pdf-parser.js 3.0.8 → 4.5.2

package/dist/objects.d.ts

@@ -1,59 +1,2 @@
-//#region src/objects.d.ts
-type PdfObject = {
-  kind: 'null';
-} | {
-  kind: 'bool';
-  value: boolean;
-} | {
-  kind: 'number';
-  value: number;
-} | {
-  kind: 'name';
-  name: string;
-} | {
-  kind: 'string';
-  bytes: Uint8Array<ArrayBuffer>;
-  hex: boolean;
-} | {
-  kind: 'array';
-  items: PdfObject[];
-} | {
-  kind: 'dict';
-  entries: Map<string, PdfObject>;
-} | {
-  kind: 'stream';
-  dict: PdfDict;
-  raw: Uint8Array<ArrayBuffer>;
-} | {
-  kind: 'ref';
-  num: number;
-  gen: number;
-};
-type PdfDict = Extract<PdfObject, {
-  kind: 'dict';
-}>;
-type PdfArray = Extract<PdfObject, {
-  kind: 'array';
-}>;
-type PdfStream = Extract<PdfObject, {
-  kind: 'stream';
-}>;
-declare function pdfNull(): PdfObject;
-declare function pdfBool(value: boolean): PdfObject;
-declare function pdfNum(value: number): PdfObject;
-declare function pdfName(name: string): PdfObject;
-declare function pdfHexString(bytes: Uint8Array<ArrayBuffer>): PdfObject;
-declare function pdfLiteralString(bytes: Uint8Array<ArrayBuffer>): PdfObject;
-declare function pdfArray(items: PdfObject[]): PdfObject;
-declare function pdfDict(entries: Record<string, PdfObject> | Map<string, PdfObject>): PdfDict;
-declare function pdfStream(dict: PdfDict, raw: Uint8Array<ArrayBuffer>): PdfObject;
-declare function pdfRef(num: number, gen: number): PdfObject;
-declare function isName(obj: PdfObject | undefined, name: string): boolean;
-declare function asNumber(obj: PdfObject | undefined): number | undefined;
-declare function asName(obj: PdfObject | undefined): string | undefined;
-declare function asBool(obj: PdfObject | undefined): boolean | undefined;
-declare function asArray(obj: PdfObject | undefined): PdfObject[] | undefined;
-declare function asDict(obj: PdfObject | undefined): PdfDict | undefined;
-declare function dictGet(dict: PdfDict, key: string): PdfObject | undefined;
-//#endregion
+import { _ as pdfName, a as asArray, b as pdfRef, c as asName, d as isName, f as pdfArray, g as pdfLiteralString, h as pdfHexString, i as PdfStream, l as asNumber, m as pdfDict, n as PdfDict, o as asBool, p as pdfBool, r as PdfObject, s as asDict, t as PdfArray, u as dictGet, v as pdfNull, x as pdfStream, y as pdfNum } from "./objects-B7HdMeWS.js";
 export { PdfArray, PdfDict, PdfObject, PdfStream, asArray, asBool, asDict, asName, asNumber, dictGet, isName, pdfArray, pdfBool, pdfDict, pdfHexString, pdfLiteralString, pdfName, pdfNull, pdfNum, pdfRef, pdfStream };

