2021-04-30 18:33:13 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/*
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-04 19:53:38 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								 * Copyright (c) 2021-2022, Matthew Olsson <mattco@serenityos.org>
							 | 
						
					
						
							
								
									
										
										
										
											2021-04-30 18:33:13 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * SPDX-License-Identifier: BSD-2-Clause
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#pragma once
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-05 17:30:55 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								#include <AK/SourceLocation.h>
							 | 
						
					
						
							
								
									
										
										
										
											2021-11-28 11:52:48 +01:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								#include <AK/WeakPtr.h>
							 | 
						
					
						
							
								
									
										
										
										
											2021-04-30 18:33:13 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#include <LibPDF/Object.h>
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-25 15:00:11 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								#include <LibPDF/Operator.h>
							 | 
						
					
						
							
								
									
										
										
										
											2021-04-30 18:33:13 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#include <LibPDF/Reader.h>
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#include <LibPDF/XRefTable.h>
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								namespace PDF {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2022-08-15 11:45:24 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								template<typename T, typename... Args>
							 | 
						
					
						
							
								
									
										
										
										
											2022-10-17 00:06:11 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								static NonnullRefPtr<T> make_object(Args... args)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								requires(IsBaseOf<Object, T>)
							 | 
						
					
						
							
								
									
										
										
										
											2022-08-15 11:45:24 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    return adopt_ref(*new T(forward<Args>(args)...));
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2021-04-30 18:33:13 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								class Document;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2022-08-15 11:45:24 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								class Parser {
							 | 
						
					
						
							
								
									
										
										
										
											2021-04-30 18:33:13 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								public:
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-25 15:00:11 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    static PDFErrorOr<Vector<Operator>> parse_operators(Document*, ReadonlyBytes);
							 | 
						
					
						
							
								
									
										
										
										
											2021-05-10 10:39:19 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2022-08-15 11:45:24 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    Parser(ReadonlyBytes);
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-22 19:22:45 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    Parser(Document*, ReadonlyBytes);
							 | 
						
					
						
							
								
									
										
										
										
											2021-04-30 18:33:13 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2021-11-28 11:52:48 +01:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    void set_document(WeakPtr<Document> const&);
							 | 
						
					
						
							
								
									
										
										
										
											2021-05-10 10:33:32 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2023-12-16 17:49:34 +03:30
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    ByteString parse_comment();
							 | 
						
					
						
							
								
									
										
										
										
											2021-04-30 18:33:13 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2022-08-15 13:34:08 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    void move_by(size_t count) { m_reader.move_by(count); }
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    void move_to(size_t offset) { m_reader.move_to(offset); }
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2022-08-20 09:00:51 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    enum class CanBeIndirectValue {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								        No,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								        Yes
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    };
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<Value> parse_value(CanBeIndirectValue = CanBeIndirectValue::Yes);
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-05 17:30:55 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<Value> parse_possible_indirect_value_or_ref();
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-21 11:26:31 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<NonnullRefPtr<IndirectValue>> parse_indirect_value(u32 index, u32 generation);
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-05 17:30:55 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<NonnullRefPtr<IndirectValue>> parse_indirect_value();
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<Value> parse_number();
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<NonnullRefPtr<NameObject>> parse_name();
							 | 
						
					
						
							
								
									
										
										
										
											2023-10-25 00:02:55 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<NonnullRefPtr<StringObject>> parse_string();
							 | 
						
					
						
							
								
									
										
										
										
											2023-12-16 17:49:34 +03:30
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<ByteString> parse_literal_string();
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<ByteString> parse_hex_string();
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-05 17:30:55 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<NonnullRefPtr<ArrayObject>> parse_array();
							 | 
						
					
						
							
								
									
										
										
										
											2023-12-09 22:06:51 -05:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<HashMap<DeprecatedFlyString, Value>> parse_dict_contents_until(char const*);
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-05 17:30:55 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<NonnullRefPtr<DictObject>> parse_dict();
							 | 
						
					
						
							
								
									
										
										
										
											2023-12-18 22:23:41 -05:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<void> unfilter_stream(NonnullRefPtr<StreamObject>);
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-05 17:30:55 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<NonnullRefPtr<StreamObject>> parse_stream(NonnullRefPtr<DictObject> dict);
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-25 15:00:11 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<Vector<Operator>> parse_operators();
							 | 
						
					
						
							
								
									
										
										
										
											2021-05-10 10:39:19 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2023-07-10 15:49:48 -04:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    void set_filters_enabled(bool enabled)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								        m_enable_filters = enabled;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    }
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2023-07-11 12:02:16 -04:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    void set_encryption_enabled(bool enabled)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								        m_enable_encryption = enabled;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    }
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-21 11:26:31 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    void push_reference(Reference const& ref) { m_current_reference_stack.append(ref); }
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    void pop_reference() { m_current_reference_stack.take_last(); }
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2023-07-11 12:02:16 -04:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								protected:
							 | 
						
					
						
							
								
									
										
										
										
											2023-12-21 22:01:43 -05:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    PDFErrorOr<NonnullRefPtr<StreamObject>> parse_inline_image();
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-05 17:30:55 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    Error error(
							 | 
						
					
						
							
								
									
										
										
										
											2023-12-16 17:49:34 +03:30
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								        ByteString const& message
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-05 17:30:55 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								#ifdef PDF_DEBUG
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								        ,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								        SourceLocation loc = SourceLocation::current()
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#endif
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    ) const;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2021-04-30 18:33:13 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								    Reader m_reader;
							 | 
						
					
						
							
								
									
										
										
										
											2021-11-28 11:52:48 +01:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    WeakPtr<Document> m_document;
							 | 
						
					
						
							
								
									
										
										
										
											2022-03-21 11:26:31 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    Vector<Reference> m_current_reference_stack;
							 | 
						
					
						
							
								
									
										
										
										
											2023-07-11 10:16:16 -04:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    bool m_enable_encryption { true };
							 | 
						
					
						
							
								
									
										
										
										
											2023-10-20 22:39:36 -04:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								    bool m_enable_filters { true };
							 | 
						
					
						
							
								
									
										
										
										
											2021-04-30 18:33:13 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								};
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2021-05-26 22:52:05 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								};
							 |