summaryrefslogtreecommitdiff
path: root/Master/texmf-dist/source/latex/l3experimental/l3dt/l3dt.dtx
blob: 63a3218f06aa9e13249308d0330c74cde80496bd (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
% \iffalse meta-comment
%
%% File l3dt.dtx Copyright (C) 2011-2013 The LaTeX3 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
%%
%%    http://www.latex-project.org/lppl.txt
%%
%% This file is part of the "l3experimental bundle" (The Work in LPPL)
%% and all files in that bundle must be distributed together.
%%
%% The released version of this bundle is available from CTAN.
%%
%% -----------------------------------------------------------------------
%%
%% The development version of the bundle can be found at
%%
%%    http://www.latex-project.org/svnroot/experimental/trunk/
%%
%% for those people who are interested.
%%
%%%%%%%%%%%
%% NOTE: %%
%%%%%%%%%%%
%%
%%   Snapshots taken from the repository represent work in progress and may
%%   not work or may contain conflicting material!  We therefore ask
%%   people _not_ to put them into distributions, archives, etc. without
%%   prior consultation with the LaTeX Project Team.
%%
%% -----------------------------------------------------------------------
%%
%
%<*driver|package>
\RequirePackage{expl3}
\GetIdInfo$Id: l3dt.dtx 4420 2013-01-08 20:00:04Z joseph $
  {L3 Experimental data tables}
%</driver|package>
%<*driver>
\documentclass[full]{l3doc}
\begin{document}
  \DocInput{\jobname.dtx}
\end{document}
%</driver>
% \fi
%
% \title{^^A
%   The \pkg{l3dt} package\\ Data tables^^A
%   \thanks{This file describes v\ExplFileVersion,
%      last revised \ExplFileDate.}^^A
% }
%
% \author{^^A
%  The \LaTeX3 Project\thanks
%    {^^A
%      E-mail:
%        \href{mailto:latex-team@latex-project.org}
%          {latex-team@latex-project.org}^^A
%    }^^A
% }
%
% \date{Released \ExplFileDate}
%
% \maketitle
%
% \begin{documentation}
%
% \LaTeX3 implements a \enquote{data table} variable type, which is made up of
% a series of rows each of which contain a number of key--value pairs. Thus
% a data table is in effect an array of property lists. The rows of the table
% are stored in a fixed order, and are numbered consecutively from one.
% In the same way, the order of keys (columns) is recorded in a sequence-like
% manner, again indexed from one.
%
% Within each row in a data table each entry must have a unique \meta{key}: if
% an entry is added to a row within a data table which already contains the
% \meta{key} then the new entry will overwrite the existing one. The
% \meta{keys} are compared on a string basis, using the same method as
% \cs{str_if_eq:nn}.
%
% \section{Creating and initialising data tables}
%
% \begin{function}{\dt_new:N}
%   \begin{syntax}
%     \cs{dt_new:N} \meta{data table}
%   \end{syntax}
%   Creates a new \meta{data table} or raises an error if the name is
%   already taken. The declaration is global. The \meta{property lists} will
%   initially contain no entries.
% \end{function}
%
% \begin{function}{\dt_clear:N, \dt_gclear:N}
%   \begin{syntax}
%     \cs{dt_clear:N} \meta{data table}
%   \end{syntax}
%   Clears all entries and keys from the \meta{data table}.
% \end{function}
%
% \begin{function}{\dt_clear_new:N, \dt_gclear_new:N}
%   \begin{syntax}
%     \cs{dt_clear_new:N} \meta{data table}
%   \end{syntax}
%   Ensures that the \meta{data table} exists globally by applying
%   \cs{dt_new:N} if necessary, then applies \cs{dt_(g)clear:N} to leave
%   the table empty.
% \end{function}
%
% \begin{function}{\dt_set_eq:NN, \dt_gset_eq:NN}
%   \begin{syntax}
%     \cs{dt_set_eq:NN} \meta{data table1} \meta{data table2}
%   \end{syntax}
%   Sets the content of \meta{data table1} equal to that of
%   \meta{data table2}.
% \end{function}
%
% \section{Adding data}
%
% \begin{function}{\dt_add_key:Nn, \dt_gadd_key:Nn}
%   \begin{syntax}
%     \cs{dt_add_key:Nn} \meta{dt} \Arg{key}
%   \end{syntax}
%   Adds the \meta{key} to the list of those in the \meta{data table}. The
%   \meta{key} will be converted to a string using \cs{tl_to_str:n}, and thus
%   category codes in the \meta{key} are ignored. If the \meta{key} is already
%   present in the \meta{data table} then no action is taken.
% \end{function}
%
% \begin{function}{\dt_add_row:N, \dt_gadd_row:N}
%   \begin{syntax}
%     \cs{dt_add_row:N} \meta{dt}
%   \end{syntax}
%   Adds a new row to the \meta{data table}. This will initially contain
%   no entries: all keys will be be blank.
