1 | /* |
2 | * Copyright (c) 2023. |
3 | * |
4 | * This software is free software; |
5 | * |
6 | * You can redistribute it or modify it under terms of the MIT, Apache License or Zlib license |
7 | */ |
8 | |
9 | //!Miscellaneous stuff |
10 | #![allow (dead_code)] |
11 | |
12 | use alloc::format; |
13 | use core::cmp::max; |
14 | use core::fmt; |
15 | |
16 | use zune_core::bytestream::{ZByteReader, ZReaderTrait}; |
17 | use zune_core::colorspace::ColorSpace; |
18 | use zune_core::log::trace; |
19 | |
20 | use crate::components::{ComponentID, SampleRatios}; |
21 | use crate::errors::DecodeErrors; |
22 | use crate::huffman::HuffmanTable; |
23 | use crate::JpegDecoder; |
24 | |
25 | /// Start of baseline DCT Huffman coding |
26 | |
27 | pub const START_OF_FRAME_BASE: u16 = 0xffc0; |
28 | |
29 | /// Start of another frame |
30 | |
31 | pub const START_OF_FRAME_EXT_SEQ: u16 = 0xffc1; |
32 | |
33 | /// Start of progressive DCT encoding |
34 | |
35 | pub const START_OF_FRAME_PROG_DCT: u16 = 0xffc2; |
36 | |
37 | /// Start of Lossless sequential Huffman coding |
38 | |
39 | pub const START_OF_FRAME_LOS_SEQ: u16 = 0xffc3; |
40 | |
41 | /// Start of extended sequential DCT arithmetic coding |
42 | |
43 | pub const START_OF_FRAME_EXT_AR: u16 = 0xffc9; |
44 | |
45 | /// Start of Progressive DCT arithmetic coding |
46 | |
47 | pub const START_OF_FRAME_PROG_DCT_AR: u16 = 0xffca; |
48 | |
49 | /// Start of Lossless sequential Arithmetic coding |
50 | |
51 | pub const START_OF_FRAME_LOS_SEQ_AR: u16 = 0xffcb; |
52 | |
53 | /// Undo run length encoding of coefficients by placing them in natural order |
54 | #[rustfmt::skip] |
55 | pub const UN_ZIGZAG: [usize; 64 + 16] = [ |
56 | 0, 1, 8, 16, 9, 2, 3, 10, |
57 | 17, 24, 32, 25, 18, 11, 4, 5, |
58 | 12, 19, 26, 33, 40, 48, 41, 34, |
59 | 27, 20, 13, 6, 7, 14, 21, 28, |
60 | 35, 42, 49, 56, 57, 50, 43, 36, |
61 | 29, 22, 15, 23, 30, 37, 44, 51, |
62 | 58, 59, 52, 45, 38, 31, 39, 46, |
63 | 53, 60, 61, 54, 47, 55, 62, 63, |
64 | // Prevent overflowing |
65 | 63, 63, 63, 63, 63, 63, 63, 63, |
66 | 63, 63, 63, 63, 63, 63, 63, 63 |
67 | ]; |
68 | |
69 | /// Align data to a 16 byte boundary |
70 | #[repr (align(16))] |
71 | #[derive (Clone)] |
72 | |
73 | pub struct Aligned16<T: ?Sized>(pub T); |
74 | |
75 | impl<T> Default for Aligned16<T> |
76 | where |
77 | T: Default |
78 | { |
79 | fn default() -> Self { |
80 | Aligned16(T::default()) |
81 | } |
82 | } |
83 | |
84 | /// Align data to a 32 byte boundary |
85 | #[repr (align(32))] |
86 | #[derive (Clone)] |
87 | pub struct Aligned32<T: ?Sized>(pub T); |
88 | |
89 | impl<T> Default for Aligned32<T> |
90 | where |
91 | T: Default |
92 | { |
93 | fn default() -> Self { |
94 | Aligned32(T::default()) |
95 | } |
96 | } |
97 | |
98 | /// Markers that identify different Start of Image markers |
99 | /// They identify the type of encoding and whether the file use lossy(DCT) or |
100 | /// lossless compression and whether we use Huffman or arithmetic coding schemes |
101 | #[derive (Eq, PartialEq, Copy, Clone)] |
102 | #[allow (clippy::upper_case_acronyms)] |
103 | pub enum SOFMarkers { |
104 | /// Baseline DCT markers |
105 | BaselineDct, |
106 | /// SOF_1 Extended sequential DCT,Huffman coding |
107 | ExtendedSequentialHuffman, |
108 | /// Progressive DCT, Huffman coding |
109 | ProgressiveDctHuffman, |
110 | /// Lossless (sequential), huffman coding, |
111 | LosslessHuffman, |
112 | /// Extended sequential DEC, arithmetic coding |