package/dist/optional-content.cjs

@@ -0,0 +1,63 @@
+Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
+const require_objects = require("./objects.cjs");
+const require_pdf_text = require("./pdf-text.cjs");
+//#region src/optional-content.ts
+function readOptionalContent(catalog, resolver, sink) {
+	const ocProperties = resolver.resolveDict(require_objects.dictGet(catalog, "OCProperties"));
+	if (ocProperties === void 0) return {
+		layers: [],
+		layerNameOf: () => void 0
+	};
+	const nameByGroup = /* @__PURE__ */ new Map();
+	const groups = [];
+	const ocgs = require_objects.asArray(resolver.resolve(require_objects.dictGet(ocProperties, "OCGs"))) ?? [];
+	for (const ocgRef of ocgs) {
+		const ocg = resolver.resolveDict(ocgRef);
+		if (ocg === void 0) {
+			sink({
+				code: "pdf/ocg-unresolved",
+				severity: "warning",
+				message: "an entry in /OCProperties /OCGs did not resolve to a dictionary; skipping it"
+			});
+			continue;
+		}
+		const nameObj = require_objects.dictGet(ocg, "Name");
+		const name = nameObj?.kind === "string" ? require_pdf_text.decodePdfString(nameObj.bytes) : mintLayerName(nameByGroup);
+		nameByGroup.set(ocg, name);
+		groups.push({
+			dict: ocg,
+			name
+		});
+	}
+	const defaultConfig = resolver.resolveDict(require_objects.dictGet(ocProperties, "D"));
+	const baseVisible = defaultConfig === void 0 || require_objects.asName(require_objects.dictGet(defaultConfig, "BaseState")) !== "OFF";
+	const groupsIn = (key) => {
+		const set = /* @__PURE__ */ new Set();
+		const arr = defaultConfig === void 0 ? void 0 : require_objects.asArray(resolver.resolve(require_objects.dictGet(defaultConfig, key)));
+		if (arr !== void 0) for (const entry of arr) {
+			const dict = resolver.resolveDict(entry);
+			if (dict !== void 0) set.add(dict);
+		}
+		return set;
+	};
+	const on = groupsIn("ON");
+	const off = groupsIn("OFF");
+	return {
+		layers: groups.map(({ dict, name }) => ({
+			name,
+			visible: off.has(dict) ? false : on.has(dict) ? true : baseVisible
+		})),
+		layerNameOf: (obj) => {
+			const dict = resolver.resolveDict(obj);
+			return dict !== void 0 ? nameByGroup.get(dict) : void 0;
+		}
+	};
+}
+function mintLayerName(taken) {
+	const used = new Set(taken.values());
+	let n = 1;
+	while (used.has(`layer${n}`)) n++;
+	return `layer${n}`;
+}
+//#endregion
+exports.readOptionalContent = readOptionalContent;

package/dist/optional-content.d.cts

@@ -0,0 +1,12 @@
+import { PdfDiagnosticSink } from "./diagnostics.cjs";
+import { n as PdfDict, r as PdfObject } from "./objects-B7HdMeWS.cjs";
+import { PdfObjectResolver } from "./interpret.cjs";
+import { LayoutLayer } from "./layout.cjs";
+//#region src/optional-content.d.ts
+interface OptionalContentContext {
+  readonly layers: readonly LayoutLayer[];
+  readonly layerNameOf: (obj: PdfObject | undefined) => string | undefined;
+}
+declare function readOptionalContent(catalog: PdfDict, resolver: PdfObjectResolver, sink: PdfDiagnosticSink): OptionalContentContext;
+//#endregion
+export { OptionalContentContext, readOptionalContent };

package/dist/optional-content.d.ts

@@ -0,0 +1,12 @@
+import { PdfDiagnosticSink } from "./diagnostics.js";
+import { n as PdfDict, r as PdfObject } from "./objects-B7HdMeWS.js";
+import { PdfObjectResolver } from "./interpret.js";
+import { LayoutLayer } from "./layout.js";
+//#region src/optional-content.d.ts
+interface OptionalContentContext {
+  readonly layers: readonly LayoutLayer[];
+  readonly layerNameOf: (obj: PdfObject | undefined) => string | undefined;
+}
+declare function readOptionalContent(catalog: PdfDict, resolver: PdfObjectResolver, sink: PdfDiagnosticSink): OptionalContentContext;
+//#endregion
+export { OptionalContentContext, readOptionalContent };

