pdf-codec 3.3.0 → 3.4.0

package/dist/structure.cjs

@@ -0,0 +1,184 @@
+Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
+const require_objects = require("./objects.cjs");
+const require_pdf_text = require("./pdf-text.cjs");
+//#region src/structure.ts
+function readStructure(catalog, pages, resolver, sink) {
+	const root = resolver.resolveDict(require_objects.dictGet(catalog, "StructTreeRoot"));
+	if (root === void 0) return {
+		tree: [],
+		ownerOf: () => void 0
+	};
+	const roleMap = /* @__PURE__ */ new Map();
+	const roleMapDict = resolver.resolveDict(require_objects.dictGet(root, "RoleMap"));
+	for (const [custom, standard] of roleMapDict?.entries ?? []) {
+		const target = require_objects.asName(standard);
+		if (target !== void 0) roleMap.set(custom, target);
+	}
+	const classAttributes = /* @__PURE__ */ new Map();
+	const classMapDict = resolver.resolveDict(require_objects.dictGet(root, "ClassMap"));
+	for (const [className, value] of classMapDict?.entries ?? []) {
+		const attributes = elementAttributes(resolver.resolveDict(value));
+		if (attributes !== void 0) classAttributes.set(className, attributes);
+	}
+	const idByDict = /* @__PURE__ */ new Map();
+	const tree = [];
+	const visited = /* @__PURE__ */ new Set();
+	let minted = 0;
+	const readElement = (dict) => {
+		if (visited.has(dict)) {
+			sink({
+				code: "pdf/structure-cycle",
+				severity: "warning",
+				message: "the structure tree contains a cycle; stopping descent at the repeated element"
+			});
+			return;
+		}
+		visited.add(dict);
+		const ownType = require_objects.asName(require_objects.dictGet(dict, "S"));
+		if (ownType === void 0) {
+			sink({
+				code: "pdf/structure-element-invalid",
+				severity: "warning",
+				message: "a structure element carries no /S type; skipping it"
+			});
+			return;
+		}
+		minted += 1;
+		const id = `struct${minted}`;
+		idByDict.set(dict, id);
+		const children = [];
+		for (const kid of elementKids(dict, resolver)) {
+			const child = readElement(kid);
+			if (child !== void 0) children.push(child);
+		}
+		return {
+			id,
+			type: roleMap.get(ownType) ?? ownType,
+			...mergedAttributes(dict, resolver, classAttributes),
+			children
+		};
+	};
+	for (const kid of elementKids(root, resolver)) {
+		const element = readElement(kid);
+		if (element !== void 0) tree.push(element);
+	}
+	return {
+		tree,
+		ownerOf: parentTreeOwners(root, pages, resolver, idByDict, sink)
+	};
+}
+const ATTRIBUTE_KEYS = [
+	"title",
+	"language",
+	"alt",
+	"actualText"
+];
+const ATTRIBUTE_SOURCES = [
+	["title", "T"],
+	["language", "Lang"],
+	["alt", "Alt"],
+	["actualText", "ActualText"]
+];
+function elementAttributes(dict) {
+	if (dict === void 0) return;
+	const stringField = (key) => {
+		const obj = require_objects.dictGet(dict, key);
+		return obj?.kind === "string" ? require_pdf_text.decodePdfString(obj.bytes) : void 0;
+	};
+	const values = /* @__PURE__ */ new Map();
+	for (const [field, source] of ATTRIBUTE_SOURCES) {
+		const value = stringField(source);
+		if (value !== void 0) values.set(field, value);
+	}
+	if (values.size === 0) return;
+	return Object.fromEntries(values);
+}
+function mergedAttributes(dict, resolver, classAttributes) {
+	const own = elementAttributes(dict);
+	const classAttributeList = [];
+	for (const classObj of require_objects.asArray(require_objects.dictGet(dict, "C")) ?? []) {
+		const name = require_objects.asName(classObj);
+		const attributes = name !== void 0 ? classAttributes.get(name) : elementAttributes(resolver.resolveDict(classObj));
+		if (attributes !== void 0) classAttributeList.push(attributes);
+	}
+	const values = /* @__PURE__ */ new Map();
+	for (const key of ATTRIBUTE_KEYS) {
+		const ownValue = own?.[key];
+		if (ownValue !== void 0) {
+			values.set(key, ownValue);
+			continue;
+		}
+		for (const attributes of classAttributeList) {
+			const value = attributes[key];
+			if (value !== void 0) {
+				values.set(key, value);
+				break;
+			}
+		}
+	}
+	return Object.fromEntries(values);
+}
+function elementKids(dict, resolver) {
+	const kids = [];
+	const k = require_objects.dictGet(dict, "K");
+	for (const kid of require_objects.asArray(k) ?? (k !== void 0 ? [k] : [])) {
+		const resolved = resolver.resolveDict(kid);
+		if (resolved !== void 0 && require_objects.dictGet(resolved, "S") !== void 0) kids.push(resolved);
+	}
+	return kids;
+}
+function numberTreeEntries(node, resolver, visited) {
+	if (visited.has(node)) return [];
+	visited.add(node);
+	const entries = [];
+	const nums = require_objects.asArray(require_objects.dictGet(node, "Nums"));
+	for (let i = 0; nums !== void 0 && i + 1 < nums.length; i += 2) {
+		const key = require_objects.asNumber(nums[i]);
+		if (key !== void 0) entries.push([key, nums[i + 1]]);
+	}
+	for (const kid of require_objects.asArray(require_objects.dictGet(node, "Kids")) ?? []) {
+		const dict = resolver.resolveDict(kid);
+		if (dict !== void 0) entries.push(...numberTreeEntries(dict, resolver, visited));
+	}
+	return entries;
+}
+function parentTreeOwners(root, pages, resolver, idByDict, sink) {
+	const documentTree = resolver.resolveDict(require_objects.dictGet(root, "ParentTree"));
+	if (documentTree === void 0) {
+		if (require_objects.dictGet(root, "ParentTree") !== void 0) sink({
+			code: "pdf/parent-tree-unresolved",
+			severity: "warning",
+			message: "the /ParentTree did not resolve to a dictionary; no marked-content ownership can be associated"
+		});
+		return () => void 0;
+	}
+	const entries = new Map(numberTreeEntries(documentTree, resolver, /* @__PURE__ */ new Set()));
+	const owners = /* @__PURE__ */ new Map();
+	pages.forEach((page, pageIndex) => {
+		const structParents = require_objects.asNumber(require_objects.dictGet(page, "StructParents"));
+		if (structParents === void 0) return;
+		const entry = entries.get(structParents);
+		if (entry === void 0) {
+			sink({
+				code: "pdf/parent-tree-missing-entry",
+				severity: "warning",
+				pageIndex,
+				message: `page ${pageIndex} declares /StructParents ${structParents} but the /ParentTree carries no entry for that key; its marked content resolves to no owner`
+			});
+			return;
+		}
+		const mcidOwners = require_objects.asArray(entry);
+		if (mcidOwners === void 0) return;
+		mcidOwners.forEach((element, mcid) => {
+			const id = idByDict.get(resolver.resolveDict(element) ?? NEVER_DICT);
+			if (id !== void 0) owners.set(`${pageIndex}:${mcid}`, id);
+		});
+	});
+	return (pageIndex, mcid) => owners.get(`${pageIndex}:${mcid}`);
+}
+const NEVER_DICT = {
+	kind: "dict",
+	entries: /* @__PURE__ */ new Map()
+};
+//#endregion
+exports.readStructure = readStructure;