% \end{function}
%
% \begin{function}{\dt_put:Nnn, \dt_gput:Nnn}
%   \begin{syntax}
%     \cs{dt_put:Nnn} \meta{dt} \Arg{key} \Arg{value}
%   \end{syntax}
%   Adds an entry to the current row of the \meta{data table} which may be
%   accessed using the \meta{key} and which has \meta{value}. Both the
%   \meta{key} and \meta{value} may contain any \meta{balanced text}. The
%   \meta{key} is stored after processing with \cs{tl_to_str:n}, meaning
%   that category codes are ignored. If the \meta{key} is already present
%   in the current row of the \meta{data table}, the existing entry is
%   overwritten by the new \meta{value}.
% \end{function}
%
% \begin{function}{\dt_put:Nnnn, \dt_gput:Nnnn}
%   \begin{syntax}
%     \cs{dt_put:Nnnn} \meta{dt} \Arg{row} \Arg{key} \Arg{value}
%   \end{syntax}
%   Adds an entry to the \meta{row} of the \meta{data table} which may be
%   accessed using the \meta{key} and which has \meta{value}. Both the
%   \meta{key} and \meta{value} may contain any \meta{balanced text}. The
%   \meta{key} is stored after processing with \cs{tl_to_str:n}, meaning
%   that category codes are ignored. If the \meta{key} is already present
%   in the \meta{row} of the \meta{data table}, the existing entry is
%   overwritten by the new \meta{value}. The \meta{row} should be given as
%   an \meta{integer expression}.
% \end{function}
%
% \section{Removing data}
%
% \begin{function}{\dt_remove:Nn, \dt_gremove:Nn}
%   \begin{syntax}
%     \cs{dt_remove:Nn} \meta{dt} \Arg{key}
%   \end{syntax}
%   Deletes any entry from the current row of the \meta{data table} with
%   the \meta{key}. The \meta{key} is compared after processing with
%   \cs{tl_to_str:n}, meaning that category codes are ignored. Deleting of
%   all entries from a row does not delete the row itself.
% \end{function}
%
% \begin{function}{\dt_remove:Nnn, \dt_gremove:Nnn}
%   \begin{syntax}
%     \cs{dt_remove:Nnn} \meta{dt} \Arg{row} \Arg{key}
%   \end{syntax}
%   Deletes any entry from the \meta{row} of the \meta{data table} with
%   the \meta{key}. The \meta{key} is compared after processing with
%   \cs{tl_to_str:n}, meaning that category codes are ignored. The \meta{row}
%   may be given as an \meta{integer expression}. Deleting of
%   all entries from a row does not delete the row itself.
% \end{function}
%
% \begin{function}{\dt_remove_key:Nn, \dt_gremove_key:Nn}
%   \begin{syntax}
%     \cs{dt_remove_key:N} \meta{data table} \Arg{key}
%   \end{syntax}
%   Removes the \meta{key} from the \meta{data table} if it is present.
%   The \meta{key} and any associated \meta{value} will be removed from any
%   row that it is found in.
% \end{function}
%
% \begin{function}{\dt_remove_row:Nn, \dt_gremove_row:Nn}
%   \begin{syntax}
%     \cs{dt_remove_row:Nn} \meta{data table} \Arg{row}
%   \end{syntax}
%   Removes the \meta{row} (given as an \meta{integer expressions}) from the
%   \meta{data table}. The remaining rows of the table will be renumbered
%   such that they are sequential.
% \end{function}
%
% \section{Recovering information}
%
% \begin{function}[EXP]{\dt_keys:N}
%   \begin{syntax}
%     \cs{dt_keys:N} \meta{dt}
%   \end{syntax}
%   Leaves the number of keys in the \meta{data table} in the input
%   stream as an \meta{integer denotation}.
% \end{function}
%
% \begin{function}[EXP]{\dt_rows:N}
%   \begin{syntax}
%     \cs{dt_rows:N} \meta{dt}
%   \end{syntax}
%   Leaves the number of rows in the \meta{data table} in the input
%   stream as an \meta{integer denotation}.
% \end{function}
%
% \begin{function}{\dt_get:NnN}
%   \begin{syntax}
%     \cs{dt_get:NnnN} \meta{dt} \Arg{key} \meta{tl var}
%   \end{syntax}
%   Recovers the \meta{value} stored with \meta{key} from the current row in
%   the \meta{data table}, and places this in the \meta{token list variable}.
%   If the \meta{key} is not found in the \meta{row} of the  \meta{data table}
%   then the \meta{token list variable} will contain the special marker
%   \cs{q_no_value}. The \meta{token list variable} is set within the
%   current \TeX{} group. The \meta{row} should be given as an
%   \meta{integer expression}. See also \cs{dt_get:NnNTF}.
% \end{function}
%
% \begin{function}[TF]{\dt_get:NnN}
%   \begin{syntax}
%     \cs{dt_get:NnnNTF} \meta{dt} \Arg{key} \meta{tl var} \Arg{true code} \Arg{false code}
%   \end{syntax}
%   Recovers the \meta{value} stored with \meta{key} from the current row in
%   the \meta{data table}, and places this in the \meta{token list variable}.
%   If the \meta{key} is not found in the \meta{row} of the  \meta{data table}
%   then the \meta{token list variable} will contain the special marker
%   \cs{q_no_value}. The \meta{token list variable} is set within the
%   current \TeX{} group. The \meta{row} should be given as an
%   \meta{integer expression}. Once the \meta{token list variable} has been
%   assigned either the \meta{true code} or \meta{false code} will be left in
%   the input stream, depending on whether the \meta{key} was found.
