Thomas G. Lane | 36a4ccc | 1994-09-24 00:00:00 +0000 | [diff] [blame^] | 1 | USING THE IJG JPEG LIBRARY |
| 2 | |
| 3 | Copyright (C) 1994, Thomas G. Lane. |
| 4 | This file is part of the Independent JPEG Group's software. |
| 5 | For conditions of distribution and use, see the accompanying README file. |
| 6 | |
| 7 | |
| 8 | This file describes how to use the IJG JPEG library within an application |
| 9 | program. Read it if you want to write a program that uses the library. |
| 10 | |
| 11 | The file example.c provides heavily commented skeleton code for calling the |
| 12 | JPEG library. Also see jpeglib.h (the include file to be used by application |
| 13 | programs) for full details about data structures and function parameter lists. |
| 14 | The library source code, of course, is the ultimate reference. |
| 15 | |
| 16 | Note that there have been *major* changes from the application interface |
| 17 | presented by IJG version 4 and earlier versions. The old design had several |
| 18 | inherent limitations, and it had accumulated a lot of cruft as we added |
| 19 | features while trying to minimize application-interface changes. We have |
| 20 | sacrificed backward compatibility in the version 5 rewrite, but we think the |
| 21 | improvements justify this. |
| 22 | |
| 23 | |
| 24 | TABLE OF CONTENTS |
| 25 | ----------------- |
| 26 | |
| 27 | Overview: |
| 28 | Functions provided by the library |
| 29 | Outline of typical usage |
| 30 | Basic library usage: |
| 31 | Data formats |
| 32 | Compression details |
| 33 | Decompression details |
| 34 | Mechanics of usage: include files, linking, etc |
| 35 | Advanced features: |
| 36 | Compression parameter selection |
| 37 | Decompression parameter selection |
| 38 | Special color spaces |
| 39 | Error handling |
| 40 | Compressed data handling (source and destination managers) |
| 41 | I/O suspension |
| 42 | Abbreviated datastreams and multiple images |
| 43 | Special markers |
| 44 | Raw (downsampled) image data |
| 45 | Progress monitoring |
| 46 | Memory management |
| 47 | Library compile-time options |
| 48 | Portability considerations |
| 49 | Notes for MS-DOS implementors |
| 50 | |
| 51 | You should read at least the overview and basic usage sections before trying |
| 52 | to program with the library. The sections on advanced features can be read |
| 53 | if and when you need them. |
| 54 | |
| 55 | |
| 56 | OVERVIEW |
| 57 | ======== |
| 58 | |
| 59 | Functions provided by the library |
| 60 | --------------------------------- |
| 61 | |
| 62 | The IJG JPEG library provides C code to read and write JPEG-compressed image |
| 63 | files. The surrounding application program receives or supplies image data a |
| 64 | scanline at a time, using a straightforward uncompressed image format. All |
| 65 | details of color conversion and other preprocessing/postprocessing can be |
| 66 | handled by the library. |
| 67 | |
| 68 | The library includes a substantial amount of code that is not covered by the |
| 69 | JPEG standard but is necessary for typical applications of JPEG. These |
| 70 | functions preprocess the image before JPEG compression or postprocess it after |
| 71 | decompression. They include colorspace conversion, downsampling/upsampling, |
| 72 | and color quantization. The application indirectly selects use of this code |
| 73 | by specifying the format in which it wishes to supply or receive image data. |
| 74 | For example, if colormapped output is requested, then the decompression |
| 75 | library automatically invokes color quantization. |
| 76 | |
| 77 | A wide range of quality vs. speed tradeoffs are possible in JPEG processing, |
| 78 | and even more so in decompression postprocessing. The decompression library |
| 79 | provides multiple implementations that cover most of the useful tradeoffs, |
| 80 | ranging from very-high-quality down to fast-preview operation. On the |
| 81 | compression side we have generally not provided low-quality choices, since |
| 82 | compression is normally less time-critical. It should be understood that the |
| 83 | low-quality modes may not meet the JPEG standard's accuracy requirements; |
| 84 | nonetheless, they are useful for viewers. |
| 85 | |
| 86 | A word about functions *not* provided by the library. We handle a subset of |
| 87 | the ISO JPEG standard; most baseline and extended-sequential JPEG processes |
| 88 | are supported. (Our subset includes all features now in common use.) |
| 89 | Unsupported ISO options include: |
| 90 | * Progressive storage (may be supported in future versions) |
| 91 | * Hierarchical storage |
| 92 | * Lossless JPEG |
| 93 | * Arithmetic entropy coding (unsupported for legal reasons) |
| 94 | * DNL marker |
| 95 | * Nonintegral subsampling ratios |
| 96 | We support both 8- and 12-bit data precision, but this is a compile-time |
| 97 | choice rather than a run-time choice; hence it is difficult to use both |
| 98 | precisions in a single application. |
| 99 | |
| 100 | By itself, the library handles only interchange JPEG datastreams --- in |
| 101 | particular the widely used JFIF file format. The library can be used by |
| 102 | surrounding code to process interchange or abbreviated JPEG datastreams that |
| 103 | are embedded in more complex file formats. (For example, we anticipate that |
| 104 | Sam Leffler's LIBTIFF library will use this code to support the revised TIFF |
| 105 | JPEG format.) |
| 106 | |
| 107 | |
| 108 | Outline of typical usage |
| 109 | ------------------------ |
| 110 | |
| 111 | The rough outline of a JPEG compression operation is: |
| 112 | |
| 113 | Allocate and initialize a JPEG compression object |
| 114 | Specify the destination for the compressed data (eg, a file) |
| 115 | Set parameters for compression, including image size & colorspace |
| 116 | jpeg_start_compress(...); |
| 117 | while (scan lines remain to be written) |
| 118 | jpeg_write_scanlines(...); |
| 119 | jpeg_finish_compress(...); |
| 120 | Release the JPEG compression object |
| 121 | |
| 122 | A JPEG compression object holds parameters and working state for the JPEG |
| 123 | library. We make creation/destruction of the object separate from starting |
| 124 | or finishing compression of an image; the same object can be re-used for a |
| 125 | series of image compression operations. This makes it easy to re-use the |
| 126 | same parameter settings for a sequence of images. Re-use of a JPEG object |
| 127 | also has important implications for processing abbreviated JPEG datastreams, |
| 128 | as discussed later. |
| 129 | |
| 130 | The image data to be compressed is supplied to jpeg_write_scanlines() from |
| 131 | in-memory buffers. If the application is doing file-to-file compression, |
| 132 | reading image data from the source file is the application's responsibility. |
| 133 | The library emits compressed data by calling a "data destination manager", |
| 134 | which typically will write the data into a file; but the application can |
| 135 | provide its own destination manager to do something else. |
| 136 | |
| 137 | Similarly, the rough outline of a JPEG decompression operation is: |
| 138 | |
| 139 | Allocate and initialize a JPEG decompression object |
| 140 | Specify the source of the compressed data (eg, a file) |
| 141 | Call jpeg_read_header() to obtain image info |
| 142 | Set parameters for decompression |
| 143 | jpeg_start_decompress(...); |
| 144 | while (scan lines remain to be read) |
| 145 | jpeg_read_scanlines(...); |
| 146 | jpeg_finish_decompress(...); |
| 147 | Release the JPEG decompression object |
| 148 | |
| 149 | This is comparable to the compression outline except that reading the |
| 150 | datastream header is a separate step. This is helpful because information |
| 151 | about the image's size, colorspace, etc is available when the application |
| 152 | selects decompression parameters. For example, the application can choose an |
| 153 | output scaling ratio that will fit the image into the available screen size. |
| 154 | |
| 155 | The decompression library obtains compressed data by calling a data source |
| 156 | manager, which typically will read the data from a file; but other behaviors |
| 157 | can be obtained with a custom source manager. Decompressed data is delivered |
| 158 | into in-memory buffers passed to jpeg_read_scanlines(). |
| 159 | |
| 160 | It is possible to abort an incomplete compression or decompression operation |
| 161 | by calling jpeg_abort(); or, if you do not need to retain the JPEG object, |
| 162 | simply release it by calling jpeg_destroy(). |
| 163 | |
| 164 | JPEG compression and decompression objects are two separate struct types. |
| 165 | However, they share some common fields, and certain routines such as |
| 166 | jpeg_destroy() can work on either type of object. |
| 167 | |
| 168 | The JPEG library has no static variables: all state is in the compression |
| 169 | or decompression object. Therefore it is possible to process multiple |
| 170 | compression and decompression operations concurrently, using multiple JPEG |
| 171 | objects. |
| 172 | |
| 173 | Both compression and decompression can be done in an incremental memory-to- |
| 174 | memory fashion, if suitable source/destination managers are used. However, |
| 175 | there are some restrictions on the processing that can be done in this mode. |
| 176 | See the section on "I/O suspension" for more details. |
| 177 | |
| 178 | |
| 179 | BASIC LIBRARY USAGE |
| 180 | =================== |
| 181 | |
| 182 | Data formats |
| 183 | ------------ |
| 184 | |
| 185 | Before diving into procedural details, it is helpful to understand the |
| 186 | image data format that the JPEG library expects or returns. |
| 187 | |
| 188 | The standard input image format is a rectangular array of pixels, with each |
| 189 | pixel having the same number of "component" values (color channels). You |
| 190 | must specify how many components there are and the colorspace interpretation |
| 191 | of the components. Most applications will use RGB data (three components |
| 192 | per pixel) or grayscale data (one component per pixel). |
| 193 | |
| 194 | Note that there is no provision for colormapped input. You can feed in a |
| 195 | colormapped image by expanding it to full-color format. However JPEG often |
| 196 | doesn't work very well with colormapped source data, because of dithering |
| 197 | noise. This is discussed in more detail in the JPEG FAQ and the other |
| 198 | references mentioned in the README file. |
| 199 | |
| 200 | Pixels are stored by scanlines, with each scanline running from left to |
| 201 | right. The component values for each pixel are adjacent in the row; for |
| 202 | example, R,G,B,R,G,B,R,G,B,... for 24-bit RGB color. Each scanline is an |
| 203 | array of data type JSAMPLE --- which is typically "unsigned char", unless |
| 204 | you've changed jmorecfg.h. (You can also change the RGB pixel layout, say |
| 205 | to B,G,R order, by modifying jmorecfg.h. But see the restrictions listed in |
| 206 | that file before doing so.) |
| 207 | |
| 208 | A 2-D array of pixels is formed by making a list of pointers to the starts of |
| 209 | scanlines; so the scanlines need not be physically adjacent in memory. Even |
| 210 | if you process just one scanline at a time, you must make a one-element |
| 211 | pointer array to serve this purpose. Pointers to JSAMPLE rows are of type |
| 212 | JSAMPROW, and the pointer to the pointer array is of type JSAMPARRAY. |
| 213 | |
| 214 | The library accepts or supplies one or more complete scanlines per call. |
| 215 | It is not possible to process part of a row at a time. Scanlines are always |
| 216 | processed top-to-bottom. You can process an entire image in one call if you |
| 217 | have it all in memory, but usually it's simplest to process one scanline at |
| 218 | a time. |
| 219 | |
| 220 | For best results, source data values should have the precision specified by |
| 221 | BITS_IN_JSAMPLE (normally 8 bits). For instance, if you choose to compress |
| 222 | data that's only 6 bits/channel, you should left-justify each value in a |
| 223 | byte before passing it to the compressor. If you need to compress data |
| 224 | that has more than 8 bits/channel, compile with BITS_IN_JSAMPLE = 12. |
| 225 | (See "Library compile-time options", later.) |
| 226 | |
| 227 | The data format returned by the decompressor is the same in all details, |
| 228 | except that colormapped data is supported. If you request colormapped |
| 229 | output then the returned data array contains a single JSAMPLE per pixel; |
| 230 | its value is an index into a color map. The color map is represented as |
| 231 | a 2-D JSAMPARRAY in which each row holds the values of one color component, |
| 232 | that is, colormap[i][j] is the value of the i'th color component for pixel |
| 233 | value (map index) j. Note that since the colormap indexes are stored in |
| 234 | JSAMPLEs, the maximum number of colors is limited by the size of JSAMPLE |
| 235 | (ie, at most 256 colors for an 8-bit JPEG library). |
| 236 | |
| 237 | |
| 238 | Compression details |
| 239 | ------------------- |
| 240 | |
| 241 | Here we revisit the JPEG compression outline given in the overview. |
| 242 | |
| 243 | 1. Allocate and initialize a JPEG compression object. |
| 244 | |
| 245 | A JPEG compression object is a "struct jpeg_compress_struct" (plus a bunch of |
| 246 | subsidiary structures which are allocated via malloc(), but the application |
| 247 | doesn't control those directly). This struct can be just a local variable in |
| 248 | the calling routine, if a single routine is going to execute the whole JPEG |
| 249 | compression sequence. Otherwise it can be static or allocated from malloc(). |
| 250 | |
| 251 | You will also need a structure representing a JPEG error handler. The part |
| 252 | of this that the library cares about is a "struct jpeg_error_mgr". If you |
| 253 | are providing your own error handler, you'll typically want to embed the |
| 254 | jpeg_error_mgr struct in a larger structure; this is discussed later under |
| 255 | "Error handling". For now we'll assume you are just using the default error |
| 256 | handler. The default error handler will print JPEG error/warning messages |
| 257 | on stderr, and it will call exit() if a fatal error occurs. |
| 258 | |
| 259 | You must initialize the error handler structure, store a pointer to it into |
| 260 | the JPEG object's "err" field, and then call jpeg_create_compress() to |
| 261 | initialize the rest of the JPEG object. |
| 262 | |
| 263 | Typical code for this step, if you are using the default error handler, is |
| 264 | |
| 265 | struct jpeg_compress_struct cinfo; |
| 266 | struct jpeg_error_mgr jerr; |
| 267 | ... |
| 268 | cinfo.err = jpeg_std_error(&jerr); |
| 269 | jpeg_create_compress(&cinfo); |
| 270 | |
| 271 | jpeg_create_compress allocates a small amount of memory, so it could fail |
| 272 | if you are out of memory. In that case it will exit via the error handler; |
| 273 | that's why the error handler must be initialized first. |
| 274 | |
| 275 | |
| 276 | 2. Specify the destination for the compressed data (eg, a file). |
| 277 | |
| 278 | As previously mentioned, the JPEG library delivers compressed data to a |
| 279 | "data destination" module. The library includes one data destination |
| 280 | module which knows how to write to a stdio stream. You can use your own |
| 281 | destination module if you want to do something else, as discussed later. |
| 282 | |
| 283 | If you use the standard destination module, you must open the target stdio |
| 284 | stream beforehand. Typical code for this step looks like: |
| 285 | |
| 286 | FILE * outfile; |
| 287 | ... |
| 288 | if ((outfile = fopen(filename, "wb")) == NULL) { |
| 289 | fprintf(stderr, "can't open %s\n", filename); |
| 290 | exit(1); |
| 291 | } |
| 292 | jpeg_stdio_dest(&cinfo, outfile); |
| 293 | |
| 294 | where the last line invokes the standard destination module. |
| 295 | |
| 296 | WARNING: it is critical that the binary compressed data be delivered to the |
| 297 | output file unchanged. On non-Unix systems the stdio library may perform |
