summaryrefslogtreecommitdiff
blob: dc1064a18bf3951731eb0a49e7066acfd3755ee8 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
/* Copyright (C) 2001-2021 Artifex Software, Inc.
   All Rights Reserved.

   This software is provided AS-IS with no warranty, either express or
   implied.

   This software is distributed under license and may not be copied,
   modified or distributed except as expressly authorized under the terms
   of the license contained in the file LICENSE in this distribution.

   Refer to licensing information at http://www.artifex.com or contact
   Artifex Software, Inc.,  1305 Grant Avenue - Suite 200, Novato,
   CA 94945, U.S.A., +1(415)492-9861, for further information.
*/


/* Color halftone rendering for Ghostscript imaging library */
#include "memory_.h"
#include "gx.h"
#include "gserrors.h"
#include "gsutil.h"		/* for id generation */
#include "gxarith.h"
#include "gxfixed.h"
#include "gxmatrix.h"
#include "gxdevice.h"
#include "gxcmap.h"
#include "gxdcolor.h"
#include "gxgstate.h"
#include "gzht.h"
#include "gsserial.h"
#include "gxdevsop.h"

/* Define whether to force use of the slow code, for testing. */
#define USE_SLOW_CODE 0

/* Define the size of the tile buffer allocated on the stack. */
#define tile_longs_LARGE 256
#define tile_longs_SMALL 64
#if ARCH_SMALL_MEMORY
#  define tile_longs_allocated tile_longs_SMALL
#  define tile_longs tile_longs_SMALL
#else
#  define tile_longs_allocated tile_longs_LARGE
#  ifdef DEBUG
#    define tile_longs\
       (gs_debug_c('.') ? tile_longs_SMALL : tile_longs_LARGE)
#  else
#    define tile_longs tile_longs_LARGE
#  endif
#endif

/* Define the colored halftone device color type. */
gs_private_st_ptrs1(st_dc_ht_colored, gx_device_color, "dc_ht_colored",
    dc_ht_colored_enum_ptrs, dc_ht_colored_reloc_ptrs, colors.colored.c_ht);
static dev_color_proc_save_dc(gx_dc_ht_colored_save_dc);
static dev_color_proc_get_dev_halftone(gx_dc_ht_colored_get_dev_halftone);
static dev_color_proc_load(gx_dc_ht_colored_load);
static dev_color_proc_fill_rectangle(gx_dc_ht_colored_fill_rectangle);
static dev_color_proc_equal(gx_dc_ht_colored_equal);
static dev_color_proc_write(gx_dc_ht_colored_write);
static dev_color_proc_read(gx_dc_ht_colored_read);
const gx_device_color_type_t gx_dc_type_data_ht_colored = {
    &st_dc_ht_colored,
    gx_dc_ht_colored_save_dc, gx_dc_ht_colored_get_dev_halftone,
    gx_dc_ht_get_phase,
    gx_dc_ht_colored_load, gx_dc_ht_colored_fill_rectangle,
    gx_dc_default_fill_masked, gx_dc_ht_colored_equal,
    gx_dc_ht_colored_write, gx_dc_ht_colored_read,
    gx_dc_ht_colored_get_nonzero_comps
};
#undef gx_dc_type_ht_colored
const gx_device_color_type_t *const gx_dc_type_ht_colored =
    &gx_dc_type_data_ht_colored;
#define gx_dc_type_ht_colored (&gx_dc_type_data_ht_colored)

/* save information about the operand device color */
static void
gx_dc_ht_colored_save_dc(const gx_device_color * pdevc,
                         gx_device_color_saved * psdc)
{
    psdc->type = pdevc->type;
    memcpy( psdc->colors.colored.c_base,
            pdevc->colors.colored.c_base,
            sizeof(psdc->colors.colored.c_base) );
    memcpy( psdc->colors.colored.c_level,
            pdevc->colors.colored.c_level,
            sizeof(psdc->colors.colored.c_base) );
    psdc->colors.colored.alpha = pdevc->colors.colored.alpha;
    psdc->phase = pdevc->phase;
}

/* get the halftone used for the operand device color */
static const gx_device_halftone *
gx_dc_ht_colored_get_dev_halftone(const gx_device_color * pdevc)
{
    return pdevc->colors.colored.c_ht;
}

/* Compare two colored halftones for equality. */
static bool
gx_dc_ht_colored_equal(const gx_device_color * pdevc1,
                       const gx_device_color * pdevc2)
{
    uint num_comp = pdevc1->colors.colored.num_components;

    if (pdevc2->type != pdevc1->type ||
        pdevc1->colors.colored.c_ht != pdevc2->colors.colored.c_ht ||
        pdevc1->colors.colored.alpha != pdevc2->colors.colored.alpha ||
        pdevc1->phase.x != pdevc2->phase.x ||
        pdevc1->phase.y != pdevc2->phase.y ||
        num_comp != pdevc2->colors.colored.num_components
        )
        return false;
    return
        !memcmp(pdevc1->colors.colored.c_base,
                pdevc2->colors.colored.c_base,
                num_comp * sizeof(pdevc1->colors.colored.c_base[0])) &&
        !memcmp(pdevc1->colors.colored.c_level,
                pdevc2->colors.colored.c_level,
                num_comp * sizeof(pdevc1->colors.colored.c_level[0]));
}

/*
 * Flags to indicate the pieces of a colored halftone that are included
 * in its string representation. The first byte of the string holds this
 * set of flags.
 *
 * The case alpha = gx_max_color_value is by far the most common, so
 * special treatment is provided for this case.
 *
 * The halftone is never transmitted as part of a device color, so there
 * is no flag for it.
 */
static const int   dc_ht_colored_has_base = 0x01;
static const int   dc_ht_colored_has_level = 0x02;
static const int   dc_ht_colored_has_alpha = 0x04;
static const int   dc_ht_colored_alpha_is_max = 0x08;

/*
 * Serialize a device color that uses a traditional colored halftone.
 *
 * The number of components of a device color must match that of the
 * process color model, so it is not transmitted.
 *
 * The most common situation in which this routine is used is for 1-bit
 * per component devices. In that case, base[i] will always be 0 or 1,
 * and thus may be fit in a single bit.
 *
 * In many situations, one or more of the color component intensity
 * levels will be 0. The plane_mask field identifies those components
 * where this is not the case. By tansmitting the plane_mask, only those
 * components with non-zero levels need be transmitted.
 *
 * The case alpha = gx_max_color_value is by far the most common, so
 * special treatment is provided for this case.
 *
 *
 * Operands:
 *
 *  pdevc       pointer to device color to be serialized
 *
 *  psdc        pointer ot saved version of last serialized color (for
 *              this band)
 *
 *  dev         pointer to the current device, used to retrieve process
 *              color model information
 *
 *  pdata       pointer to buffer in which to write the data
 *
 *  psize       pointer to a location that, on entry, contains the size of
 *              the buffer pointed to by pdata; on return, the size of
 *              the data required or actually used will be written here.
 *
 * Returns:
 *  1, with *psize set to 0, if *psdc and *pdevc represent the same color
 *
 *  0, with *psize set to the amount of data written, if everything OK
 *
 *  gs_error_rangecheck, with *psize set to the size of buffer required,
 *  if *psize was not large enough
 *
 *  < 0, != gs_error_rangecheck, in the event of some other error
 *  (currently none); in this case *psize is not changed.
 */
static int
gx_dc_ht_colored_write(
    const gx_device_color *         pdevc,
    const gx_device_color_saved *   psdc0,
    const gx_device *               dev,
    int64_t			    offset,
    byte *                          pdata,
    uint *                          psize )
{
    int                             req_size = 1;
    int                             flag_bits = 0;
    int                             num_comps = dev->color_info.num_components;
    int                             depth = dev->color_info.depth;
    gx_color_index                  plane_mask = pdevc->colors.colored.plane_mask;
    gx_color_value                  alpha = pdevc->colors.colored.alpha;
    const gx_device_color_saved *   psdc = psdc0;
    byte *                          pdata0 = pdata;