%   See also \cs{dt_get:NnN}.
% \end{function}
%
% \begin{function}{\dt_get:NnnN}
%   \begin{syntax}
%     \cs{dt_get:NnnN} \meta{dt} \Arg{row} \Arg{key} \meta{tl var}
%   \end{syntax}
%   Recovers the \meta{value} stored with \meta{key} from \meta{row} in the
%   \meta{data table}, and places this in the \meta{token list variable}. If
%   the \meta{key} is not found in the \meta{row} of the  \meta{data table}
%   then the \meta{token list variable} will contain the special marker
%   \cs{q_no_value}. The \meta{token list variable} is set within the
%   current \TeX{} group. The \meta{row} should be given as an
%   \meta{integer expression}. See also \cs{dt_get:NnnNTF}.
% \end{function}
%
% \begin{function}[TF]{\dt_get:NnnN}
%   \begin{syntax}
%     \cs{dt_get:NnnNTF} \meta{dt} \Arg{row} \Arg{key} \meta{tl var} \Arg{true code} \Arg{false code}
%   \end{syntax}
%   Recovers the \meta{value} stored with \meta{key} from \meta{row} in the
%   \meta{data table}, and places this in the \meta{token list variable}. If
%   the \meta{key} is not found in the \meta{row} of the  \meta{data table}
%   then the \meta{token list variable} will contain the special marker
%   \cs{q_no_value}. The \meta{token list variable} is set within the
%   current \TeX{} group. The \meta{row} should be given as an
%   \meta{integer expression}. Once the \meta{token list variable} has been
%   assigned either the \meta{true code} or \meta{false code} will be left in
%   the input stream, depending on whether the \meta{key} was found.
%   See also \cs{dt_get:NnnN}.
% \end{function}
%
% \section{Mapping to data tables}
%
% \begin{function}{\dt_map_variables:Nnn}
%   \begin{syntax}
%     \cs{dt_map_variables:Nnn} \meta{data table} \Arg{key--variable mapping} \Arg{code}
%   \end{syntax}
%   Applies the \meta{code} to each \meta{row} of the \meta{data table}. The
%   \meta{keys} of the \meta{data table} are mapped to variables by the
%   \meta{key--variable mapping}, which should be a key--value list of the form
%   \begin{verbatim}
%     key-a = \l_a_tl ,
%     key-b = \l_b_tl
%     ...
%   \end{verbatim}
%   It is not necessary to map all of the \meta{keys} in a \meta{data table}
%   to variables. If there is not \meta{value} for a \meta{key} in a row,
%   the variable will contain the marker \cs{q_no_value}. Assignment of the
%   \meta{variables} is local to the current \TeX{} group. The mapping
%   to rows is ordered.
% \end{function}
%
% \begin{variable}{\g_dt_map_level_int}
%   The nesting level of the data table mapping is available as
%   \cs{g_dt_map_level_int}. Within a mapping, the \texttt{int} variable
%   \cs{l_dt_map_\meta{level}_row_int} is available so that the row
%   number being operated on is available. Thus
%   \begin{verbatim}
%     \int_use:c { l_dt_map_ \int_use:N \g_dt_map_level_int _row_int }
%   \end{verbatim}
%   will give the current row for the current mapping.
% \end{variable}
%
% \begin{function}[rEXP]{\dt_map_break:, \dt_map_break:n}
%   \begin{syntax}
%     \cs{dt_map_break:}
%     \cs{dt_map_break:n} \Arg{tokens}
%   \end{syntax}
%   Used to terminate a \cs{dt_map_\ldots} function before all
%   entries in the \meta{data table} have been processed. This will
%   normally take place within a conditional statement, for example
%   \begin{verbatim}
%     \dt_map_variables:Nn \l_my_dt { a = \l_my_tl }
%       {
%         \str_if_eq:VnTF \l_my_tl { bingo }
%           { \dt_map_break: }
%           {
%             % Do something useful
%           }
%       }
%   \end{verbatim}
%   The \texttt{:n} variant will insert the \meta{tokens} into the input stream
%   after the mapping terminates. Use outside of a \cs{dt_map_\ldots} scenario
%   will lead to low level \TeX{} errors.
% \end{function}
%
% \section{Data table conditionals}
%
% \begin{function}[EXP, pTF]{\dt_if_empty:N}
%   \begin{syntax}
%     \cs{dt_if_empty_p:N} \meta{dt}
%     \cs{dt_if_empty:NTF} \meta{dt} \Arg{true code} \Arg{false code}
%   \end{syntax}
%   Tests if the \meta{dt} is empty, containing no keys and no rows.
% \end{function}
%
% \begin{function}[EXP, pTF]{\dt_if_in:Nn}
%   \begin{syntax}
%     \cs{dt_if_in_p:Nn} \meta{dt} \Arg{key}
%     \cs{dt_if_in:NnTF} \meta{dt} \Arg{key} \Arg{true code} \Arg{false code}
%   \end{syntax}
%   Tests if the \meta{key} is present in the \meta{data table} at all,
%   \emph{i.e.} if it is one of the columns of the table. This test will be
%   \texttt{true} even if none of the rows contain an entry for the \meta{key}.
