summaryrefslogtreecommitdiff
path: root/macros/latex/contrib/l3kernel/l3unicode.dtx
blob: 27bf195fc6654f7479207a581d347df92fd11704 (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
% \iffalse meta-comment
%
%% File: l3unicode.dtx
%
% Copyright (C) 2018-2023 The LaTeX Project
%
% It may be distributed and/or modified under the conditions of the
% LaTeX Project Public License (LPPL), either version 1.3c of this
% license or (at your option) any later version.  The latest version
% of this license is in the file
%
%    https://www.latex-project.org/lppl.txt
%
% This file is part of the "l3kernel bundle" (The Work in LPPL)
% and all files in that bundle must be distributed together.
%
% -----------------------------------------------------------------------
%
% The development version of the bundle can be found at
%
%    https://github.com/latex3/latex3
%
% for those people who are interested.
%
%<*driver>
\documentclass[full,kernel]{l3doc}
\begin{document}
  \DocInput{\jobname.dtx}
\end{document}
%</driver>
% \fi
%
% \title{^^A
%   The \pkg{l3unicode} package: Unicode support functions^^A
% }
%
% \author{^^A
%  The \LaTeX{} Project\thanks
%    {^^A
%      E-mail:
%        \href{mailto:latex-team@latex-project.org}
%          {latex-team@latex-project.org}^^A
%    }^^A
% }
%
% \date{Released 2023-01-24}
%
% \maketitle
%
% \begin{documentation}
%
% This module provides Unicode-specific functions along with loading data
% from a range of Unicode Consortium files. Most of the code here is
% internal, but there are a small set of public functions. These work with
% Unicode \meta{codepoints} and are designed to give useable results with
% both Unicode-aware and $8$-bit engines.
%
% \begin{function}[EXP, added = 2022-10-09, updated = 2022-11-09]
%   {\codepoint_generate:nn}
%   \begin{syntax}
%      \cs{codepoint_generate:nn} \Arg{codepoint} \Arg{catcode}
%   \end{syntax}
%   Generates one or more character tokens representing the \meta{codepoint}.
%   With Unicode engines, exactly one character token will be generated, and
%   this will have the \meta{catcode} specified as the second argument:
%   \begin{itemize}
%     \item $1$ (begin group)
%     \item $2$ (end group)
%     \item $3$ (math toggle)
%     \item $4$ (alignment)
%     \item $6$ (parameter)
%     \item $7$ (math superscript)
%     \item $8$ (math subscript)
%     \item $10$ (space)
%     \item $11$ (letter)
%     \item $12$ (other)
%     \item $13$ (active)
%   \end{itemize}
%   For $8$-bit engines, between one and four character tokens will be
%   produced: these will be the bytes of the UTF-8 representation of the
%   \meta{codepoint}. For all codepoints outside of the classical ASCII
%   range, the generated character tokens will be active (category code
%   $13$); for codepoints in the ASCII range, the given \meta{catcode}
%   will be used. To allow the result of this function to be used
%   inside a expansion context, the result is protected by \cs{exp_not:n}.
% \end{function}
%
% \begin{function}[EXP, added = 2022-10-09]
%   {\codepoint_str_generate:n}
%   \begin{syntax}
%      \cs{codepoint_str_generate:n} \Arg{codepoint}
%   \end{syntax}
%   Generates one or more character tokens representing the \meta{codepoint}.
%   With Unicode engines, exactly one character token will be generated.
%   For $8$-bit engines, between one and four character tokens will be
%   produced: these will be the bytes of the UTF-8 representation of the
%   \meta{codepoint}. All of the generated character tokens will be of
%   category code $12$, except any spaces (codepoint $32$), which will be
%   category code $10$.
% \end{function}
%
% \begin{function}[added = 2022-10-09, EXP]{\codepoint_to_nfd:n}
%   \begin{syntax}
%     \cs{codepoint_to_nfd:n} \Arg{codepoint}
%   \end{syntax}
%   Converts the \meta{codepoint} to the Unicode Normalization
%   Form Canonical Decomposition. The generated character(s) will have
%   the current category code as they would if typed in directly for Unicode
%   engines; for $8$-bit engines, active characters are used for all codepoints
%   outside of the ASCII range.
% \end{function}
%
% \end{documentation}
%
% \begin{implementation}
%
% \section{\pkg{l3unicode} implementation}
%
%    \begin{macrocode}
%<*package>
%    \end{macrocode}
%
%    \begin{macrocode}
%<@@=codepoint>
%    \end{macrocode}
%
% \subsection{User functions}
%
% \begin{macro}[EXP]{\codepoint_str_generate:n}
% \begin{macro}[EXP]{\@@_str_generate:nnnn}
% \begin{macro}[EXP]{\codepoint_generate:nn}
% \begin{macro}[EXP]{\@@_generate:nnnn}
% \begin{macro}[EXP]{\@@_generate:n}
%   Conversion of a codepoint to a character (Unicode engines) or to one
%   or more bytes ($8$-bit engines) is required. For loading the data,
%   all that is needed is the form which creates strings: these are outside
%   the group as they will also be used when looking up data in the hash
%   table storage at point-of-use. Later, we will also need functions that
%   can generate character tokens for document use: those are defined below,
%   in the data recovery setup.