| 298 | newline translation or otherwise corrupt binary data. To suppress this |
| 299 | behavior, you may need to use a "b" option to fopen (as shown above), or use |
| 300 | setmode() or another routine to put the stdio stream in binary mode. See |
| 301 | cjpeg.c and djpeg.c for code that has been found to work on many systems. |
| 302 | |
| 303 | You can select the data destination after setting other parameters (step 3), |
| 304 | if that's more convenient. You may not change the destination between |
| 305 | calling jpeg_start_compress() and jpeg_finish_compress(). |
| 306 | |
| 307 | |
| 308 | 3. Set parameters for compression, including image size & colorspace. |
| 309 | |
| 310 | You must supply information about the source image by setting the following |
| 311 | fields in the JPEG object (cinfo structure): |
| 312 | |
| 313 | image_width Width of image, in pixels |
| 314 | image_height Height of image, in pixels |
| 315 | input_components Number of color channels (samples per pixel) |
| 316 | in_color_space Color space of source image |
| 317 | |
| 318 | The image dimensions are, hopefully, obvious. JPEG supports image dimensions |
| 319 | of 1 to 64K pixels in either direction. The input color space is typically |
| 320 | RGB or grayscale, and input_components is 3 or 1 accordingly. (See "Special |
| 321 | color spaces", later, for more info.) The in_color_space field must be |
| 322 | assigned one of the J_COLOR_SPACE enum constants, typically JCS_RGB or |
| 323 | JCS_GRAYSCALE. |
| 324 | |
| 325 | JPEG has a large number of compression parameters that determine how the |
| 326 | image is encoded. Most applications don't need or want to know about all |
| 327 | these parameters. You can set all the parameters to reasonable defaults by |
| 328 | calling jpeg_set_defaults(); then, if there are particular values you want |
| 329 | to change, you can do so after that. The "Compression parameter selection" |
| 330 | section tells about all the parameters. |
| 331 | |
| 332 | You must set in_color_space correctly before calling jpeg_set_defaults(), |
| 333 | because the defaults depend on the source image colorspace. However the |
| 334 | other three source image parameters need not be valid until you call |
| 335 | jpeg_start_compress(). There's no harm in calling jpeg_set_defaults() more |
| 336 | than once, if that happens to be convenient. |
| 337 | |
| 338 | Typical code for a 24-bit RGB source image is |
| 339 | |
| 340 | cinfo.image_width = Width; /* image width and height, in pixels */ |
| 341 | cinfo.image_height = Height; |
| 342 | cinfo.input_components = 3; /* # of color components per pixel */ |
| 343 | cinfo.in_color_space = JCS_RGB; /* colorspace of input image */ |
| 344 | |
| 345 | jpeg_set_defaults(&cinfo); |
| 346 | /* Make optional parameter settings here */ |
| 347 | |
| 348 | |
| 349 | 4. jpeg_start_compress(...); |
| 350 | |
| 351 | After you have established the data destination and set all the necessary |
| 352 | source image info and other parameters, call jpeg_start_compress() to begin |
| 353 | a compression cycle. This will initialize internal state, allocate working |
| 354 | storage, and emit the first few bytes of the JPEG datastream header. |
| 355 | |
| 356 | Typical code: |
| 357 | |
| 358 | jpeg_start_compress(&cinfo, TRUE); |
| 359 | |
| 360 | The "TRUE" parameter ensures that a complete JPEG interchange datastream |
| 361 | will be written. This is appropriate in most cases. If you think you might |
| 362 | want to use an abbreviated datastream, read the section on abbreviated |
| 363 | datastreams, below. |
| 364 | |
| 365 | Once you have called jpeg_start_compress(), you may not alter any JPEG |
| 366 | parameters or other fields of the JPEG object until you have completed |
| 367 | the compression cycle. |
| 368 | |
| 369 | |
| 370 | 5. while (scan lines remain to be written) |
| 371 | jpeg_write_scanlines(...); |
| 372 | |
| 373 | Now write all the required image data by calling jpeg_write_scanlines() |
| 374 | one or more times. You can pass one or more scanlines in each call, up |
| 375 | to the total image height. In most applications it is convenient to pass |
| 376 | just one or a few scanlines at a time. The expected format for the passed |
| 377 | data is discussed under "Data formats", above. |
| 378 | |
| 379 | Image data should be written in top-to-bottom scanline order. The JPEG spec |
| 380 | contains some weasel wording about how top and bottom are application-defined |
| 381 | terms (a curious interpretation of the English language...) but if you want |
| 382 | your files to be compatible with everyone else's, you WILL use top-to-bottom |
| 383 | order. If the source data must be read in bottom-to-top order, you can use |
| 384 | the JPEG library's virtual array mechanism to invert the data efficiently. |
| 385 | Examples of this can be found in the sample application cjpeg. |
| 386 | |
| 387 | The library maintains a count of the number of scanlines written so far |
| 388 | in the next_scanline field of the JPEG object. Usually you can just use |
| 389 | this variable as the loop counter, so that the loop test looks like |
| 390 | "while (cinfo.next_scanline < cinfo.image_height)". |
| 391 | |
| 392 | Code for this step depends heavily on the way that you store the source data. |
| 393 | example.c shows the following code for the case of a full-size 2-D source |
| 394 | array containing 3-byte RGB pixels: |
| 395 | |
| 396 | JSAMPROW row_pointer[1]; /* pointer to a single row */ |
| 397 | int row_stride; /* physical row width in buffer */ |
| 398 | |
| 399 | row_stride = image_width * 3; /* JSAMPLEs per row in image_buffer */ |
| 400 | |
| 401 | while (cinfo.next_scanline < cinfo.image_height) { |
| 402 | row_pointer[0] = & image_buffer[cinfo.next_scanline * row_stride]; |
| 403 | jpeg_write_scanlines(&cinfo, row_pointer, 1); |
| 404 | } |
| 405 | |
| 406 | jpeg_write_scanlines() returns the number of scanlines actually written. |
| 407 | This will normally be equal to the number passed in, so you can usually |
| 408 | ignore the return value. It is different in just two cases: |
| 409 | * If you try to write more scanlines than the declared image height, |
| 410 | the additional scanlines are ignored. |
| 411 | * If you use a suspending data destination manager, output buffer overrun |
| 412 | will cause the compressor to return before accepting all the passed lines. |
| 413 | This feature is discussed under "I/O suspension", below. The normal |
| 414 | stdio destination manager will NOT cause this to happen. |
| 415 | In any case, the return value is the same as the change in the value of |
| 416 | next_scanline. |
| 417 | |
| 418 | |
| 419 | 6. jpeg_finish_compress(...); |
| 420 | |
| 421 | After all the image data has been written, call jpeg_finish_compress() to |
| 422 | complete the compression cycle. This step is ESSENTIAL to ensure that the |
| 423 | last bufferload of data is written to the data destination. |
| 424 | jpeg_finish_compress() also releases working memory associated with the JPEG |
| 425 | object. |
| 426 | |
| 427 | Typical code: |
| 428 | |
| 429 | jpeg_finish_compress(&cinfo); |
| 430 | |
| 431 | If using the stdio destination manager, don't forget to close the output |
| 432 | stdio stream if necessary. |
| 433 | |
| 434 | If you have requested a multi-pass operating mode, such as Huffman code |
| 435 | optimization, jpeg_finish_compress() will perform the additional passes using |
| 436 | data buffered by the first pass. In this case jpeg_finish_compress() may take |
| 437 | quite a while to complete. With the default compression parameters, this will |
| 438 | not happen. |
| 439 | |
| 440 | It is an error to call jpeg_finish_compress() before writing the necessary |
| 441 | total number of scanlines. If you wish to abort compression, call |
| 442 | jpeg_abort() as discussed below. |
| 443 | |
| 444 | After completing a compression cycle, you may dispose of the JPEG object |
| 445 | as discussed next, or you may use it to compress another image. In that case |
| 446 | return to step 2, 3, or 4 as appropriate. If you do not change the |
| 447 | destination manager, the new datastream will be written to the same target. |
| 448 | If you do not change any JPEG parameters, the new datastream will be written |
| 449 | with the same parameters as before. Note that you can change the input image |
| 450 | dimensions freely between cycles, but if you change the input colorspace, you |
| 451 | should call jpeg_set_defaults() to adjust for the new colorspace; and then |
| 452 | you'll need to repeat all of step 3. |
| 453 | |
| 454 | |
| 455 | 7. Release the JPEG compression object. |
| 456 | |
| 457 | When you are done with a JPEG compression object, destroy it by calling |
| 458 | jpeg_destroy_compress(). This will free all subsidiary memory. Or you can |
| 459 | call jpeg_destroy() which works for either compression or decompression |
| 460 | objects --- this may be more convenient if you are sharing code between |
| 461 | compression and decompression cases. (Actually, these routines are equivalent |
| 462 | except for the declared type of the passed pointer. To avoid gripes from |
| 463 | ANSI C compilers, pass a j_common_ptr to jpeg_destroy().) |
| 464 | |
| 465 | If you allocated the jpeg_compress_struct structure from malloc(), freeing |
| 466 | it is your responsibility --- jpeg_destroy() won't. Ditto for the error |
| 467 | handler structure. |
| 468 | |
| 469 | Typical code: |
| 470 | |
| 471 | jpeg_destroy_compress(&cinfo); |
| 472 | |
| 473 | |
| 474 | 8. Aborting. |
| 475 | |
| 476 | If you decide to abort a compression cycle before finishing, you can clean up |
| 477 | in either of two ways: |
| 478 | |
| 479 | * If you don't need the JPEG object any more, just call |
| 480 | jpeg_destroy_compress() or jpeg_destroy() to release memory. This is |
| 481 | legitimate at any point after calling jpeg_create_compress() --- in fact, |
| 482 | it's safe even if jpeg_create_compress() fails. |
| 483 | |
| 484 | * If you want to re-use the JPEG object, call jpeg_abort_compress(), or |
| 485 | jpeg_abort() which works on both compression and decompression objects. |
| 486 | This will return the object to an idle state, releasing any working memory. |
| 487 | jpeg_abort() is allowed at any time after successful object creation. |
| 488 | |
| 489 | Note that cleaning up the data destination, if required, is your |
| 490 | responsibility. |
| 491 | |
| 492 | |
| 493 | Decompression details |
| 494 | --------------------- |
| 495 | |
| 496 | Here we revisit the JPEG decompression outline given in the overview. |
| 497 | |
| 498 | 1. Allocate and initialize a JPEG decompression object. |
| 499 | |
| 500 | This is just like initialization for compression, as discussed above, |
| 501 | except that the object is a "struct jpeg_decompress_struct" and you |
| 502 | call jpeg_create_decompress(). Error handling is exactly the same. |
| 503 | |
| 504 | Typical code: |
| 505 | |
| 506 | struct jpeg_decompress_struct cinfo; |
| 507 | struct jpeg_error_mgr jerr; |
| 508 | ... |
| 509 | cinfo.err = jpeg_std_error(&jerr); |
| 510 | jpeg_create_decompress(&cinfo); |
| 511 | |
| 512 | (Both here and in the IJG code, we usually use variable name "cinfo" for |
| 513 | both compression and decompression objects.) |
| 514 | |
| 515 | |
| 516 | 2. Specify the source of the compressed data (eg, a file). |
| 517 | |
| 518 | As previously mentioned, the JPEG library reads compressed data from a "data |
| 519 | source" module. The library includes one data source module which knows how |
| 520 | to read from a stdio stream. You can use your own source module if you want |
| 521 | to do something else, as discussed later. |
| 522 | |
| 523 | If you use the standard source module, you must open the source stdio stream |
| 524 | beforehand. Typical code for this step looks like: |
| 525 | |
| 526 | FILE * infile; |
| 527 | ... |
| 528 | if ((infile = fopen(filename, "rb")) == NULL) { |
| 529 | fprintf(stderr, "can't open %s\n", filename); |
| 530 | exit(1); |
| 531 | } |
| 532 | jpeg_stdio_src(&cinfo, infile); |
| 533 | |
| 534 | where the last line invokes the standard source module. |
| 535 | |
| 536 | WARNING: it is critical that the binary compressed data be read unchanged. |
| 537 | On non-Unix systems the stdio library may perform newline translation or |
| 538 | otherwise corrupt binary data. To suppress this behavior, you may need to use |
| 539 | a "b" option to fopen (as shown above), or use setmode() or another routine to |
| 540 | put the stdio stream in binary mode. See cjpeg.c and djpeg.c for code that |
| 541 | has been found to work on many systems. |
| 542 | |
| 543 | You may not change the data source between calling jpeg_read_header() and |
| 544 | jpeg_finish_decompress(). If you wish to read a series of JPEG images from |
| 545 | a single source file, you should repeat the jpeg_read_header() to |
| 546 | jpeg_finish_decompress() sequence without reinitializing either the JPEG |
| 547 | object or the data source module; this prevents buffered input data from |
| 548 | being discarded. |
| 549 | |
| 550 | |
| 551 | 3. Call jpeg_read_header() to obtain image info. |
| 552 | |
| 553 | Typical code for this step is just |
| 554 | |
| 555 | jpeg_read_header(&cinfo, TRUE); |
| 556 | |
| 557 | This will read the source datastream header markers, up to the beginning |
| 558 | of the compressed data proper. On return, the image dimensions and other |
| 559 | info have been stored in the JPEG object. The application may wish to |
| 560 | consult this information before selecting decompression parameters. |
| 561 | |
| 562 | More complex code is necessary if |
| 563 | * A suspending data source is used --- in that case jpeg_read_header() |
| 564 | may return before it has read all the header data. See "I/O suspension", |
| 565 | below. The normal stdio source manager will NOT cause this to happen. |
| 566 | * Abbreviated JPEG files are to be processed --- see the section on |
| 567 | abbreviated datastreams. Standard applications that deal only in |
| 568 | interchange JPEG files need not be concerned with this case either. |
| 569 | |
| 570 | It is permissible to stop at this point if you just wanted to find out the |
| 571 | image dimensions and other header info for a JPEG file. In that case, |
| 572 | call jpeg_destroy() when you are done with the JPEG object, or call |
| 573 | jpeg_abort() to return it to an idle state before selecting a new data |
| 574 | source and reading another header. |
| 575 | |
| 576 | |
| 577 | 4. Set parameters for decompression. |
| 578 | |
| 579 | jpeg_read_header() sets appropriate default decompression parameters based on |
| 580 | the properties of the image (in particular, its colorspace). However, you |
| 581 | may well want to alter these defaults before beginning the decompression. |
| 582 | For example, the default is to produce full color output from a color file. |
| 583 | If you want colormapped output you must ask for it. Other options allow the |
| 584 | returned image to be scaled and allow various speed/quality tradeoffs to be |
| 585 | selected. "Decompression parameter selection", below, gives details. |
| 586 | |
| 587 | If the defaults are appropriate, nothing need be done at this step. |
| 588 | |
| 589 | Note that all default values are set by each call to jpeg_read_header(). |
| 590 | If you reuse a decompression object, you cannot expect your parameter |
| 591 | settings to be preserved across cycles, as you can for compression. |
| 592 | You must adjust parameter values each time. |
| 593 | |
| 594 | |
| 595 | 5. jpeg_start_decompress(...); |
| 596 | |
| 597 | Once the parameter values are satisfactory, call jpeg_start_decompress() to |
| 598 | begin decompression. This will initialize internal state, allocate working |
| 599 | memory, and prepare for returning data. |
| 600 | |
| 601 | Typical code is just |
| 602 | |
| 603 | jpeg_start_decompress(&cinfo); |
| 604 | |
| 605 | If you have requested a multi-pass operating mode, such as 2-pass color |
| 606 | quantization, jpeg_start_decompress() will do everything needed before data |
