-
Notifications
You must be signed in to change notification settings - Fork 0
/
ZeissQuickStartCZIReader.java
4808 lines (4189 loc) · 225 KB
/
ZeissQuickStartCZIReader.java
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
/*
* #%L
* OME Bio-Formats package for reading and converting biological file formats.
* %%
* Copyright (C) 2005 - 2017 Open Microscopy Environment:
* - Board of Regents of the University of Wisconsin-Madison
* - Glencoe Software, Inc.
* - University of Dundee
* %%
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as
* published by the Free Software Foundation, either version 2 of the
* License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public
* License along with this program. If not, see
* <http://www.gnu.org/licenses/gpl-2.0.html>.
* #L%
*/
package ch.epfl.biop.formats.in;
/*
* #%L
* OME Bio-Formats package for reading and converting biological file formats.
* %%
* Copyright (C) 2005 - 2017 Open Microscopy Environment:
* - Board of Regents of the University of Wisconsin-Madison
* - Glencoe Software, Inc.
* - University of Dundee
* %%
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as
* published by the Free Software Foundation, either version 2 of the
* License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public
* License along with this program. If not, see
* <http://www.gnu.org/licenses/gpl-2.0.html>.
* #L%
*/
import loci.common.ByteArrayHandle;
import loci.common.Constants;
import loci.common.DataTools;
import loci.common.DateTools;
import loci.common.Location;
import loci.common.RandomAccessInputStream;
import loci.common.Region;
import loci.common.services.DependencyException;
import loci.common.services.ServiceException;
import loci.common.xml.XMLTools;
import loci.formats.CoreMetadata;
import loci.formats.FormatException;
import loci.formats.FormatReader;
import loci.formats.FormatTools;
import loci.formats.MetadataTools;
import loci.formats.codec.CodecOptions;
import loci.formats.codec.JPEGCodec;
import loci.formats.codec.JPEGXRCodec;
import loci.formats.codec.LZWCodec;
import loci.formats.codec.ZstdCodec;
import loci.formats.in.DynamicMetadataOptions;
import loci.formats.in.MetadataOptions;
import ch.epfl.biop.formats.in.libczi.LibCZI;
import loci.formats.meta.DummyMetadata;
import loci.formats.meta.MetadataStore;
import loci.formats.tiff.IFD;
import loci.formats.tiff.TiffParser;
import ome.units.UNITS;
import ome.units.quantity.Length;
import ome.units.quantity.Power;
import ome.units.quantity.Pressure;
import ome.units.quantity.Temperature;
import ome.units.quantity.Time;
import ome.units.unit.Unit;
import ome.xml.model.enums.AcquisitionMode;
import ome.xml.model.enums.Binning;
import ome.xml.model.enums.IlluminationType;
import ome.xml.model.primitives.Color;
import ome.xml.model.primitives.NonNegativeInteger;
import ome.xml.model.primitives.PercentFraction;
import ome.xml.model.primitives.PositiveFloat;
import ome.xml.model.primitives.PositiveInteger;
import ome.xml.model.primitives.Timestamp;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.w3c.dom.Element;
import org.w3c.dom.NamedNodeMap;
import org.w3c.dom.Node;
import org.w3c.dom.NodeList;
import org.xml.sax.SAXException;
import javax.xml.parsers.DocumentBuilder;
import java.io.ByteArrayInputStream;
import java.io.File;
import java.io.IOException;
import java.lang.annotation.Retention;
import java.lang.annotation.RetentionPolicy;
import java.lang.ref.SoftReference;
import java.lang.reflect.Field;
import java.util.ArrayDeque;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Collection;
import java.util.Comparator;
import java.util.Deque;
import java.util.HashMap;
import java.util.HashSet;
import java.util.Iterator;
import java.util.LinkedHashMap;
import java.util.List;
import java.util.Map;
import java.util.Objects;
import java.util.Optional;
import java.util.Set;
import java.util.concurrent.atomic.AtomicLong;
import java.util.concurrent.atomic.AtomicReference;
import java.util.concurrent.locks.Lock;
import java.util.concurrent.locks.ReentrantLock;
import java.util.function.Function;
import java.util.stream.Collectors;
import static ch.epfl.biop.formats.in.libczi.LibCZI.JPEG;
import static ch.epfl.biop.formats.in.libczi.LibCZI.JPEGXR;
import static ch.epfl.biop.formats.in.libczi.LibCZI.LZW;
import static ch.epfl.biop.formats.in.libczi.LibCZI.UNCOMPRESSED;
import static ch.epfl.biop.formats.in.libczi.LibCZI.ZSTD_0;
import static ch.epfl.biop.formats.in.libczi.LibCZI.ZSTD_1;
/**
* ZeissCZIReader is the file format reader for Zeiss .czi files.