%    \begin{macrocode}
\bool_lazy_or:nnTF
  { \sys_if_engine_luatex_p: }
  { \sys_if_engine_xetex_p: }
  {
    \cs_new:Npn \codepoint_str_generate:n #1
      {
        \int_compare:nNnTF {#1} = { `\  }
          { ~ }
          { \char_generate:nn {#1} { 12 } }
      }
   \cs_new:Npn \codepoint_generate:nn #1#2
      {
        \int_compare:nNnTF {#1} = { `\  }
          { ~ }
          {
            \__kernel_exp_not:w \exp_after:wN \exp_after:wN \exp_after:wN
              { \char_generate:nn {#1} {#2} }
          }
      }
  }
  {
    \cs_new:Npn \codepoint_str_generate:n #1
      {
        \int_compare:nNnTF {#1} = { `\  }
          { ~ }
          {
            \use:e
              {
                \exp_not:N \@@_str_generate:nnnn
                  \__kernel_codepoint_to_bytes:n {#1}
              }
          }
      }
    \cs_new:Npn \@@_str_generate:nnnn #1#2#3#4
      {
        \char_generate:nn {#1} { 12 }
        \tl_if_blank:nF {#2}
          {
            \char_generate:nn {#2} { 12 }
            \tl_if_blank:nF {#3}
              {
                \char_generate:nn {#3} { 12 }
                \tl_if_blank:nF {#4}
                  { \char_generate:nn {#4} { 12 } }
              }
          }
      }
    \cs_new:Npn \codepoint_generate:nn #1#2
      {
        \int_compare:nNnTF {#1} = { `\  }
          { ~ }
          {
            \int_compare:nNnTF {#1} < { "80 }
              {
                \__kernel_exp_not:w \exp_after:wN \exp_after:wN \exp_after:wN
                  { \char_generate:nn {#1} {#2} }
              }
              {
                \use:e
                  {
                    \exp_not:N \@@_generate:nnnn
                      \__kernel_codepoint_to_bytes:n {#1}
                  }
              }
          }
      }
    \cs_new:Npn \@@_generate:nnnn #1#2#3#4
      {
        \__kernel_exp_not:w \exp_after:wN
          {
            \tex_expanded:D
              {
                \@@_generate:n {#1}
                \@@_generate:n {#2}
                \tl_if_blank:nF {#3}
                  {
                    \@@_generate:n {#3}
                    \tl_if_blank:nF {#4}
                      { \@@_generate:n {#4} }
                  }
              }
          }
      }
     \cs_new:Npn \@@_generate:n #1
       {
         \__kernel_exp_not:w \exp_after:wN \exp_after:wN \exp_after:wN
           { \char_generate:nn {#1} { 13 } }
       }
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}[EXP]{\__kernel_codepoint_to_bytes:n}
% \begin{macro}[EXP]{\@@_to_bytes_auxi:n}
% \begin{macro}[EXP]{\@@_to_bytes_auxii:Nnn}
% \begin{macro}[EXP]{\@@_to_bytes_auxiii:n}
% \begin{macro}[EXP]
%   {
%     \@@_to_bytes_outputi:nw   ,
%     \@@_to_bytes_outputii:nw  ,
%     \@@_to_bytes_outputiii:nw ,
%     \@@_to_bytes_outputiv:nw
%   }
% \begin{macro}[EXP]
%   {\@@_to_bytes_output:nnn, \@@_to_bytes_output:fnn}
% \begin{macro}[EXP]{\@@_to_bytes_end:}
%   This code converts a codepoint into the correct UTF-8 representation.
%   In terms of the algorithm itself, see
%   \url{https://en.wikipedia.org/wiki/UTF-8} for the octet pattern.