package/dist/structure.d.cts

@@ -0,0 +1,12 @@
+import { PdfDiagnosticSink } from "./diagnostics.cjs";
+import { PdfDict } from "./objects.cjs";
+import { PdfObjectResolver } from "./interpret.cjs";
+import { LayoutStructureElement } from "./layout.cjs";
+//#region src/structure.d.ts
+interface StructureContext {
+  readonly tree: readonly LayoutStructureElement[];
+  readonly ownerOf: (pageIndex: number, mcid: number) => string | undefined;
+}
+declare function readStructure(catalog: PdfDict, pages: readonly PdfDict[], resolver: PdfObjectResolver, sink: PdfDiagnosticSink): StructureContext;
+//#endregion
+export { StructureContext, readStructure };

package/dist/structure.d.ts

@@ -0,0 +1,12 @@
+import { PdfDiagnosticSink } from "./diagnostics.js";
+import { PdfDict } from "./objects.js";
+import { PdfObjectResolver } from "./interpret.js";
+import { LayoutStructureElement } from "./layout.js";
+//#region src/structure.d.ts
+interface StructureContext {
+  readonly tree: readonly LayoutStructureElement[];
+  readonly ownerOf: (pageIndex: number, mcid: number) => string | undefined;
+}
+declare function readStructure(catalog: PdfDict, pages: readonly PdfDict[], resolver: PdfObjectResolver, sink: PdfDiagnosticSink): StructureContext;
+//#endregion
+export { StructureContext, readStructure };