% \end{function}
%
% \begin{function}[EXP, pTF]{\dt_if_in_row:Nnn}
%   \begin{syntax}
%     \cs{dt_if_in_row_p:Nnn} \meta{dt} \Arg{row} \Arg{key}
%     \cs{dt_if_in_row:NnnTF} \meta{dt} \Arg{row} \Arg{key} \Arg{true code} \Arg{false code}
%   \end{syntax}
%   Tests if the \meta{key} is present in the \meta{row} of the
%   \meta{data table}. The \meta{row} may be given as an \meta{integer
%   expression}.
% \end{function}
%
% \begin{function}[EXP, pTF]{\dt_if_in_row:Nn}
%   \begin{syntax}
%     \cs{dt_if_in_row_p:Nn} \meta{dt} \Arg{key}
%     \cs{dt_if_in_row:NnTF} \meta{dt} \Arg{key} \Arg{true code} \Arg{false code}
%   \end{syntax}
%   Tests if the \meta{key} is present in the current row of
%   \meta{data table}.
% \end{function}
%
% \section{Variables}
%
% \begin{variable}{\c_empty_dt}
%   A permanently empty data table.
% \end{variable}
%
% \begin{variable}{\l_tmpa_dt, \l_tmpb_dt, \g_tmpa_dt, \g_tmpb_dt}
%   Scratch data tables for general use: these are never used by the kernel.
% \end{variable}
%
% \end{documentation}
%
% \begin{implementation}
%
% \section{\pkg{l3dt} implementation}
%
%    \begin{macrocode}
%<*initex|package>
%    \end{macrocode}
%
%    \begin{macrocode}
%<@@=dt>
%    \end{macrocode}
%
%    \begin{macrocode}
%<*package>
\ProvidesExplPackage
  {\ExplFileName}{\ExplFileDate}{\ExplFileVersion}{\ExplFileDescription}
%</package>
%    \end{macrocode}
%
% \subsection{Structures}
%
% The structure of a data table must allow each row (record) to contain only
% some of the keys, and for the keys to be removed after the table
% is initialised. It also needs to ensure that a unique match can be made to
% every item in the table. At the same time, it is desirable to keep all of
% the information about the table in a single \TeX{} macro. This can be
% achieved by packing the data into a structure in which each key and row is
% numbered:
% \begin{quote}
%   \Arg{rows} \\
%   \Arg{columns} \\
%   \cs{q_@@} \meta{key_1} \cs{q_@@} \meta{key_2} \cs{q_@@} \ldots \\
%   \cs{q_nil} \\
%   \cs{q_@@_header} \\
%   \cs{q_@@_row} \\
%   \meta{row_1} \\
%   \cs{q_@@} \meta{key_1} \cs{q_@@} \Arg{data_{1,1}} \\
%   \cs{q_@@} \meta{key_2} \cs{q_@@} \Arg{data_{1,2}} \\
%   \ldots \\
%   \cs{q_@@} \\
%   \cs{q_nil} \\
%   \cs{q_@@_row} \\
%   \meta{row_2} \\
%   \cs{q_@@} \meta{key_1} \cs{q_@@} \Arg{data_{2,1}} \\
%   \cs{q_@@} \meta{key_2} \cs{q_@@} \Arg{data_{2,2}} \\
%   \ldots \\
%   \cs{q_@@} \\
%   \cs{q_nil} \\
%   \cs{q_@@_row} \\
%   \ldots \\
%   \cs{q_@@_row}
% \end{quote}
%
% \begin{variable}{\q_@@, \q_@@_row, \q_@@_header}
%   The quarks are set up.
%    \begin{macrocode}
\quark_new:N \q_@@
\quark_new:N \q_@@_row
\quark_new:N \q_@@_header
%    \end{macrocode}
% \end{variable}
%
% \begin{variable}{\c_empty_dt}
%   A permanently-empty data table, which therefore contains only the minimum
%   number of items necessary to comply with the structure above.
%    \begin{macrocode}
\tl_const:Nn \c_empty_dt
  {
    { 0 }
    { 0 }
    \q_@@
    \q_nil
    \q_@@_header
    \q_@@_row
  }
%    \end{macrocode}
% \end{variable}
%
% \subsection{Allocation and initialisation}
%
% \begin{macro}{\dt_new:N}
%   Internally, data tables are token lists, but an empty dt
%   is not an empty tl.
%    \begin{macrocode}
\cs_new_protected:Npn \dt_new:N #1 { \cs_new_eq:NN #1 \c_empty_dt }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\dt_clear:N, \dt_gclear:N}
%   The same idea for  clearing.
%    \begin{macrocode}
\cs_new_protected:Npn \dt_clear:N #1  { \cs_set_eq:NN  #1 \c_empty_dt }
\cs_new_protected:Npn \dt_gclear:N #1 { \cs_gset_eq:NN #1 \c_empty_dt }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}
%   {\dt_clear_new:N, \dt_gclear_new:N}
%   Once again a simple copy from the token list functions.