    if (offset != 0)
        return_error(gs_error_unregistered); /* Not implemented yet. */

    /* sanity check */
    if (pdevc->colors.colored.num_components != num_comps)
        return_error(gs_error_unregistered); /* Must not happen. */

    /* check if saved color is of the same type */
    if (psdc != 0 && psdc->type != pdevc->type)
        psdc = 0;

    /* calculate the size required */
    if ( psdc == 0                                                         ||
         memcmp( pdevc->colors.colored.c_base,
                 psdc->colors.colored.c_base,
                 num_comps * sizeof(pdevc->colors.colored.c_base[0]) ) != 0  ) {
        flag_bits |= dc_ht_colored_has_base;
        if (num_comps == depth)     /* 1 bit / component */
            req_size += (num_comps + 7) >> 3;
        else
            req_size += num_comps * sizeof(pdevc->colors.colored.c_base[0]);
    }

    plane_mask = pdevc->colors.colored.plane_mask;
    if ( psdc == 0                                                          ||
         memcmp( pdevc->colors.colored.c_level,
                 psdc->colors.colored.c_level,
                 num_comps * sizeof(pdevc->colors.colored.c_level[0]) ) != 0  ) {
        gx_color_index  comp_bit;
        int             i;
        uint            tmp_mask;

        flag_bits |= dc_ht_colored_has_level;
        if (num_comps > 8 * sizeof(uint)) {
            tmp_mask = (uint)plane_mask;
            req_size += enc_u_sizew(tmp_mask);
            tmp_mask = (uint)((plane_mask >> (8*sizeof(uint)-1)) >> 1);
            req_size += enc_u_sizew(tmp_mask);
        } else {
            tmp_mask = (uint)plane_mask;
            req_size += enc_u_sizew(tmp_mask);
        }
        for (i = 0, comp_bit = 0x1; i < num_comps; i++, comp_bit <<= 1) {
            if ((plane_mask & comp_bit) != 0)
                req_size += enc_u_sizew(pdevc->colors.colored.c_level[i]);
        }
    }

    if (psdc == 0 || alpha != psdc->colors.colored.alpha) {
        if (alpha == gx_max_color_value)
            flag_bits |= dc_ht_colored_alpha_is_max;
        else {
            flag_bits |= dc_ht_colored_has_alpha;
            req_size += enc_u_sizew(alpha);
        }
    }

    /* see if there is anything to do */
    if (flag_bits == 0) {
        *psize = 0;
        return 1;
    }

    /* see if enough space is available */
    if (req_size > *psize) {
        *psize = req_size;
        return_error(gs_error_rangecheck);
    }

    /* write out the flag byte */
    *pdata++ = (byte)flag_bits;

    /* write out such other parts of the device color as required */
    if ((flag_bits & dc_ht_colored_has_base) != 0) {
        if (num_comps == depth) {
            gx_color_index  base_mask = 0;
            int             num_bytes = (num_comps + 7) >> 3;
            int             i;

            for (i = 0; i < num_comps; i++) {
                if (pdevc->colors.colored.c_base[i] != 0)
                    base_mask |= (gx_color_index)1 << i;
            }
            for (i = 0; i < num_bytes; i++, base_mask >>= 8)
                *pdata++ = (byte)base_mask;
        } else {
            memcpy( pdata,
                    pdevc->colors.colored.c_base,
                    num_comps * sizeof(pdevc->colors.colored.c_base[0]) );
            pdata += num_comps * sizeof(pdevc->colors.colored.c_base[0]);
        }
    }

    if ((flag_bits & dc_ht_colored_has_level) != 0) {
        gx_color_index  code_bit;
        int             i;
        uint            tmp_mask;

        if (num_comps > 8 * sizeof(uint)) {
            tmp_mask = (uint)plane_mask;
            enc_u_putw(tmp_mask, pdata);
            tmp_mask = (uint)((plane_mask >> (8*sizeof(uint)-1))>>1);
            enc_u_putw(tmp_mask, pdata);
        } else {
            tmp_mask = (uint)plane_mask;
            enc_u_putw(tmp_mask, pdata);
        }
        for (i = 0, code_bit = 0x1; i < num_comps; i++, code_bit <<= 1) {
            if ((plane_mask & code_bit) != 0)
                enc_u_putw(pdevc->colors.colored.c_level[i], pdata);
        }
    }

    if ((flag_bits & dc_ht_colored_has_alpha) != 0)
        enc_u_putw(alpha, pdata);

    *psize = pdata - pdata0;
    return 0;
}

/*
 * Reconstruct a device color from its serial representation.
 *
 * Operands:
 *
 *  pdevc       pointer to the location in which to write the
 *              reconstructed device color
 *
 *  pgs         pointer to the current gs_gstate (to access the
 *              current halftone)
 *
 *  prior_devc  pointer to the current device color (this is provided
 *              separately because the device color is not part of the
 *              gs_gstate)
 *
 *  dev         pointer to the current device, used to retrieve process
 *              color model information
 *
 *  pdata       pointer to the buffer to be read
 *
 *  size        size of the buffer to be read; this should be large
 *              enough to hold the entire color description
 *
 *  mem         pointer to the memory to be used for allocations
 *              (ignored here)
 *
 * Returns:
 *
 *  # of bytes read if everthing OK, < 0 in the event of an error
 */
static int
gx_dc_ht_colored_read(
    gx_device_color *       pdevc,
    const gs_gstate * pgs,
    const gx_device_color * prior_devc,
    const gx_device *       dev,
    int64_t		    offset,
    const byte *            pdata,
    uint                    size,
    gs_memory_t *           mem )       /* ignored */
{
    gx_device_color         devc;
    int                     num_comps = dev->color_info.num_components;
    int                     depth = dev->color_info.depth;
    const byte *            pdata0 = pdata;
    int                     flag_bits;

    if (offset != 0)
        return_error(gs_error_unregistered); /* Not implemented yet. */