| 607 | output can begin. In this case jpeg_start_decompress() may take quite a while |
| 608 | to complete. With a single-scan (fully interleaved) JPEG file and default |
| 609 | decompression parameters, this will not happen; jpeg_start_decompress() will |
| 610 | return quickly. |
| 611 | |
| 612 | After this call, the final output image dimensions, including any requested |
| 613 | scaling, are available in the JPEG object; so is the selected colormap, if |
| 614 | colormapped output has been requested. Useful fields include |
| 615 | |
| 616 | output_width image width and height, as scaled |
| 617 | output_height |
| 618 | out_color_components # of color components in out_color_space |
| 619 | output_components # of color components returned per pixel |
| 620 | colormap the selected colormap, if any |
| 621 | actual_number_of_colors number of entries in colormap |
| 622 | |
| 623 | output_components is 1 (a colormap index) when quantizing colors; otherwise it |
| 624 | equals out_color_components. It is the number of JSAMPLE values that will be |
| 625 | emitted per pixel in the output arrays. |
| 626 | |
| 627 | Typically you will need to allocate data buffers to hold the incoming image. |
| 628 | You will need output_width * output_components JSAMPLEs per scanline in your |
| 629 | output buffer, and a total of output_height scanlines will be returned. |
| 630 | |
| 631 | Note: if you are using the JPEG library's internal memory manager to allocate |
| 632 | data buffers (as djpeg does), then the manager's protocol requires that you |
| 633 | request large buffers *before* calling jpeg_start_decompress(). This is a |
| 634 | little tricky since the output_XXX fields are not normally valid then. You |
| 635 | can make them valid by calling jpeg_calc_output_dimensions() after setting the |
| 636 | relevant parameters (scaling, output color space, and quantization flag). |
| 637 | |
| 638 | |
| 639 | 6. while (scan lines remain to be read) |
| 640 | jpeg_read_scanlines(...); |
| 641 | |
| 642 | Now you can read the decompressed image data by calling jpeg_read_scanlines() |
| 643 | one or more times. At each call, you pass in the maximum number of scanlines |
| 644 | to be read (ie, the height of your working buffer); jpeg_read_scanlines() |
| 645 | will return up to that many lines. The return value is the number of lines |
| 646 | actually read. The format of the returned data is discussed under "Data |
| 647 | formats", above. |
| 648 | |
| 649 | Image data is returned in top-to-bottom scanline order. If you must write |
| 650 | out the image in bottom-to-top order, you can use the JPEG library's virtual |
| 651 | array mechanism to invert the data efficiently. Examples of this can be |
| 652 | found in the sample application djpeg. |
| 653 | |
| 654 | The library maintains a count of the number of scanlines returned so far |
| 655 | in the output_scanline field of the JPEG object. Usually you can just use |
| 656 | this variable as the loop counter, so that the loop test looks like |
| 657 | "while (cinfo.output_scanline < cinfo.output_height)". (Note that the test |
| 658 | should NOT be against image_height, unless you never use scaling. The |
| 659 | image_height field is the height of the original unscaled image.) |
| 660 | |
| 661 | If you don't use a suspending data source, it is safe to assume that |
| 662 | jpeg_read_scanlines() reads at least one scanline per call, until the |
| 663 | bottom of the image has been reached. If you use a buffer larger than one |
| 664 | scanline, it is NOT safe to assume that jpeg_read_scanlines() fills it. |
| 665 | In any case, the return value is the same as the change in the value of |
| 666 | output_scanline. |
| 667 | |
| 668 | |
| 669 | 7. jpeg_finish_decompress(...); |
| 670 | |
| 671 | After all the image data has been read, call jpeg_finish_decompress() to |
| 672 | complete the decompression cycle. This causes working memory associated |
| 673 | with the JPEG object to be released. |
| 674 | |
| 675 | Typical code: |
| 676 | |
| 677 | jpeg_finish_decompress(&cinfo); |
| 678 | |
| 679 | If using the stdio source manager, don't forget to close the source stdio |
| 680 | stream if necessary. |
| 681 | |
| 682 | It is an error to call jpeg_finish_decompress() before reading the correct |
| 683 | total number of scanlines. If you wish to abort compression, call |
| 684 | jpeg_abort() as discussed below. |
| 685 | |
| 686 | After completing a decompression cycle, you may dispose of the JPEG object as |
| 687 | discussed next, or you may use it to decompress another image. In that case |
| 688 | return to step 2 or 3 as appropriate. If you do not change the source |
| 689 | manager, the next image will be read from the same source. |
| 690 | |
| 691 | |
| 692 | 8. Release the JPEG decompression object. |
| 693 | |
| 694 | When you are done with a JPEG decompression object, destroy it by calling |
| 695 | jpeg_destroy_decompress() or jpeg_destroy(). The previous discussion of |
| 696 | destroying compression objects applies here too. |
| 697 | |
| 698 | Typical code: |
| 699 | |
| 700 | jpeg_destroy_decompress(&cinfo); |
| 701 | |
| 702 | |
| 703 | 9. Aborting. |
| 704 | |
| 705 | You can abort a decompression cycle by calling jpeg_destroy_decompress() or |
| 706 | jpeg_destroy() if you don't need the JPEG object any more, or |
| 707 | jpeg_abort_decompress() or jpeg_abort() if you want to reuse the object. |
| 708 | The previous discussion of aborting compression cycles applies here too. |
| 709 | |
| 710 | |
| 711 | Mechanics of usage: include files, linking, etc |
| 712 | ----------------------------------------------- |
| 713 | |
| 714 | Applications using the JPEG library should include the header file jpeglib.h |
| 715 | to obtain declarations of data types and routines. Before including |
| 716 | jpeglib.h, include system headers that define at least the typedefs FILE and |
| 717 | size_t. On ANSI-conforming systems, including <stdio.h> is sufficient; on |
| 718 | older Unix systems, you may need <sys/types.h> to define size_t. |
| 719 | |
| 720 | If the application needs to refer to individual JPEG library error codes, also |
| 721 | include jerror.h to define those symbols. |
| 722 | |
| 723 | jpeglib.h indirectly includes the files jconfig.h and jmorecfg.h. If you are |
| 724 | installing the JPEG header files in a system directory, you will want to |
| 725 | install all four files: jpeglib.h, jerror.h, jconfig.h, jmorecfg.h. |
| 726 | |
| 727 | The most convenient way to include the JPEG code into your executable program |
| 728 | is to prepare a library file ("libjpeg.a", or a corresponding name on non-Unix |
| 729 | machines) and reference it at your link step. If you use only half of the |
| 730 | library (only compression or only decompression), only that much code will be |
| 731 | included from the library, unless your linker is hopelessly brain-damaged. |
| 732 | The supplied makefiles build libjpeg.a automatically (see install.doc). |
| 733 | |
| 734 | On some systems your application may need to set up a signal handler to ensure |
| 735 | that temporary files are deleted if the program is interrupted. This is most |
| 736 | critical if you are on MS-DOS and use the jmemdos.c memory manager back end; |
| 737 | it will try to grab extended memory for temp files, and that space will NOT be |
| 738 | freed automatically. See cjpeg.c or djpeg.c for an example signal handler. |
| 739 | |
| 740 | It may be worth pointing out that the core JPEG library does not actually |
| 741 | require the stdio library: only the default source/destination managers and |
| 742 | error handler need it. You can use the library in a stdio-less environment |
| 743 | if you replace those modules and use jmemnobs.c (or another memory manager of |
| 744 | your own devising). More info about the minimum system library requirements |
| 745 | may be found in jinclude.h. |
| 746 | |
| 747 | |
| 748 | ADVANCED FEATURES |
| 749 | ================= |
| 750 | |
| 751 | Compression parameter selection |
| 752 | ------------------------------- |
| 753 | |
| 754 | This section describes all the optional parameters you can set for JPEG |
| 755 | compression, as well as the "helper" routines provided to assist in this |
| 756 | task. Proper setting of some parameters requires detailed understanding |
| 757 | of the JPEG standard; if you don't know what a parameter is for, it's best |
| 758 | not to mess with it! See REFERENCES in the README file for pointers to |
| 759 | more info about JPEG. |
| 760 | |
| 761 | It's a good idea to call jpeg_set_defaults() first, even if you plan to set |
| 762 | all the parameters; that way your code is more likely to work with future JPEG |
| 763 | libraries that have additional parameters. For the same reason, we recommend |
| 764 | you use a helper routine where one is provided, in preference to twiddling |
| 765 | cinfo fields directly. |
| 766 | |
| 767 | The helper routines are: |
| 768 | |
| 769 | jpeg_set_defaults (j_compress_ptr cinfo) |
| 770 | This routine sets all JPEG parameters to reasonable defaults, using |
| 771 | only the input image's color space (field in_color_space, which must |
| 772 | already be set in cinfo). Many applications will only need to use |
| 773 | this routine and perhaps jpeg_set_quality(). |
| 774 | |
| 775 | jpeg_set_colorspace (j_compress_ptr cinfo, J_COLOR_SPACE colorspace) |
| 776 | Sets the JPEG file's colorspace (field jpeg_color_space) as specified, |
| 777 | and sets other color-space-dependent parameters appropriately. See |
| 778 | "Special color spaces", below, before using this. A large number of |
| 779 | parameters, including all per-component parameters, are set by this |
| 780 | routine; if you want to twiddle individual parameters you should call |
| 781 | jpeg_set_colorspace() before rather than after. |
| 782 | |
| 783 | jpeg_default_colorspace (j_compress_ptr cinfo) |
| 784 | Selects an appropriate JPEG colorspace based on cinfo->in_color_space, |
| 785 | and calls jpeg_set_colorspace(). This is actually a subroutine of |
| 786 | jpeg_set_defaults(). It's broken out in case you want to change |
| 787 | just the colorspace-dependent JPEG parameters. |
| 788 | |
| 789 | jpeg_set_quality (j_compress_ptr cinfo, int quality, boolean force_baseline) |
| 790 | Constructs JPEG quantization tables appropriate for the indicated |
| 791 | quality setting. The quality value is expressed on the 0..100 scale |
| 792 | recommended by IJG (cjpeg's "-quality" switch uses this routine). |
| 793 | Note that the exact mapping from quality values to tables may change |
| 794 | in future IJG releases as more is learned about DCT quantization. |
| 795 | If the force_baseline parameter is TRUE, then the quantization table |
| 796 | entries are constrained to the range 1..255 for full JPEG baseline |
| 797 | compatibility. In the current implementation, this only makes a |
| 798 | difference for quality settings below 25, and it effectively prevents |
| 799 | very small/low quality files from being generated. The IJG decoder |
| 800 | is capable of reading the non-baseline files generated at low quality |
| 801 | settings when force_baseline is FALSE, but other decoders may not be. |
| 802 | |
| 803 | jpeg_set_linear_quality (j_compress_ptr cinfo, int scale_factor, |
| 804 | boolean force_baseline) |
| 805 | Same as jpeg_set_quality() except that the generated tables are the |
| 806 | sample tables given in the JPEC spec section K.1, multiplied by the |
| 807 | specified scale factor (which is expressed as a percentage; thus |
| 808 | scale_factor = 100 reproduces the spec's tables). Note that larger |
| 809 | scale factors give lower quality. This entry point is useful for |
| 810 | conforming to the Adobe PostScript DCT conventions, but we do not |
| 811 | recommend linear scaling as a user-visible quality scale otherwise. |
| 812 | force_baseline again constrains the computed table entries to 1..255. |
| 813 | |
| 814 | int jpeg_quality_scaling (int quality) |
| 815 | Converts a value on the IJG-recommended quality scale to a linear |
| 816 | scaling percentage. Note that this routine may change or go away |
| 817 | in future releases --- IJG may choose to adopt a scaling method that |
| 818 | can't be expressed as a simple scalar multiplier, in which case the |
| 819 | premise of this routine collapses. Caveat user. |
| 820 | |
| 821 | jpeg_add_quant_table (j_compress_ptr cinfo, int which_tbl, |
| 822 | const unsigned int *basic_table, |
| 823 | int scale_factor, boolean force_baseline)); |
| 824 | Allows an arbitrary quantization table to be created. which_tbl |
| 825 | indicates which table slot to fill. basic_table points to an array |
| 826 | of 64 unsigned ints given in JPEG zigzag order. These values are |
| 827 | multiplied by scale_factor/100 and then clamped to the range 1..65535 |
| 828 | (or to 1..255 if force_baseline is TRUE). |
| 829 | |
| 830 | |
| 831 | Compression parameters (cinfo fields) include: |
| 832 | |
| 833 | boolean optimize_coding |
| 834 | TRUE causes the compressor to compute optimal Huffman coding tables |
| 835 | for the image. This requires an extra pass over the data and |
| 836 | therefore costs a good deal of space and time. The default is |
| 837 | FALSE, which tells the compressor to use the supplied or default |
| 838 | Huffman tables. In most cases optimal tables save only a few percent |
| 839 | of file size compared to the default tables. Note that when this is |
| 840 | TRUE, you need not supply Huffman tables at all, and any you do |
| 841 | supply will be overwritten. |
| 842 | |
| 843 | int smoothing_factor |
| 844 | If non-zero, the input image is smoothed; the value should be 1 for |
| 845 | minimal smoothing to 100 for maximum smoothing. Consult jcsample.c |
| 846 | for details of the smoothing algorithm. The default is zero. |
| 847 | |
| 848 | J_DCT_METHOD dct_method |
| 849 | Selects the algorithm used for the DCT step. Choices are: |
| 850 | JDCT_ISLOW: slow but accurate integer algorithm |
| 851 | JDCT_IFAST: faster, less accurate integer method |
| 852 | JDCT_FLOAT: floating-point method |
| 853 | JDCT_DEFAULT: default method (normally JDCT_ISLOW) |
| 854 | JDCT_FASTEST: fastest method (normally JDCT_IFAST) |
| 855 | The floating-point method is the most accurate, but may give slightly |
| 856 | different results on different machines due to varying roundoff |
| 857 | behavior. The integer methods should give the same results on all |
| 858 | machines. On machines with sufficiently fast FP hardware, the |
| 859 | floating-point method may also be the fastest. The IFAST method is |
| 860 | considerably less accurate than the other two; its use is not |
| 861 | recommended if high quality is a concern. JDCT_DEFAULT and |
| 862 | JDCT_FASTEST are macros configurable by each installation. |
| 863 | |
| 864 | unsigned int restart_interval |
| 865 | int restart_in_rows |
| 866 | To emit restart markers in the JPEG file, set one of these nonzero. |
| 867 | Set restart_interval to specify the exact interval in MCU blocks. |
| 868 | Set restart_in_rows to specify the interval in MCU rows. (If |
| 869 | restart_in_rows is not 0, then restart_interval is set after the |
| 870 | image width in MCUs is computed.) Defaults are zero (no restarts). |
| 871 | |
| 872 | J_COLOR_SPACE jpeg_color_space |
| 873 | int num_components |
| 874 | The JPEG color space and corresponding number of components; see |
| 875 | "Special color spaces", below, for more info. We recommend using |
| 876 | jpeg_set_color_space() if you want to change these. |
| 877 | |
| 878 | boolean write_JFIF_header |
| 879 | If TRUE, a JFIF APP0 marker is emitted. jpeg_set_defaults() and |
| 880 | jpeg_set_colorspace() set this TRUE if a JFIF-legal JPEG color space |
| 881 | (ie, YCbCr or grayscale) is selected, otherwise FALSE. |
| 882 | |
| 883 | UINT8 density_unit |
| 884 | UINT16 X_density |
| 885 | UINT16 Y_density |
| 886 | The resolution information to be written into the JFIF marker; |
| 887 | not used otherwise. density_unit may be 0 for unknown, |
| 888 | 1 for dots/inch, or 2 for dots/cm. The default values are 0,1,1 |
| 889 | indicating square pixels of unknown size. |