%    \begin{macrocode}
\cs_new_protected:Npn \dt_clear_new:N #1
  { \cs_if_exist:NTF #1 { \dt_clear:N #1 } { \dt_new:N #1 } }
\cs_new_protected:Npn \dt_gclear_new:N #1
  { \cs_if_exist:NTF #1 { \dt_gclear:N #1 } { \dt_new:N #1 } }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\dt_set_eq:NN, \dt_gset_eq:NN}
%   Once again, these are simply copies from the token list functions.
%    \begin{macrocode}
\cs_new_eq:NN \dt_set_eq:NN  \tl_set_eq:NN
\cs_new_eq:NN \dt_gset_eq:NN \tl_gset_eq:NN
%    \end{macrocode}
% \end{macro}
%
% \begin{variable}{\l_tmpa_dt, \l_tmpb_dt, \g_tmpa_dt, \g_tmpb_dt}
%   Scratch tables.
%    \begin{macrocode}
\dt_new:N \l_tmpa_dt
\dt_new:N \l_tmpb_dt
\dt_new:N \g_tmpa_dt
\dt_new:N \g_tmpb_dt
%    \end{macrocode}
% \end{variable}
%
% \subsection{Splitting functions}
%
% \begin{macro}[aux]{\@@_split:nnnn}
% \begin{macro}[aux]{\@@_split:w}
%   Two general auxiliaries. The \texttt{nnnn} function is used to apply the
%   \texttt{T} branch if a match is found and the \texttt{F} branch otherwise.
%   The \texttt{w} function is general purpose, and is used to define the
%   matching parameter set.
%    \begin{macrocode}
\cs_new_protected:Npn \@@_split:nnnn #1#2#3#4 { #3 #2 }
\cs_new_protected:Npn \@@_split:w { }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}[int,EXP]{\@@_split_header:NT}
% \begin{macro}[aux,EXP]{\@@_split_header:wn}
%   Splits the header from the table, inserting the code required to then
%   process the split table. The \cs{q_nil} is also removed from the end of the
%   header, as it is essentially a distraction here.
%    \begin{macrocode}
\cs_new:Npn \@@_split_header:NT #1#2
  { \exp_after:wN \@@_split_header:wn #1 \q_stop {#2} }
\cs_new:Npn \@@_split_header:wn #1 \q_nil \q_@@_header #2 \q_stop #3
  { #3 {#1} { \q_@@_header #2 } }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}[int]{\@@_split_key:nnTF}
% \begin{macro}[aux]{\@@_split_key_aux:nnTF}
%   Here, the split is made for a partial list within a row. The row is
%   basically the same as a property list, so the split here is almost
%   identical to that in \cs{prop_split_aux:NnTF}. The row-end data is set up
%   such that it will not interfere with this process.
%    \begin{macrocode}
\cs_new_protected:Npn \@@_split_key:nnTF #1#2
  { \exp_args:No \@@_split_key_aux:nnTF { \tl_to_str:n {#2} } {#1} }
\cs_new_protected:Npn \@@_split_key_aux:nnTF #1#2
  {
    \cs_set_protected:Npn \@@_split:w
      ##1 \q_@@ #1 \q_@@ ##2##3##4 \q_mark ##5 \q_stop
      { \@@_split:nnnn ##3 { { ##1 \q_@@ } {##2} {##4} } }
    \@@_split:w #2 \q_mark
      \q_@@ #1 \q_@@ { } { ? \use_ii:nn { } } \q_mark \q_stop
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}[int]{\@@_split_key_list:NnTF}
% \begin{macro}[aux]{\@@_split_key_list_aux:NnTF}
%   Finding a key in the header uses a similar approach to finding a key in
%   a property list. Here, if the key is found there will always be at least
%   one token between \cs{q_@@} and \cs{q_@@_header} due to the \cs{q_nil}
%   which is part of a new table. The use of |##1##2##3| in \cs{@@_split:w}
%   here is to deal with the overall number of rows and keys. The set up here
%   means that these will always be unbraced then rebraced: simply grabbing
%   |##1##2| to include this and anything before the key of interest will give
%   variable results depending on whether the match is to the very first key or not.
%    \begin{macrocode}
\cs_new_protected:Npn \@@_split_key_list:NnTF #1#2
  { \exp_args:NNo \@@_split_key_list_aux:NnTF #1 { \tl_to_str:n {#2} } }
\cs_new_protected:Npn \@@_split_key_list_aux:NnTF #1#2
  {
    \cs_set_protected:Npn \@@_split:w
      ##1##2##3 \q_@@ #2 \q_@@ ##4##5 \q_@@_header ##6 \q_mark ##7 \q_stop
      {
        \@@_split:nnnn ##4
          { { {##1} {##2} ##3 \q_@@ } { ##4##5 \q_@@_header ##6 } }
      }
    \exp_after:wN \@@_split:w #1 \q_mark
      \q_@@ #2 \q_@@ { ? \use_ii:nn { } } \q_@@_header \q_mark \q_stop
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}[int]{\@@_split_row:NnTF}
% \begin{macro}[aux]{\@@_split_row_aux:NnTF, \@@_split_row_aux:NfTF}
%   The usual approach, here using the fact that each row start with row number
%   and ends with \cs{q_nil} so there will always be at least one token to be
%   absorbed as |##2|. The only odd thing to watch here is that the row
%   number is evaluated so that higher-level functions in the main do not
%   need to have an \texttt{f}-type variant.