113 | ExtendedSequentialDctArithmetic, |
114 | /// Progressive DCT, arithmetic coding, |
115 | ProgressiveDctArithmetic, |
116 | /// Lossless ( sequential), arithmetic coding |
117 | LosslessArithmetic |
118 | } |
119 | |
120 | impl Default for SOFMarkers { |
121 | fn default() -> Self { |
122 | Self::BaselineDct |
123 | } |
124 | } |
125 | |
126 | impl SOFMarkers { |
127 | /// Check if a certain marker is sequential DCT or not |
128 | |
129 | pub fn is_sequential_dct(self) -> bool { |
130 | matches!( |
131 | self, |
132 | Self::BaselineDct |
133 | | Self::ExtendedSequentialHuffman |
134 | | Self::ExtendedSequentialDctArithmetic |
135 | ) |
136 | } |
137 | |
138 | /// Check if a marker is a Lossles type or not |
139 | |
140 | pub fn is_lossless(self) -> bool { |
141 | matches!(self, Self::LosslessHuffman | Self::LosslessArithmetic) |
142 | } |
143 | |
144 | /// Check whether a marker is a progressive marker or not |
145 | |
146 | pub fn is_progressive(self) -> bool { |
147 | matches!( |
148 | self, |
149 | Self::ProgressiveDctHuffman | Self::ProgressiveDctArithmetic |
150 | ) |
151 | } |
152 | |
153 | /// Create a marker from an integer |
154 | |
155 | pub fn from_int(int: u16) -> Option<SOFMarkers> { |
156 | match int { |
157 | START_OF_FRAME_BASE => Some(Self::BaselineDct), |
158 | START_OF_FRAME_PROG_DCT => Some(Self::ProgressiveDctHuffman), |
159 | START_OF_FRAME_PROG_DCT_AR => Some(Self::ProgressiveDctArithmetic), |
160 | START_OF_FRAME_LOS_SEQ => Some(Self::LosslessHuffman), |
161 | START_OF_FRAME_LOS_SEQ_AR => Some(Self::LosslessArithmetic), |
162 | START_OF_FRAME_EXT_SEQ => Some(Self::ExtendedSequentialHuffman), |
163 | START_OF_FRAME_EXT_AR => Some(Self::ExtendedSequentialDctArithmetic), |
164 | _ => None |
165 | } |
166 | } |
167 | } |
168 | |
169 | impl fmt::Debug for SOFMarkers { |
170 | fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
171 | match &self { |
172 | Self::BaselineDct => write!(f, "Baseline DCT" ), |
173 | Self::ExtendedSequentialHuffman => { |
174 | write!(f, "Extended sequential DCT, Huffman Coding" ) |
175 | } |
176 | Self::ProgressiveDctHuffman => write!(f, "Progressive DCT,Huffman Encoding" ), |
177 | Self::LosslessHuffman => write!(f, "Lossless (sequential) Huffman encoding" ), |
178 | Self::ExtendedSequentialDctArithmetic => { |
179 | write!(f, "Extended sequential DCT, arithmetic coding" ) |
180 | } |
181 | Self::ProgressiveDctArithmetic => write!(f, "Progressive DCT, arithmetic coding" ), |
182 | Self::LosslessArithmetic => write!(f, "Lossless (sequential) arithmetic coding" ) |
183 | } |
184 | } |
185 | } |
186 | |
187 | /// Read `buf.len()*2` data from the underlying `u8` buffer and convert it into |
188 | /// u16, and store it into `buf` |
189 | /// |
190 | /// # Arguments |
191 | /// - reader: A mutable reference to the underlying reader. |
192 | /// - buf: A mutable reference to a slice containing u16's |
193 | #[inline ] |
194 | pub fn read_u16_into<T>(reader: &mut ZByteReader<T>, buf: &mut [u16]) -> Result<(), DecodeErrors> |
195 | where |
196 | T: ZReaderTrait |
197 | { |
198 | for i: &mut u16 in buf { |
199 | *i = reader.get_u16_be_err()?; |
200 | } |
201 | |
202 | Ok(()) |
203 | } |
204 | |
205 | /// Set up component parameters. |
206 | /// |
207 | /// This modifies the components in place setting up details needed by other |
208 | /// parts fo the decoder. |
209 | pub(crate) fn setup_component_params<T: ZReaderTrait>( |