package/dist/structure.js

@@ -0,0 +1,183 @@
+import { asArray, asName, asNumber, dictGet } from "./objects.js";
+import { decodePdfString } from "./pdf-text.js";
+//#region src/structure.ts
+function readStructure(catalog, pages, resolver, sink) {
+	const root = resolver.resolveDict(dictGet(catalog, "StructTreeRoot"));
+	if (root === void 0) return {
+		tree: [],
+		ownerOf: () => void 0
+	};
+	const roleMap = /* @__PURE__ */ new Map();
+	const roleMapDict = resolver.resolveDict(dictGet(root, "RoleMap"));
+	for (const [custom, standard] of roleMapDict?.entries ?? []) {
+		const target = asName(standard);
+		if (target !== void 0) roleMap.set(custom, target);
+	}
+	const classAttributes = /* @__PURE__ */ new Map();
+	const classMapDict = resolver.resolveDict(dictGet(root, "ClassMap"));
+	for (const [className, value] of classMapDict?.entries ?? []) {
+		const attributes = elementAttributes(resolver.resolveDict(value));
+		if (attributes !== void 0) classAttributes.set(className, attributes);
+	}
+	const idByDict = /* @__PURE__ */ new Map();
+	const tree = [];
+	const visited = /* @__PURE__ */ new Set();
+	let minted = 0;
+	const readElement = (dict) => {
+		if (visited.has(dict)) {
+			sink({
+				code: "pdf/structure-cycle",
+				severity: "warning",
+				message: "the structure tree contains a cycle; stopping descent at the repeated element"
+			});
+			return;
+		}
+		visited.add(dict);
+		const ownType = asName(dictGet(dict, "S"));
+		if (ownType === void 0) {
+			sink({
+				code: "pdf/structure-element-invalid",
+				severity: "warning",
+				message: "a structure element carries no /S type; skipping it"
+			});
+			return;
+		}
+		minted += 1;
+		const id = `struct${minted}`;
+		idByDict.set(dict, id);
+		const children = [];
+		for (const kid of elementKids(dict, resolver)) {
+			const child = readElement(kid);
+			if (child !== void 0) children.push(child);
+		}
+		return {
+			id,
+			type: roleMap.get(ownType) ?? ownType,
+			...mergedAttributes(dict, resolver, classAttributes),
+			children
+		};
+	};
+	for (const kid of elementKids(root, resolver)) {
+		const element = readElement(kid);
+		if (element !== void 0) tree.push(element);
+	}
+	return {
+		tree,
+		ownerOf: parentTreeOwners(root, pages, resolver, idByDict, sink)
+	};
+}
+const ATTRIBUTE_KEYS = [
+	"title",
+	"language",
+	"alt",
+	"actualText"
+];
+const ATTRIBUTE_SOURCES = [
+	["title", "T"],
+	["language", "Lang"],
+	["alt", "Alt"],
+	["actualText", "ActualText"]
+];
+function elementAttributes(dict) {
+	if (dict === void 0) return;
+	const stringField = (key) => {
+		const obj = dictGet(dict, key);
+		return obj?.kind === "string" ? decodePdfString(obj.bytes) : void 0;
+	};
+	const values = /* @__PURE__ */ new Map();
+	for (const [field, source] of ATTRIBUTE_SOURCES) {
+		const value = stringField(source);
+		if (value !== void 0) values.set(field, value);
+	}
+	if (values.size === 0) return;
+	return Object.fromEntries(values);
+}
+function mergedAttributes(dict, resolver, classAttributes) {
+	const own = elementAttributes(dict);
+	const classAttributeList = [];
+	for (const classObj of asArray(dictGet(dict, "C")) ?? []) {
+		const name = asName(classObj);
+		const attributes = name !== void 0 ? classAttributes.get(name) : elementAttributes(resolver.resolveDict(classObj));
+		if (attributes !== void 0) classAttributeList.push(attributes);
+	}