%    \begin{macrocode}
\cs_new_protected:Npn \@@_split_row:NnTF #1#2
  { \@@_split_row_aux:NfTF #1 { \int_eval:n {#2} } }
\cs_new_protected:Npn \@@_split_row_aux:NnTF #1#2
  {
    \cs_set_protected:Npn \@@_split:w
      ##1 \q_@@_row #2 \q_@@ ##2##3 \q_@@_row ##4 \q_mark ##5 \q_stop
      {
        \@@_split:nnnn ##2
          { { ##1 \q_@@_row } { #2 \q_@@ ##2##3 } {##4} }
      }
    \exp_after:wN \@@_split:w #1 \q_mark
      \q_@@_row #2 \q_@@ { ? \use_ii:nn { } } \q_@@_row \q_mark \q_stop
  }
\cs_generate_variant:Nn \@@_split_row_aux:NnTF { Nf }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \subsection{Adding and removing data}
%
% \begin{macro}{\dt_add_key:Nn, \dt_gadd_key:Nn}
% \begin{macro}[aux]{\@@_add_key:NNn}
% \begin{macro}[aux]{\@@_add_key:NNnnn}
% \begin{macro}[aux]{\@@_add_key:NNnnn}
%   Here, there are two stages. If the key is already present in the list of
%   known keys then no action is taken, and the split list is thrown away.
%   On the other hand, if the key is not present then the header and body
%   are separated and the key is added to the end of the list of known keys
%   (hence keys are ordered). The \cs{@@_split_header:Nn} function will have
%   removed the \cs{q_nil} from the header, and so it is put back in here.
%    \begin{macrocode}
\cs_new_protected_nopar:Npn \dt_add_key:Nn { \@@_add_key:NNn \tl_set:Nx }
\cs_new_protected_nopar:Npn \dt_gadd_key:Nn { \@@_add_key:NNn \tl_gset:Nx }
\cs_new_protected:Npn \@@_add_key:NNn #1#2#3
  {
    \@@_split_key_list:NnTF #2 {#3}
      { \use_none:nn }
      {
        \@@_split_header:NT #2
          { \@@_add_key:NNnnn #1 #2 {#3} }
      }
  }
\cs_new_protected:Npn \@@_add_key:NNnnn #1#2#3#4#5
  { \@@_add_key:NNnnnwnn #1 #2 #4 \q_stop {#3} {#5} }
\cs_new_protected:Npn \@@_add_key:NNnnnwnn #1#2#3#4#5 \q_stop #6#7
  {
    #1 #2
      {
        {#3}
        { \int_eval:n { #4 + \c_one } }
        \exp_not:n {#5}
        \tl_to_str:n {#6}
        \exp_not:n { \q_@@ \q_nil #7 }
      }
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\dt_add_row:N, \dt_gadd_row:N}
% \begin{macro}[aux]{\@@_add_row:NN}
% \begin{macro}[aux]{\@@_add_row:NnN, \@@_add_row:NfN}
% \begin{macro}[aux,EXP]{\@@_add_row:nw}
%   Adding a row means incrementing the total number and adding the structure
%   of an empty row. As finding the rows will get slow for large tables, this
%   is only done once.
%    \begin{macrocode}
\cs_new_protected_nopar:Npn \dt_add_row:N { \@@_add_row:NN \tl_set:Nx }
\cs_new_protected_nopar:Npn \dt_gadd_row:N { \@@_add_row:NN \tl_gset:Nx }
\cs_new_protected:Npn \@@_add_row:NN #1#2
  { \@@_add_row:NfN #1 { \int_eval:n { \dt_rows:N #2 + \c_one } } #2 }
\cs_new_protected:Npn \@@_add_row:NnN #1#2#3
  {
    #1 #3
      {
        {#2}
        \exp_after:wN \@@_add_row:nw #3 \q_stop
        #2
        \exp_not:n { \q_@@ \q_nil \q_@@_row }
      }
  }
\cs_generate_variant:Nn \@@_add_row:NnN { Nf } 
\cs_new:Npn \@@_add_row:nw #1#2 \q_stop { \exp_not:n {#2} }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\dt_put:Nnn, \dt_gput:Nnn}
%   Adding to the current row is simply a special case of adding to an
%   arbitrary row.
%    \begin{macrocode}
\cs_new_protected:Npn \dt_put:Nnn #1
  { \dt_put:Nnnn #1 { \dt_rows:N #1 } }
\cs_new_protected:Npn \dt_gput:Nnn #1
  { \dt_gput:Nnnn #1 { \dt_rows:N #1 } }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\dt_put:Nnnn, \dt_gput:Nnnn}
% \begin{macro}[aux]{\@@_put:NNNnnn}
% \begin{macro}[aux]{\@@_put:NNnnnnn}
% \begin{macro}[aux]{\@@_put_update:NNnnnnnnn}
% \begin{macro}[aux]{\@@_put_add_to_row:NNnnnnn}
% \begin{macro}[aux, EXP]{\@@_put_add_to_row_aux:w}
%   Adding to a row is a slightly complex procedure. The lead-off is the
%   standard combination across the local and global routes.