| 890 | |
| 891 | boolean write_Adobe_marker |
| 892 | If TRUE, an Adobe APP14 marker is emitted. jpeg_set_defaults() and |
| 893 | jpeg_set_colorspace() set this TRUE if JPEG color space RGB, CMYK, |
| 894 | or YCCK is selected, otherwise FALSE. It is generally a bad idea |
| 895 | to set both write_JFIF_header and write_Adobe_marker. In fact, |
| 896 | you probably shouldn't change the default settings at all --- the |
| 897 | default behavior ensures that the JPEG file's color space can be |
| 898 | recognized by the decoder. |
| 899 | |
| 900 | JQUANT_TBL * quant_tbl_ptrs[NUM_QUANT_TBLS] |
| 901 | Pointers to coefficient quantization tables, one per table slot, |
| 902 | or NULL if no table is defined for a slot. Usually these should |
| 903 | be set via one of the above helper routines; jpeg_add_quant_table() |
| 904 | is general enough to define any quantization table. The other |
| 905 | routines will set up table slot 0 for luminance quality and table |
| 906 | slot 1 for chrominance. |
| 907 | |
| 908 | JHUFF_TBL * dc_huff_tbl_ptrs[NUM_HUFF_TBLS] |
| 909 | JHUFF_TBL * ac_huff_tbl_ptrs[NUM_HUFF_TBLS] |
| 910 | Pointers to Huffman coding tables, one per table slot, or NULL if |
| 911 | no table is defined for a slot. Slots 0 and 1 are filled with the |
| 912 | JPEG sample tables by jpeg_set_defaults(). If you need to allocate |
| 913 | more table structures, jpeg_alloc_huff_table() may be used. |
| 914 | Note that optimal Huffman tables can be computed for an image |
| 915 | by setting optimize_coding, as discussed above; there's seldom |
| 916 | any need to mess with providing your own Huffman tables. |
| 917 | |
| 918 | There are some additional cinfo fields which are not documented here |
| 919 | because you currently can't change them; for example, you can't set |
| 920 | arith_code TRUE because arithmetic coding is unsupported. |
| 921 | |
| 922 | |
| 923 | Per-component parameters are stored in the struct cinfo.comp_info[i] for |
| 924 | component number i. Note that components here refer to components of the |
| 925 | JPEG color space, *not* the source image color space. A suitably large |
| 926 | comp_info[] array is allocated by jpeg_set_defaults(); if you choose not |
| 927 | to use that routine, it's up to you to allocate the array. |
| 928 | |
| 929 | int component_id |
| 930 | The one-byte identifier code to be recorded in the JPEG file for |
| 931 | this component. For the standard color spaces, we recommend you |
| 932 | leave the default values alone. |
| 933 | |
| 934 | int h_samp_factor |
| 935 | int v_samp_factor |
| 936 | Horizontal and vertical sampling factors for the component; must |
| 937 | be 1..4 according to the JPEG standard. Note that larger sampling |
| 938 | factors indicate a higher-resolution component; many people find |
| 939 | this behavior quite unintuitive. The default values are 2,2 for |
| 940 | luminance components and 1,1 for chrominance components, except |
| 941 | for grayscale where 1,1 is used. |
| 942 | |
| 943 | int quant_tbl_no |
| 944 | Quantization table number for component. The default value is |
| 945 | 0 for luminance components and 1 for chrominance components. |
| 946 | |
| 947 | int dc_tbl_no |
| 948 | int ac_tbl_no |
| 949 | DC and AC entropy coding table numbers. The default values are |
| 950 | 0 for luminance components and 1 for chrominance components. |
| 951 | |
| 952 | int component_index |
| 953 | Must equal the component's index in comp_info[]. |
| 954 | |
| 955 | |
| 956 | Decompression parameter selection |
| 957 | --------------------------------- |
| 958 | |
| 959 | Decompression parameter selection is somewhat simpler than compression |
| 960 | parameter selection, since all of the JPEG internal parameters are |
| 961 | recorded in the source file and need not be supplied by the application. |
| 962 | (Unless you are working with abbreviated files, in which case see |
| 963 | "Abbreviated datastreams", below.) Decompression parameters control |
| 964 | the postprocessing done on the image to deliver it in a format suitable |
| 965 | for the application's use. Many of the parameters control speed/quality |
| 966 | tradeoffs, in which faster decompression may be obtained at the price of |
| 967 | a poorer-quality image. The defaults select the highest quality (slowest) |
| 968 | processing. |
| 969 | |
| 970 | The following fields in the JPEG object are set by jpeg_read_header() and |
| 971 | may be useful to the application in choosing decompression parameters: |
| 972 | |
| 973 | JDIMENSION image_width Width and height of image |
| 974 | JDIMENSION image_height |
| 975 | int num_components Number of color components |
| 976 | J_COLOR_SPACE jpeg_color_space Colorspace of image |
| 977 | boolean saw_JFIF_marker TRUE if a JFIF APP0 marker was seen |
| 978 | UINT8 density_unit Resolution data from JFIF marker |
| 979 | UINT16 X_density |
| 980 | UINT16 Y_density |
| 981 | boolean saw_Adobe_marker TRUE if an Adobe APP14 marker was seen |
| 982 | UINT8 Adobe_transform Color transform code from Adobe marker |
| 983 | |
| 984 | The JPEG color space, unfortunately, is something of a guess since the JPEG |
| 985 | standard proper does not provide a way to record it. In practice most files |
| 986 | adhere to the JFIF or Adobe conventions, and the decoder will recognize these |
| 987 | correctly. See "Special color spaces", below, for more info. |
| 988 | |
| 989 | |
| 990 | The decompression parameters that determine the basic properties of the |
| 991 | returned image are: |
| 992 | |
| 993 | J_COLOR_SPACE out_color_space |
| 994 | Output color space. jpeg_read_header() sets an appropriate default |
| 995 | based on jpeg_color_space; typically it will be RGB or grayscale. |
| 996 | The application can change this field to request output in a different |
| 997 | colorspace. For example, set it to JCS_GRAYSCALE to get grayscale |
| 998 | output from a color file. (This is useful for previewing: grayscale |
| 999 | output is faster than full color since the color components need not |
| 1000 | be processed.) Note that not all possible color space transforms are |
| 1001 | currently implemented; you may need to extend jdcolor.c if you want an |
| 1002 | unusual conversion. |
| 1003 | |
| 1004 | unsigned int scale_num, scale_denom |
| 1005 | Scale the image by the fraction scale_num/scale_denom. Default is |
| 1006 | 1/1, or no scaling. Currently, the only supported scaling ratios |
| 1007 | are 1/1, 1/2, 1/4, and 1/8. (The library design allows for arbitrary |
| 1008 | scaling ratios but this is not likely to be implemented any time soon.) |
| 1009 | Smaller scaling ratios permit significantly faster decoding since |
| 1010 | fewer pixels need be processed and a simpler IDCT method can be used. |
| 1011 | |
| 1012 | boolean quantize_colors |
| 1013 | If set TRUE, colormapped output will be delivered. Default is FALSE, |
| 1014 | meaning that full-color output will be delivered. |
| 1015 | |
| 1016 | The next three parameters are relevant only if quantize_colors is TRUE. |
| 1017 | |
| 1018 | int desired_number_of_colors |
| 1019 | Maximum number of colors to use in generating a library-supplied color |
| 1020 | map (the actual number of colors is returned in a different field). |
| 1021 | Default 256. Ignored when the application supplies its own color map. |
| 1022 | |
| 1023 | boolean two_pass_quantize |
| 1024 | If TRUE, an extra pass over the image is made to select a custom color |
| 1025 | map for the image. This usually looks a lot better than the one-size- |
| 1026 | fits-all colormap that is used otherwise. Default is TRUE. Ignored |
| 1027 | when the application supplies its own color map. |
| 1028 | |
| 1029 | J_DITHER_MODE dither_mode |
| 1030 | Selects color dithering method. Supported values are: |
| 1031 | JDITHER_NONE no dithering: fast, very low quality |
| 1032 | JDITHER_ORDERED ordered dither: moderate speed and quality |
| 1033 | JDITHER_FS Floyd-Steinberg dither: slow, high quality |
| 1034 | Default is JDITHER_FS. (At present, ordered dither is implemented |
| 1035 | only in the single-pass, standard-colormap case. If you ask for |
| 1036 | ordered dither when two_pass_quantize is TRUE or when you supply |
| 1037 | an external color map, you'll get F-S dithering.) |
| 1038 | |
| 1039 | When quantize_colors is TRUE, the target color map is described by the next |
| 1040 | two fields. colormap is set to NULL by jpeg_read_header(). The application |
| 1041 | can supply a color map by setting colormap non-NULL and setting |
| 1042 | actual_number_of_colors to the map size. Otherwise, jpeg_start_decompress() |
| 1043 | selects a suitable color map and sets these two fields itself. |
| 1044 | [Implementation restriction: at present, an externally supplied colormap is |
| 1045 | only accepted for 3-component output color spaces.] |
| 1046 | |
| 1047 | JSAMPARRAY colormap |
| 1048 | The color map, represented as a 2-D pixel array of out_color_components |
| 1049 | rows and actual_number_of_colors columns. Ignored if not quantizing. |
| 1050 | |
| 1051 | int actual_number_of_colors |
| 1052 | The number of colors in the color map. |
| 1053 | |
| 1054 | Additional decompression parameters that the application may set include: |
| 1055 | |
| 1056 | J_DCT_METHOD dct_method |
| 1057 | Selects the algorithm used for the DCT step. Choices are the same |
| 1058 | as described above for compression. |
| 1059 | |
| 1060 | boolean do_fancy_upsampling |
| 1061 | If TRUE, do careful upsampling of chroma components. If FALSE, |
| 1062 | a faster but sloppier method is used. Default is TRUE. The visual |
| 1063 | impact of the sloppier method is often very small. |
| 1064 | |
| 1065 | |
| 1066 | The output image dimensions are given by the following fields. These are |
| 1067 | computed from the source image dimensions and the decompression parameters |
| 1068 | by jpeg_start_decompress(). You can also call jpeg_calc_output_dimensions() |
| 1069 | to obtain the values that will result from the current parameter settings. |
| 1070 | This can be useful if you are trying to pick a scaling ratio that will get |
| 1071 | close to a desired target size. It's also important if you are using the |
| 1072 | JPEG library's memory manager to allocate output buffer space, because you |
| 1073 | are supposed to request such buffers *before* jpeg_start_decompress(). |
| 1074 | |
| 1075 | JDIMENSION output_width Actual dimensions of output image. |
| 1076 | JDIMENSION output_height |
| 1077 | int out_color_components Number of color components in out_color_space. |
| 1078 | int output_components Number of color components returned. |
| 1079 | int rec_outbuf_height Recommended height of scanline buffer. |
| 1080 | |
| 1081 | When quantizing colors, output_components is 1, indicating a single color map |
| 1082 | index per pixel. Otherwise it equals out_color_components. The output arrays |
| 1083 | are required to be output_width * output_components JSAMPLEs wide. |
| 1084 | |
| 1085 | rec_outbuf_height is the recommended minimum height (in scanlines) of the |
| 1086 | buffer passed to jpeg_read_scanlines(). If the buffer is smaller, the |
| 1087 | library will still work, but time will be wasted due to unnecessary data |
| 1088 | copying. In high-quality modes, rec_outbuf_height is always 1, but some |
| 1089 | faster, lower-quality modes set it to larger values (typically 2 to 4). |
| 1090 | If you are going to ask for a high-speed processing mode, you may as well |
| 1091 | go to the trouble of honoring rec_outbuf_height so as to avoid data copying. |
| 1092 | |
| 1093 | |
| 1094 | Special color spaces |
| 1095 | -------------------- |
| 1096 | |
| 1097 | The JPEG standard itself is "color blind" and doesn't specify any particular |
| 1098 | color space. It is customary to convert color data to a luminance/chrominance |
| 1099 | color space before compressing, since this permits greater compression. The |
| 1100 | existing de-facto JPEG file format standards specify YCbCr or grayscale data |
| 1101 | (JFIF), or grayscale, RGB, YCbCr, CMYK, or YCCK (Adobe). For special |
| 1102 | applications such as multispectral images, other color spaces can be used, |
| 1103 | but it must be understood that such files will be unportable. |
| 1104 | |
| 1105 | The JPEG library can handle the most common colorspace conversions (namely |
| 1106 | RGB <=> YCbCr and CMYK <=> YCCK). It can also deal with data of an unknown |
| 1107 | color space, passing it through without conversion. If you deal extensively |
| 1108 | with an unusual color space, you can easily extend the library to understand |
| 1109 | additional color spaces and perform appropriate conversions. |
| 1110 | |
| 1111 | For compression, the source data's color space is specified by field |
| 1112 | in_color_space. This is transformed to the JPEG file's color space given |
| 1113 | by jpeg_color_space. jpeg_set_defaults() chooses a reasonable JPEG color |
| 1114 | space depending on in_color_space, but you can override this by calling |
| 1115 | jpeg_set_colorspace(). Of course you must select a supported transformation. |
| 1116 | jccolor.c currently supports the following transformations: |
| 1117 | RGB => YCbCr |
| 1118 | RGB => GRAYSCALE |
| 1119 | YCbCr => GRAYSCALE |
| 1120 | CMYK => YCCK |
| 1121 | plus the null transforms: GRAYSCALE => GRAYSCALE, RGB => RGB, |
| 1122 | YCbCr => YCbCr, CMYK => CMYK, YCCK => YCCK, and UNKNOWN => UNKNOWN. |
| 1123 | |
| 1124 | The de-facto file format standards (JFIF and Adobe) specify APPn markers that |
| 1125 | indicate the color space of the JPEG file. It is important to ensure that |
| 1126 | these are written correctly, or omitted if the JPEG file's color space is not |
| 1127 | one of the ones supported by the de-facto standards. jpeg_set_colorspace() |
| 1128 | will set the compression parameters to include or omit the APPn markers |
| 1129 | properly, so long as it is told the truth about the JPEG color space. |
| 1130 | For example, if you are writing some random 3-component color space without |
| 1131 | conversion, don't try to fake out the library by setting in_color_space and |
| 1132 | jpeg_color_space to JCS_YCbCr; use JCS_UNKNOWN. You may want to write an |
| 1133 | APPn marker of your own devising to identify the colorspace --- see "Special |
| 1134 | markers", below. |
| 1135 | |
| 1136 | When told that the color space is UNKNOWN, the library will default to using |
| 1137 | luminance-quality compression parameters for all color components. You may |
| 1138 | well want to change these parameters. See the source code for |
| 1139 | jpeg_set_colorspace(), in jcparam.c, for details. |
| 1140 | |
| 1141 | For decompression, the JPEG file's color space is given in jpeg_color_space, |
| 1142 | and this is transformed to the output color space out_color_space. |
| 1143 | jpeg_read_header's setting of jpeg_color_space can be relied on if the file |
| 1144 | conforms to JFIF or Adobe conventions, but otherwise it is no better than a |
| 1145 | guess. If you know the JPEG file's color space for certain, you can override |
| 1146 | jpeg_read_header's guess by setting jpeg_color_space. jpeg_read_header also |
| 1147 | selects a default output color space based on (its guess of) jpeg_color_space; |
| 1148 | set out_color_space to override this. Again, you must select a supported |
| 1149 | transformation. jdcolor.c currently supports |
| 1150 | YCbCr => GRAYSCALE |
| 1151 | YCbCr => RGB |
| 1152 | YCCK => CMYK |
| 1153 | as well as the null transforms. |
| 1154 | |
| 1155 | The two-pass color quantizer, jquant2.c, is specialized to handle RGB data |
| 1156 | (it weights distances appropriately for RGB colors). You'll need to modify |
| 1157 | the code if you want to use it for non-RGB output color spaces. Note that |
| 1158 | jquant2.c is used to map to an application-supplied colormap as well as for |
| 1159 | the normal two-pass colormap selection process. |
| 1160 | |
| 1161 | CAUTION: it appears that Adobe Photoshop writes inverted data in CMYK JPEG |
| 1162 | files: 0 represents 100% ink coverage, rather than 0% ink as you'd expect. |
| 1163 | This is arguably a bug in Photoshop, but if you need to work with Photoshop |