%    \begin{macrocode}
\cs_new:Npn \__kernel_codepoint_to_bytes:n #1
  {
    \exp_args:Nf \@@_to_bytes_auxi:n
      { \int_eval:n {#1} }
  }
\cs_new:Npn \@@_to_bytes_auxi:n #1
  {
    \if_int_compare:w #1 > "80 \exp_stop_f:
      \if_int_compare:w #1 < "800 \exp_stop_f:
        \@@_to_bytes_outputi:nw
          { \@@_to_bytes_auxii:Nnn C {#1} { 64 } }
        \@@_to_bytes_outputii:nw
          { \@@_to_bytes_auxiii:n {#1} }
      \else:
        \if_int_compare:w #1 < "10000 \exp_stop_f:
          \@@_to_bytes_outputi:nw
            { \@@_to_bytes_auxii:Nnn E {#1} { 64 * 64 } }
          \@@_to_bytes_outputii:nw
            {
              \@@_to_bytes_auxiii:n
                { \int_div_truncate:nn {#1} { 64 } }
            }
          \@@_to_bytes_outputiii:nw
            { \@@_to_bytes_auxiii:n {#1} }
        \else:
          \@@_to_bytes_outputi:nw
            {
              \@@_to_bytes_auxii:Nnn F
                 {#1} { 64 * 64 * 64 }
            }
          \@@_to_bytes_outputii:nw
            {
              \@@_to_bytes_auxiii:n
                { \int_div_truncate:nn {#1} { 64 * 64 } }
            }
          \@@_to_bytes_outputiii:nw
            {
              \@@_to_bytes_auxiii:n
                { \int_div_truncate:nn {#1} { 64 } }
            }
          \@@_to_bytes_outputiv:nw
            { \@@_to_bytes_auxiii:n {#1} }
        \fi:
      \fi:
    \else:
      \@@_to_bytes_outputi:nw {#1}
    \fi:
    \@@_to_bytes_end: { } { } { } { }
  }
\cs_new:Npn \@@_to_bytes_auxii:Nnn #1#2#3
  {  "#10 + \int_div_truncate:nn {#2} {#3} }
\cs_new:Npn \@@_to_bytes_auxiii:n #1
  { \int_mod:nn {#1} { 64 } + 128 }
\cs_new:Npn \@@_to_bytes_outputi:nw
  #1 #2 \@@_to_bytes_end: #3
  { \@@_to_bytes_output:fnn { \int_eval:n {#1} } { } {#2} }
\cs_new:Npn \@@_to_bytes_outputii:nw
  #1 #2 \@@_to_bytes_end: #3#4
  { \@@_to_bytes_output:fnn { \int_eval:n {#1} } { {#3} } {#2} }
\cs_new:Npn \@@_to_bytes_outputiii:nw
  #1 #2 \@@_to_bytes_end: #3#4#5
  {
    \@@_to_bytes_output:fnn
      { \int_eval:n {#1} } { {#3} {#4} } {#2}
  }
\cs_new:Npn \@@_to_bytes_outputiv:nw
  #1 #2 \@@_to_bytes_end: #3#4#5#6
  {
    \@@_to_bytes_output:fnn
      { \int_eval:n {#1} } { {#3} {#4} {#5} } {#2}
  }
\cs_new:Npn \@@_to_bytes_output:nnn #1#2#3
  {
    #3
    \@@_to_bytes_end: #2 {#1}
  }
\cs_generate_variant:Nn \@@_to_bytes_output:nnn { f }
\cs_new:Npn \@@_to_bytes_end: { }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}[EXP]{\codepoint_to_nfd:n}
% \begin{macro}[EXP]{\@@_to_nfd:nn}
% \begin{macro}[EXP]{\@@_to_nfd:nnn}
% \begin{macro}[EXP]{\@@_to_nfd:nnnn}
% \begin{macro}[EXP]{\@@_to_nfd_generate:nn}
% \begin{macro}[EXP]{\@@_to_nfd_generate:n}
% \begin{macro}[EXP]{\@@_to_nfd_generate:nnnn}
%   Converted to NFD is a potentially-recursive process: the key is to
%   check if we get the input codepoint back again. As far as possible,
%   we use the same path for all engines.
%    \begin{macrocode}
\cs_new:Npn \codepoint_to_nfd:n #1
  { \@@_to_nfd:nn {#1} { \char_value_catcode:n {#1} } }
\cs_new:Npn \@@_to_nfd:nn #1#2
  {
    \exp_args:Ne \@@_to_nfd:nnn
      { \@@_nfd:n {#1} } {#1} {#2}
  }
\cs_new:Npn \@@_to_nfd:nnn #1#2#3 { \@@_to_nfd:nnnn #1 {#2} {#3} }
\cs_new:Npn \@@_to_nfd:nnnn #1#2#3#4
  {
    \int_compare:nNnTF {#1} = {#3}
      { \codepoint_generate:nn {#1} {#4} }
      {
        \@@_to_nfd:nn {#1} {#4}
        \tl_if_blank:nF {#2}
          { \@@_to_nfd:nn {#2} {#4} }
      }
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \subsection{Data loader}
%
% Text operations requires data from the Unicode Consortium. Data read into
% Unicode engine formats is at best a small part of what we need, so there
% is a loader here to set  up the appropriate data structures.