210 | img: &mut JpegDecoder<T> |
211 | ) -> Result<(), DecodeErrors> { |
212 | let img_width = img.width(); |
213 | let img_height = img.height(); |
214 | |
215 | // in case of adobe app14 being present, zero may indicate |
216 | // either CMYK if components are 4 or RGB if components are 3, |
217 | // see https://docs.oracle.com/javase/6/docs/api/javax/imageio/metadata/doc-files/jpeg_metadata.html |
218 | // so since we may not know how many number of components |
219 | // we have when decoding app14, we have to defer that check |
220 | // until now. |
221 | // |
222 | // We know adobe app14 was present since it's the only one that can modify |
223 | // input colorspace to be CMYK |
224 | if img.components.len() == 3 && img.input_colorspace == ColorSpace::CMYK { |
225 | img.input_colorspace = ColorSpace::RGB; |
226 | } |
227 | |
228 | for component in &mut img.components { |
229 | // compute interleaved image info |
230 | // h_max contains the maximum horizontal component |
231 | img.h_max = max(img.h_max, component.horizontal_sample); |
232 | // v_max contains the maximum vertical component |
233 | img.v_max = max(img.v_max, component.vertical_sample); |
234 | img.mcu_width = img.h_max * 8; |
235 | img.mcu_height = img.v_max * 8; |
236 | // Number of MCU's per width |
237 | img.mcu_x = (usize::from(img.info.width) + img.mcu_width - 1) / img.mcu_width; |
238 | // Number of MCU's per height |
239 | img.mcu_y = (usize::from(img.info.height) + img.mcu_height - 1) / img.mcu_height; |
240 | |
241 | if img.h_max != 1 || img.v_max != 1 { |
242 | // interleaved images have horizontal and vertical sampling factors |
243 | // not equal to 1. |
244 | img.is_interleaved = true; |
245 | } |
246 | // Extract quantization tables from the arrays into components |
247 | let qt_table = *img.qt_tables[component.quantization_table_number as usize] |
248 | .as_ref() |
249 | .ok_or_else(|| { |
250 | DecodeErrors::DqtError(format!( |
251 | "No quantization table for component {:?}" , |
252 | component.component_id |
253 | )) |
254 | })?; |
255 | |
256 | let x = (usize::from(img_width) * component.horizontal_sample + img.h_max - 1) / img.h_max; |
257 | let y = (usize::from(img_height) * component.horizontal_sample + img.h_max - 1) / img.v_max; |
258 | component.x = x; |
259 | component.w2 = img.mcu_x * component.horizontal_sample * 8; |
260 | // probably not needed. :) |
261 | component.y = y; |
262 | component.quantization_table = qt_table; |
263 | // initially stride contains its horizontal sub-sampling |
264 | component.width_stride *= img.mcu_x * 8; |
265 | } |
266 | { |
267 | // Sampling factors are one thing that suck |
268 | // this fixes a specific problem with images like |
269 | // |
270 | // (2 2) None |
271 | // (2 1) H |
272 | // (2 1) H |
273 | // |
274 | // The images exist in the wild, the images are not meant to exist |
275 | // but they do, it's just an annoying horizontal sub-sampling that |
276 | // I don't know why it exists. |
277 | // But it does |
278 | // So we try to cope with that. |
279 | // I am not sure of how to explain how to fix it, but it involved a debugger |
280 | // and to much coke(the legal one) |
281 | // |
282 | // If this wasn't present, self.upsample_dest would have the wrong length |
283 | let mut handle_that_annoying_bug = false; |
284 | |
285 | if let Some(y_component) = img |
286 | .components |
287 | .iter() |
288 | .find(|c| c.component_id == ComponentID::Y) |