* See @see <a href="https://zeiss.github.io/">CZI reference documentation</a>
* <p>
* Essentially, all data is stored into subblocks where each subblock location is specified by its dimension indices.
* There are standard spatial and time dimensions, as well as extra ones necessary to describe channel, scenes,
* acquisition modalities, etc:
* <p>
* X,Y,Z, // 3 spaces dimension
* T, Time
* M, Mosaic but why is there no trace of it in libczi documentation ???
* C, Channel
* R, Rotation
* I, Illumination
* H, Phase
* V, View
* B, Block = deprecated
* S Scene
* <p>
*
* A subblock may represent a lower resolution level. How to know this ? Because its stored size (x or y) is lower
* than its size (x or y). Its downscaling factor can thus be computed the ratio between stored size and size.
* For convenience, this reader adds the downscaling factor as an extra dimension named 'PY'
* <p>
* A CZI file consists of several segments. The majority of segments are data subblocks, as described before. But other
* segments are present. Essentially this reader reads the {@link LibCZI.FileHeaderSegment} that
* contains some metadata as well as the location of the {@link LibCZI.SubBlockDirectorySegment}
*
* The SubBlockDirectorySegment is a critical segment because it contains the dimension indices and file location of all
* data subblocks. Thus, by reading this segment only, there is no need to go through all file segments while
* initializing the reader.
* <p>
* Using this initial reading of the directory segment, all dimensions and all dimension ranges are known in advance.
* This is used to compute the number of core series of the reader, as well as the resolution levels. This is done
* by creating a core series signature {@link CoreSignature} where the dimension are sorted according to a priority
* {@link ZeissQuickStartCZIReader#dimensionPriority(String)}. If autostitching is true, all mosaics belong to the same core
* series. If autostitching is false, each mosaic is split into different core series.
* <p>
* (core series (or core index) = series + resolution level)
* <p>
* Notes:
* 1. It is assumed that all subblocks from a single core index
* have the same compression type {@link ZeissQuickStartCZIReader#coreIndexToCompression}
*
* 2. This reader is not thread safe, you can use memoization or {@link ZeissQuickStartCZIReader#copy()}
* to get a new reader and perform parallel reading.
* <p>
* 3. This reader is optimized for low memory footprint. It has been tested to work on Tb
* czi size files. To save memory, the data structures used for reading are trimmed to the minimal amount of data
* necessary for the reading after the reader has been initialized To illustrate this point, for a 6Tb dataset, each 'int'
* saved per block saves 7Mb (in RAM and in memo file). Trimming down libczi dimension entries
* to {@link MinDimEntry} leads to a memo file of around 100Mb for a 4Tb czi file. Its initialisation
* takes below a minute, with memo building. Then a few seconds to generate a new reader from a memo file is sufficient.
* <p>
* 4. Even with memoization, at runtime, a reader for a multi Tb file will take around 300Mb on the heap. While this
* is reasonable for a single reader, it becomes an issue to create multiple readers for parallel reading: 10 readers
* will take 3 Gb. Thus the method {@link ZeissQuickStartCZIReader#copy()} exist in order to create a new reader from an
* existing one, which saves memory because it reuses all fields from the previous reader. Using this method, 10
* readers can be created to read in parallel en single czi file, but it will use only the memory of one reader.
* WARNING: calling {@link ZeissQuickStartCZIReader#close()} on one of these readers will prevent the use
* of all the other readers created with the copy method!
* <p>
* 5. This reader has an optimisation for lattice-light sheet fast reader initialisation: the metadata for each subblock
* is not read for each plane, but only for the first plane of each time point, and then the timestamp of each plane is
* linearly interpolated. This prevents a lot of random accesses to the file when it is initialized (up to a factor 1000
* if there are 1000 planes). This a small cost associated to this choice: the linear interpolation is not alway exact,
* and timestamps shift up to 10 ms may exist.
* <p>
* The annotation {@link CopyByRef} is used to annotate the fields that should be initialized in the duplicated reader
* using the reference of the model one, see the constructor with the reader in argument.
* <p>
* 5. This reader uses the class {@link LibCZI} which contains the czi data structure translated to Java and which
* contains very no logic related to the reader itself.
* <p>
* TODO:
* - get optimal tile size should vary depending on compression: on raw data it's easy to partially read planes,
* but for compressed data that's much harder so it would be better to read the whole block rather that decompressing
* it multiple times the same block to extract a partial region.