%
% Where we need data for most or all of the Unicode range, we use the two-stage
% table approach recommended by the Unicode Consortium and demonstrated in a
% model implementation in Python in
% \url{https://www.strchr.com/multi-stage_tables}. This approach uses the
% \texttt{intarray} (\texttt{fontdimen}-based) data type as it is fast for
% random access and avoids significant hash table usage. In contrast, where
% only a small subset of codepoints are required, storage as macros is
% preferable. There is also some consideration of the effort needed to load
% data: see for example the grapheme breaking information, which would be 
% problematic to convert into a two-stage table but which can be used with
% reasonable performance in a small number of comma lists (at the cost that
% breaking at higher codepoint Hangul characters will be slightly slow).
%
% \begin{variable}{\c_@@_block_size_int}
%   Choosing the block size for the blocks in the two-stage approach is
%   non-trivial: depending on the data stored, the optimal size for
%   memory usage will vary. At the same time, for us there is also the
%   question of load-time: larger blocks require longer comma lists
%   as intermediates, so are slower. As this is going to be needed
%   to use the data, we set it up outside of the group for clarity.
%    \begin{macrocode}
\int_const:Nn \c_@@_block_size_int { 64 }
%    \end{macrocode}
% \end{variable}
%
% Parsing the data files can be the same way for all engines, but where they
% are stored as character tokens, the construction method depends on whether
% they are Unicode or $8$-bit internally. Parsing is therefore done by common
% functions, with some data storage using engine-specific auxiliaries.
%
% As only the data needs to remain at the end of this process, everything
% is set up inside a group. The only thing that is outside is creating a
% stream: they are global anyway and it is best to force a stream for
% all engines.
%
% \begin{variable}{\g_@@_data_ior}
%    \begin{macrocode}
\ior_new:N \g_@@_data_ior
%    \end{macrocode}
% \end{variable}
%
% We need some setup for the two-part table approach. The number of blocks we
% need will be variable, but the resulting size of the stage one table
% is predictable. For performance reasons, we therefore create the stage one 
% tables now so they can be used immediately, and will later rename them as a
% constant tables. For each two-stage table construction, we need a comma
% list to hold the partial block and a couple of integers to track where
% we are up to. To avoid burning registers, the latter are stored in macros
% and are \enquote{fake} integers. We also avoid any \texttt{new} functions,
% keeping as much as possible local.
%
% As we need both positive and negative values, case data requires one
% two-stage table for each transformation. In contrasts, general Unicode
% properties could be stored in one table with appropriate combination rules:
% that is not done at present but is likely to be added over time. Here, all
% that is needed is additional entries into the comma-list to create the
% structures.
%
% Notice that in the standard \pkg{expl3} way we are indexes position not
% offset: that does mean a little work later.
%    \begin{macrocode}
\group_begin:
  \clist_map_inline:nn
    { uppercase , lowercase }
    {
      \cs_set_nopar:cpn { l_@@_ #1 _block_clist } { }
      \cs_set_nopar:cpn { l_@@_ #1 _block_tl } { 1 }
      \cs_set_nopar:cpn { l_@@_ #1 _pos_tl } { 0 }
      \intarray_new:cn { g_@@_ #1 _index_intarray }
        { \int_div_truncate:nn { "110000 } \c_@@_block_size_int }
    }
%    \end{macrocode}
%  We need an integer value when matching the current block to those we have
%  already seen, and a way to track codepoints for handling ranges. Again,
%  we avoid using up registers or creating global names.
%    \begin{macrocode}
  \cs_set_nopar:Npn \l_@@_next_codepoint_fint_tl { 0 }
  \cs_set_nopar:Npn \l_@@_matched_block_tl { 0 }
%    \end{macrocode}
% Parse the main Unicode data file and pull out the NFD and case changing
% data. The NFD data is stored on using the hash table approach and can yield
% a predictable number of code points: one or two. We also need the case data,
% which will be modified further below. To allow for finding rangs, the
% description of the codepoint needs to be carried forward.
%    \begin{macrocode}
  \cs_set_protected:Npn \@@_data_auxi:w
    #1 ; #2 ; #3 ; #4 ; #5 ; #6 ; #7 ; #8 ; #9 ;
    {
      \tl_if_blank:nF {#6}
        {
          \tl_if_head_eq_charcode:nNF {#6}  < % >
            { \@@_data_auxii:w #1 ; #6 ~ \q_stop }
        }
      \@@_data_auxiii:w #1 ; #2 ;
    }
  \cs_set_protected:Npn \@@_data_auxii:w #1 ; #2 ~ #3 \q_stop
    {
      \tl_const:cx
        { c_@@_nfd_ \codepoint_str_generate:n {"#1} _tl }
        {
          {"#2}
          { \tl_if_blank:nF {#3} {"#3} }
        }
    }
%    \end{macrocode}
% The case data is going to be stored as an offset from the parent character,
% rather than an absolute value. We therefore deal with that plus the situation
% where a codepoint has no mapping data in oen shot.