%    \begin{macrocode}
\cs_new_protected_nopar:Npn \dt_put:Nnnn
  { \@@_put:NNNnnn \dt_add_key:Nn \tl_set:Nx }
\cs_new_protected_nopar:Npn \dt_gput:Nnnn
  { \@@_put:NNNnnn \dt_gadd_key:Nn \tl_gset:Nx }
%    \end{macrocode}
%   Add the key to the list those known, if necessary, then check that the
%   row requested makes sense.
%    \begin{macrocode}
\cs_new_protected:Npn \@@_put:NNNnnn #1#2#3#4#5#6
  {
    #1 #3 {#5}
    \@@_split_row:NnTF #3 {#4}
      { \@@_put:NNnnnnn #2 #3 {#5} {#6} }
      {
        \__msg_kernel_error:nnxxx { dt } { unknown-row }
          { \token_to_str:N #3 } { \int_eval:n {#4} } { \dt_rows:N #3 }
      }
  }
%    \end{macrocode}
%   At this stage, the arguments are
%   \begin{enumerate}
%     \item the set function \cs{tl_(g)set:Nx},
%     \item the data table,
%     \item the key,
%     \item the value,
%     \item the data table before the row,
%     \item the extracted data table row,
%     \item the data table after the row.
%   \end{enumerate}
%   Splitting on the key will then leave three further items in the input
%   stack if the key is already present. So there is some care needed sending
%   the parameters forward without running out of \TeX{} arguments.
%    \begin{macrocode}
\cs_new_protected:Npn \@@_put:NNnnnnn #1#2#3#4#5#6#7
  {
    \@@_split_key:nnTF {#6} {#3}
      { \@@_put_update:NNnnnnnnn #1 #2 {#3} {#4} {#5} {#7} }
      { \@@_put_add_to_row:NNnnnnn #1 #2 {#3} {#4} {#5} {#6} {#7} }
  }
%    \end{macrocode}
%   The arguments here are
%   \begin{enumerate}
%     \item the set function \cs{tl_(g)set:Nx},
%     \item the data table,
%     \item the key,
%     \item the value,
%     \item the data table before the row,
%     \item the data table after the row,
%     \item the row before the key,
%     \item the current value for the key
%     \item the row after the key.
%   \end{enumerate}
%   What happens here is a reconstruction of the table: everything except
%   |#8| is needed. To try to keep things clear, there are a few more
%   \cs{exp_not:n} here than formally required.
%    \begin{macrocode}
\cs_new_protected:Npn \@@_put_update:NNnnnnnnn #1#2#3#4#5#6#7#8#9
  {
    #1 #2
      {
        \exp_not:n { #5 #7 }
        \tl_to_str:n {#3}
        \exp_not:n { \q_@@ {#4} \q_@@ #9 \q_@@_row #6 }
      }
  }
%    \end{macrocode}
%   A slightly more complex case when adding an item. The arguments here are
%   identical to those for \cs{@@_put:NNnnnnnn}. The row has not been
%   split, so the \cs{q_nil} there is removed and re-added to come after the
%   new content.
%    \begin{macrocode}
\cs_new_protected:Npn \@@_put_add_to_row:NNnnnnn #1#2#3#4#5#6#7
  {
    #1 #2
      {
        \exp_not:n {#5}
        \exp_not:o { \@@_put_add_to_row_aux:w #6 }
        \tl_to_str:n {#3}
        \exp_not:n { \q_@@ {#4} \q_@@ \q_nil \q_@@_row #7 }
      }
  }
\cs_new:Npn \@@_put_add_to_row_aux:w #1 \q_nil {#1}
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}[EXP]{\dt_keys:N}
% \begin{macro}[EXP, aux]{\@@_keys:nnw}
% \begin{macro}[EXP]{\dt_rows:N}
%  The number of rows in a dt is the very first entry. Getting the number of
%  keys is almost the same, except a custom auxiliary is needed.
%    \begin{macrocode}
\cs_new:Npn \dt_keys:N #1 { \exp_after:wN \@@_keys:nnw #1 \q_stop }
\cs_new:Npn \@@_keys:nnw #1#2#3 \q_stop {#2}
\cs_new:Npn \dt_rows:N #1
  { \exp_after:wN \use_i_delimit_by_q_stop:nw #1 \q_stop }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \subsection{Removing data}
%
% \begin{macro}{\dt_remove:Nn,\dt_gremove:Nn}
%   Deleting to the current row is simply a special case of deleting to an
%   arbitrary row.
%    \begin{macrocode}
\cs_new_protected:Npn \dt_remove:Nn #1
  { \dt_remove:Nnn #1 { \dt_rows:N #1 } }
\cs_new_protected:Npn \dt_gremove:Nn #1
  { \dt_gremove:Nnn #1 { \dt_rows:N #1 } }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\dt_remove:Nnn, \dt_gremove:Nnn}
% \begin{macro}[aux]{\dt_remove_aux:NNnn}
% \begin{macro}[aux]{\dt_remove_aux:NNnnnn}
% \begin{macro}[aux]{\dt_remove_aux:NNnnnnn}
%   Deleting a single entry from a single row means first splitting by row,
%   then splitting by key, and finally doing the assignment. If the row or the
%   key are not present then the entire function does nothing at all.