| 1164 | CMYK files, you will have to deal with it in your application. We cannot |
| 1165 | "fix" this in the library by inverting the data during the CMYK<=>YCCK |
| 1166 | transform, because that would break other applications, notably Ghostscript. |
| 1167 | Photoshop versions prior to 3.0 write EPS files containing JPEG-encoded CMYK |
| 1168 | data in the same inverted-YCCK representation used in bare JPEG files, but |
| 1169 | the surrounding PostScript code performs an inversion using the PS image |
| 1170 | operator. I am told that Photoshop 3.0 will write uninverted YCCK in |
| 1171 | EPS/JPEG files, and will omit the PS-level inversion. (But the data |
| 1172 | polarity used in bare JPEG files will not change in 3.0.) In either case, |
| 1173 | the JPEG library must not invert the data itself, or else Ghostscript would |
| 1174 | read these EPS files incorrectly. |
| 1175 | |
| 1176 | |
| 1177 | Error handling |
| 1178 | -------------- |
| 1179 | |
| 1180 | When the default error handler is used, any error detected inside the JPEG |
| 1181 | routines will cause a message to be printed on stderr, followed by exit(). |
| 1182 | You can supply your own error handling routines to override this behavior |
| 1183 | and to control the treatment of nonfatal warnings and trace/debug messages. |
| 1184 | The file example.c illustrates the most common case, which is to have the |
| 1185 | application regain control after an error rather than exiting. |
| 1186 | |
| 1187 | The JPEG library never writes any message directly; it always goes through |
| 1188 | the error handling routines. Three classes of messages are recognized: |
| 1189 | * Fatal errors: the library cannot continue. |
| 1190 | * Warnings: the library can continue, but the data is corrupt, and a |
| 1191 | damaged output image is likely to result. |
| 1192 | * Trace/informational messages. These come with a trace level indicating |
| 1193 | the importance of the message; you can control the verbosity of the |
| 1194 | program by adjusting the maximum trace level that will be displayed. |
| 1195 | |
| 1196 | You may, if you wish, simply replace the entire JPEG error handling module |
| 1197 | (jerror.c) with your own code. However, you can avoid code duplication by |
| 1198 | only replacing some of the routines depending on the behavior you need. |
| 1199 | This is accomplished by calling jpeg_std_error() as usual, but then overriding |
| 1200 | some of the method pointers in the jpeg_error_mgr struct, as illustrated by |
| 1201 | example.c. |
| 1202 | |
| 1203 | All of the error handling routines will receive a pointer to the JPEG object |
| 1204 | (a j_common_ptr which points to either a jpeg_compress_struct or a |
| 1205 | jpeg_decompress_struct; if you need to tell which, test the is_decompressor |
| 1206 | field). This struct includes a pointer to the error manager struct in its |
| 1207 | "err" field. Frequently, custom error handler routines will need to access |
| 1208 | additional data which is not known to the JPEG library or the standard error |
| 1209 | handler. The most convenient way to do this is to embed either the JPEG |
| 1210 | object or the jpeg_error_mgr struct in a larger structure that contains |
| 1211 | additional fields; then casting the passed pointer provides access to the |
| 1212 | additional fields. Again, see example.c for one way to do it. |
| 1213 | |
| 1214 | The individual methods that you might wish to override are: |
| 1215 | |
| 1216 | error_exit (j_common_ptr cinfo) |
| 1217 | Receives control for a fatal error. Information sufficient to |
| 1218 | generate the error message has been stored in cinfo->err; call |
| 1219 | output_message to display it. Control must NOT return to the caller; |
| 1220 | generally this routine will exit() or longjmp() somewhere. |
| 1221 | Typically you would override this routine to get rid of the exit() |
| 1222 | default behavior. Note that if you continue processing, you should |
| 1223 | clean up the JPEG object with jpeg_abort() or jpeg_destroy(). |
| 1224 | |
| 1225 | output_message (j_common_ptr cinfo) |
| 1226 | Actual output of any JPEG message. Override this to send messages |
| 1227 | somewhere other than stderr. Note that this method does not know |
| 1228 | how to generate a message, only where to send it. |
| 1229 | |
| 1230 | format_message (j_common_ptr cinfo, char * buffer) |
| 1231 | Constructs a readable error message string based on the error info |
| 1232 | stored in cinfo->err. This method is called by output_message. Few |
| 1233 | applications should need to override this method. One possible |
| 1234 | reason for doing so is to implement dynamic switching of error message |
| 1235 | language. |
| 1236 | |
| 1237 | emit_message (j_common_ptr cinfo, int msg_level) |
| 1238 | Decide whether or not to emit a warning or trace message; if so, |
| 1239 | calls output_message. The main reason for overriding this method |
| 1240 | would be to abort on warnings. msg_level is -1 for warnings, |
| 1241 | 0 and up for trace messages. |
| 1242 | |
| 1243 | Only error_exit() and emit_message() are called from the rest of the JPEG |
| 1244 | library; the other two are internal to the error handler. |
| 1245 | |
| 1246 | The actual message texts are stored in an array of strings which is pointed to |
| 1247 | by the field err->jpeg_message_table. The messages are numbered from 0 to |
| 1248 | err->last_jpeg_message, and it is these code numbers that are used in the |
| 1249 | JPEG library code. You could replace the message texts (for instance, with |
| 1250 | messages in French or German) by changing the message table pointer. See |
| 1251 | jerror.h for the default texts. CAUTION: this table will almost certainly |
| 1252 | change or grow from one library version to the next. |
| 1253 | |
| 1254 | It may be useful for an application to add its own message texts that are |
| 1255 | handled by the same mechanism. The error handler supports a second "add-on" |
| 1256 | message table for this purpose. To define an addon table, set the pointer |
| 1257 | err->addon_message_table and the message numbers err->first_addon_message and |
| 1258 | err->last_addon_message. If you number the addon messages beginning at 1000 |
| 1259 | or so, you won't have to worry about conflicts with the library's built-in |
| 1260 | messages. See the sample applications cjpeg/djpeg for an example of using |
| 1261 | addon messages (the addon messages are defined in cderror.h). |
| 1262 | |
| 1263 | Actual invocation of the error handler is done via macros defined in jerror.h: |
| 1264 | ERREXITn(...) for fatal errors |
| 1265 | WARNMSn(...) for corrupt-data warnings |
| 1266 | TRACEMSn(...) for trace and informational messages. |
| 1267 | These macros store the message code and any additional parameters into the |
| 1268 | error handler struct, then invoke the error_exit() or emit_message() method. |
| 1269 | The variants of each macro are for varying numbers of additional parameters. |
| 1270 | The additional parameters are inserted into the generated message using |
| 1271 | standard printf() format codes. |
| 1272 | |
| 1273 | See jerror.h and jerror.c for further details. |
| 1274 | |
| 1275 | |
| 1276 | Compressed data handling (source and destination managers) |
| 1277 | ---------------------------------------------------------- |
| 1278 | |
| 1279 | The JPEG compression library sends its compressed data to a "destination |
| 1280 | manager" module. The default destination manager just writes the data to a |
| 1281 | stdio stream, but you can provide your own manager to do something else. |
| 1282 | Similarly, the decompression library calls a "source manager" to obtain the |
| 1283 | compressed data; you can provide your own source manager if you want the data |
| 1284 | to come from somewhere other than a stdio stream. |
| 1285 | |
| 1286 | In both cases, compressed data is processed a bufferload at a time: the |
| 1287 | destination or source manager provides a work buffer, and the library invokes |
| 1288 | the manager only when the buffer is filled or emptied. (You could define a |
| 1289 | one-character buffer to force the manager to be invoked for each byte, but |
| 1290 | that would be rather inefficient.) The buffer's size and location are |
| 1291 | controlled by the manager, not by the library. For example, if you desired to |
| 1292 | decompress a JPEG datastream that was all in memory, you could just make the |
| 1293 | buffer pointer and length point to the original data in memory. Then the |
| 1294 | buffer-reload procedure would be invoked only if the decompressor ran off the |
| 1295 | end of the datastream, which would indicate an erroneous datastream. |
| 1296 | |
| 1297 | The work buffer is defined as an array of datatype JOCTET, which is generally |
| 1298 | "char" or "unsigned char". On a machine where char is not exactly 8 bits |
| 1299 | wide, you must define JOCTET as a wider data type and then modify the data |
| 1300 | source and destination modules to transcribe the work arrays into 8-bit units |
| 1301 | on external storage. |
| 1302 | |
| 1303 | A data destination manager struct contains a pointer and count defining the |
| 1304 | next byte to write in the work buffer and the remaining free space: |
| 1305 | |
| 1306 | JOCTET * next_output_byte; /* => next byte to write in buffer */ |
| 1307 | size_t free_in_buffer; /* # of byte spaces remaining in buffer */ |
| 1308 | |
| 1309 | The library increments the pointer and decrements the count until the buffer |
| 1310 | is filled. The manager's empty_output_buffer method must reset the pointer |
| 1311 | and count. The manager is expected to remember the buffer's starting address |
| 1312 | and total size in private fields not visible to the library. |
| 1313 | |
| 1314 | A data destination manager provides three methods: |
| 1315 | |
| 1316 | init_destination (j_compress_ptr cinfo) |
| 1317 | Initialize destination. This is called by jpeg_start_compress() |
| 1318 | before any data is actually written. It must initialize |
| 1319 | next_output_byte and free_in_buffer. free_in_buffer must be |
| 1320 | initialized to a positive value. |
| 1321 | |
| 1322 | empty_output_buffer (j_compress_ptr cinfo) |
| 1323 | This is called whenever the buffer has filled (free_in_buffer |
| 1324 | reaches zero). In typical applications, it should write out the |
| 1325 | *entire* buffer (use the saved start address and buffer length; |
| 1326 | ignore the current state of next_output_byte and free_in_buffer). |
| 1327 | Then reset the pointer & count to the start of the buffer, and |
| 1328 | return TRUE indicating that the buffer has been dumped. |
| 1329 | free_in_buffer must be set to a positive value when TRUE is |
| 1330 | returned. A FALSE return should only be used when I/O suspension is |
| 1331 | desired (this operating mode is discussed in the next section). |
| 1332 | |
| 1333 | term_destination (j_compress_ptr cinfo) |
| 1334 | Terminate destination --- called by jpeg_finish_compress() after all |
| 1335 | data has been written. In most applications, this must flush any |
| 1336 | data remaining in the buffer. Use either next_output_byte or |
| 1337 | free_in_buffer to determine how much data is in the buffer. |
| 1338 | |
| 1339 | term_destination() is NOT called by jpeg_abort() or jpeg_destroy(). If you |
| 1340 | want the destination manager to be cleaned up during an abort, you must do it |
| 1341 | yourself. |
| 1342 | |
| 1343 | You will also need code to create a jpeg_destination_mgr struct, fill in its |
| 1344 | method pointers, and insert a pointer to the struct into the "dest" field of |
| 1345 | the JPEG compression object. This can be done in-line in your setup code if |
| 1346 | you like, but it's probably cleaner to provide a separate routine similar to |
| 1347 | the jpeg_stdio_dest() routine of the supplied destination manager. |
| 1348 | |
| 1349 | Decompression source managers follow a parallel design, but with some |
| 1350 | additional frammishes. The source manager struct contains a pointer and count |
| 1351 | defining the next byte to read from the work buffer and the number of bytes |
| 1352 | remaining: |
| 1353 | |
| 1354 | const JOCTET * next_input_byte; /* => next byte to read from buffer */ |
| 1355 | size_t bytes_in_buffer; /* # of bytes remaining in buffer */ |
| 1356 | |
| 1357 | The library increments the pointer and decrements the count until the buffer |
| 1358 | is emptied. The manager's fill_input_buffer method must reset the pointer and |
| 1359 | count. In most applications, the manager must remember the buffer's starting |
| 1360 | address and total size in private fields not visible to the library. |
| 1361 | |
| 1362 | A data source manager provides five methods: |
| 1363 | |
| 1364 | init_source (j_decompress_ptr cinfo) |
| 1365 | Initialize source. This is called by jpeg_read_header() before any |
| 1366 | data is actually read. Unlike init_destination(), it may leave |
| 1367 | bytes_in_buffer set to 0 (in which case a fill_input_buffer() call |
| 1368 | will occur immediately). |
| 1369 | |
| 1370 | fill_input_buffer (j_decompress_ptr cinfo) |
| 1371 | This is called whenever bytes_in_buffer has reached zero and more |
| 1372 | data is wanted. In typical applications, it should read fresh data |
| 1373 | into the buffer (ignoring the current state of next_input_byte and |
| 1374 | bytes_in_buffer), reset the pointer & count to the start of the |
| 1375 | buffer, and return TRUE indicating that the buffer has been reloaded. |
| 1376 | It is not necessary to fill the buffer entirely, only to obtain at |
| 1377 | least one more byte. bytes_in_buffer MUST be set to a positive value |
| 1378 | if TRUE is returned. A FALSE return should only be used when I/O |
| 1379 | suspension is desired (this mode is discussed in the next section). |
| 1380 | |
| 1381 | skip_input_data (j_decompress_ptr cinfo, long num_bytes) |
| 1382 | Skip num_bytes worth of data. The buffer pointer and count should |
| 1383 | be advanced over num_bytes input bytes, refilling the buffer as |
| 1384 | needed. This is used to skip over a potentially large amount of |
| 1385 | uninteresting data (such as an APPn marker). In some applications |
| 1386 | it may be possible to optimize away the reading of the skipped data, |
| 1387 | but it's not clear that being smart is worth much trouble; large |
| 1388 | skips are uncommon. bytes_in_buffer may be zero on return. |
| 1389 | A zero or negative skip count should be treated as a no-op. |
| 1390 | |
| 1391 | resync_to_restart (j_decompress_ptr cinfo) |
| 1392 | This routine is called only when the decompressor has failed to find |
| 1393 | a restart (RSTn) marker where one is expected. Its mission is to |
| 1394 | find a suitable point for resuming decompression. For most |
| 1395 | applications, we recommend that you just use the default resync |
| 1396 | procedure, jpeg_resync_to_restart(). However, if you are able to back |
| 1397 | up in the input data stream, or if you have a-priori knowledge about |
| 1398 | the likely location of restart markers, you may be able to do better. |
| 1399 | Read the read_restart_marker() and jpeg_resync_to_restart() routines |
| 1400 | in jdmarker.c if you think you'd like to implement your own resync |
| 1401 | procedure. |
| 1402 | |
| 1403 | term_source (j_decompress_ptr cinfo) |
| 1404 | Terminate source --- called by jpeg_finish_decompress() after all |
| 1405 | data has been read. Often a no-op. |
| 1406 | |
| 1407 | For both fill_input_buffer() and skip_input_data(), there is no such thing |
| 1408 | as an EOF return. If the end of the file has been reached, the routine has |
| 1409 | a choice of exiting via ERREXIT() or inserting fake data into the buffer. |
| 1410 | In most cases, generating a warning message and inserting a fake EOI marker |
| 1411 | is the best course of action --- this will allow the decompressor to output |
| 1412 | however much of the image is there. In pathological cases, the decompressor |
| 1413 | may swallow the EOI and again demand data ... just keep feeding it fake EOIs. |
| 1414 | jdatasrc.c illustrates the recommended error recovery behavior. |