%    \begin{macrocode}
  \cs_set_protected:Npn \@@_data_auxiii:w
    #1 ; #2 ; #3 ; #4 ; #5 ; #6 ; #7 ; #8 ~ \q_stop
    {
      \use:e
        {
          \@@_data_auxiv:w
            #1 ; #2 ;
            \@@_data_offset:nn {#1} {#6} ;
            \@@_data_offset:nn {#1} {#7} ;
            #8;
        }
    }
  \cs_set:Npn \@@_data_offset:nn #1#2
    {
      \tl_if_blank:nTF {#2}
        { 0 }
        { \int_eval:n { "#2 - "#1 } }
    }
%    \end{macrocode}
% To deal with ranges, we track the position of the next codepoint expected.
% If there is a gap, we deal with that separately: it could be a range or
% an unused part of the Unicode space. As such, we deal with the current
% codepoint here whether or not there is range to fill in. Upper- and 
% lowercase data go into the two-stage table, any titlecase exception is
% just stored in a macro. The data for the codepoint is added to the current
% block, and if that is now complete we move on to save the block. The
% case exceptions are all stored as codepoints, with a fixed number of
% balanced text as we know that there are never more than three.
%    \begin{macrocode}
  \cs_set_protected:Npn \@@_data_auxiv:w #1 ; #2 ; #3 ; #4 ; #5 ;
    {
      \int_compare:nNnT {"#1} > \l_@@_next_codepoint_fint_tl
        {
          \@@_data_auxvi:nnnw {#1} {#3} {#4}
            #2 Last> \q_stop
        }
      \@@_add:nn { uppercase } {#3}
      \@@_add:nn { lowercase } {#4}
      \int_compare:nNnF {#3} = { \@@_data_offset:nn {#1} {#5} }
        {
          \tl_const:cx
            { c_@@_titlecase_ \codepoint_str_generate:n {"#1} _tl }
            { {"#5} { } { } }
        }
      \tl_set:Nx \l_@@_next_codepoint_fint_tl
        { \int_eval:n { "#1 + 1 } }
    }
  \cs_set_protected:Npn \@@_add:nn #1#2
    {
      \clist_put_right:cn { l_@@_ #1 _block_clist } {#2}
      \int_compare:nNnT { \clist_count:c { l_@@_ #1 _block_clist } }
        = \c_@@_block_size_int
        { \@@_save_blocks:nn {#1} { 1 } }
    }
%    \end{macrocode}
%  Distinguish between a range and a gap, and pass on the appropriate value(s).
%    \begin{macrocode}
  \cs_set_protected:Npn \@@_data_auxvi:nnnw #1#2#3#4 Last> #5 \q_stop
    {
       \tl_if_blank:nTF {#5}
         {
           \@@_range:nnn {#1} { uppercase } { 0 }
           \@@_range:nnn {#1} { lowercase } { 0 }
         }
         {
           \@@_range:nnn {#1} { uppercase } {#2}
           \@@_range:nnn {#1} { lowercase } {#3}
         }      
    }
%    \end{macrocode}
%  Calculated the length of the range and the space remaining in the current
%  block.
%    \begin{macrocode}
  \cs_set_protected:Npn \@@_range:nnn #1
    {
      \exp_args:Nf \@@_range_aux:nnn
        { \int_eval:n { "#1 - \l_@@_next_codepoint_fint_tl } }
    }
  \cs_set_protected:Npn \@@_range_aux:nnn #1#2
    {
      \exp_args:Nf \@@_range:nnnn
        {
          \int_min:nn
            {#1}
            {
              \c_@@_block_size_int 
              - \clist_count:c { l_@@_ #2 _block_clist }
            }
        }
        {#1} {#2}
    }
%    \end{macrocode}
%   Here we want to do three things: add to and possibly complete the current
%   block, add complete blocks quickly, then finish up the range in a final
%   open block. We need to avoid as far as possible avoid dealing with ever
%   single codepoint, so the middle step is optimised.
%    \begin{macrocode}
  \cs_set_protected:Npn \@@_range:nnnn #1#2#3#4
    {
      \prg_replicate:nn {#1}
        { \clist_put_right:cn { l_@@_ #3 _block_clist } {#4} }
    \int_compare:nNnT { \clist_count:c { l_@@_ #3 _block_clist } }
      = \c_@@_block_size_int
      { \@@_save_blocks:nn {#3} { 1 } }
     \int_compare:nNnF
       { \int_div_truncate:nn { #2 - #1 } \c_@@_block_size_int } = 0
       {
         \tl_set:cx { l_@@_ #3 _block_clist }
           {
             \exp_args:NNe \use:nn \use_none:n
               { \prg_replicate:nn { \c_@@_block_size_int } { , #4 } }
           }
         \@@_save_blocks:nn {#3}
           { \int_div_truncate:nn { (#2 - #1) } \c_@@_block_size_int }
        }
     \prg_replicate:nn
       { \int_mod:nn { #2 - #1 } \c_@@_block_size_int }
       { \clist_put_right:cx { l_@@_ #3 _block_clist } {#4} }
    }
%    \end{macrocode}
%   To allow rapid comparison, each completed block is stored locally as a
%   comma list: once all of the blocks have been created, they are converted
%   into an \texttt{intarray} in one step. The aim here is to check the current
%   block against those we've already used, and either match to an existing
%   block or save a new block.