* Missing features:
* - add two methods that map forth and back czi dimension indices to bio-formats series
* - add a method that returns a 3D matrix per series (for lattice skewed dataset?) take care with version
* Issues:
* - some absolute path are stored in the reader, thus the memo fails if the file is moved
* - improve: slide preview and label image are stored directly in the reader as a byte array. That does not look optimal
* but loading these bytes on demand is quite tedious: hard to explain, but a reader is created inside the reader and
* maps the file 'temporarily' to a fake file. That's pretty clever and convenient, but prevents (most probably)
* lazy loading AND memoization functionality.
* <p>
* TODO: ask how to get rid of absolute file path in memo that do not crash the reader when the file is moved
*
*/
public class ZeissQuickStartCZIReader extends FormatReader {
final static Logger logger = LoggerFactory.getLogger(ZeissQuickStartCZIReader.class);
// -- Constants --
public static final String ALLOW_AUTOSTITCHING_KEY = "zeissczi.autostitch";
public static final boolean ALLOW_AUTOSTITCHING_DEFAULT = true;
public static final String STITCHSCENES_KEY = "zeissczi.stitchscenes";
public static final boolean STITCHSCENES_DEFAULT = false;
public static final String INCLUDE_ATTACHMENTS_KEY = "zeissczi.attachments";
public static final boolean INCLUDE_ATTACHMENTS_DEFAULT = true;
public static final String TRIM_DIMENSIONS_KEY = "zeissczi.trim_dimensions";
public static final boolean TRIM_DIMENSIONS_DEFAULT = false;
public static final String RELATIVE_POSITIONS_KEY = "zeissczi.relative_positions";
public static final boolean RELATIVE_POSITIONS_DEFAULT = false;
private static final String CZI_MAGIC_STRING = "ZISRAWFILE";
private static final int BUFFER_SIZE = 512;
// A string identifier for an extra dimension: the resolution level. It's not directly part of the CZI format,
// at least not written as a dimension entry
private static final String RESOLUTION_LEVEL_DIMENSION = "PY";
// A string identifier for an extra dimension: the file part. It's not directly part of the CZI format,
// at least not written as a dimension entry
private static final String FILE_PART_DIMENSION = "PA";
// -- Fields --
// bio-formats core index to x origin, in the Zeiss 2D coordinates system, common to all planes. Unit: pixel (highest resolution level)
@CopyByRef
private List<Integer> coreIndexToOx = new ArrayList<>();
// bio-formats core index to y origin, in the Zeiss 2D coordinates system, common to all planes. Unit: pixel (highest resolution level)
@CopyByRef
private List<Integer> coreIndexToOy = new ArrayList<>();
// bio-formats core index the compression factor of the series.
@CopyByRef
private List<Integer> coreIndexToCompression = new ArrayList<>();
// bio-formats core index the compression factor of the series.
@CopyByRef
private List<CoreSignature> coreIndexToSignature = new ArrayList<>();
// bio-formats core index the downscaling factor of the series.
@CopyByRef
private List<Integer> coreIndexToDownscaleFactor = new ArrayList<>();
// Maps file part to the filename, in case of multipart file
@CopyByRef
private List<String> filePartToFileName = new ArrayList<>(); // TODO: Find a way to not store the absolutepath
@CopyByRef
private Map<Integer, Integer> coreIndexToSeries = new HashMap<>();
// This array has to be taken out of the metadata initializer because of the way IFormatReader#get8BitLookupTable
// and IFormatReader#get16BitLookupTable work
@CopyByRef
ArrayList<MetadataInitializer.Channel> channels = new ArrayList<>();
// streamCurrentSeries is a temp field that should maybe be changed when setSeries is called
transient int streamCurrentPart = -1;
// previous channel has a value set by the last bytes being called, this is a weird behaviour IMO
// but it behaves as expected to make the methods IFormatReader#get8BitLookupTable and
// IFormatReader#get16BitLookupTable work
transient int previousChannel = 0;
// Core map structure for fast access to blocks:
// - first key: bio-formats core index
// - second key: czt index
@CopyByRef
private List< // CoreIndex
HashMap<CZTKey, // CZT
List<MinDimEntry>>>
coreIndexToTZCToMinimalBlocks = new ArrayList<>();
@CopyByRef
int nIlluminations, nRotations, nPhases;
@CopyByRef
boolean hasPyramid = false;
// ------------------------ METADATA FIELDS
@CopyByRef
private MetadataStore store;
@CopyByRef
private ArrayList<byte[]> extraImages = new ArrayList<>();
@CopyByRef
int maxBlockSizeX = -1;
@CopyByRef
int maxBlockSizeY = -1;
//----------------- CACHE
/**
* While the reader is not thread safe, the cache should be, because
* it is shared between multiple readers which can coexist in different threads
* if using the {@link ZeissQuickStartCZIReader#copy()} method to duplicate the reader
*/
@CopyByRef
transient SubBlockLRUCache subBlockLRUCache = new SubBlockLRUCache(10 * 1024 * 1024, 400 * 1024 * 1024 );
@CopyByRef
transient Lock cacheLock = new ReentrantLock();
@CopyByRef
transient Set<MinDimEntry> subBlocksCurrentlyLoading = new HashSet<>();
@CopyByRef
transient boolean useCache = true;
// -- Constructor --
final static String FORMAT = "Zeiss CZI (Quick Start)";
final static String SUFFIX = "czi";
/** Constructs a new Zeiss .czi reader. */
public ZeissQuickStartCZIReader() {
super(FORMAT, SUFFIX);
domains = new String[] {FormatTools.LM_DOMAIN, FormatTools.HISTOLOGY_DOMAIN};
suffixSufficient = false;
suffixNecessary = false;
}
/** Duplicates 'that' reader for parallel reading.