| 1415 | |
| 1416 | term_source() is NOT called by jpeg_abort() or jpeg_destroy(). If you want |
| 1417 | the source manager to be cleaned up during an abort, you must do it yourself. |
| 1418 | |
| 1419 | You will also need code to create a jpeg_source_mgr struct, fill in its method |
| 1420 | pointers, and insert a pointer to the struct into the "src" field of the JPEG |
| 1421 | decompression object. This can be done in-line in your setup code if you |
| 1422 | like, but it's probably cleaner to provide a separate routine similar to the |
| 1423 | jpeg_stdio_src() routine of the supplied source manager. |
| 1424 | |
| 1425 | For more information, consult the stdio source and destination managers |
| 1426 | in jdatasrc.c and jdatadst.c. |
| 1427 | |
| 1428 | |
| 1429 | I/O suspension |
| 1430 | -------------- |
| 1431 | |
| 1432 | Some applications need to use the JPEG library as an incremental memory-to- |
| 1433 | memory filter: when the compressed data buffer is filled or emptied, they want |
| 1434 | control to return to the outer loop, rather than expecting that the buffer can |
| 1435 | be flushed or reloaded within the data source/destination manager subroutine. |
| 1436 | The library supports this need by providing an "I/O suspension" mode, which we |
| 1437 | describe in this section. |
| 1438 | |
| 1439 | The I/O suspension mode is a limited solution: it works only in the simplest |
| 1440 | operating modes (namely single-pass processing of single-scan JPEG files), and |
| 1441 | it has several other restrictions which are documented below. Furthermore, |
| 1442 | nothing is guaranteed about the maximum amount of time spent in any one call |
| 1443 | to the library, so a single-threaded application may still have response-time |
| 1444 | problems. If you need multi-pass processing or guaranteed response time, we |
| 1445 | suggest you "bite the bullet" and implement a real multi-tasking capability. |
| 1446 | |
| 1447 | To use I/O suspension, cooperation is needed between the calling application |
| 1448 | and the data source or destination manager; you will always need a custom |
| 1449 | source/destination manager. (Please read the previous section if you haven't |
| 1450 | already.) The basic idea is that the empty_output_buffer() or |
| 1451 | fill_input_buffer() routine is a no-op, merely returning FALSE to indicate |
| 1452 | that it has done nothing. Upon seeing this, the JPEG library suspends |
| 1453 | operation and returns to its caller. The surrounding application is |
| 1454 | responsible for emptying or refilling the work buffer before calling the JPEG |
| 1455 | library again. |
| 1456 | |
| 1457 | Compression suspension: |
| 1458 | |
| 1459 | For compression suspension, use an empty_output_buffer() routine that |
| 1460 | returns FALSE; typically it will not do anything else. This will cause the |
| 1461 | compressor to return to the caller of jpeg_write_scanlines(), with the |
| 1462 | return value indicating that not all the supplied scanlines have been |
| 1463 | accepted. The application must make more room in the output buffer, adjust |
| 1464 | the buffer pointer/count appropriately, and then call jpeg_write_scanlines() |
| 1465 | again, pointing to the first unconsumed scanline. |
| 1466 | |
| 1467 | When forced to suspend, the compressor will backtrack to a convenient stopping |
| 1468 | point (usually the start of the current MCU); it will regenerate some output |
| 1469 | data when restarted. Therefore, although empty_output_buffer() is only called |
| 1470 | when the buffer is filled, you should NOT dump out the entire buffer, only the |
| 1471 | data up to the current position of next_output_byte/free_in_buffer. The data |
| 1472 | beyond that point will be regenerated after resumption. |
| 1473 | |
| 1474 | Because of the backtracking behavior, a good-size output buffer is essential |
| 1475 | for efficiency; you don't want the compressor to suspend often. (In fact, an |
| 1476 | overly small buffer could lead to infinite looping, if a single MCU required |
| 1477 | more data than would fit in the buffer.) We recommend a buffer of at least |
| 1478 | several Kbytes. You may want to insert explicit code to ensure that you don't |
| 1479 | call jpeg_write_scanlines() unless there is a reasonable amount of space in |
| 1480 | the output buffer; in other words, flush the buffer before trying to compress |
| 1481 | more data. |
| 1482 | |
| 1483 | The JPEG compressor does not support suspension while it is trying to write |
| 1484 | JPEG markers at the beginning and end of the file. This means that |
| 1485 | * At the beginning of a compression operation, there must be enough free |
| 1486 | space in the output buffer to hold the header markers (typically 600 or |
| 1487 | so bytes). The recommended buffer size is bigger than this anyway, so |
| 1488 | this is not a problem as long as you start with an empty buffer. However, |
| 1489 | this restriction might catch you if you insert large special markers, such |
| 1490 | as a JFIF thumbnail image. |
| 1491 | * When you call jpeg_finish_compress(), there must be enough space in the |
| 1492 | output buffer to emit any buffered data and the final EOI marker. In the |
| 1493 | current implementation, half a dozen bytes should suffice for this, but |
| 1494 | for safety's sake we recommend ensuring that at least 100 bytes are free |
| 1495 | before calling jpeg_finish_compress(). |
| 1496 | Furthermore, since jpeg_finish_compress() cannot suspend, you cannot request |
| 1497 | multi-pass operating modes such as Huffman code optimization or multiple-scan |
| 1498 | output. That would imply that a large amount of data would be written inside |
| 1499 | jpeg_finish_compress(), which would certainly trigger a buffer overrun. |
| 1500 | |
| 1501 | Decompression suspension: |
| 1502 | |
| 1503 | For decompression suspension, use a fill_input_buffer() routine that simply |
| 1504 | returns FALSE (except perhaps during error recovery, as discussed below). |
| 1505 | This will cause the decompressor to return to its caller with an indication |
| 1506 | that suspension has occurred. This can happen at three places: |
| 1507 | * jpeg_read_header(): will return JPEG_SUSPENDED. |
| 1508 | * jpeg_read_scanlines(): will return the number of scanlines already |
| 1509 | completed (possibly 0). |
| 1510 | * jpeg_finish_decompress(): will return FALSE, rather than its usual TRUE. |
| 1511 | The surrounding application must recognize these cases, load more data into |
| 1512 | the input buffer, and repeat the call. In the case of jpeg_read_scanlines(), |
| 1513 | adjust the passed pointers to reflect any scanlines successfully read. |
| 1514 | |
| 1515 | Just as with compression, the decompressor will typically backtrack to a |
| 1516 | convenient restart point before suspending. The data beyond the current |
| 1517 | position of next_input_byte/bytes_in_buffer must NOT be discarded; it will |
| 1518 | be re-read upon resumption. In most implementations, you'll need to shift |
| 1519 | this data down to the start of your work buffer and then load more data |
| 1520 | after it. Again, this behavior means that a several-Kbyte work buffer is |
| 1521 | essential for decent performance; furthermore, you should load a reasonable |
| 1522 | amount of new data before resuming decompression. (If you loaded, say, |
| 1523 | only one new byte each time around, you could waste a LOT of cycles.) |
| 1524 | |
| 1525 | The skip_input_data() source manager routine requires special care in a |
| 1526 | suspension scenario. This routine is NOT granted the ability to suspend the |
| 1527 | decompressor; it can decrement bytes_in_buffer to zero, but no more. If the |
| 1528 | requested skip distance exceeds the amount of data currently in the input |
| 1529 | buffer, then skip_input_data() must set bytes_in_buffer to zero and record the |
| 1530 | additional skip distance somewhere else. The decompressor will immediately |
| 1531 | call fill_input_buffer(), which will return FALSE, which will cause a |
| 1532 | suspension return. The surrounding application must then arrange to discard |
| 1533 | the right number of bytes before it resumes loading the input buffer. (Yes, |
| 1534 | this design is rather baroque, but it avoids complexity in the far more common |
| 1535 | case where a non-suspending source manager is used.) |
| 1536 | |
| 1537 | If the input data has been exhausted, we recommend that you emit a warning |
| 1538 | and insert dummy EOI markers just as a non-suspending data source manager |
| 1539 | would do. This can be handled either in the surrounding application logic or |
| 1540 | within fill_input_buffer(); the latter is probably more efficient. If |
| 1541 | fill_input_buffer() knows that no more data is available, it can set the |
| 1542 | pointer/count to point to a dummy EOI marker and then return TRUE just as |
| 1543 | though it had read more data in a non-suspending situation. |
| 1544 | |
| 1545 | The decompressor does not support suspension within jpeg_start_decompress(). |
| 1546 | This means that you cannot use suspension with any multi-pass processing mode |
| 1547 | (eg, two-pass color quantization or multiple-scan JPEG files). In single-pass |
| 1548 | modes, jpeg_start_decompress() reads no data and thus need never suspend. |
| 1549 | |
| 1550 | The decompressor does not attempt to suspend within any JPEG marker; it will |
| 1551 | backtrack to the start of the marker. Hence the input buffer must be large |
| 1552 | enough to hold the longest marker in the file. We recommend at least a 2K |
| 1553 | buffer. The buffer would need to be 64K to allow for arbitrary COM or APPn |
| 1554 | markers, but the decompressor does not actually try to read these; it just |
| 1555 | skips them by calling skip_input_data(). If you provide a special marker |
| 1556 | handling routine that does look at such markers, coping with buffer overflow |
| 1557 | is your problem. Ordinary JPEG markers should normally not exceed a few |
| 1558 | hundred bytes each (DHT tables are typically the longest). For robustness |
| 1559 | against damaged marker length counts, you may wish to insert a test in your |
| 1560 | application for the case that the input buffer is completely full and yet the |
| 1561 | decoder has suspended without consuming any data --- otherwise, if this |
| 1562 | situation did occur, it would lead to an endless loop. |
| 1563 | |
| 1564 | |
| 1565 | Abbreviated datastreams and multiple images |
| 1566 | ------------------------------------------- |
| 1567 | |
| 1568 | A JPEG compression or decompression object can be reused to process multiple |
| 1569 | images. This saves a small amount of time per image by eliminating the |
| 1570 | "create" and "destroy" operations, but that isn't the real purpose of the |
| 1571 | feature. Rather, reuse of an object provides support for abbreviated JPEG |
| 1572 | datastreams. Object reuse can also simplify processing a series of images in |
| 1573 | a single input or output file. This section explains these features. |
| 1574 | |
| 1575 | A JPEG file normally contains several hundred bytes worth of quantization |
| 1576 | and Huffman tables. In a situation where many images will be stored or |
| 1577 | transmitted with identical tables, this may represent an annoying overhead. |
| 1578 | The JPEG standard therefore permits tables to be omitted. The standard |
| 1579 | defines three classes of JPEG datastreams: |
| 1580 | * "Interchange" datastreams contain an image and all tables needed to decode |
| 1581 | the image. These are the usual kind of JPEG file. |
| 1582 | * "Abbreviated image" datastreams contain an image, but are missing some or |
| 1583 | all of the tables needed to decode that image. |
| 1584 | * "Abbreviated table specification" (henceforth "tables-only") datastreams |
| 1585 | contain only table specifications. |
| 1586 | To decode an abbreviated image, it is necessary to load the missing table(s) |
| 1587 | into the decoder beforehand. This can be accomplished by reading a separate |
| 1588 | tables-only file. A variant scheme uses a series of images in which the first |
| 1589 | image is an interchange (complete) datastream, while subsequent ones are |
| 1590 | abbreviated and rely on the tables loaded by the first image. It is assumed |
| 1591 | that once the decoder has read a table, it will remember that table until a |
| 1592 | new definition for the same table number is encountered. |
| 1593 | |
| 1594 | It is the application designer's responsibility to figure out how to associate |
| 1595 | the correct tables with an abbreviated image. While abbreviated datastreams |
| 1596 | can be useful in a closed environment, their use is strongly discouraged in |
| 1597 | any situation where data exchange with other applications might be needed. |
| 1598 | Caveat designer. |
| 1599 | |
| 1600 | The JPEG library provides support for reading and writing any combination of |
| 1601 | tables-only datastreams and abbreviated images. In both compression and |
| 1602 | decompression objects, a quantization or Huffman table will be retained for |
| 1603 | the lifetime of the object, unless it is overwritten by a new table definition. |
| 1604 | |
| 1605 | |
| 1606 | To create abbreviated image datastreams, it is only necessary to tell the |
| 1607 | compressor not to emit some or all of the tables it is using. Each |
| 1608 | quantization and Huffman table struct contains a boolean field "sent_table", |
| 1609 | which normally is initialized to FALSE. For each table used by the image, the |
| 1610 | header-writing process emits the table and sets sent_table = TRUE unless it is |
| 1611 | already TRUE. (In normal usage, this prevents outputting the same table |
| 1612 | definition multiple times, as would otherwise occur because the chroma |
| 1613 | components typically share tables.) Thus, setting this field to TRUE before |
| 1614 | calling jpeg_start_compress() will prevent the table from being written at |
| 1615 | all. |
| 1616 | |
| 1617 | If you want to create a "pure" abbreviated image file containing no tables, |
| 1618 | just call "jpeg_suppress_tables(&cinfo, TRUE)" after constructing all the |
| 1619 | tables. If you want to emit some but not all tables, you'll need to set the |
| 1620 | individual sent_table fields directly. |
| 1621 | |
| 1622 | To create an abbreviated image, you must also call jpeg_start_compress() |
| 1623 | with a second parameter of FALSE, not TRUE. Otherwise jpeg_start_compress() |
| 1624 | will force all the sent_table fields to FALSE. (This is a safety feature to |
| 1625 | prevent abbreviated images from being created accidentally.) |
| 1626 | |
| 1627 | To create a tables-only file, perform the same parameter setup that you |
| 1628 | normally would, but instead of calling jpeg_start_compress() and so on, call |
| 1629 | jpeg_write_tables(&cinfo). This will write an abbreviated datastream |
| 1630 | containing only SOI, DQT and/or DHT markers, and EOI. All the quantization |
| 1631 | and Huffman tables that are currently defined in the compression object will |
| 1632 | be emitted unless their sent_tables flag is already TRUE, and then all the |
| 1633 | sent_tables flags will be set TRUE. |
| 1634 | |
| 1635 | A sure-fire way to create matching tables-only and abbreviated image files |
| 1636 | is to proceed as follows: |
| 1637 | |
| 1638 | create JPEG compression object |
| 1639 | set JPEG parameters |
| 1640 | set destination to tables-only file |
| 1641 | jpeg_write_tables(&cinfo); |
| 1642 | set destination to image file |
| 1643 | jpeg_start_compress(&cinfo, FALSE); |
| 1644 | write data... |
| 1645 | jpeg_finish_compress(&cinfo); |
| 1646 | |
| 1647 | Since the JPEG parameters are not altered between writing the table file and |
| 1648 | the abbreviated image file, the same tables are sure to be used. Of course, |
| 1649 | you can repeat the jpeg_start_compress() ... jpeg_finish_compress() sequence |
| 1650 | many times to produce many abbreviated image files matching the table file. |
| 1651 | |
| 1652 | You cannot suppress output of the computed Huffman tables when Huffman |