%    \begin{macrocode}
  \cs_set_protected:Npn \@@_save_blocks:nn #1#2
    {
      \tl_set_eq:Nc \l_@@_matched_block_tl { l_@@_ #1 _block_tl }
      \int_step_inline:nn { \tl_use:c { l_@@_ #1 _block_tl } - 1 }
        {
          \tl_if_eq:ccT { l_@@_ #1 _block_clist }
            { l_@@_ #1 _block_ ##1 _clist }
            { \tl_set:Nn \l_@@_matched_block_tl {##1} }
        }
      \int_compare:nNnT
        { \tl_use:c { l_@@_ #1 _block_tl } } = \l_@@_matched_block_tl
          {
            \clist_set_eq:cc
              {
                l_@@_ #1 _block_
                \tl_use:c { l_@@_ #1 _block_tl } _clist
              }
              { l_@@_ #1 _block_clist }
            \tl_set:cx { l_@@_ #1 _block_tl }
              { \int_eval:n { \tl_use:c { l_@@_ #1 _block_tl } + 1 } }
          }
        \prg_replicate:nn {#2}
          {
            \tl_set:cx { l_@@_ #1 _pos_tl }
              { \int_eval:n { \tl_use:c { l_@@_ #1 _pos_tl } + 1 } }
            \exp_args:Nc \__kernel_intarray_gset:Nnn
              { g_@@_ #1 _index_intarray }
              { \tl_use:c { l_@@_ #1 _pos_tl } }
              \l_@@_matched_block_tl
          }
      \clist_clear:c { l_@@_ #1 _block_clist }
    }
%    \end{macrocode}
% Close out the final block, rename the first stage table, then combine all
% of the block comma-lists into one large second-stage table with offsets.
% As we use an index not an offet, there is a little back-and-forward to do.
%    \begin{macrocode}
  \cs_set_protected:Npn \@@_finalise_blocks:
    {
      \clist_map_inline:nn { uppercase , lowercase }
        {
          \@@_range:nnn { 110000 } {##1} { 0 }
          \@@_finalise_blocks:n {##1}
        }
   }
  \cs_set_protected:Npn \@@_finalise_blocks:n #1
    {
      \cs_gset_eq:cc { c_@@_ #1 _index_intarray } { g_@@_ #1 _index_intarray }
      \cs_undefine:c { g_@@_ #1 _index_intarray }
      \intarray_new:cn { g_@@_ #1 _blocks_intarray }
         { ( \tl_use:c { l_@@_ #1 _block_tl } - 1 ) * \c_@@_block_size_int }
      \int_step_inline:nn { \tl_use:c { l_@@_ #1 _block_tl } - 1 }
        {
          \exp_args:Nv \@@_finalise_blocks:nnn
            { l_@@_ #1 _block_ ##1 _clist }
            {##1} {#1}
        }
      \cs_gset_eq:cc { c_@@_ #1 _blocks_intarray }
        { g_@@_ #1 _blocks_intarray }
      \cs_undefine:c { g_@@_ #1 _blocks_intarray }
    }
  \cs_set_protected:Npn \@@_finalise_blocks:nnn #1#2#3
    {
      \exp_args:Nnf \@@_finalise_blocks:nnnw { 1 }
        { \int_eval:n { ( #2 - 1 ) * \c_@@_block_size_int } }
        {#3}
        #1 , \q_recursion_tail , \q_recursion_stop
    }
  \cs_set_protected:Npn \@@_finalise_blocks:nnnw #1#2#3#4 ,
    {
      \quark_if_recursion_tail_stop:n {#4}
      \intarray_gset:cnn { g_@@_ #3 _blocks_intarray }
        { #1 + #2 }
        {#4}
      \exp_args:Nf \@@_finalise_blocks:nnnw
        { \int_eval:n { #1 + 1 } } {#2} {#3}
    }
%    \end{macrocode}
%  With the setup done, read the main data file: it's easiest to do that as
%  a token list with spaces retained.
