sourcePath: zod.z.string().optional()
 });
 const LayoutImageSchema = zod.z.object({
@@ -24,6 +27,7 @@ const LayoutImageSchema = zod.z.object({
 	widthPt: zod.z.number().positive(),
 	heightPt: zod.z.number().positive(),
 	rotationDeg: zod.z.number().optional(),
+	layer: zod.z.string().optional(),
 	sourcePath: zod.z.string().optional()
 });
 const LayoutRectSchema = zod.z.object({
@@ -37,6 +41,7 @@ const LayoutRectSchema = zod.z.object({
 		color: document_schema_js.ColorSchema,
 		widthPt: zod.z.number().positive()
 	}).optional(),
+	layer: zod.z.string().optional(),
 	sourcePath: zod.z.string().optional()
 });
 const LayoutLineSchema = zod.z.object({
@@ -48,6 +53,7 @@ const LayoutLineSchema = zod.z.object({
 	color: document_schema_js.ColorSchema,
 	widthPt: zod.z.number().positive(),
 	style: document_schema_js.ContentStrokeStyleSchema.optional(),
+	layer: zod.z.string().optional(),
 	sourcePath: zod.z.string().optional()
 });
 const LayoutEllipseSchema = zod.z.object({
@@ -61,6 +67,7 @@ const LayoutEllipseSchema = zod.z.object({
 		color: document_schema_js.ColorSchema,
 		widthPt: zod.z.number().positive()
 	}).optional(),
+	layer: zod.z.string().optional(),
 	sourcePath: zod.z.string().optional()
 });
 const LayoutPathSegmentSchema = zod.z.discriminatedUnion("kind", [zod.z.object({
@@ -92,6 +99,7 @@ const LayoutPathSchema = zod.z.object({
 		widthPt: zod.z.number().positive()
 	}).optional(),
 	style: document_schema_js.ContentStrokeStyleSchema.optional(),
+	layer: zod.z.string().optional(),
 	sourcePath: zod.z.string().optional()
 });
 const LayoutLinkSchema = zod.z.object({
@@ -101,8 +109,18 @@ const LayoutLinkSchema = zod.z.object({
 	yPt: zod.z.number(),
 	widthPt: zod.z.number().nonnegative(),
 	heightPt: zod.z.number().nonnegative(),
+	title: zod.z.string().optional(),
 	sourcePath: zod.z.string().optional()
 });
+const LayoutInternalLinkSchema = zod.z.object({
+	kind: zod.z.literal("internalLink"),
+	destination: zod.z.string(),
+	xPt: zod.z.number(),
+	yPt: zod.z.number(),
+	widthPt: zod.z.number().nonnegative(),
+	heightPt: zod.z.number().nonnegative(),
+	title: zod.z.string().optional()
+});
 const LayoutItemSchema = zod.z.discriminatedUnion("kind", [
 	LayoutTextSchema,
 	LayoutImageSchema,
@@ -110,12 +128,44 @@ const LayoutItemSchema = zod.z.discriminatedUnion("kind", [
 	LayoutLineSchema,
 	LayoutEllipseSchema,
 	LayoutPathSchema,
-	LayoutLinkSchema
+	LayoutLinkSchema,
+	LayoutInternalLinkSchema
 ]);
+const LayoutAnnotationQuadSchema = zod.z.tuple([
+	zod.z.object({
+		xPt: zod.z.number(),
+		yPt: zod.z.number()
+	}),
+	zod.z.object({
+		xPt: zod.z.number(),
+		yPt: zod.z.number()
+	}),
+	zod.z.object({
+		xPt: zod.z.number(),
+		yPt: zod.z.number()
+	}),
+	zod.z.object({
+		xPt: zod.z.number(),
+		yPt: zod.z.number()
+	})
+]);
+const LayoutAnnotationSchema = zod.z.object({
+	subtype: zod.z.string(),
+	xPt: zod.z.number(),
+	yPt: zod.z.number(),
+	widthPt: zod.z.number().nonnegative(),
+	heightPt: zod.z.number().nonnegative(),
+	contents: zod.z.string().optional(),
+	author: zod.z.string().optional(),
+	modifiedIso: zod.z.string().optional(),
+	quads: zod.z.array(LayoutAnnotationQuadSchema).optional(),
+	source: document_schema_js.SourceResidueSchema.optional()
+});
 const LayoutPageSchema = zod.z.object({
 	widthPt: zod.z.number().positive(),
 	heightPt: zod.z.number().positive(),
 	items: zod.z.array(LayoutItemSchema),
+	annotations: zod.z.array(LayoutAnnotationSchema).optional(),
 	notes: zod.z.string().optional()
 });
 const LayoutImageAssetSchema = zod.z.object({
@@ -124,21 +174,101 @@ const LayoutImageAssetSchema = zod.z.object({
 	widthPx: zod.z.number().int().positive(),