289 | { |
290 | if y_component.horizontal_sample == 2 || y_component.vertical_sample == 2 { |
291 | handle_that_annoying_bug = true; |
292 | } |
293 | } |
294 | if handle_that_annoying_bug { |
295 | for comp in &mut img.components { |
296 | if (comp.component_id != ComponentID::Y) |
297 | && (comp.horizontal_sample != 1 || comp.vertical_sample != 1) |
298 | { |
299 | comp.fix_an_annoying_bug = 2; |
300 | } |
301 | } |
302 | } |
303 | } |
304 | |
305 | if img.is_mjpeg { |
306 | fill_default_mjpeg_tables( |
307 | img.is_progressive, |
308 | &mut img.dc_huffman_tables, |
309 | &mut img.ac_huffman_tables |
310 | ); |
311 | } |
312 | |
313 | Ok(()) |
314 | } |
315 | |
316 | ///Calculate number of fill bytes added to the end of a JPEG image |
317 | /// to fill the image |
318 | /// |
319 | /// JPEG usually inserts padding bytes if the image width cannot be evenly divided into |
320 | /// 8 , 16 or 32 chunks depending on the sub sampling ratio. So given a sub-sampling ratio, |
321 | /// and the actual width, this calculates the padded bytes that were added to the image |
322 | /// |
323 | /// # Params |
324 | /// -actual_width: Actual width of the image |
325 | /// -sub_sample: Sub sampling factor of the image |
326 | /// |
327 | /// # Returns |
328 | /// The padded width, this is how long the width is for a particular image |
329 | pub fn calculate_padded_width(actual_width: usize, sub_sample: SampleRatios) -> usize { |
330 | match sub_sample { |
331 | SampleRatios::None | SampleRatios::V => { |
332 | // None+V sends one MCU row, so that's a simple calculation |
333 | ((actual_width + 7) / 8) * 8 |
334 | } |
335 | SampleRatios::H | SampleRatios::HV => { |
336 | // sends two rows, width can be expanded by up to 15 more bytes |
337 | ((actual_width + 15) / 16) * 16 |
338 | } |
339 | } |
340 | } |
341 | |
342 | // https://www.loc.gov/preservation/digital/formats/fdd/fdd000063.shtml |
343 | // "Avery Lee, writing in the rec.video.desktop newsgroup in 2001, commented that "MJPEG, or at |
344 | // least the MJPEG in AVIs having the MJPG fourcc, is restricted JPEG with a fixed -- and |
345 | // *omitted* -- Huffman table. The JPEG must be YCbCr colorspace, it must be 4:2:2, and it must |
346 | // use basic Huffman encoding, not arithmetic or progressive.... You can indeed extract the |
347 | // MJPEG frames and decode them with a regular JPEG decoder, but you have to prepend the DHT |
348 | // segment to them, or else the decoder won't have any idea how to decompress the data. |
349 | // The exact table necessary is given in the OpenDML spec."" |
350 | pub fn fill_default_mjpeg_tables( |
351 | is_progressive: bool, dc_huffman_tables: &mut [Option<HuffmanTable>], |
352 | ac_huffman_tables: &mut [Option<HuffmanTable>] |
353 | ) { |
354 | // Section K.3.3 |
355 | trace!("Filling with default mjpeg tables" ); |
356 | |
357 | if dc_huffman_tables[0].is_none() { |
358 | // Table K.3 |
359 | dc_huffman_tables[0] = Some( |
360 | HuffmanTable::new_unfilled( |
361 | &[ |
362 | 0x00, 0x00, 0x01, 0x05, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x00, 0x00, 0x00, |
363 | 0x00, 0x00, 0x00, 0x00 |
364 | ], |
365 | &[ |
366 | 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0A, 0x0B |
367 | ], |
368 | true, |
369 | is_progressive |
370 | ) |
371 | .unwrap() |
372 | ); |
373 | } |
374 | if dc_huffman_tables[1].is_none() { |
375 | // Table K.4 |
376 | dc_huffman_tables[1] = Some( |