%    \begin{macrocode}
  \ior_open:Nn \g_@@_data_ior { UnicodeData.txt }
  \group_begin:
    \char_set_catcode_space:n { `\  }%
    \ior_map_variable:NNn \g_@@_data_ior \l_@@_tmpa_tl
      {%
        \if_meaning:w \l_@@_tmpa_tl \c_space_tl
          \exp_after:wN \ior_map_break:
        \fi:
        \exp_after:wN \@@_data_auxi:w \l_@@_tmpa_tl \q_stop
      }%
    \@@_finalise_blocks:
  \group_end:
\group_end:
%    \end{macrocode}
%
% \begin{macro}[EXP]{\__kernel_codepoint_data:nn}
% \begin{macro}[EXP]{\@@_data:nnn}
%   Recover data from a two-stage table: entirely generic as this applies to
%   all tables (as we use the same block size for all of them). Notice that
%   as we use indices not offsets we have to shuffle out-by-one issues. This
%   function is needed \emph{before} loading the special casing data, as there
%   we need to be able to check the standard case mappings.
%    \begin{macrocode}
\cs_new:Npn \__kernel_codepoint_data:nn #1#2
  {
    \exp_args:Nf \@@_data:nnn
      {
        \int_eval:n
          {
            \c_@@_block_size_int *
              (
                \intarray_item:cn { c_@@_ #1 _index_intarray }
                  {
                    \int_div_truncate:nn {#2}
                      \c_@@_block_size_int
                    + 1
                  }
                  - 1
              )
          }
      }
      {#2} {#1}
  }
\cs_new:Npn \@@_data:nnn #1#2#3
  {
    \intarray_item:cn { c_@@_ #3 _blocks_intarray }
      { #1 + \int_mod:nn {#2} \c_@@_block_size_int + 1 }
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% The other data files all use C-style comments so we have to worry about
% |#| tokens (and reading as strings). The set up for case folding is in two
% parts. For the basic (core) mappings, folding is the same as lower casing in
% most positions so only store the differences. For the more complex foldings,
% always store the result, splitting up the two or three code points in the input
% as required.
%    \begin{macrocode}
\group_begin:
  \ior_open:Nn \g_@@_data_ior { CaseFolding.txt }
  \cs_set_protected:Npn \@@_data_auxi:w #1 ;~ #2 ;~ #3 ; #4 \q_stop
    {
      \if:w \tl_head:n { #2 ? } C
        \reverse_if:N \if_int_compare:w
          \int_eval:n { \__kernel_codepoint_data:nn { lowercase } {"#1} + "#1 }
            = "#3 ~
          \tl_const:cx
            { c_@@_casefold_ \codepoint_str_generate:n {"#1} _tl }
            { {"#3} { } { } }
        \fi:
      \else:
        \if:w \tl_head:n { #2 ? } F
          \@@_data_auxii:w #1 ~ #3 ~ \q_stop
        \fi:
      \fi:
    }
%    \end{macrocode}
% Here, |#4| can have have a trailing space, so we tidy up a bit at the cost of
% speed for these small number of cases it applies to.
%    \begin{macrocode}
  \cs_set_protected:Npn \@@_data_auxii:w #1 ~ #2 ~ #3 ~ #4 \q_stop
    {
      \tl_const:cx { c_@@_casefold_ \codepoint_str_generate:n {"#1} _tl }
        {
          {"#2}
          {"#3}
          { \tl_if_blank:nF {#4} { " \int_to_Hex:n {"#4} } }
        }
    }
  \ior_str_map_inline:Nn \g_@@_data_ior
    {
      \reverse_if:N \if:w \c_hash_str \tl_head:w #1 \c_hash_str \q_stop
        \@@_data_auxi:w #1 \q_stop
      \fi:
    }
  \ior_close:N \g_@@_data_ior
%    \end{macrocode}
% For upper- and lowercasing special situations, there is a bit more to
% do as we also have titlecasing to consider, plus we need to stop part-way
% through the file.