    /* if prior information is available, use it */
    if (prior_devc != 0 && prior_devc->type == gx_dc_type_ht_colored)
        devc = *prior_devc;
    else
        memset(&devc, 0, sizeof(devc));   /* clear pointers */
    devc.type = gx_dc_type_ht_colored;

    /* the number of components is determined by the color model */
    devc.colors.colored.num_components = num_comps;
    devc.colors.colored.c_ht = pgs->dev_ht;

    /*
     * Verify that we have at least the flag bits. For performance
     * reasons, the routines that convert serialized representations
     * of integers do not check buffer size. Hence, in many cases below,
     * only a very rough check is made to verify that we have not
     * exhausted the buffer. This should not cause a problem in
     * practice.
     */
    if (size == 0)
        return_error(gs_error_rangecheck);
    size--;
    flag_bits = *pdata++;

    /* read the other components provided */
    if ((flag_bits & dc_ht_colored_has_base) != 0) {
        if (depth == num_comps) {
            gx_color_index  base_mask = 0;
            int             num_bytes = (num_comps + 7) >> 3;
            int             i, shift = 0;

            if (size < num_bytes)
                return_error(gs_error_rangecheck);
            size -= num_bytes;
            for (i = 0; i < num_bytes; i++, shift += 8)
                base_mask |= (gx_color_index)(*pdata++) << shift;
            for (i = 0; i < num_comps; i++, base_mask >>= 1)
                devc.colors.colored.c_base[i] = base_mask & 0x1;
        } else {
            if (size < num_comps)
                return_error(gs_error_rangecheck);
            size -= num_comps;
            memcpy(devc.colors.colored.c_base, pdata, num_comps);
            pdata += num_comps;
        }
    }

    if ((flag_bits & dc_ht_colored_has_level) != 0) {
        const byte *    pdata_start = pdata;
        gx_color_index  plane_mask;
        uint            tmp_mask;
        int             i;

        if (size < 1)
            return_error(gs_error_rangecheck);

        if (num_comps > 8 * sizeof(uint)) {
            enc_u_getw(tmp_mask, pdata);
            plane_mask = (gx_color_index)tmp_mask;
            enc_u_getw(tmp_mask, pdata);
            plane_mask = (((gx_color_index)tmp_mask)<<(8 * sizeof(uint)-1))<<1;
        } else {
            enc_u_getw(tmp_mask, pdata);
            plane_mask = (gx_color_index)tmp_mask;
        }
        devc.colors.colored.plane_mask = plane_mask;
        for (i = 0; i < num_comps; i++, plane_mask >>= 1) {
            if ((plane_mask & 0x1) != 0) {
                if (size - (pdata - pdata_start) < 1)
                    return_error(gs_error_rangecheck);
                enc_u_getw(devc.colors.colored.c_level[i], pdata);
            } else
                devc.colors.colored.c_level[i] = 0;
        }
        size -= pdata - pdata_start;
    }

    if ((flag_bits & dc_ht_colored_alpha_is_max) != 0)
        devc.colors.colored.alpha = gx_max_color_value;
    else if ((flag_bits & dc_ht_colored_has_alpha) != 0) {
        const byte *    pdata_start = pdata;

        if (size < 1)
            return_error(gs_error_rangecheck);
        enc_u_getw(devc.colors.colored.alpha, pdata);
        size -= pdata - pdata_start;
    }

    /* set the phase as required (select value is arbitrary) */
    color_set_phase_mod( &devc,
                         pgs->screen_phase[0].x,
                         pgs->screen_phase[0].y,
                         pgs->dev_ht->lcm_width,
                         pgs->dev_ht->lcm_height );

    /* everything looks OK */
    *pdevc = devc;
    return pdata - pdata0;
}

/*
 * Get the nonzero components of a coloredhalftone. This is used to
 * distinguish components that are given zero intensity due to halftoning
 * from those for which the original color intensity was in fact zero.
 *
 * An original component intensity of zero will yield a c_base value of
 * 0 and a c_level of 0. The plane_mask field already contains the latter
 * information, so we need only add those components for which c_base is
 * non-zero.
 */
int
gx_dc_ht_colored_get_nonzero_comps(
    const gx_device_color * pdevc,
    const gx_device *       dev_ignored,
    gx_color_index *        pcomp_bits )
{
    int                     i, ncomps =  pdevc->colors.colored.num_components;
    gx_color_index          comp_bits = pdevc->colors.colored.plane_mask;

    for (i = 0; i < ncomps; i++) {
        if (pdevc->colors.colored.c_base[i] != 0)
            comp_bits |= ((gx_color_index)1) << i;
    }
    *pcomp_bits = comp_bits;

    return 0;
}

/*
 * Define an abbreviation for a heavily used value: the maximum number of
 * of device colorants (device colors).
 */
#define MAX_DCC GX_DEVICE_COLOR_MAX_COMPONENTS
/*
 * Define a size for the "colors" array.  For less than 5 colors, there are
 * 2**n values stored (for a maximum of 16 values).  For 5 or more colors, we
 * only store 2 values per color so the array size can be 2 * MAX_DCC.  Use which
 * ever is larger for the array size.
 */
#define MAX_DCC_16 (2 * MAX_DCC < 16 ? 16 : 2 * MAX_DCC)

/* Forward references. */
/* Use a typedef to attempt to work around overly picky compilers. */
typedef gx_color_value gx_color_value_array[MAX_DCC];
typedef struct color_values_pair_s {
    gx_color_value_array values[2];
} color_values_pair_t;
#define SET_HT_COLORS_PROC(proc)\
  int proc(\
              color_values_pair_t *pvp,\
              gx_color_index colors[MAX_DCC_16],\
              const gx_const_strip_bitmap *sbits[MAX_DCC],\
              const gx_device_color *pdevc,\
              gx_device *dev,\
              gx_ht_cache *caches[MAX_DCC],\
              int nplanes\
              )

static SET_HT_COLORS_PROC(set_ht_colors_le_4);
static SET_HT_COLORS_PROC(set_cmyk_1bit_colors);
static SET_HT_COLORS_PROC(set_ht_colors_gt_4);

#define SET_COLOR_HT_PROC(proc)\
  void proc(\
                byte *dest_data, /* the output tile */\
                uint dest_raster, /* ibid. */\
                int px,	/* the initial phase of the output tile */\
                int py,\
                int w,	/* how much of the tile to set */\
                int h,\
                int depth,	/* depth of tile (4, 8, 16, 24, 32) */\
                int special,	/* >0 means special 1-bit CMYK */\
                int nplanes,\
                gx_color_index plane_mask,	/* which planes are halftoned */\
                gx_device *dev,		/* in case we are mapping lazily */\
                const color_values_pair_t *pvp,	/* color values ditto */\
                gx_color_index colors[MAX_DCC],	/* the actual colors for the tile, */\
                                /* actually [nplanes] */\
                const gx_const_strip_bitmap * sbits[MAX_DCC]	/* the bitmaps for the planes, */\
                                /* actually [nplanes] */\
                )

static SET_COLOR_HT_PROC(set_color_ht_le_4);
static SET_COLOR_HT_PROC(set_color_ht_gt_4);