* Creating another reader using this constructor allows to keep a very low memory footprint
* because all immutable objects are re-used by reference.
* WARNING: calling {@link ZeissQuickStartCZIReader#close()} on this or that reader will prevent the use
* of the other reader created with this constructor
* WARNING: 'that' reader should have been initialized with setId before creating another reader
* */
public ZeissQuickStartCZIReader(ZeissQuickStartCZIReader that) {
super(FORMAT, SUFFIX);
domains = new String[] {FormatTools.LM_DOMAIN, FormatTools.HISTOLOGY_DOMAIN};
suffixSufficient = false;
suffixNecessary = false;
if ((that.currentId == null)||(that.currentId.equals(""))) {
throw new RuntimeException("Do not duplicate this reader from a model if the model has not been initialized");
}
this.streamCurrentPart = -1;
// Copy all annotated fields from this class (does not do anything with the inherited ones)
Field[] fields = ZeissQuickStartCZIReader.class.getDeclaredFields();
for (Field field:fields) {
if (field.isAnnotationPresent(CopyByRef.class)) {
try {
field.set(this,field.get(that));
} catch (IllegalAccessException e) {
throw new RuntimeException(e);
}
}
}
// Fields of the super class
this.flattenedResolutions = that.flattenedResolutions;
this.metadataOptions = that.metadataOptions;
this.currentId = that.currentId;
this.core = that.core;
this.metadataStore = that.metadataStore;
this.filterMetadata = that.filterMetadata;
this.datasetDescription = that.datasetDescription;
this.group = that.group;
this.hasCompanionFiles = that.hasCompanionFiles;
this.indexedAsRGB = that.indexedAsRGB;
this.normalizeData = that.normalizeData;
// Set state, just in case
this.setCoreIndex(that.getCoreIndex());
}
/* @see loci.formats.IFormatReader#close(boolean) */
@Override
public void close(boolean fileOnly) throws IOException {
super.close(fileOnly);
if (!fileOnly) {
coreIndexToTZCToMinimalBlocks.clear(); // ZE big one! Hum, problem if another reader uses it... But ok
store = null;
// getStream().close(); done in the super method call
coreIndexToOx.clear();
coreIndexToOy.clear();
coreIndexToCompression.clear();
coreIndexToSignature.clear();
coreIndexToDownscaleFactor.clear();
filePartToFileName.clear();
coreIndexToSeries.clear();
coreIndexToTZCToMinimalBlocks.clear(); // The big one!
extraImages.clear(); // Can be big as well
cacheLock.lock();
subBlockLRUCache.clear();
subBlocksCurrentlyLoading.clear();
cacheLock.unlock();
}
}
/* @see loci.formats.FormatReader#initFile(String) */
@Override
protected ArrayList<String> getAvailableOptions() {
ArrayList<String> optionsList = super.getAvailableOptions();
optionsList.add(ALLOW_AUTOSTITCHING_KEY);
optionsList.add(STITCHSCENES_KEY);
optionsList.add(INCLUDE_ATTACHMENTS_KEY);
optionsList.add(TRIM_DIMENSIONS_KEY);
optionsList.add(RELATIVE_POSITIONS_KEY);
return optionsList;
}
// -- ZeissCZI-specific methods --
public boolean allowAutostitching() {
MetadataOptions options = getMetadataOptions();
if (options instanceof DynamicMetadataOptions) {
return ((DynamicMetadataOptions) options).getBoolean(
ALLOW_AUTOSTITCHING_KEY, ALLOW_AUTOSTITCHING_DEFAULT);
}
return ALLOW_AUTOSTITCHING_DEFAULT;
}
public boolean stitchScenes() {
if (!allowAutostitching()) return false; // we can't stitch scenes if autostitch is disabled
MetadataOptions options = getMetadataOptions();
if (options instanceof DynamicMetadataOptions) {
return ((DynamicMetadataOptions) options).getBoolean(
STITCHSCENES_KEY, STITCHSCENES_DEFAULT);
}
return STITCHSCENES_DEFAULT;
}
public boolean canReadAttachments() {
MetadataOptions options = getMetadataOptions();
if (options instanceof DynamicMetadataOptions) {
return ((DynamicMetadataOptions) options).getBoolean(
INCLUDE_ATTACHMENTS_KEY, INCLUDE_ATTACHMENTS_DEFAULT);
}
return INCLUDE_ATTACHMENTS_DEFAULT;
}
// -- IFormatReader API methods --
/** @see loci.formats.IFormatReader#isThisType(RandomAccessInputStream) */
@Override
public boolean isThisType(RandomAccessInputStream stream) throws IOException {