%    \begin{macrocode}
  \ior_open:Nn \g_@@_data_ior { SpecialCasing.txt }
  \cs_set_protected:Npn \@@_data_auxi:w
    #1 ;~ #2 ;~ #3 ;~ #4 ; #5 \q_stop
    {
      \use:n { \@@_data_auxii:w #1 ~ lower ~ #2 ~ } ~ \q_stop
      \use:n { \@@_data_auxii:w #1 ~ upper ~ #4 ~ } ~ \q_stop
      \str_if_eq:nnF {#3} {#4}
        { \use:n { \@@_data_auxii:w #1 ~ title ~ #3 ~ } ~ \q_stop }
    }
  \cs_set_protected:Npn \@@_data_auxii:w
    #1 ~ #2 ~ #3 ~ #4 ~ #5 \q_stop
    {
      \tl_if_empty:nF {#4}
        {
          \tl_const:cx { c_@@_ #2 case_ \codepoint_str_generate:n {"#1} _tl }
            {
              {"#3}
              {"#4}
              { \tl_if_blank:nF {#5} {"#5} }
            }
        }
    }
  \ior_str_map_inline:Nn \g_@@_data_ior
    {
      \str_if_eq:eeTF { \tl_head:w #1 \c_hash_str \q_stop } { \c_hash_str }
        {
          \str_if_eq:eeT
            {#1}
            { \c_hash_str \c_space_tl Conditional~Mappings }
            { \ior_map_break: }
        }
        { \@@_data_auxi:w #1 \q_stop }
    }
  \ior_close:N \g_@@_data_ior
\group_end:
%    \end{macrocode}
%
% \begin{macro}[EXP]{\__kernel_codepoint_case:nn}
% \begin{macro}[EXP]{\@@_case:nnn}
% \begin{macro}[EXP]
%   {\@@_uppercase:n, \@@_lowercase:n, \@@_titlecase:n, \@@_casefold:n}
% \begin{macro}[EXP]{\@_case:nn}
%   With the core data files loaded, there is now a need to provide access to
%   this information for other modules. That is done here such that case
%   folding can also be covered. At this level, all that needs to be returned
%   is the
%    \begin{macrocode}
\cs_new:Npn \__kernel_codepoint_case:nn #1#2
  {
    \exp_args:Ne \@@_case:nnn
      { \codepoint_str_generate:n {#2} } {#1} {#2}
  }
\cs_new:Npn \@@_case:nnn #1#2#3
  {
    \cs_if_exist:cTF { c_@@_ #2 _ #1 _tl }
      {
        \tl_use:c
          { c_@@_ #2 _ #1 _tl }
      }
      { \use:c { @@_ #2 :n } {#3} }
  }
\cs_new:Npn \@@_uppercase:n { \@@_case:nn { uppercase } }
\cs_new:Npn \@@_lowercase:n { \@@_case:nn { lowercase } }
\cs_new:Npn \@@_titlecase:n { \@@_case:nn { uppercase } }
\cs_new:Npn \@@_casefold:n  { \@@_case:nn { lowercase } }
\cs_new:Npn \@@_case:nn #1#2
  {
    { \int_eval:n { \__kernel_codepoint_data:nn {#1} {#2} + #2 } }
    { }
    { }
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}[EXP]{\@@_nfd:n}
% \begin{macro}[EXP]{\@@_nfd:nn}
%   A simple interface.
%    \begin{macrocode}
\cs_new:Npn \@@_nfd:n #1
  { \exp_args:Ne \@@_nfd:nn { \codepoint_str_generate:n {#1} } {#1} }
\cs_new:Npn \@@_nfd:nn #1#2
  {
    \tl_if_exist:cTF { c_@@_nfd_ #1 _tl }
      { \tl_use:c { c_@@_nfd_ #1 _tl } }
      { {#2} { } }
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
%    \begin{macrocode}
%<@@=text>
%    \end{macrocode}
%
%  Read the Unicode grapheme data. This is quite easy to handle and we only need
%  codepoints, not characters, so there is no need to worry about the engine in use.
%  As reading as a string is most convenient, we have to do some work to remove
%  spaces: the hardest part of the entire process!
%    \begin{macrocode}
\ior_new:N \g_@@_data_ior
\group_begin:
  \ior_open:Nn \g_@@_data_ior { GraphemeBreakProperty.txt }
  \cs_set_nopar:Npn \l_@@_tmpa_str { }
  \cs_set_nopar:Npn \l_@@_tmpb_str { }
  \cs_set_protected:Npn \@@_data_auxi:w #1 ;~ #2 ~ #3 \q_stop
    {
      \str_if_eq:VnF \l_@@_tmpb_str {#2}
        {
          \str_if_empty:NF \l_@@_tmpb_str
            {
              \clist_const:cx { c_@@_grapheme_ \l_@@_tmpb_str _clist }
                { \exp_after:wN \use_none:n \l_@@_tmpa_str }
              \cs_set_nopar:Npn \l_@@_tmpa_str { }
            }
          \cs_set_nopar:Npn \l_@@_tmpb_str {#2}
        }
      \@@_data_auxii:w #1 .. #1 .. #1 \q_stop
    }
  \cs_set_protected:Npn \@@_data_auxii:w #1 .. #2 .. #3 \q_stop
    {
      \cs_set_nopar:Npx \l_@@_tmpa_str
        {
          \l_@@_tmpa_str ,
          \tl_trim_spaces:n {#1} .. \tl_trim_spaces:n {#2}
        }
    }
  \ior_str_map_inline:Nn \g_@@_data_ior
    {
      \str_if_eq:eeF { \tl_head:w #1 \c_hash_str \q_stop } { \c_hash_str }
        {
          \tl_if_blank:nF {#1}
            { \@@_data_auxi:w #1 \q_stop }
        }
    }
  \ior_close:N \g_@@_data_ior
\group_end:    
%    \end{macrocode}
%
%    \begin{macrocode}
%</package>
%    \end{macrocode}
%
% \end{implementation}
%
% \PrintIndex