/* Prepare to use a colored halftone, by loading the default cache. */
static int
gx_dc_ht_colored_load(gx_device_color * pdevc, const gs_gstate * pgs,
                      gx_device * ignore_dev, gs_color_select_t select)
{
    /* TO_DO_DEVICEN */

    return 0;
}

/* Fill a rectangle with a colored halftone. */
/* Note that we treat this as "texture" for RasterOp. */
static int
gx_dc_ht_colored_fill_rectangle(const gx_device_color * pdevc,
                                int x, int y, int w, int h,
                                gx_device * dev, gs_logical_operation_t lop,
                                const gx_rop_source_t * source)
{
#if defined(PACIFY_VALGRIND) || defined(MEMENTO)
    ulong tbits[tile_longs_allocated] = { 0 };
#else
    ulong tbits[tile_longs_allocated];
#endif
    const uint tile_bytes = tile_longs * size_of(long);
    gx_strip_bitmap tiles;
    gx_rop_source_t no_source;
    const gx_device_halftone *pdht = pdevc->colors.colored.c_ht;
    int depth = dev->color_info.depth;
    int nplanes = dev->color_info.num_components;

    SET_HT_COLORS_PROC((*set_ht_colors)) =
        (
#if USE_SLOW_CODE
         set_ht_colors_gt_4
#else
         (dev_proc(dev, dev_spec_op)(dev, gxdso_is_std_cmyk_1bit, NULL, 0) > 0) ?
            set_cmyk_1bit_colors :
            nplanes <= 4 ? set_ht_colors_le_4 :
                          set_ht_colors_gt_4
#endif
         );
    SET_COLOR_HT_PROC((*set_color_ht)) =
        (
#if !USE_SLOW_CODE
         !(pdevc->colors.colored.plane_mask & ~(gx_color_index)15) &&
          set_ht_colors != set_ht_colors_gt_4 ?
           set_color_ht_le_4 :
#endif
         set_color_ht_gt_4);
    color_values_pair_t vp;
    gx_color_index colors[MAX_DCC_16];
    const gx_const_strip_bitmap *sbits[MAX_DCC];
    gx_ht_cache *caches[MAX_DCC];
    int special;
    int code = 0;
    int raster;
    uint size_x;
    int dw, dh;
    int lw = pdht->lcm_width, lh = pdht->lcm_height;
    bool no_rop;
    int i;
    int origx, origy;

    /* This routine cannot build 3bit chunky halftones, as 3 bit
     * things don't pack nicely into bytes or words. Accordingly
     * treat 3 bit things as 4 bit things. This is appropriate when
     * generating halftones for planar. */
    if (depth == 3)
        depth = 4;

    if (w <= 0 || h <= 0)
        return 0;
    origx = x;
    origy = y;
    if ((w | h) >= 16) {
        /* It's worth taking the trouble to check the clipping box. */
        gs_fixed_rect cbox;
        int t;

        dev_proc(dev, get_clipping_box)(dev, &cbox);
        if ((t = fixed2int(cbox.p.x)) > x) {
            if ((w += x - t) <= 0)
                return 0;
            x = t;
        }
        if ((t = fixed2int(cbox.p.y)) > y) {
            if ((h += y - t) <= 0)
                return 0;
            y = t;
        }
        if ((t = fixed2int(cbox.q.x)) < x + w)
            if ((w = t - x) <= 0)
                return 0;
        if ((t = fixed2int(cbox.q.y)) < y + h)
            if ((h = t - y) <= 0)
                return 0;
    }
    /* Colored halftone patterns are unconditionally opaque. */
    lop &= ~lop_T_transparent;
    if (pdht->components == 0) {
        caches[0] = caches[1] = caches[2] = caches[3] = pdht->order.cache;
        for (i = 4; i < nplanes; ++i)
            caches[i] = pdht->order.cache;
    } else {
        gx_ht_order_component *pocs = pdht->components;

        for (i = 0; i < nplanes; ++i)
            caches[i] = pocs[i].corder.cache;
    }
    special = set_ht_colors(&vp, colors, sbits, pdevc, dev, caches, nplanes);
    no_rop = source == NULL && lop_no_S_is_T(lop);
    if ((!no_rop) && (source == NULL))
        set_rop_no_source(source, no_source, dev);

    /*
     * If the LCM of the plane cell sizes is smaller than the rectangle
     * being filled, compute a single tile and let tile_rectangle do the
     * replication.
     */
    if ((w > lw || h > lh) &&
        (raster = bitmap_raster(lw * depth)) <= tile_bytes / lh
        ) {
        /*
         * The only reason we need to do fit_fill here is that if the
         * device is a clipper, the caller might be counting on it to do
         * all necessary clipping.  Actually, we should clip against the
         * device's clipping box, not the default....
         */
        fit_fill(dev, x, y, w, h);
        /* Check to make sure we still have a big rectangle. */
        if (w > lw || h > lh) {
            tiles.data = (byte *)tbits;
            tiles.raster = raster;
            tiles.rep_width = tiles.size.x = lw;
            tiles.rep_height = tiles.size.y = lh;
            tiles.id = gs_next_ids(pdht->rc.memory, 1);
            tiles.rep_shift = tiles.shift = 0;
            tiles.num_planes = 1;
            set_color_ht((byte *)tbits, raster, 0, 0, lw, lh, depth,
                         special, nplanes, pdevc->colors.colored.plane_mask,
                         dev, &vp, colors, sbits);
            if (no_rop)
                return (*dev_proc(dev, strip_tile_rectangle)) (dev, &tiles,
                                                               x, y, w, h,
                                                               gx_no_color_index,
                                                               gx_no_color_index,
                                                               pdevc->phase.x,
                                                               pdevc->phase.y);
            if (source->planar_height == 0)
                return (*dev_proc(dev, strip_copy_rop))
                               (dev,
                                source->sdata + (y - origy) * source->sraster,
                                source->sourcex + (x - origx),
                                source->sraster, source->id,
                                (source->use_scolors ? source->scolors : NULL),
                                &tiles, NULL,
                                x, y, w, h,
                                pdevc->phase.x, pdevc->phase.y, lop);
            else
                return (*dev_proc(dev, strip_copy_rop2))
                               (dev,
                                source->sdata + (y - origy) * source->sraster,
                                source->sourcex + (x - origx),
                                source->sraster, source->id,
                                (source->use_scolors ? source->scolors : NULL),
                                &tiles, NULL,
                                x, y, w, h,
                                pdevc->phase.x, pdevc->phase.y, lop,
                                source->planar_height);
        }
    }
    size_x = w * depth;
    raster = bitmap_raster(size_x);
    if (raster > tile_bytes) {
        /*
         * We can't even do an entire line at once.  See above for
         * why we do the X equivalent of fit_fill here.
         */
        if (x < 0)
            w += x, x = 0;
        if (x > dev->width - w)
            w = dev->width - x;
        if (w <= 0)
            return 0;
        size_x = w * depth;
        raster = bitmap_raster(size_x);
        if (raster > tile_bytes) {
            /* We'll have to do a partial line. */
            dw = tile_bytes * 8 / depth;
            size_x = dw * depth;
            raster = bitmap_raster(size_x);
            dh = 1;
            goto fit;
        }
    }
    /* Do as many lines as will fit. */
    dw = w;
    dh = tile_bytes / raster;
    if (dh > h)
        dh = h;
fit:				/* Now the tile will definitely fit. */
    if (!no_rop) {
        tiles.data = (byte *)tbits;
        tiles.id = gx_no_bitmap_id;
        tiles.raster = raster;
        tiles.rep_width = tiles.size.x = size_x / depth;
        tiles.rep_shift = tiles.shift = 0;
        tiles.num_planes = 1;
    }
    while (w) {
        int cy = y, ch = dh, left = h;