final int blockLen = 10;
if (!FormatTools.validStream(stream, blockLen, true)) return false;
String check = stream.readString(blockLen);
return check.equals(CZI_MAGIC_STRING);
}
/** @see loci.formats.IFormatReader#get8BitLookupTable() */
@Override
public byte[][] get8BitLookupTable() throws FormatException, IOException {
if ((getPixelType() != FormatTools.INT8 &&
getPixelType() != FormatTools.UINT8) || previousChannel == -1 ||
previousChannel >= channels.size())
{
return null;
}
byte[][] lut = new byte[3][256];
String color = channels.get(previousChannel).color;
if (color != null) {
color = normalizeColor(color);
try {
int colorValue = Integer.parseInt(color, 16);
int redMax = (colorValue & 0xff0000) >> 16;
int greenMax = (colorValue & 0xff00) >> 8;
int blueMax = colorValue & 0xff;
for (int i=0; i<lut[0].length; i++) {
lut[0][i] = (byte) (redMax * (i / 255.0));
lut[1][i] = (byte) (greenMax * (i / 255.0));
lut[2][i] = (byte) (blueMax * (i / 255.0));
}
return lut;
}
catch (NumberFormatException e) {
return null;
}
} else return null;
}
/** @see loci.formats.IFormatReader#get16BitLookupTable() */
@Override
public short[][] get16BitLookupTable() throws FormatException, IOException {
if ((getPixelType() != FormatTools.INT16 &&
getPixelType() != FormatTools.UINT16) || previousChannel == -1 ||
previousChannel >= channels.size())
{
return null;
}
short[][] lut = new short[3][65536];
String color = channels.get(previousChannel).color;
if (color != null) {
color = normalizeColor(color);
try {
int colorValue = Integer.parseInt(color, 16);
int redMax = (colorValue & 0xff0000) >> 16;
int greenMax = (colorValue & 0xff00) >> 8;
int blueMax = colorValue & 0xff;
redMax = (int) (65535 * (redMax / 255.0));
greenMax = (int) (65535 * (greenMax / 255.0));
blueMax = (int) (65535 * (blueMax / 255.0));
for (int i=0; i<lut[0].length; i++) {
lut[0][i] = (short) ((int) (redMax * (i / 65535.0)) & 0xffff);
lut[1][i] = (short) ((int) (greenMax * (i / 65535.0)) & 0xffff);
lut[2][i] = (short) ((int) (blueMax * (i / 65535.0)) & 0xffff);
}
return lut;
}
catch (NumberFormatException e) {
return null;
}
}
else return null;
}
static private String normalizeColor(String color) {
String c = color.replaceAll("#", "");
if (c.length() > 6) {
c = c.substring(2, Math.min(8, c.length()));
LOGGER.debug("Replaced color {} with {}", color, c);
}
return c;
}
private void swapRGBIfnecessary(byte[] buf, int compression, int bpp, int pixel) {
if (isRGB() && compression != JPEGXR) {
// channels are stored in BGR order; red and blue channels need switching
// JPEG-XR data has already been reversed during decompression
int redOffset = bpp * 2;
int index = 0;
int nloops=buf.length/pixel;
for (int i=0; i<nloops; i++) {
for (int b=0; b<bpp; b++) {
int blueIndex = index + b;
int redIndex = index + redOffset + b;
byte red = buf[redIndex];
buf[redIndex] = buf[blueIndex];
buf[blueIndex] = red;
}
index+=pixel;
}
}
}
private byte[] readRawPixelData(
CZTKey key,
MinDimEntry block,
int compression,
int storedSizeX,
int storedSizeY,
RandomAccessInputStream s, Region tile, byte[] buf,
int bpp, int totalBpp) throws FormatException, IOException {
//s.order(isLittleEndian()); -> unnecessary because it is already set when calling the method
if ((useCache)&&(compression!=UNCOMPRESSED)) {
cacheLock.lock(); // acquires lock -> it will be the only thread reading cache properties below
if (subBlockLRUCache.containsKey(block)) {
byte[] bytes = subBlockLRUCache.get(block).get();
if (bytes!=null) {
// - cache hit for block
subBlockLRUCache.touch(block, bytes); // Updates order
cacheLock.unlock();
return bytes;
}
}
// - block not in cache
if (subBlocksCurrentlyLoading.contains(block)) {
// There's already, in a different thread, a reader instance
// computing the same block. We need to wait for it to finish
try {
do {
// - waiting for block to be computed
cacheLock.unlock();
synchronized (subBlocksCurrentlyLoading) {
subBlocksCurrentlyLoading.wait();
}
cacheLock.lock();
// - is the right block being computed ?