package/dist/optional-content.js

@@ -0,0 +1,62 @@
+import { asArray, asName, dictGet } from "./objects.js";
+import { decodePdfString } from "./pdf-text.js";
+//#region src/optional-content.ts
+function readOptionalContent(catalog, resolver, sink) {
+	const ocProperties = resolver.resolveDict(dictGet(catalog, "OCProperties"));
+	if (ocProperties === void 0) return {
+		layers: [],
+		layerNameOf: () => void 0
+	};
+	const nameByGroup = /* @__PURE__ */ new Map();
+	const groups = [];
+	const ocgs = asArray(resolver.resolve(dictGet(ocProperties, "OCGs"))) ?? [];
+	for (const ocgRef of ocgs) {
+		const ocg = resolver.resolveDict(ocgRef);
+		if (ocg === void 0) {
+			sink({
+				code: "pdf/ocg-unresolved",
+				severity: "warning",
+				message: "an entry in /OCProperties /OCGs did not resolve to a dictionary; skipping it"
+			});
+			continue;
+		}
+		const nameObj = dictGet(ocg, "Name");
+		const name = nameObj?.kind === "string" ? decodePdfString(nameObj.bytes) : mintLayerName(nameByGroup);
+		nameByGroup.set(ocg, name);
+		groups.push({
+			dict: ocg,
+			name
+		});
+	}
+	const defaultConfig = resolver.resolveDict(dictGet(ocProperties, "D"));
+	const baseVisible = defaultConfig === void 0 || asName(dictGet(defaultConfig, "BaseState")) !== "OFF";
+	const groupsIn = (key) => {
+		const set = /* @__PURE__ */ new Set();
+		const arr = defaultConfig === void 0 ? void 0 : asArray(resolver.resolve(dictGet(defaultConfig, key)));
+		if (arr !== void 0) for (const entry of arr) {
+			const dict = resolver.resolveDict(entry);
+			if (dict !== void 0) set.add(dict);
+		}
+		return set;
+	};
+	const on = groupsIn("ON");
+	const off = groupsIn("OFF");
+	return {
+		layers: groups.map(({ dict, name }) => ({
+			name,
+			visible: off.has(dict) ? false : on.has(dict) ? true : baseVisible
+		})),
+		layerNameOf: (obj) => {
+			const dict = resolver.resolveDict(obj);
+			return dict !== void 0 ? nameByGroup.get(dict) : void 0;
+		}
+	};
+}
+function mintLayerName(taken) {
+	const used = new Set(taken.values());
+	let n = 1;
+	while (used.has(`layer${n}`)) n++;
+	return `layer${n}`;
+}
+//#endregion
+export { readOptionalContent };

package/dist/parse.cjs

@@ -1,6 +1,6 @@
 Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
-const require_lexer = require("./lexer.cjs");
 const require_objects = require("./objects.cjs");
+const require_lexer = require("./lexer.cjs");
 //#region src/parse.ts
 function parseValue(reader, sink) {
 	const token = require_lexer.nextToken(reader);

package/dist/parse.d.cts

@@ -1,6 +1,6 @@
-import { ByteReader } from "./bytes/reader.cjs";
 import { PdfDiagnosticSink } from "./diagnostics.cjs";
-import { PdfObject } from "./objects.cjs";
+import { r as PdfObject } from "./objects-B7HdMeWS.cjs";
+import { ByteReader } from "./bytes/reader.cjs";
 //#region src/parse.d.ts
 declare function parseValue(reader: ByteReader, sink: PdfDiagnosticSink): PdfObject | undefined;
 interface ParsedIndirectObject {

package/dist/parse.d.ts

@@ -1,6 +1,6 @@
-import { ByteReader } from "./bytes/reader.js";
 import { PdfDiagnosticSink } from "./diagnostics.js";
-import { PdfObject } from "./objects.js";
+import { r as PdfObject } from "./objects-B7HdMeWS.js";
+import { ByteReader } from "./bytes/reader.js";
 //#region src/parse.d.ts
 declare function parseValue(reader: ByteReader, sink: PdfDiagnosticSink): PdfObject | undefined;
 interface ParsedIndirectObject {

package/dist/parse.js

@@ -1,5 +1,5 @@
-import { nextToken } from "./lexer.js";
 import { dictGet, pdfArray, pdfBool, pdfDict, pdfHexString, pdfLiteralString, pdfName, pdfNull, pdfNum, pdfRef, pdfStream } from "./objects.js";
+import { nextToken } from "./lexer.js";
 //#region src/parse.ts
 function parseValue(reader, sink) {
 	const token = nextToken(reader);