        for (;;) {
            set_color_ht((byte *)tbits, raster,
                         x + pdevc->phase.x, cy + pdevc->phase.y,
                         dw, ch, depth, special, nplanes,
                         pdevc->colors.colored.plane_mask,
                         dev, &vp, colors, sbits);
            if (no_rop) {
                code = (*dev_proc(dev, copy_color))
                    (dev, (byte *)tbits, 0, raster, gx_no_bitmap_id,
                     x, cy, dw, ch);
            } else {
                tiles.rep_height = tiles.size.y = ch;
                if (source->planar_height == 0)
                    code = (*dev_proc(dev, strip_copy_rop))
                          (dev, source->sdata + source->sraster * (cy-origy),
                           source->sourcex + (x - origx),
                           source->sraster,
                           source->id,
                           (source->use_scolors ? source->scolors : NULL),
                           &tiles, NULL, x, cy, dw, ch, 0, 0, lop);
                else
                    code = (*dev_proc(dev, strip_copy_rop2))
                          (dev, source->sdata + source->sraster * (cy-origy),
                           source->sourcex + (x - origx),
                           source->sraster,
                           source->id,
                           (source->use_scolors ? source->scolors : NULL),
                           &tiles, NULL, x, cy, dw, ch, 0, 0, lop,
                           source->planar_height);
            }
            if (code < 0)
                return code;
            if (!(left -= ch))
                break;
            cy += ch;
            if (ch > left)
                ch = left;
        }
        if (!(w -= dw))
            break;
        x += dw;
        if (dw > w)
            dw = w;
    }
    return code;
}

/* ---------------- Color table setup ---------------- */

/*
 * We could cache this if we had a place to store it.  Even a 1-element
 * cache would help performance substantially.
 *   Key: device + c_base/c_level of device color
 *   Value: colors table
 */

/*
 * We construct color halftone tiles out of multiple "planes".
 * Each plane specifies halftoning for one component (R/G/B, C/M/Y/K,
 * or DeviceN components).
 */

static const struct {
    ulong pad;			/* to get bytes aligned properly */
    byte bytes[sizeof(ulong) * 8];	/* 8 is arbitrary */
} ht_no_bitmap_data = { 0 };
static const gx_const_strip_bitmap ht_no_bitmap = {
    &ht_no_bitmap_data.bytes[0], sizeof(ulong),
    {sizeof(ulong) * 8, sizeof(ht_no_bitmap_data.bytes) / sizeof(ulong)},
    gx_no_bitmap_id, 1, 1, 0, 0
};

/* Set the color value(s) and halftone mask for one plane. */
static inline void set_plane_color(int i, color_values_pair_t *pvp, const gx_device_color * pdc,
    const gx_const_strip_bitmap * sbits[MAX_DCC], gx_ht_cache * caches[MAX_DCC],
    gx_color_value max_color, bool invert)
{
    uint q = pdc->colors.colored.c_base[i];
    uint r = pdc->colors.colored.c_level[i];

    pvp->values[0][i] = fractional_color(q, max_color);
    if (r == 0)
        pvp->values[1][i] = pvp->values[0][i], sbits[i] = &ht_no_bitmap;
    else if (!invert) {
        pvp->values[1][i] = fractional_color(q + 1, max_color);
        sbits[i] = (const gx_const_strip_bitmap *) &gx_render_ht(caches[i], r)->tiles;
    } else {
        const gx_device_halftone *pdht = pdc->colors.colored.c_ht;
        int nlevels =
            (pdht->components ? pdht->components[i].corder.num_levels : pdht->order.num_levels);
        pvp->values[1][i] = pvp->values[0][i];
        pvp->values[0][i] = fractional_color(q + 1, max_color);
        sbits[i] = (const gx_const_strip_bitmap *) &gx_render_ht(caches[i], nlevels - r)->tiles;
    }
}

/* Set up the colors and the individual plane halftone bitmaps. */
static int
set_ht_colors_le_4(color_values_pair_t *pvp /* only used internally */,
                   gx_color_index colors[MAX_DCC_16] /* 16 used */,
                   const gx_const_strip_bitmap * sbits[MAX_DCC],
                   const gx_device_color * pdc, gx_device * dev,
                   gx_ht_cache * caches[MAX_DCC], int nplanes)
{
    gx_color_value max_color = dev->color_info.dither_colors - 1;
    gx_color_value cvalues[4];
    /*
     * NB: the halftone orders are all set up for an additive color space.
     *     To make these work with a subtractive device space such as CMYK,
     *     it is necessary to invert both the color level and the color
     *     pair. Note that if the original color was provided an additive
     *     space, this will reverse (in an approximate sense) the color
     *     conversion performed to express the color in the device space.
     */
    bool invert = dev->color_info.polarity == GX_CINFO_POLARITY_SUBTRACTIVE;

    set_plane_color(0, pvp, pdc, sbits, caches, max_color, invert);
    if (nplanes >= 2) {
        set_plane_color(1, pvp, pdc, sbits, caches, max_color, invert);
    }
    if (nplanes >= 3) {
        set_plane_color(2, pvp, pdc, sbits, caches, max_color, invert);
    }
    if (nplanes == 3) {
        gx_color_value alpha = pdc->colors.colored.alpha;

        if (alpha == gx_max_color_value) {
#define M(i)\
  cvalues[0] = pvp->values[(i) & 1][0];\
  cvalues[1] = pvp->values[((i) & 2) >> 1][1];\
  cvalues[2] = pvp->values[(i) >> 2][2];\
  colors[i] = dev_proc(dev, encode_color)(dev, cvalues);