} while (subBlocksCurrentlyLoading.contains(block));
// Yes -> the right block has been computed
} catch (InterruptedException e) {
throw new RuntimeException(e);
}
byte[] bytes = subBlockLRUCache.get(block).get();
if (bytes!=null) {
// - the block has been computed
subBlockLRUCache.touch(block, bytes); // put on top, for LRU cache
cacheLock.unlock();
return bytes;
} else {
// weird: the block is not there in the end... maybe it's been removed from the cache
synchronized (subBlocksCurrentlyLoading) {
subBlocksCurrentlyLoading.add(block);
}
}
} else {
// This thread will compute this block
synchronized (subBlocksCurrentlyLoading) {
subBlocksCurrentlyLoading.add(block);
}
}
cacheLock.unlock();
}
LibCZI.SubBlockSegment subBlock = LibCZI.getBlock(s, block.filePosition);
long blockDataOffset = subBlock.dataOffset;
long blockDataSize = subBlock.data.dataSize;
if ((key.t!=block.dimensionStartT)||(key.z!=block.dimensionStartZ)) { // Line scan with multiple T or Z per subblock
if (compression!=UNCOMPRESSED) throw new FormatException("Compression is not supported with line scans.");
blockDataOffset+= (long) ((key.t - block.dimensionStartT) + (key.z - block.dimensionStartZ)) *totalBpp*Math.max(storedSizeX,storedSizeY);
}
s.seek(blockDataOffset);
if (compression == UNCOMPRESSED) {
if (buf == null) {
buf = new byte[(int) blockDataSize];
}
if (tile != null) {
readPlane(s, tile.x, tile.y, tile.width, tile.height,0,storedSizeX,storedSizeY,buf);
}
else {
s.readFully(buf);
}
swapRGBIfnecessary(buf, UNCOMPRESSED, bpp, totalBpp);
return buf;
}
byte[] data = new byte[(int) blockDataSize];
s.read(data);
int bytesPerPixel = FormatTools.getBytesPerPixel(getPixelType());
CodecOptions options = new CodecOptions();
options.interleaved = isInterleaved();
options.littleEndian = isLittleEndian();
options.bitsPerSample = bytesPerPixel * 8;
options.maxBytes = block.storedSizeX * block.storedSizeY * getRGBChannelCount() * bytesPerPixel; // The maximal size is the one of the subblock
switch (compression) {
case JPEG:
data = new JPEGCodec().decompress(data, options);
break;
case LZW:
data = new LZWCodec().decompress(data, options);
break;
case JPEGXR:
options.width = storedSizeX;
options.height = storedSizeY;
options.maxBytes = options.width * options.height * getRGBChannelCount() * bytesPerPixel;
try {
byte[] decompressed = new JPEGXRCodec().decompress(data, options);
data = fixUnexpectedJPEGXRDimensions(data, decompressed,
block.storedSizeX,
block.storedSizeY,
totalBpp, getFillColor());
}
catch (FormatException e) {
if (data.length == options.maxBytes) {
logger.debug("Invalid JPEG-XR compression flag");
}
else {
logger.warn("Could not decompress block; some pixels may be 0", e);
data = new byte[options.maxBytes];
}
}
break;
case ZSTD_0:
data = new ZstdCodec().decompress(data);
break;
case ZSTD_1:
boolean highLowUnpacking = false;
int pointer;
try (RandomAccessInputStream stream = new RandomAccessInputStream(data)) {
int sizeOfHeader = readVarint(stream);
while (stream.getFilePointer() < sizeOfHeader) {
int chunkID = readVarint(stream);
// only one chunk ID defined so far
if (chunkID == 1) {
int payload = stream.read();
highLowUnpacking = (payload & 1) == 1;
} else {
throw new FormatException("Invalid chunk ID: " + chunkID);
}
}
// safe cast because stream wraps a byte array
pointer = (int) stream.getFilePointer();
}