package/dist/pdf-text.cjs

@@ -0,0 +1,21 @@
+Object.defineProperty(exports, Symbol.toStringTag, { value: "Module" });
+//#region src/pdf-text.ts
+function decodePdfString(bytes) {
+	if (bytes.length >= 2 && bytes[0] === 254 && bytes[1] === 255) {
+		let out = "";
+		for (let i = 2; i + 1 < bytes.length; i += 2) out += String.fromCharCode((bytes[i] ?? 0) << 8 | (bytes[i + 1] ?? 0));
+		return out;
+	}
+	return Array.from(bytes, (b) => String.fromCharCode(b)).join("");
+}
+const PDF_DATE_PATTERN = /^D:(\d{4})(\d{2})?(\d{2})?(\d{2})?(\d{2})?(\d{2})?([+\-Z])?(\d{2})?'?(\d{2})?'?$/;
+function parsePdfDate(raw) {
+	if (raw === void 0) return;
+	const match = PDF_DATE_PATTERN.exec(raw);
+	if (match === null) return;
+	const [, year, month = "01", day = "01", hour = "00", minute = "00", second = "00", tzSign, tzHour = "00", tzMinute = "00"] = match;
+	return `${year}-${month}-${day}T${hour}:${minute}:${second}${tzSign === void 0 || tzSign === "Z" ? "Z" : `${tzSign}${tzHour}:${tzMinute}`}`;
+}
+//#endregion
+exports.decodePdfString = decodePdfString;
+exports.parsePdfDate = parsePdfDate;

package/dist/pdf-text.d.cts

@@ -0,0 +1,5 @@
+//#region src/pdf-text.d.ts
+declare function decodePdfString(bytes: Uint8Array<ArrayBuffer>): string;
+declare function parsePdfDate(raw: string | undefined): string | undefined;
+//#endregion
+export { decodePdfString, parsePdfDate };

package/dist/pdf-text.d.ts

@@ -0,0 +1,5 @@
+//#region src/pdf-text.d.ts
+declare function decodePdfString(bytes: Uint8Array<ArrayBuffer>): string;
+declare function parsePdfDate(raw: string | undefined): string | undefined;
+//#endregion
+export { decodePdfString, parsePdfDate };

package/dist/pdf-text.js

@@ -0,0 +1,19 @@
+//#region src/pdf-text.ts
+function decodePdfString(bytes) {
+	if (bytes.length >= 2 && bytes[0] === 254 && bytes[1] === 255) {
+		let out = "";
+		for (let i = 2; i + 1 < bytes.length; i += 2) out += String.fromCharCode((bytes[i] ?? 0) << 8 | (bytes[i + 1] ?? 0));
+		return out;
+	}
+	return Array.from(bytes, (b) => String.fromCharCode(b)).join("");
+}
+const PDF_DATE_PATTERN = /^D:(\d{4})(\d{2})?(\d{2})?(\d{2})?(\d{2})?(\d{2})?([+\-Z])?(\d{2})?'?(\d{2})?'?$/;
+function parsePdfDate(raw) {
+	if (raw === void 0) return;
+	const match = PDF_DATE_PATTERN.exec(raw);
+	if (match === null) return;
+	const [, year, month = "01", day = "01", hour = "00", minute = "00", second = "00", tzSign, tzHour = "00", tzMinute = "00"] = match;
+	return `${year}-${month}-${day}T${hour}:${minute}:${second}${tzSign === void 0 || tzSign === "Z" ? "Z" : `${tzSign}${tzHour}:${tzMinute}`}`;
+}
+//#endregion
+export { decodePdfString, parsePdfDate };

package/dist/predictors.d.cts

@@ -1,5 +1,5 @@
 import { PdfDiagnosticSink } from "./diagnostics.cjs";
-import { PdfDict } from "./objects.cjs";
+import { n as PdfDict } from "./objects-B7HdMeWS.cjs";
 //#region src/predictors.d.ts
 interface PredictorParams {
   readonly predictor: number;

package/dist/predictors.d.ts

@@ -1,5 +1,5 @@
 import { PdfDiagnosticSink } from "./diagnostics.js";
-import { PdfDict } from "./objects.js";
+import { n as PdfDict } from "./objects-B7HdMeWS.js";
 //#region src/predictors.d.ts
 interface PredictorParams {
   readonly predictor: number;