 	heightPx: zod.z.number().int().positive()
 });
+const LayoutDestinationTargetSchema = zod.z.object({
+	kind: zod.z.enum([
+		"xyz",
+		"fit",
+		"fitH",
+		"fitV",
+		"fitR",
+		"fitB",
+		"fitBH",
+		"fitBV"
+	]),
+	leftPt: zod.z.number().optional(),
+	topPt: zod.z.number().optional(),
+	bottomPt: zod.z.number().optional(),
+	rightPt: zod.z.number().optional(),
+	zoom: zod.z.number().optional()
+});
+const LayoutDestinationSchema = zod.z.object({
+	name: zod.z.string(),
+	pageIndex: zod.z.number().int().nonnegative(),
+	target: LayoutDestinationTargetSchema
+});
+const LayoutOutlineItemSchema = zod.z.lazy(() => zod.z.object({
+	title: zod.z.string(),
+	destination: zod.z.string().optional(),
+	children: zod.z.array(LayoutOutlineItemSchema)
+}));
+const LayoutAttachmentSchema = zod.z.object({
+	name: zod.z.string(),
+	description: zod.z.string().optional(),
+	mimeType: zod.z.string().optional(),
+	base64: zod.z.string()
+});
+const LayoutFormWidgetSchema = zod.z.object({
+	pageIndex: zod.z.number().int().nonnegative(),
+	xPt: zod.z.number(),
+	yPt: zod.z.number(),
+	widthPt: zod.z.number().nonnegative(),
+	heightPt: zod.z.number().nonnegative()
+});
+const LayoutFormFieldSchema = zod.z.lazy(() => zod.z.object({
+	name: zod.z.string(),
+	fieldType: zod.z.enum([
+		"text",
+		"checkbox",
+		"radio",
+		"button",
+		"listbox",
+		"combobox",
+		"signature",
+		"group"
+	]),
+	value: zod.z.string().optional(),
+	checked: zod.z.boolean().optional(),
+	options: zod.z.array(zod.z.string()).optional(),
+	alias: zod.z.string().optional(),
+	readOnly: zod.z.boolean().optional(),
+	widgets: zod.z.array(LayoutFormWidgetSchema),
+	children: zod.z.array(LayoutFormFieldSchema)
+}));
+const LayoutLayerSchema = zod.z.object({
+	name: zod.z.string(),
+	visible: zod.z.boolean()
+});
 const LayoutDocumentSchema = zod.z.object({
 	formatVersion: zod.z.literal(1),
 	metadata: document_schema_js.LayoutMetadataSchema,
 	pages: zod.z.array(LayoutPageSchema),
-	images: zod.z.record(zod.z.string(), LayoutImageAssetSchema)
+	images: zod.z.record(zod.z.string(), LayoutImageAssetSchema),
+	destinations: zod.z.array(LayoutDestinationSchema).optional(),
+	outline: zod.z.array(LayoutOutlineItemSchema).optional(),
+	attachments: zod.z.array(LayoutAttachmentSchema).optional(),
+	layers: zod.z.array(LayoutLayerSchema).optional(),
+	form: zod.z.array(LayoutFormFieldSchema).optional(),
+	source: zod.z.record(zod.z.string(), document_schema_js.SourceResidueSchema).optional()
 });
 //#endregion
 exports.LAYOUT_FORMAT_VERSION = LAYOUT_FORMAT_VERSION;
+exports.LayoutAnnotationQuadSchema = LayoutAnnotationQuadSchema;
+exports.LayoutAnnotationSchema = LayoutAnnotationSchema;
+exports.LayoutAttachmentSchema = LayoutAttachmentSchema;
+exports.LayoutDestinationSchema = LayoutDestinationSchema;
+exports.LayoutDestinationTargetSchema = LayoutDestinationTargetSchema;
 exports.LayoutDocumentSchema = LayoutDocumentSchema;
 exports.LayoutEllipseSchema = LayoutEllipseSchema;
+exports.LayoutFormFieldSchema = LayoutFormFieldSchema;
+exports.LayoutFormWidgetSchema = LayoutFormWidgetSchema;
 exports.LayoutImageAssetSchema = LayoutImageAssetSchema;
 exports.LayoutImageSchema = LayoutImageSchema;
+exports.LayoutInternalLinkSchema = LayoutInternalLinkSchema;
 exports.LayoutItemSchema = LayoutItemSchema;
+exports.LayoutLayerSchema = LayoutLayerSchema;
 exports.LayoutLineSchema = LayoutLineSchema;
 exports.LayoutLinkSchema = LayoutLinkSchema;
+exports.LayoutOutlineItemSchema = LayoutOutlineItemSchema;
 exports.LayoutPageSchema = LayoutPageSchema;
 exports.LayoutPathSchema = LayoutPathSchema;
 exports.LayoutPathSegmentSchema = LayoutPathSegmentSchema;