+	const values = /* @__PURE__ */ new Map();
+	for (const key of ATTRIBUTE_KEYS) {
+		const ownValue = own?.[key];
+		if (ownValue !== void 0) {
+			values.set(key, ownValue);
+			continue;
+		}
+		for (const attributes of classAttributeList) {
+			const value = attributes[key];
+			if (value !== void 0) {
+				values.set(key, value);
+				break;
+			}
+		}
+	}
+	return Object.fromEntries(values);
+}
+function elementKids(dict, resolver) {
+	const kids = [];
+	const k = dictGet(dict, "K");
+	for (const kid of asArray(k) ?? (k !== void 0 ? [k] : [])) {
+		const resolved = resolver.resolveDict(kid);
+		if (resolved !== void 0 && dictGet(resolved, "S") !== void 0) kids.push(resolved);
+	}
+	return kids;
+}
+function numberTreeEntries(node, resolver, visited) {
+	if (visited.has(node)) return [];
+	visited.add(node);
+	const entries = [];
+	const nums = asArray(dictGet(node, "Nums"));
+	for (let i = 0; nums !== void 0 && i + 1 < nums.length; i += 2) {
+		const key = asNumber(nums[i]);
+		if (key !== void 0) entries.push([key, nums[i + 1]]);
+	}
+	for (const kid of asArray(dictGet(node, "Kids")) ?? []) {
+		const dict = resolver.resolveDict(kid);
+		if (dict !== void 0) entries.push(...numberTreeEntries(dict, resolver, visited));
+	}
+	return entries;
+}
+function parentTreeOwners(root, pages, resolver, idByDict, sink) {
+	const documentTree = resolver.resolveDict(dictGet(root, "ParentTree"));
+	if (documentTree === void 0) {
+		if (dictGet(root, "ParentTree") !== void 0) sink({
+			code: "pdf/parent-tree-unresolved",
+			severity: "warning",
+			message: "the /ParentTree did not resolve to a dictionary; no marked-content ownership can be associated"
+		});
+		return () => void 0;
+	}
+	const entries = new Map(numberTreeEntries(documentTree, resolver, /* @__PURE__ */ new Set()));
+	const owners = /* @__PURE__ */ new Map();
+	pages.forEach((page, pageIndex) => {
+		const structParents = asNumber(dictGet(page, "StructParents"));
+		if (structParents === void 0) return;
+		const entry = entries.get(structParents);
+		if (entry === void 0) {
+			sink({
+				code: "pdf/parent-tree-missing-entry",
+				severity: "warning",
+				pageIndex,
+				message: `page ${pageIndex} declares /StructParents ${structParents} but the /ParentTree carries no entry for that key; its marked content resolves to no owner`
+			});
+			return;
+		}
+		const mcidOwners = asArray(entry);
+		if (mcidOwners === void 0) return;
+		mcidOwners.forEach((element, mcid) => {
+			const id = idByDict.get(resolver.resolveDict(element) ?? NEVER_DICT);
+			if (id !== void 0) owners.set(`${pageIndex}:${mcid}`, id);
+		});
+	});
+	return (pageIndex, mcid) => owners.get(`${pageIndex}:${mcid}`);
+}
+const NEVER_DICT = {
+	kind: "dict",
+	entries: /* @__PURE__ */ new Map()
+};
+//#endregion
+export { readStructure };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pdf-codec",
-  "version": "3.3.0",
+  "version": "3.4.0",
   "description": "Hand-written, dependency-minimal PDF codec: parses arbitrary real-world PDFs and generates new ones, built on its own codec-owned LayoutDocument item model and Zod 4 codecs.",
   "type": "module",
   "repository": {
@@ -98,7 +98,7 @@
   "packageManager": "pnpm@11.6.0",
   "dependencies": {
     "byte-codec": "^1.1.13",
-    "document-schema.js": "^4.8.0",
+    "document-schema.js": "^4.9.0",
     "fflate": "^0.8.3",
     "zod": "^4.4.3"
   },