byte[] decoded = new ZstdCodec().decompress(data, pointer, data.length - pointer);
// ZSTD_1 implies high/low byte unpacking, so it would be weird
// if this flag were unset
if (highLowUnpacking) {
data = new byte[decoded.length];
int secondHalf = decoded.length / 2;
for (int i=0; i<decoded.length; i++) {
boolean even = i % 2 == 0;
int offset = i / 2;
data[i] = even ? decoded[offset] : decoded[secondHalf + offset];
}
}
else {
logger.warn("ZSTD-1 compression used, but no high/low byte unpacking");
data = decoded;
}
break;
case 104: // camera-specific packed pixels
data = decode12BitCamera(data, options.maxBytes);
// reverse column ordering
for (int row=0; row<getSizeY(); row++) {
for (int col=0; col<getSizeX()/2; col++) {
int left = row * getSizeX() * 2 + col * 2;
int right = row * getSizeX() * 2 + (getSizeX() - col - 1) * 2;
byte left1 = data[left];
byte left2 = data[left + 1];
data[left] = data[right];
data[left + 1] = data[right + 1];
data[right] = left1;
data[right + 1] = left2;
}
}
break;
case 504: // camera-specific packed pixels
data = decode12BitCamera(data, options.maxBytes);
break;
case 121: // See https://zenodo.org/records/10708864 40_Dual.czi
int nBytesExpected = storedSizeX*storedSizeY*bpp;
if (data.length > nBytesExpected) {
byte[] r = new byte[nBytesExpected];
System.arraycopy(data, 0, r, 0, nBytesExpected);
data = r;
}
break;
}
if (buf != null && buf.length >= data.length) {
System.arraycopy(data, 0, buf, 0, data.length);
swapRGBIfnecessary(buf, compression, bpp, totalBpp);
if (useCache) {
cacheLock.lock();
// Block just computed
subBlockLRUCache.touch(block, buf);
// Put in cache
subBlocksCurrentlyLoading.remove(block);
cacheLock.unlock();
synchronized (subBlocksCurrentlyLoading) {
subBlocksCurrentlyLoading.notifyAll(); // Wake the threads waiting for this block
}
}
return buf;
}
swapRGBIfnecessary(data, compression, bpp, totalBpp);
if (useCache) {
cacheLock.lock();
subBlockLRUCache.touch(block, data);
subBlocksCurrentlyLoading.remove(block);
cacheLock.unlock();
synchronized (subBlocksCurrentlyLoading) {
subBlocksCurrentlyLoading.notifyAll();
}
}
return data;
}
private static byte[] fixUnexpectedJPEGXRDimensions(byte[] compressed, byte[] uncompressed,
int storedSizeX, int storedSizeY,
int totalBpp, Byte fillColor) {
// Sometimes (see post https://forum.image.sc/t/would-anyone-have-a-palm-czi-example-file/85900/12),
// decompressed subblock does not return the number of pixels expected from subblock field storedSizeX or storedSizeY
// to find an example that will use this correction, execute the code below
// on the Young-Mouse czi image from - resolution level 6 https://zenodo.org/records/10577621:
/* ZeissQuickStartCZIReader r = new ZeissQuickStartCZIReader();
r.setId("image path to \\Young_mouse.czi");
r.setSeries(5);
r.openPlane(0,0,0,5947,2168); */
int expectedRawDataSize = storedSizeX*storedSizeY*totalBpp;
if (uncompressed.length!=expectedRawDataSize) {
// We got an issue
try {
int[] result = getWidthAndHeightFromJPEGXRBytes(compressed); // It is possible to get the dimension of the decompressed subblock
// thanks to a header present in the compressed bytes (check JPEGXR specs in the getWidthAndHeight method)
int w = result[0];
int h = result[1];
if ((w>storedSizeX)||(h>storedSizeY)) { // No handling of a decompressed subblock bigger than expected. Only smaller.