| 1653 | optimization is selected. (If you could, there'd be no way to decode the |
| 1654 | image...) Generally, you don't want to set optimize_coding = TRUE when |
| 1655 | you are trying to produce abbreviated files. |
| 1656 | |
| 1657 | In some cases you might want to compress an image using tables which are |
| 1658 | not stored in the application, but are defined in an interchange or |
| 1659 | tables-only file readable by the application. This can be done by setting up |
| 1660 | a JPEG decompression object to read the specification file, then copying the |
| 1661 | tables into your compression object. |
| 1662 | |
| 1663 | |
| 1664 | To read abbreviated image files, you simply need to load the proper tables |
| 1665 | into the decompression object before trying to read the abbreviated image. |
| 1666 | If the proper tables are stored in the application program, you can just |
| 1667 | allocate the table structs and fill in their contents directly. More commonly |
| 1668 | you'd want to read the tables from a tables-only file. The jpeg_read_header() |
| 1669 | call is sufficient to read a tables-only file. You must pass a second |
| 1670 | parameter of FALSE to indicate that you do not require an image to be present. |
| 1671 | Thus, the typical scenario is |
| 1672 | |
| 1673 | create JPEG decompression object |
| 1674 | set source to tables-only file |
| 1675 | jpeg_read_header(&cinfo, FALSE); |
| 1676 | set source to abbreviated image file |
| 1677 | jpeg_read_header(&cinfo, TRUE); |
| 1678 | set decompression parameters |
| 1679 | jpeg_start_decompress(&cinfo); |
| 1680 | read data... |
| 1681 | jpeg_finish_decompress(&cinfo); |
| 1682 | |
| 1683 | In some cases, you may want to read a file without knowing whether it contains |
| 1684 | an image or just tables. In that case, pass FALSE and check the return value |
| 1685 | from jpeg_read_header(): it will be JPEG_HEADER_OK if an image was found, |
| 1686 | JPEG_HEADER_TABLES_ONLY if only tables were found. (A third return value, |
| 1687 | JPEG_SUSPENDED, is possible when using a suspending data source manager.) |
| 1688 | Note that jpeg_read_header() will not complain if you read an abbreviated |
| 1689 | image for which you haven't loaded the missing tables; the missing-table check |
| 1690 | occurs in jpeg_start_decompress(). |
| 1691 | |
| 1692 | |
| 1693 | It is possible to read a series of images from a single source file by |
| 1694 | repeating the jpeg_read_header() ... jpeg_finish_decompress() sequence, |
| 1695 | without releasing/recreating the JPEG object or the data source module. |
| 1696 | (If you did reinitialize, any partial bufferload left in the data source |
| 1697 | buffer at the end of one image would be discarded, causing you to lose the |
| 1698 | start of the next image.) When you use this method, stored tables are |
| 1699 | automatically carried forward, so some of the images can be abbreviated images |
| 1700 | that depend on tables from earlier images. |
| 1701 | |
| 1702 | If you intend to write a series of images into a single destination file, |
| 1703 | you might want to make a specialized data destination module that doesn't |
| 1704 | flush the output buffer at term_destination() time. This would speed things |
| 1705 | up by some trifling amount. Of course, you'd need to remember to flush the |
| 1706 | buffer after the last image. You can make the later images be abbreviated |
| 1707 | ones by passing FALSE to jpeg_start_compress(). |
| 1708 | |
| 1709 | |
| 1710 | Special markers |
| 1711 | --------------- |
| 1712 | |
| 1713 | Some applications may need to insert or extract special data in the JPEG |
| 1714 | datastream. The JPEG standard provides marker types "COM" (comment) and |
| 1715 | "APP0" through "APP15" (application) to hold application-specific data. |
| 1716 | Unfortunately, the use of these markers is not specified by the standard. |
| 1717 | COM markers are fairly widely used to hold user-supplied text. The JFIF file |
| 1718 | format spec uses APP0 markers with specified initial strings to hold certain |
| 1719 | data. Adobe applications use APP14 markers beginning with the string "Adobe" |
| 1720 | for miscellaneous data. Other APPn markers are rarely seen, but might |
| 1721 | contain almost anything. |
| 1722 | |
| 1723 | If you wish to store user-supplied text, we recommend you use COM markers |
| 1724 | and place readable 7-bit ASCII text in them. Newline conventions are not |
| 1725 | standardized --- expect to find LF (Unix style), CR/LF (DOS style), or CR |
| 1726 | (Mac style). A robust COM reader should be able to cope with random binary |
| 1727 | garbage, including nulls, since some applications generate COM markers |
| 1728 | containing non-ASCII junk. (But yours should not be one of them.) |
| 1729 | |
| 1730 | For program-supplied data, use an APPn marker, and be sure to begin it with an |
| 1731 | identifying string so that you can tell whether the marker is actually yours. |
| 1732 | It's probably best to avoid using APP0 or APP14 for any private markers. |
| 1733 | |
| 1734 | Keep in mind that at most 65533 bytes can be put into one marker, but you |
| 1735 | can have as many markers as you like. |
| 1736 | |
| 1737 | By default, the JPEG compression library will write a JFIF APP0 marker if the |
| 1738 | selected JPEG colorspace is grayscale or YCbCr, or an Adobe APP14 marker if |
| 1739 | the selected colorspace is RGB, CMYK, or YCCK. You can disable this, but |
| 1740 | we don't recommend it. The decompression library will recognize JFIF and |
| 1741 | Adobe markers and will set the JPEG colorspace properly when one is found. |
| 1742 | |
| 1743 | You can write special markers immediately following the datastream header by |
| 1744 | calling jpeg_write_marker() after jpeg_start_compress() and before the first |
| 1745 | call to jpeg_write_scanlines(). When you do this, the markers appear after |
| 1746 | the SOI and the JFIF APP0 and Adobe APP14 markers (if written), but before |
| 1747 | all else. Write the marker type parameter as "JPEG_COM" for COM or |
| 1748 | "JPEG_APP0 + n" for APPn. (Actually, jpeg_write_marker will let you write |
| 1749 | any marker type, but we don't recommend writing any other kinds of marker.) |
| 1750 | For example, to write a user comment string pointed to by comment_text: |
| 1751 | jpeg_write_marker(cinfo, JPEG_COM, comment_text, strlen(comment_text)); |
| 1752 | Or if you prefer to synthesize the marker byte sequence yourself, you can |
| 1753 | just cram it straight into the data destination module. |
| 1754 | |
| 1755 | For decompression, you can supply your own routine to process COM or APPn |
| 1756 | markers by calling jpeg_set_marker_processor(). Usually you'd call this |
| 1757 | after creating a decompression object and before calling jpeg_read_header(), |
| 1758 | because the markers of interest will normally be scanned by jpeg_read_header. |
| 1759 | Once you've supplied a routine, it will be used for the life of that |
| 1760 | decompression object. A separate routine may be registered for COM and for |
| 1761 | each APPn marker code. |
| 1762 | |
| 1763 | A marker processor routine must have the signature |
| 1764 | boolean jpeg_marker_parser_method (j_decompress_ptr cinfo) |
| 1765 | Although the marker code is not explicitly passed, the routine can find it |
| 1766 | in cinfo->unread_marker. At the time of call, the marker proper has been |
| 1767 | read from the data source module. The processor routine is responsible for |
| 1768 | reading the marker length word and the remaining parameter bytes, if any. |
| 1769 | Return TRUE to indicate success. (FALSE should be returned only if you are |
| 1770 | using a suspending data source and it tells you to suspend. See the standard |
| 1771 | marker processors in jdmarker.c for appropriate coding methods if you need to |
| 1772 | use a suspending data source.) |
| 1773 | |
| 1774 | If you override the default APP0 or APP14 processors, it is up to you to |
| 1775 | recognize JFIF and Adobe markers if you want colorspace recognition to occur |
| 1776 | properly. We recommend copying and extending the default processors if you |
| 1777 | want to do that. |
| 1778 | |
| 1779 | A simple example of an external COM processor can be found in djpeg.c. |
| 1780 | |
| 1781 | |
| 1782 | Raw (downsampled) image data |
| 1783 | ---------------------------- |
| 1784 | |
| 1785 | Some applications need to supply already-downsampled image data to the JPEG |
| 1786 | compressor, or to receive raw downsampled data from the decompressor. The |
| 1787 | library supports this requirement by allowing the application to write or |
| 1788 | read raw data, bypassing the normal preprocessing or postprocessing steps. |
| 1789 | The interface is different from the standard one and is somewhat harder to |
| 1790 | use. If your interest is merely in bypassing color conversion, we recommend |
| 1791 | that you use the standard interface and simply set jpeg_color_space = |
| 1792 | in_color_space (or jpeg_color_space = out_color_space for decompression). |
| 1793 | The mechanism described in this section is necessary only to supply or |
| 1794 | receive downsampled image data, in which not all components have the same |
| 1795 | dimensions. |
| 1796 | |
| 1797 | |
| 1798 | To compress raw data, you must supply the data in the colorspace to be used |
| 1799 | in the JPEG file (please read the earlier section on Special color spaces) |
| 1800 | and downsampled to the sampling factors specified in the JPEG parameters. |
| 1801 | You must supply the data in the format used internally by the JPEG library, |
| 1802 | namely a JSAMPIMAGE array. This is an array of pointers to two-dimensional |
| 1803 | arrays, each of type JSAMPARRAY. Each 2-D array holds the values for one |
| 1804 | color component. This structure is necessary since the components are of |
| 1805 | different sizes. If the image dimensions are not a multiple of the MCU size, |
| 1806 | you must also pad the data correctly (usually, this is done by replicating |
| 1807 | the last column and/or row). The data must be padded to a multiple of a DCT |
| 1808 | block in each component: that is, each downsampled row must contain a |
| 1809 | multiple of 8 valid samples, and there must be a multiple of 8 sample rows |
| 1810 | for each component. (For applications such as conversion of digital TV |
| 1811 | images, the standard image size is usually a multiple of the DCT block size, |
| 1812 | so that no padding need actually be done.) |
| 1813 | |
| 1814 | The procedure for compression of raw data is basically the same as normal |
| 1815 | compression, except that you call jpeg_write_raw_data() in place of |
| 1816 | jpeg_write_scanlines(). Before calling jpeg_start_compress(), you must do |
| 1817 | the following: |
| 1818 | * Set cinfo->raw_data_in to TRUE. (It is set FALSE by jpeg_set_defaults().) |
| 1819 | This notifies the library that you will be supplying raw data. |
| 1820 | * Ensure jpeg_color_space is correct --- an explicit jpeg_set_colorspace() |
| 1821 | call is a good idea. Note that since color conversion is bypassed, |
| 1822 | in_color_space is ignored, except that jpeg_set_defaults() uses it to |
| 1823 | choose the default jpeg_color_space setting. |
| 1824 | * Ensure the sampling factors, cinfo->comp_info[i].h_samp_factor and |
| 1825 | cinfo->comp_info[i].v_samp_factor, are correct. Since these indicate the |
| 1826 | dimensions of the data you are supplying, it's wise to set them |
| 1827 | explicitly, rather than assuming the library's defaults are what you want. |
| 1828 | |
| 1829 | To pass raw data to the library, call jpeg_write_raw_data() in place of |
| 1830 | jpeg_write_scanlines(). The two routines work similarly except that |
| 1831 | jpeg_write_raw_data takes a JSAMPIMAGE data array rather than JSAMPARRAY. |
| 1832 | The scanlines count passed to and returned from jpeg_write_raw_data is |
| 1833 | measured in terms of the component with the largest v_samp_factor. |
| 1834 | |
| 1835 | jpeg_write_raw_data() processes one MCU row per call, which is to say |
| 1836 | v_samp_factor*DCTSIZE sample rows of each component. The passed num_lines |
| 1837 | value must be at least max_v_samp_factor*DCTSIZE, and the return value will |
| 1838 | be exactly that amount (or possibly some multiple of that amount, in future |
| 1839 | library versions). This is true even on the last call at the bottom of the |
| 1840 | image; don't forget to pad your data as necessary. |
| 1841 | |
| 1842 | The required dimensions of the supplied data can be computed for each |
| 1843 | component as |
| 1844 | cinfo->comp_info[i].width_in_blocks*DCTSIZE samples per row |
| 1845 | cinfo->comp_info[i].height_in_blocks*DCTSIZE rows in image |
| 1846 | after jpeg_start_compress() has initialized those fields. If the valid data |
| 1847 | is smaller than this, it must be padded appropriately. For some sampling |
| 1848 | factors and image sizes, additional dummy DCT blocks are inserted to make |
| 1849 | the image a multiple of the MCU dimensions. The library creates such dummy |
| 1850 | blocks itself; it does not read them from your supplied data. Therefore you |
| 1851 | need never pad by more than DCTSIZE samples. An example may help here. |
| 1852 | Assume 2h2v downsampling of YCbCr data, that is |
| 1853 | cinfo->comp_info[0].h_samp_factor = 2 for Y |
| 1854 | cinfo->comp_info[0].v_samp_factor = 2 |
| 1855 | cinfo->comp_info[1].h_samp_factor = 1 for Cb |
| 1856 | cinfo->comp_info[1].v_samp_factor = 1 |
| 1857 | cinfo->comp_info[2].h_samp_factor = 1 for Cr |
| 1858 | cinfo->comp_info[2].v_samp_factor = 1 |
| 1859 | and suppose that the nominal image dimensions (cinfo->image_width and |
| 1860 | cinfo->image_height) are 101x101 pixels. Then jpeg_start_compress() will |
| 1861 | compute downsampled_width = 101 and width_in_blocks = 13 for Y, |
| 1862 | downsampled_width = 51 and width_in_blocks = 7 for Cb and Cr (and the same |
| 1863 | for the height fields). You must pad the Y data to at least 13*8 = 104 |
| 1864 | columns and rows, the Cb/Cr data to at least 7*8 = 56 columns and rows. The |
| 1865 | MCU height is max_v_samp_factor = 2 DCT rows so you must pass at least 16 |
| 1866 | scanlines on each call to jpeg_write_raw_data(), which is to say 16 actual |
| 1867 | sample rows of Y and 8 each of Cb and Cr. A total of 7 MCU rows are needed, |
| 1868 | so you must pass a total of 7*16 = 112 "scanlines". The last DCT block row |
| 1869 | of Y data is dummy, so it doesn't matter what you pass for it in the data |
| 1870 | arrays, but the scanlines count must total up to 112 so that all of the Cb |
| 1871 | and Cr data gets passed. |
| 1872 | |
| 1873 | Currently, output suspension is not supported with raw data output: an error |
| 1874 | will result if the data destination module tries to suspend. |
| 1875 | |
| 1876 | |
| 1877 | Decompression with raw data output implies bypassing all postprocessing: |
| 1878 | you cannot ask for color quantization, for instance. More seriously, you must |
| 1879 | deal with the color space and sampling factors present in the incoming file. |
| 1880 | If your application only handles, say, 2h1v YCbCr data, you must check for |
| 1881 | and fail on other color spaces or other sampling factors. |
| 1882 | |
| 1883 | To obtain raw data output, set cinfo->raw_data_out = TRUE before |
| 1884 | jpeg_start_decompress() (it is set FALSE by jpeg_read_header()). Be sure to |
| 1885 | verify that the color space and sampling factors are ones you can handle. |
| 1886 | Then call jpeg_read_raw_data() in place of jpeg_read_scanlines(). The |
| 1887 | decompression process is otherwise the same as usual. |
| 1888 | |
| 1889 | jpeg_read_raw_data() returns one MCU row per call, and thus you must pass a |
| 1890 | buffer of at least max_v_samp_factor*DCTSIZE scanlines (scanline counting is |
| 1891 | the same as for raw-data compression). The buffer you pass must be large |
| 1892 | enough to hold the actual data plus padding to DCT-block boundaries. As with |
| 1893 | compression, any entirely dummy DCT blocks are not processed so you need not |
| 1894 | allocate space for them, but the total scanline count includes them. The |