377 | HuffmanTable::new_unfilled( |
378 | &[ |
379 | 0x00, 0x00, 0x03, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x00, |
380 | 0x00, 0x00, 0x00, 0x00 |
381 | ], |
382 | &[ |
383 | 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0A, 0x0B |
384 | ], |
385 | true, |
386 | is_progressive |
387 | ) |
388 | .unwrap() |
389 | ); |
390 | } |
391 | if ac_huffman_tables[0].is_none() { |
392 | // Table K.5 |
393 | ac_huffman_tables[0] = Some( |
394 | HuffmanTable::new_unfilled( |
395 | &[ |
396 | 0x00, 0x00, 0x02, 0x01, 0x03, 0x03, 0x02, 0x04, 0x03, 0x05, 0x05, 0x04, 0x04, |
397 | 0x00, 0x00, 0x01, 0x7D |
398 | ], |
399 | &[ |
400 | 0x01, 0x02, 0x03, 0x00, 0x04, 0x11, 0x05, 0x12, 0x21, 0x31, 0x41, 0x06, 0x13, |
401 | 0x51, 0x61, 0x07, 0x22, 0x71, 0x14, 0x32, 0x81, 0x91, 0xA1, 0x08, 0x23, 0x42, |
402 | 0xB1, 0xC1, 0x15, 0x52, 0xD1, 0xF0, 0x24, 0x33, 0x62, 0x72, 0x82, 0x09, 0x0A, |
403 | 0x16, 0x17, 0x18, 0x19, 0x1A, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2A, 0x34, 0x35, |
404 | 0x36, 0x37, 0x38, 0x39, 0x3A, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49, 0x4A, |
405 | 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59, 0x5A, 0x63, 0x64, 0x65, 0x66, 0x67, |
406 | 0x68, 0x69, 0x6A, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7A, 0x83, 0x84, |
407 | 0x85, 0x86, 0x87, 0x88, 0x89, 0x8A, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, |
408 | 0x99, 0x9A, 0xA2, 0xA3, 0xA4, 0xA5, 0xA6, 0xA7, 0xA8, 0xA9, 0xAA, 0xB2, 0xB3, |
409 | 0xB4, 0xB5, 0xB6, 0xB7, 0xB8, 0xB9, 0xBA, 0xC2, 0xC3, 0xC4, 0xC5, 0xC6, 0xC7, |
410 | 0xC8, 0xC9, 0xCA, 0xD2, 0xD3, 0xD4, 0xD5, 0xD6, 0xD7, 0xD8, 0xD9, 0xDA, 0xE1, |
411 | 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7, 0xE8, 0xE9, 0xEA, 0xF1, 0xF2, 0xF3, 0xF4, |
412 | 0xF5, 0xF6, 0xF7, 0xF8, 0xF9, 0xFA |
413 | ], |
414 | false, |
415 | is_progressive |
416 | ) |
417 | .unwrap() |
418 | ); |
419 | } |
420 | if ac_huffman_tables[1].is_none() { |
421 | // Table K.6 |
422 | ac_huffman_tables[1] = Some( |
423 | HuffmanTable::new_unfilled( |
424 | &[ |
425 | 0x00, 0x00, 0x02, 0x01, 0x02, 0x04, 0x04, 0x03, 0x04, 0x07, 0x05, 0x04, 0x04, |
426 | 0x00, 0x01, 0x02, 0x77 |
427 | ], |
428 | &[ |
429 | 0x00, 0x01, 0x02, 0x03, 0x11, 0x04, 0x05, 0x21, 0x31, 0x06, 0x12, 0x41, 0x51, |
430 | 0x07, 0x61, 0x71, 0x13, 0x22, 0x32, 0x81, 0x08, 0x14, 0x42, 0x91, 0xA1, 0xB1, |
431 | 0xC1, 0x09, 0x23, 0x33, 0x52, 0xF0, 0x15, 0x62, 0x72, 0xD1, 0x0A, 0x16, 0x24, |
432 | 0x34, 0xE1, 0x25, 0xF1, 0x17, 0x18, 0x19, 0x1A, 0x26, 0x27, 0x28, 0x29, 0x2A, |
433 | 0x35, 0x36, 0x37, 0x38, 0x39, 0x3A, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49, |
434 | 0x4A, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59, 0x5A, 0x63, 0x64, 0x65, 0x66, |
435 | 0x67, 0x68, 0x69, 0x6A, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7A, 0x82, |
436 | 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8A, 0x92, 0x93, 0x94, 0x95, 0x96, |
437 | 0x97, 0x98, 0x99, 0x9A, 0xA2, 0xA3, 0xA4, 0xA5, 0xA6, 0xA7, 0xA8, 0xA9, 0xAA, |
438 | 0xB2, 0xB3, 0xB4, 0xB5, 0xB6, 0xB7, 0xB8, 0xB9, 0xBA, 0xC2, 0xC3, 0xC4, 0xC5, |
439 | 0xC6, 0xC7, 0xC8, 0xC9, 0xCA, 0xD2, 0xD3, 0xD4, 0xD5, 0xD6, 0xD7, 0xD8, 0xD9, |
440 | 0xDA, 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7, 0xE8, 0xE9, 0xEA, 0xF2, 0xF3, 0xF4, |
441 | 0xF5, 0xF6, 0xF7, 0xF8, 0xF9, 0xFA |
442 | ], |
443 | false, |
444 | is_progressive |
445 | ) |
446 | .unwrap() |
447 | ); |
448 | } |
449 | } |
450 | |