            M(0); M(1); M(2); M(3); M(4); M(5); M(6); M(7);
#undef M
        } else {
#define M(i)\
  colors[i] = dev_proc(dev, map_rgb_alpha_color)(dev, pvp->values[(i) & 1][0],\
                                     pvp->values[((i) & 2) >> 1][1],\
                                     pvp->values[(i) >> 2][2], alpha)
            M(0); M(1); M(2); M(3); M(4); M(5); M(6); M(7);
#undef M
        }
    } else if (nplanes > 3){
        set_plane_color(3, pvp, pdc, sbits, caches, max_color, invert);
        if (nplanes > 4) {
            /*
             * Set colors for any planes beyond the 4th.  Since this code
             * only handles the case of at most 4 active planes, we know
             * that any further planes are constant.
             */
            /****** DOESN'T MAP COLORS RIGHT, DOESN'T HANDLE ALPHA ******/
            int pi;

            for (pi = 4; pi < nplanes; ++pi) {
                pvp->values[1][pi] = pvp->values[0][pi] =
                    fractional_color(pdc->colors.colored.c_base[pi], max_color);
                sbits[pi] = &ht_no_bitmap;
            }
        }
        /*
         * For CMYK output, especially if the input was RGB, it's
         * common for one or more of the components to be zero.
         * Each zero component can cut the cost of color mapping in
         * half, so it's worth doing a little checking here.
         */
#define M(i)\
    cvalues[0] = pvp->values[(i) & 1][0];\
    cvalues[1] = pvp->values[((i) & 2) >> 1][1];\
    cvalues[2] = pvp->values[((i) & 4) >> 2][2];\
    cvalues[3] = pvp->values[(i) >> 3][3];\
    colors[i] = dev_proc(dev, encode_color)(dev, cvalues)

      /* We know that plane_mask <= 15. */
        switch ((int)pdc->colors.colored.plane_mask) {
            case 15:
                M(15); M(14); M(13); M(12);
                M(11); M(10); M(9); M(8);
            case 7:
                M(7); M(6); M(5); M(4);
c3:	    case 3:
                M(3); M(2);
c1:	    case 1:
                M(1);
                break;
            case 14:
                M(14); M(12); M(10); M(8);
            case 6:
                M(6); M(4);
c2:	    case 2:
                M(2);
                break;
            case 13:
                M(13); M(12); M(9); M(8);
            case 5:
                M(5); M(4);
                goto c1;
            case 12:
                M(12); M(8);
            case 4:
                M(4);
                break;
            case 11:
                M(11); M(10); M(9); M(8);
                goto c3;
            case 10:
                M(10); M(8);
                goto c2;
            case 9:
                M(9); M(8);
                goto c1;
            case 8:
                M(8);
                break;
            case 0:;
        }
        M(0);

#undef M
    }
    return 0;
}

/* Set up colors using the standard 1-bit CMYK mapping. */
static int
set_cmyk_1bit_colors(color_values_pair_t *ignore_pvp,
                     gx_color_index colors[MAX_DCC_16] /*2 used*/,
                     const gx_const_strip_bitmap * sbits[MAX_DCC /*4 used*/],
                     const gx_device_color * pdc, gx_device * dev,
                     gx_ht_cache * caches[MAX_DCC /*4 used*/],
                     int nplanes /*4*/)
{
    const gx_device_halftone *pdht = pdc->colors.colored.c_ht;
    /*
     * By reversing the order of the planes, we make the pixel values
     * line up with the color indices.  Then instead of a lookup, we
     * can compute the pixels directly using a Boolean function.
     *
     * We compute each output bit
     *   out[i] = (in[i] & mask1) | (~in[i] & mask0)
     * We store the two masks in colors[0] and colors[1], since the
     * colors array is otherwise unused in this case.  We duplicate
     * the values in all the nibbles so we can do several pixels at a time.
     */
    bits32 mask0 = 0, mask1 = 0;
#define SET_PLANE_COLOR_CMYK(i, mask)\
  BEGIN\
    uint r = pdc->colors.colored.c_level[i];\
\
    if (r == 0) {\
        if (pdc->colors.colored.c_base[i])\
            mask0 |= mask, mask1 |= mask;\
        sbits[3 - i] = &ht_no_bitmap;\
    } else {\
        int nlevels =\
            (pdht->components ?\
             pdht->components[i].corder.num_levels :\
             pdht->order.num_levels);\
\
        mask0 |= mask;\
        sbits[3 - i] = (const gx_const_strip_bitmap *)\
            &gx_render_ht(caches[i], nlevels - r)->tiles;\
    }\
  END
         /* Suppress a compiler warning about signed/unsigned constants. */
    SET_PLANE_COLOR_CMYK(0, /*0x88888888*/ (bits32)~0x77777777);
    SET_PLANE_COLOR_CMYK(1, 0x44444444);
    SET_PLANE_COLOR_CMYK(2, 0x22222222);
    SET_PLANE_COLOR_CMYK(3, 0x11111111);

#undef SET_PLANE_COLOR_CMYK
    {
        gx_ht_cache *ctemp;

        ctemp = caches[0], caches[0] = caches[3], caches[3] = ctemp;
        ctemp = caches[1], caches[1] = caches[2], caches[2] = ctemp;
    }
    colors[0] = mask0;
    colors[1] = mask1;
    return 1;
}

/*
 * Set up colors for >4 planes.  In this case, we assume that the color
 * component values are "separable".  (That we can form a gx_color_index value
 * for a color by a bit wise or of the gx_color_index values of the individual
 * components.)
 */
static int
set_ht_colors_gt_4(color_values_pair_t *pvp,
                   gx_color_index colors[MAX_DCC_16 /* 2 * nplanes */],
                   const gx_const_strip_bitmap * sbits[MAX_DCC],
                   const gx_device_color * pdc, gx_device * dev,
                   gx_ht_cache * caches[MAX_DCC], int nplanes)
{
    gx_color_value max_color = dev->color_info.dither_colors - 1;
    bool invert = dev->color_info.polarity == GX_CINFO_POLARITY_SUBTRACTIVE;
    gx_color_index plane_mask = pdc->colors.colored.plane_mask;
    int i;
    gx_color_value cv[MAX_DCC] = {0};

    /* Set the color values and halftone caches. */
    for (i = 0; i < nplanes; ++i) {
        if ((plane_mask >> i) & 1)
            set_plane_color(i, pvp, pdc, sbits, caches, max_color, invert);
        else {
            pvp->values[1][i] = pvp->values[0][i] =
                fractional_color(pdc->colors.colored.c_base[i], max_color);
            sbits[i] = &ht_no_bitmap;
        }
    }
    /*
     * Determine a gs_color_index value for each pair of component values.
     * We assume that an overall index value can be formed from the
     * bitwise or of each component.  We calculate a value for both
     * the high and low value of each component.  These are stored
     * in adjacent locations in 'colors'.
     */
    for (i = 0; i < nplanes; i++ ) {
        cv[i] = pvp->values[0][i];
        colors[2 * i] = dev_proc(dev, encode_color)(dev, cv);
        /* We only need both values for components being halftoned */
        if ((plane_mask >> i) & 1) {
            cv[i] = pvp->values[1][i];
            colors[2 * i + 1] = dev_proc(dev, encode_color)(dev, cv);
        }
        cv[i] = 0;
    }

    return 0;
}

/* ---------------- Color rendering ---------------- */

/* Define the bookkeeping structure for each plane of halftone rendering. */
typedef struct tile_cursor_s {
    int tile_shift;		/* X shift per copy of tile */
    int xoffset;
    int xshift;
    uint xbytes;
    int xbits;
    const byte *row;
    const byte *tdata;
    uint raster;
    const byte *data;
    int bit_shift;
} tile_cursor_t;