| 1895 | above example of computing buffer dimensions for raw-data compression is |
| 1896 | equally valid for decompression. |
| 1897 | |
| 1898 | Input suspension is supported with raw-data decompression: if the data source |
| 1899 | module suspends, jpeg_read_raw_data() will return 0. |
| 1900 | |
| 1901 | |
| 1902 | Progress monitoring |
| 1903 | ------------------- |
| 1904 | |
| 1905 | Some applications may need to regain control from the JPEG library every so |
| 1906 | often. The typical use of this feature is to produce a percent-done bar or |
| 1907 | other progress display. (For a simple example, see cjpeg.c or djpeg.c.) |
| 1908 | Although you do get control back frequently during the data-transferring pass |
| 1909 | (the jpeg_read_scanlines or jpeg_write_scanlines loop), any additional passes |
| 1910 | will occur inside jpeg_finish_compress or jpeg_start_decompress; those |
| 1911 | routines may take a long time to execute, and you don't get control back |
| 1912 | until they are done. |
| 1913 | |
| 1914 | You can define a progress-monitor routine which will be called periodically |
| 1915 | by the library. No guarantees are made about how often this call will occur, |
| 1916 | so we don't recommend you use it for mouse tracking or anything like that. |
| 1917 | At present, a call will occur once per MCU row, scanline, or sample row |
| 1918 | group, whichever unit is convenient for the current processing mode; so the |
| 1919 | wider the image, the longer the time between calls. (During the data |
| 1920 | transferring pass, only one call occurs per call of jpeg_read_scanlines or |
| 1921 | jpeg_write_scanlines, so don't pass a large number of scanlines at once if |
| 1922 | you want fine resolution in the progress count.) |
| 1923 | |
| 1924 | To establish a progress-monitor callback, create a struct jpeg_progress_mgr, |
| 1925 | fill in its progress_monitor field with a pointer to your callback routine, |
| 1926 | and set cinfo->progress to point to the struct. The callback will be called |
| 1927 | whenever cinfo->progress is non-NULL. (This pointer is set to NULL by |
| 1928 | jpeg_create_compress or jpeg_create_decompress; the library will not change |
| 1929 | it thereafter. So if you allocate dynamic storage for the progress struct, |
| 1930 | make sure it will live as long as the JPEG object does. Allocating from the |
| 1931 | JPEG memory manager with lifetime JPOOL_PERMANENT will work nicely.) You |
| 1932 | can use the same callback routine for both compression and decompression. |
| 1933 | |
| 1934 | The jpeg_progress_mgr struct contains four fields which are set by the library: |
| 1935 | long pass_counter; /* work units completed in this pass */ |
| 1936 | long pass_limit; /* total number of work units in this pass */ |
| 1937 | int completed_passes; /* passes completed so far */ |
| 1938 | int total_passes; /* total number of passes expected */ |
| 1939 | During any one pass, pass_counter increases from 0 up to (not including) |
| 1940 | pass_limit; the step size is not necessarily 1. Both the step size and the |
| 1941 | limit may differ from one pass to another. The expected total number of |
| 1942 | passes is in total_passes, and the number of passes already completed is in |
| 1943 | completed_passes. Thus the fraction of work completed may be estimated as |
| 1944 | completed_passes + (pass_counter/pass_limit) |
| 1945 | -------------------------------------------- |
| 1946 | total_passes |
| 1947 | ignoring the fact that the passes may not be equal amounts of work. |
| 1948 | |
| 1949 | When decompressing, the total_passes value is not trustworthy, because it |
| 1950 | depends on the number of scans in the JPEG file, which isn't always known in |
| 1951 | advance. In the current implementation, completed_passes may jump by more |
| 1952 | than one when dealing with a multiple-scan input file. About all that is |
| 1953 | really safe to assume is that when completed_passes = total_passes - 1, the |
| 1954 | current pass will be the last one. |
| 1955 | |
| 1956 | If you really need to use the callback mechanism for time-critical tasks |
| 1957 | like mouse tracking, you could insert additional calls inside some of the |
| 1958 | library's inner loops. |
| 1959 | |
| 1960 | |
| 1961 | Memory management |
| 1962 | ----------------- |
| 1963 | |
| 1964 | This section covers some key facts about the JPEG library's built-in memory |
| 1965 | manager. For more info, please read structure.doc's section about the memory |
| 1966 | manager, and consult the source code if necessary. |
| 1967 | |
| 1968 | All memory and temporary file allocation within the library is done via the |
| 1969 | memory manager. If necessary, you can replace the "back end" of the memory |
| 1970 | manager to control allocation yourself (for example, if you don't want the |
| 1971 | library to use malloc() and free() for some reason). |
| 1972 | |
| 1973 | Some data is allocated "permanently" and will not be freed until the JPEG |
| 1974 | object is destroyed. Most data is allocated "per image" and is freed by |
| 1975 | jpeg_finish_compress, jpeg_finish_decompress, or jpeg_abort. You can call the |
| 1976 | memory manager yourself to allocate structures that will automatically be |
| 1977 | freed at these times. Typical code for this is |
| 1978 | ptr = (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE, size); |
| 1979 | Use JPOOL_PERMANENT to get storage that lasts as long as the JPEG object. |
| 1980 | Use alloc_large instead of alloc_small for anything bigger than a few Kbytes. |
| 1981 | There are also alloc_sarray and alloc_barray routines that automatically |
| 1982 | build 2-D sample or block arrays. |
| 1983 | |
| 1984 | The library's minimum space requirements to process an image depend on the |
| 1985 | image's width, but not on its height, because the library ordinarily works |
| 1986 | with "strip" buffers that are as wide as the image but just a few rows high. |
| 1987 | Some operating modes (eg, two-pass color quantization) require full-image |
| 1988 | buffers. Such buffers are treated as "virtual arrays": only the current strip |
| 1989 | need be in memory, and the rest can be swapped out to a temporary file. |
| 1990 | |
| 1991 | If you use the simplest memory manager back end (jmemnobs.c), then no |
| 1992 | temporary files are used; virtual arrays are simply malloc()'d. Images bigger |
| 1993 | than memory can be processed only if your system supports virtual memory. |
| 1994 | The other memory manager back ends support temporary files of various flavors |
| 1995 | and thus work in machines without virtual memory. They may also be useful on |
| 1996 | Unix machines if you need to process images that exceed available swap space. |
| 1997 | |
| 1998 | When using temporary files, the library will make the in-memory buffers for |
| 1999 | its virtual arrays just big enough to stay within a "maximum memory" setting. |
| 2000 | Your application can set this limit by setting cinfo->mem->max_memory_to_use |
| 2001 | after creating the JPEG object. (Of course, there is still a minimum size for |
| 2002 | the buffers, so the max-memory setting is effective only if it is bigger than |
| 2003 | the minimum space needed.) If you allocate any large structures yourself, you |
| 2004 | must allocate them before jpeg_start_compress() or jpeg_start_decompress() in |
| 2005 | order to have them counted against the max memory limit. Also keep in mind |
| 2006 | that space allocated with alloc_small() is ignored, on the assumption that |
| 2007 | it's too small to be worth worrying about. |
| 2008 | |
| 2009 | If you use the jmemname.c or jmemdos.c memory manager back end, it is |
| 2010 | important to clean up the JPEG object properly to ensure that the temporary |
| 2011 | files get deleted. (This is especially crucial with jmemdos.c, where the |
| 2012 | "temporary files" may be extended-memory segments; if they are not freed, |
| 2013 | DOS will require a reboot to recover the memory.) Thus, with these memory |
| 2014 | managers, it's a good idea to provide a signal handler that will trap any |
| 2015 | early exit from your program. The handler should call either jpeg_abort() |
| 2016 | or jpeg_destroy() for any active JPEG objects. A handler is not needed with |
| 2017 | jmemnobs.c, and shouldn't be necessary with jmemansi.c either, since the C |
| 2018 | library is supposed to take care of deleting files made with tmpfile(). |
| 2019 | |
| 2020 | |
| 2021 | Library compile-time options |
| 2022 | ---------------------------- |
| 2023 | |
| 2024 | A number of compile-time options are available by modifying jmorecfg.h. |
| 2025 | |
| 2026 | The JPEG standard provides for both the baseline 8-bit DCT process and |
| 2027 | a 12-bit DCT process. 12-bit lossy JPEG is supported if you define |
| 2028 | BITS_IN_JSAMPLE as 12 rather than 8. Note that this causes JSAMPLE to be |
| 2029 | larger than a char, so it affects the surrounding application's image data. |
| 2030 | At present, a 12-bit library can handle *only* 12-bit images, not both |
| 2031 | precisions. (If you need to include both 8- and 12-bit libraries in a |
| 2032 | single application, you could probably do it by defining |
| 2033 | NEED_SHORT_EXTERNAL_NAMES for just one of the copies. You'd have to access |
| 2034 | the 8-bit and 12-bit copies from separate application source files. This is |
| 2035 | untested ... if you try it, we'd like to hear whether it works!) |
| 2036 | |
| 2037 | The maximum number of components (color channels) in the image is determined |
| 2038 | by MAX_COMPONENTS. The JPEG standard allows up to 255 components, but we |
| 2039 | expect that few applications will need more than four or so. |
| 2040 | |
| 2041 | On machines with unusual data type sizes, you may be able to improve |
| 2042 | performance or reduce memory space by tweaking the various typedefs in |
| 2043 | jmorecfg.h. In particular, on some RISC CPUs, access to arrays of "short"s |
| 2044 | is quite slow; consider trading memory for speed by making JCOEF, INT16, and |
| 2045 | UINT16 be "int" or "unsigned int". UINT8 is also a candidate to become int. |
| 2046 | You probably don't want to make JSAMPLE be int unless you have lots of memory |
| 2047 | to burn. |
| 2048 | |
| 2049 | You can reduce the size of the library by compiling out various optional |
| 2050 | functions. To do this, undefine xxx_SUPPORTED symbols as necessary. |
| 2051 | |
| 2052 | |
| 2053 | Portability considerations |
| 2054 | -------------------------- |
| 2055 | |
| 2056 | The JPEG library has been written to be extremely portable; the sample |
| 2057 | applications cjpeg and djpeg are slightly less so. This section summarizes |
| 2058 | the design goals in this area. (If you encounter any bugs that cause the |
| 2059 | library to be less portable than is claimed here, we'd appreciate hearing |
| 2060 | about them.) |
| 2061 | |
| 2062 | The code works fine on both ANSI and pre-ANSI C compilers, using any of the |
| 2063 | popular system include file setups, and some not-so-popular ones too. See |
| 2064 | install.doc for configuration procedures. |
| 2065 | |
| 2066 | The code is not dependent on the exact sizes of the C data types. As |
| 2067 | distributed, we make the assumptions that |
| 2068 | char is at least 8 bits wide |
| 2069 | short is at least 16 bits wide |
| 2070 | int is at least 16 bits wide |
| 2071 | long is at least 32 bits wide |
| 2072 | (These are the minimum requirements of the ANSI C standard.) Wider types will |
| 2073 | work fine, although memory may be used inefficiently if char is much larger |
| 2074 | than 8 bits or short is much bigger than 16 bits. The code should work |
| 2075 | equally well with 16- or 32-bit ints. |
| 2076 | |
| 2077 | In a system where these assumptions are not met, you may be able to make the |
| 2078 | code work by modifying the typedefs in jmorecfg.h. However, you will probably |
| 2079 | have difficulty if int is less than 16 bits wide, since references to plain |
| 2080 | int abound in the code. |
| 2081 | |
| 2082 | char can be either signed or unsigned, although the code runs faster if an |
| 2083 | unsigned char type is available. If char is wider than 8 bits, you will need |
| 2084 | to redefine JOCTET and/or provide custom data source/destination managers so |
| 2085 | that JOCTET represents exactly 8 bits of data on external storage. |
| 2086 | |
| 2087 | The JPEG library proper does not assume ASCII representation of characters. |
| 2088 | But some of the image file I/O modules in cjpeg/djpeg do have ASCII |
| 2089 | dependencies in file-header manipulation; so does cjpeg's select_file_type() |
| 2090 | routine. |
| 2091 | |
| 2092 | The JPEG library does not rely heavily on the C library. In particular, C |
| 2093 | stdio is used only by the data source/destination modules and the error |
| 2094 | handler, all of which are application-replaceable. (cjpeg/djpeg are more |
| 2095 | heavily dependent on stdio.) malloc and free are called only from the memory |
| 2096 | manager "back end" module, so you can use a different memory allocator by |
| 2097 | replacing that one file. |
| 2098 | |
| 2099 | The code generally assumes that C names must be unique in the first 15 |
| 2100 | characters. However, global function names can be made unique in the |
| 2101 | first 6 characters by defining NEED_SHORT_EXTERNAL_NAMES. |
| 2102 | |
| 2103 | More info about porting the code may be gleaned by reading jconfig.doc, |
| 2104 | jmorecfg.h, and jinclude.h. |
| 2105 | |
| 2106 | |
| 2107 | Notes for MS-DOS implementors |
| 2108 | ----------------------------- |
| 2109 | |
| 2110 | The IJG code is designed to work efficiently in 80x86 "small" or "medium" |
| 2111 | memory models (i.e., data pointers are 16 bits unless explicitly declared |
| 2112 | "far"; code pointers can be either size). You may be able to use small |
| 2113 | model to compile cjpeg or djpeg by itself, but you will probably have to use |
| 2114 | medium model for any larger application. This won't make much difference in |
| 2115 | performance. You *will* take a noticeable performance hit if you use a |
| 2116 | large-data memory model (perhaps 10%-25%), and you should avoid "huge" model |
| 2117 | if at all possible. |
| 2118 | |
| 2119 | The JPEG library typically needs 2Kb-3Kb of stack space. It will also |
| 2120 | malloc about 20K-30K of near heap space while executing (and lots of far |
| 2121 | heap, but that doesn't count in this calculation). This figure will vary |
| 2122 | depending on selected operating mode, and to a lesser extent on image size. |
| 2123 | Thus you have perhaps 25K available for static data and other modules' near |
| 2124 | heap space before you need to go to a larger memory model. The C library's |
| 2125 | static data will account for several K of this, but that still leaves a good |
| 2126 | deal for your needs. (If you are tight on space, you could reduce the sizes |
| 2127 | of the I/O buffers allocated by jdatasrc.c and jdatadst.c, say from 4K to |
| 2128 | 1K.) |
| 2129 | |
| 2130 | About 2K of the near heap space is "permanent" memory that will not be |
| 2131 | released until you destroy the JPEG object. This is only an issue if you |
| 2132 | save a JPEG object between compression or decompression operations. |
| 2133 | |
| 2134 | Far data space may also be a tight resource when you are dealing with large |
| 2135 | images. The most memory-intensive case is decompression with two-pass color |
| 2136 | quantization, or single-pass quantization to an externally supplied color |
| 2137 | map. This requires a 128Kb color lookup table plus strip buffers amounting |
| 2138 | to about 50 bytes per column for typical sampling ratios (eg, about 32000 |
| 2139 | bytes for a 640-pixel-wide image). You may not be able to process wide |
| 2140 | images if you have large data structures of your own. |
| 2141 | |
| 2142 | Of course, all of these concerns vanish if you use a 32-bit flat-memory-model |
| 2143 | compiler, such as DJGPP or Watcom C. We highly recommend flat model if you |
| 2144 | can use it; the JPEG library is significantly faster in flat model. |