throw new RuntimeException("Too many pixels found in a CZI JPEGXR compressed subblock.");
}
assert uncompressed.length == (w * h * totalBpp);
byte[] corrected = new byte[expectedRawDataSize];
Arrays.fill(corrected, fillColor);
// The strategy is to copy one line after another, leaving empty columns or line depending on
// the case. The color of the empty pixels is set by the fillColor argument
for (int y = 0; y < h; y++) {
System.arraycopy(uncompressed, w*totalBpp*y, corrected, storedSizeX*totalBpp*y, w*totalBpp);
}
return corrected;
} catch (FormatException | IOException e) {
throw new RuntimeException(e);
}
} else {
return uncompressed;
}
}
// see table A.4 in ITU-T T.832 https://www.itu.int/rec/T-REC-T.832/en version 2009
private static final int IMAGE_WIDTH_TAG = 0xBC80;
private static final int IMAGE_HEIGHT_TAG = 0xBC81;
private static int[] getWidthAndHeightFromJPEGXRBytes(byte[] stream) throws FormatException, IOException {
try (RandomAccessInputStream s = new RandomAccessInputStream(stream)) {
s.order(true);
s.seek(4);
long ifdPointer = s.readInt();
TiffParser p = new TiffParser(s);
IFD ifd = p.getIFD(ifdPointer);
return new int[]{ifd.getIFDIntValue(IMAGE_WIDTH_TAG), ifd.getIFDIntValue(IMAGE_HEIGHT_TAG)};
}
}
private static int readVarint(RandomAccessInputStream stream) throws IOException {
byte a = stream.readByte();
// if high bit set, read next byte
// at most 3 bytes read
if ((a & 0x80) == 0x80) {
byte b = stream.readByte();
if ((b & 0x80) == 0x80) {
byte c = stream.readByte();
return (c << 14) | ((b & 0x7f) << 7) | (a & 0x7f);
}
return (b << 7) | (a & 0x7f);
}
return a & 0xff;
}
private static byte[] decode12BitCamera(byte[] data, int maxBytes) throws IOException {
byte[] decoded = new byte[maxBytes];
RandomAccessInputStream bb = new RandomAccessInputStream(
new ByteArrayHandle(data));
byte[] fourBits = new byte[(maxBytes / 2) * 3];
int pt = 0;
while (pt < fourBits.length) {
fourBits[pt++] = (byte) bb.readBits(4);
}
bb.close();
for (int index=0; index<fourBits.length-1; index++) {
if ((index - 3) % 6 == 0) {
byte middle = fourBits[index];
byte last = fourBits[index + 1];
byte first = fourBits[index - 1];
fourBits[index + 1] = middle;
fourBits[index] = first;
fourBits[index - 1] = last;
}
}
int currentByte = 0;
for (int index=0; index<fourBits.length;) {
if (index % 3 == 0) {
decoded[currentByte++] = fourBits[index++];
}
else {
decoded[currentByte++] =
(byte) (fourBits[index++] << 4 | fourBits[index++]);
}
}
return decoded;
}
@Override
public void reopenFile() {
streamCurrentPart = -1;
}
private synchronized RandomAccessInputStream getStream(int filePart) throws IOException { // TODO : remove synchronized which is useless, a single reader is not multithreaded
if ((in != null)&&(streamCurrentPart == filePart)) {
return in;
}
if (in!=null) in.close();
streamCurrentPart = filePart;
RandomAccessInputStream ris = new RandomAccessInputStream(filePartToFileName.get(filePart), BUFFER_SIZE);
in = ris;
ris.order(isLittleEndian());
return ris;
}
@Override
public int getOptimalTileWidth() {
if (maxBlockSizeX>0) {
return Math.min(2048, maxBlockSizeX);
} else {
return Math.min(2048, getSizeX());
}
}
@Override
public int getOptimalTileHeight() {
if (maxBlockSizeY>0) {
return Math.min(2048, maxBlockSizeY);
} else {
return Math.min(2048, getSizeY());
}
}
/**
* @see loci.formats.FormatReader#getFillColor()
*
* If the fill value was set explicitly, use that.
* Otherwise, return 255 (white) for RGB data with a pyramid,
* and 0 in all other cases. RGB data with a pyramid can
* reasonably be assumed to be a brightfield slide.
*/
@Override
public Byte getFillColor() {
if (fillColor != null) {
return fillColor;
}
byte fill = (byte) 0;
if (isRGB() && (hasPyramid)) {
fill = (byte) 255;
}
return fill;
}
@Override
public byte[] openBytes(int no, byte[] buf, int x, int y, int w, int h) throws FormatException, IOException {
FormatTools.checkPlaneParameters(this, no, buf.length, x, y, w, h);
if (isThumbnailSeries()) {
// thumbnail, label, or preview image stored as an attachment
int index = getCoreIndex() - (core.size() - extraImages.size());
byte[] fullPlane = extraImages.get(index);
try (RandomAccessInputStream s = new RandomAccessInputStream(fullPlane)) {
readPlane(s, x, y, w, h, buf);
}
return buf;
}
int currentIndex = getCoreIndex();
int bpp = FormatTools.getBytesPerPixel(getPixelType());
int nCh = getRGBChannelCount();
int bytesPerPixel = (isRGB()?nCh:1) * bpp;
int baseResolution = currentIndex;
Region image = new Region(x, y, w, h);
// Because series are sorted along their resolution level, that's a way to find which
// resolution level is the lowest one - but that looks very brittle - what if you have a very
// while the downscaling is decreasing, let's decrement baseresolution
// what this assumes is that the resolution level whereby downscaling = 1 is always present
int[] czt = this.getZCTCoords(no);
previousChannel = czt[1];
CZTKey key = new CZTKey(czt[1], czt[0], czt[2]);