/*
 * Initialize one plane cursor, including setting up for the first row
 * (data and bit_shift).
 */
static void
init_tile_cursor(int i, tile_cursor_t *ptc, const gx_const_strip_bitmap *btile,
                 int endx, int lasty)
{
    int tw = btile->size.x;
    int bx = ((ptc->tile_shift = btile->shift) == 0 ? endx :
              endx + lasty / btile->size.y * ptc->tile_shift) % tw;
    int by = lasty % btile->size.y;

    ptc->xoffset = bx >> 3;
    ptc->xshift = 8 - (bx & 7);
    ptc->xbytes = (tw - 1) >> 3;
    ptc->xbits = ((tw - 1) & 7) + 1;
    ptc->tdata = btile->data;
    ptc->raster = btile->raster;
    ptc->row = ptc->tdata + by * (int)ptc->raster;
    ptc->data = ptc->row + ptc->xoffset;
    ptc->bit_shift = ptc->xshift;
    if_debug6('h', "[h]plane %d: size=%d,%d shift=%d bx=%d by=%d\n",
              i, tw, btile->size.y, btile->shift, bx, by);
}

/* Step a cursor to the next row. */
static void
wrap_shifted_cursor(tile_cursor_t *ptc, const gx_const_strip_bitmap *psbit)
{
    ptc->row += ptc->raster * (psbit->size.y - 1);
    if (ptc->tile_shift) {
        if ((ptc->xshift += ptc->tile_shift) >= 8) {
            if ((ptc->xoffset -= ptc->xshift >> 3) < 0) {
                /* wrap around in X */
                int bx = (ptc->xoffset << 3) + 8 - (ptc->xshift & 7) +
                    psbit->size.x;

                ptc->xoffset = bx >> 3;
                ptc->xshift = 8 - (bx & 7);
            } else
                ptc->xshift &= 7;
        }
    }
}
#define STEP_ROW(c, i)\
  BEGIN\
    if (c.row > c.tdata)\
      c.row -= c.raster;\
    else {	/* wrap around to end of tile */\
        wrap_shifted_cursor(&c, sbits[i]);\
    }\
    c.data = c.row + c.xoffset;\
    c.bit_shift = c.xshift;\
  END

/* Define a table for expanding 8x1 bits to 8x4. */
static const bits32 expand_8x1_to_8x4[256] = {
#define X16(c)\
  c+0, c+1, c+0x10, c+0x11, c+0x100, c+0x101, c+0x110, c+0x111,\
  c+0x1000, c+0x1001, c+0x1010, c+0x1011, c+0x1100, c+0x1101, c+0x1110, c+0x1111
    X16(0x00000000), X16(0x00010000), X16(0x00100000), X16(0x00110000),
    X16(0x01000000), X16(0x01010000), X16(0x01100000), X16(0x01110000),
    X16(0x10000000), X16(0x10010000), X16(0x10100000), X16(0x10110000),
    X16(0x11000000), X16(0x11010000), X16(0x11100000), X16(0x11110000)
#undef X16
};

/*
 * Render the combined halftone for nplanes <= 4.
 */
static void
set_color_ht_le_4(byte *dest_data, uint dest_raster, int px, int py,
                  int w, int h, int depth, int special, int nplanes,
                  gx_color_index plane_mask, gx_device *ignore_dev,
                  const color_values_pair_t *ignore_pvp,
                  gx_color_index colors[MAX_DCC_16],
                  const gx_const_strip_bitmap * sbits[MAX_DCC])
{
    /*
     * Note that the planes are specified in the order RGB or CMYK, but
     * the indices used for the internal colors array are BGR or KYMC,
     * except for the special 1-bit CMYK case.
     */
    int x, y;
    tile_cursor_t cursor[MAX_DCC];
    int dbytes = depth >> 3;
    byte *dest_row =
        dest_data + dest_raster * (h - 1) + (w * depth) / 8;

    if (special > 0) {
        /* Planes are in reverse order. */
        plane_mask =
            "\000\010\004\014\002\012\006\016\001\011\005\015\003\013\007\017"[plane_mask];
    }
    if_debug6('h',
              "[h]color_ht_le_4: x=%d y=%d w=%d h=%d plane_mask=0x%lu depth=%d\n",
              px, py, w, h, (ulong)plane_mask, depth);

    /* Do one-time cursor initialization. */
    {
        int endx = w + px;
        int lasty = h - 1 + py;

        if (plane_mask & 1)
            init_tile_cursor(0, &cursor[0], sbits[0], endx, lasty);
        if (plane_mask & 2)
            init_tile_cursor(1, &cursor[1], sbits[1], endx, lasty);
        if (plane_mask & 4)
            init_tile_cursor(2, &cursor[2], sbits[2], endx, lasty);
        if (plane_mask & 8)
            init_tile_cursor(3, &cursor[3], sbits[3], endx, lasty);
    }

    /* Now compute the actual tile. */
    for (y = h; ; dest_row -= dest_raster) {
        byte *dest = dest_row;

        --y;
        for (x = w; x > 0;) {
            bits32 indices;
            int nx, i;
            register uint bits;

/* Get the next byte's worth of bits.  Note that there may be */
/* excess bits set beyond the 8th. */
#define NEXT_BITS(c)\
  BEGIN\
    if (c.data > c.row) {\
      bits = ((c.data[-1] << 8) | *c.data) >> c.bit_shift;\
      c.data--;\
    } else {\
      bits = *c.data >> c.bit_shift;\
      c.data += c.xbytes;\
      if ((c.bit_shift -= c.xbits) < 0) {\
        bits |= *c.data << -c.bit_shift;\
        c.bit_shift += 8;\
      } else {\
        bits |= ((c.data[-1] << 8) | *c.data) >> c.bit_shift;\
        c.data--;\
      }\
    }\
  END
            if (plane_mask & 1) {
                NEXT_BITS(cursor[0]);
                indices = expand_8x1_to_8x4[bits & 0xff];
            } else
                indices = 0;
            if (plane_mask & 2) {
                NEXT_BITS(cursor[1]);
                indices |= expand_8x1_to_8x4[bits & 0xff] << 1;
            }
            if (plane_mask & 4) {
                NEXT_BITS(cursor[2]);
                indices |= expand_8x1_to_8x4[bits & 0xff] << 2;
            }
            if (plane_mask & 8) {
                NEXT_BITS(cursor[3]);
                indices |= expand_8x1_to_8x4[bits & 0xff] << 3;
            }
#undef NEXT_BITS
            nx = min(x, 8);	/* 1 <= nx <= 8 */
            x -= nx;
            switch (dbytes) {
                case 0:	/* 4 */
                    if (special > 0) {
                        /* Special 1-bit CMYK. */
                        /* Compute all the pixels at once! */
                        indices =
                            (indices & colors[1]) | (~indices & colors[0]);
                        i = nx;
                        if ((x + nx) & 1) {
                            /* First pixel is even nibble. */
                            *dest = (*dest & 0xf) +
                                ((indices & 0xf) << 4);
                            indices >>= 4;
                            --i;
                        }
                        /* Now 0 <= i <= 8. */
                        for (; (i -= 2) >= 0; indices >>= 8)
                            *--dest = (byte)indices;
                        /* Check for final odd nibble. */
                        if (i & 1)
                            *--dest = indices & 0xf;
                    } else {
                        /* Other 4-bit pixel */
                        i = nx;
                        if ((x + nx) & 1) {
                            /* First pixel is even nibble. */
                            *dest = (*dest & 0xf) +
                                ((byte)colors[indices & 0xf] << 4);
                            indices >>= 4;
                            --i;
                        }
                        /* Now 0 <= i <= 8. */
                        for (; (i -= 2) >= 0; indices >>= 8)
                            *--dest =
                                (byte)colors[indices & 0xf] +
                                ((byte)colors[(indices >> 4) & 0xf]
                                 << 4);
                        /* Check for final odd nibble. */
                        if (i & 1)
                            *--dest = (byte)colors[indices & 0xf];
                    }
                    break;
                case 4:	/* 32 */
                    for (i = nx; --i >= 0; indices >>= 4) {
                        bits32 tcolor = (bits32)colors[indices & 0xf];

                        dest -= 4;
                        dest[3] = (byte)tcolor;
                        dest[2] = (byte)(tcolor >> 8);
                        tcolor >>= 16;
                        dest[1] = (byte)tcolor;
                        dest[0] = (byte)(tcolor >> 8);
                    }
                    break;
                case 3:	/* 24 */
                    for (i = nx; --i >= 0; indices >>= 4) {
                        bits32 tcolor = (bits32)colors[indices & 0xf];

                        dest -= 3;
                        dest[2] = (byte) tcolor;
                        dest[1] = (byte)(tcolor >> 8);
                        dest[0] = (byte)(tcolor >> 16);
                    }
                    break;
                case 2:	/* 16 */
                    for (i = nx; --i >= 0; indices >>= 4) {
                        uint tcolor =
                            (uint)colors[indices & 0xf];

                        dest -= 2;
                        dest[1] = (byte)tcolor;
                        dest[0] = (byte)(tcolor >> 8);
                    }
                    break;
                case 1:	/* 8 */
                    for (i = nx; --i >= 0; indices >>= 4)
                        *--dest = (byte)colors[indices & 0xf];
                    break;
            }
        }
        if (y == 0)
            break;

        if (plane_mask & 1)
            STEP_ROW(cursor[0], 0);
        if (plane_mask & 2)
            STEP_ROW(cursor[1], 1);
        if (plane_mask & 4)
            STEP_ROW(cursor[2], 2);
        if (plane_mask & 8)
            STEP_ROW(cursor[3], 3);
    }
}

/*
 * Render the combined halftone for nplanes > 4.  This routine assumes
 * that we can form a gx_color_index value by the bitwise or index values
 * for each of the individual components.
 */
static void
set_color_ht_gt_4(byte *dest_data, uint dest_raster, int px, int py,
                  int w, int h, int depth, int special, int num_planes,
                  gx_color_index plane_mask, gx_device *dev,
                  const color_values_pair_t *pvp,
                  gx_color_index colors[MAX_DCC_16],
                  const gx_const_strip_bitmap * sbits[MAX_DCC])
{
    int x, y;
    tile_cursor_t cursor[MAX_DCC];
    int dbytes = depth >> 3;
    byte *dest_row =
        dest_data + dest_raster * (h - 1) + (w * depth) / 8;
    int pmin, pmax;
    gx_color_index base_color = 0;

    /* Compute the range of active planes. */
    if (plane_mask == 0)
        pmin = 0, pmax = -1;
    else {
        for (pmin = 0; !((plane_mask >> pmin) & 1); )
            ++pmin;
        for (pmax = 0; (plane_mask >> pmax) > 1; )
            ++pmax;
    }
    if_debug6('h',
              "[h]color_ht_gt_4: x=%d y=%d w=%d h=%d plane_mask=0x%lu depth=%d\n",
              px, py, w, h, (ulong)plane_mask, depth);

    /* Do one-time cursor initialization. */
    {
        int endx = w + px;
        int lasty = h - 1 + py;
        int i;

        for (i = pmin; i <= pmax; ++i)
            if ((plane_mask >> i) & 1)
                init_tile_cursor(i, &cursor[i], sbits[i], endx, lasty);
    }

    /* Pre-load the color value for the non halftoning planes. */
    {
        int i;

        for (i = 0; i < num_planes; ++i)
            if ((~plane_mask >> i) & 1)
                base_color |= colors[2 * i];
    }

    /* Now compute the actual tile. */
    for (y = h; ; dest_row -= dest_raster) {
        byte *dest = dest_row;
        int i;

        --y;
        for (x = w; x > 0;) {
            gx_color_index tcolor = base_color;

            for (i = pmin; i <= pmax; ++i)
                if ((plane_mask >> i) & 1) {
                    /* Get the next bit from an individual mask. */
                    tile_cursor_t *ptc = &cursor[i];
                    byte tile_bit;

b:		    if (ptc->bit_shift < 8)
                        tile_bit = *ptc->data >> ptc->bit_shift++;
                    else if (ptc->data > ptc->row) {
                        tile_bit = *--(ptc->data);
                        ptc->bit_shift = 1;
                    } else {
                        /* Wrap around. */
                        ptc->data += ptc->xbytes;
                        ptc->bit_shift = 8 - ptc->xbits;
                        goto b;
                    }
                    tcolor |= colors[2 * i + (tile_bit & 1)];
                }
            --x;
            switch (dbytes) {
                case 0:	/* 4 -- might be 2, but we don't support this */
                    if (x & 1) { /* odd nibble */
                        *--dest = (byte)tcolor;
                    } else {	/* even nibble */
                        *dest = (*dest & 0xf) + ((byte)tcolor << 4);
                    }
                    break;
                case 4:	/* 32 */
                    dest[-4] = (byte)(tcolor >> 24);
                case 3:	/* 24 */
                    dest[-3] = (byte)(tcolor >> 16);
                case 2:	/* 16 */
                    dest[-2] = (byte)(tcolor >> 8);
                case 1:	/* 8 */
                    dest[-1] = (byte)tcolor;
                    dest -= dbytes;
                    break;
            }
        }
        if (y == 0)
            break;
        for (i = pmin; i <= pmax; ++i)
            if ((plane_mask >> i) & 1)
                STEP_ROW(cursor[i], i);
    }
}