%    \begin{macrocode}
\cs_new_protected_nopar:Npn \dt_remove:Nnn { \dt_remove_aux:NNnn \tl_set:Nn }
\cs_new_protected_nopar:Npn \dt_gremove:Nnn { \dt_remove_aux:NNnn \tl_gset:Nn }
\cs_new_protected:Npn \dt_remove_aux:NNnn #1#2#3#4
  {
    \@@_split_row:NnTF #2 {#3}
      { \dt_remove_aux:NNnnnn #1 #2 {#4} }
      { }
  }
\cs_new_protected:Npn \dt_remove_aux:NNnnnn #1#2#3#4#5#6
  {
    \@@_split_key:nnTF {#5} {#3}
      { \dt_remove_aux:NNnnnnn #1 #2 {#4} {#6} }
      { }
  }
\cs_new_protected:Npn \dt_remove_aux:NNnnnnn #1#2#3#4#5#6#7
  { #1 #2 { #3 #5 #7 #4 } }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\dt_remove_key:Nn, \dt_gremove_key:Nn}
% \begin{macro}[aux]{\dt_remove_key_aux:NNn}
% \begin{macro}[aux]{\dt_remove_key_aux:nNNnn}
% \begin{macro}[aux, EXP]{\dt_remove_key_aux:w}
%   Deleting a key also removes from the table itself, so that there is no
%   need to do any awkward checks when extracting data from the table. (It's
%   likely that there will be more cases of accessing data than deleting
%   rows). The deletion mapping ignores rows entirely and just pulls out
%   matching key--value pairs, as this reduces the number of matches needed
%   to a minimum.
%    \begin{macrocode}
\cs_new_protected_nopar:Npn \dt_remove_key:Nn
  { \dt_remove_key_aux:NNn \tl_set:Nx }
\cs_new_protected_nopar:Npn \dt_gremove_key:Nn
  { \dt_remove_key_aux:NNn \tl_gset:Nx }
\cs_new_protected:Npn \dt_remove_key_aux:NNn #1#2#3
  {
    \@@_split_key_list:NnTF #2 {#3}
      { \exp_args:No \dt_remove_key_aux:nNNnn { \tl_to_str:n {#3} } #1 #2 }
      { }
  }
\cs_new_protected:Npn \dt_remove_key_aux:nNNnn #1#2#3#4#5
  { \dt_remove_key_aux:nNNnnwn {#1} #2 #3 #4 \q_stop {#5} }
\cs_new_protected:Npn \dt_remove_key_aux:nNNnnwn #1#2#3#4#5#6 \q_stop #7
  {
    \cs_set:Npn \dt_remove_key_aux:w ##1 \q_@@ #1 \q_@@ ##2 ##3
      {
        \exp_not:n {##1}
        \__quark_if_recursion_tail_break:nN {##3} \dt_map_break:
        \dt_remove_key_aux:w ##3
      }

    #2 #3
      {
        {#4}
        { \int_eval:n { #5 - \c_one } }
        \exp_not:n {#6}
        \dt_remove_key_aux:w #7 \q_@@ #1 \q_@@ { } \q_recursion_tail
          \__prg_break_point:Nn \dt_map_break: { }
      }
  }
\cs_new:Npn \dt_remove_key_aux:w { }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\dt_remove_row:Nn, \dt_gremove_row:Nn}
% \begin{macro}[aux]{\dt_remove_row_aux:NNn}
% \begin{macro}[aux]{\dt_remove_row_aux:NNnnnn}
% \begin{macro}[aux, EXP]{\dt_remove_row_aux:nw}
% \begin{macro}[aux, EXP]{\dt_remove_row_loop:nw}
%   Removing a row is a slightly complex operation as there are two stages.
%   The row itself is easy enough to remove, but then all later rows have to
%   be renumbers.
%    \begin{macrocode}
\cs_new_protected_nopar:Npn \dt_remove_row:Nn
  { \dt_remove_row_aux:NNn \tl_set:Nx }
\cs_new_protected_nopar:Npn \dt_gremove_row:Nn
  { \dt_remove_row_aux:NNn \tl_gset:Nx }
\cs_new_protected:Npn \dt_remove_row_aux:NNn #1#2#3
  {
    \@@_split_row:NnTF #2 {#3}
      { \dt_remove_row_aux:NNnnn #1 #2 }
      { }
  }
%    \end{macrocode}
%   If the code gets here, then |#3| is the table before the removed row,
%   |#4| is the removed row and |#5| is everything afterwards. The first stage
%   is to work out the new number of rows, then include all of |#3| except
%   the old number of rows. The removed row |#4| is thrown away, and then there
%   is a loop to recalculate the row numbers for all of the later rows.
%    \begin{macrocode}
\cs_new_protected:Npn \dt_remove_row_aux:NNnnn #1#2#3#4#5
  {
    #1 #2
      {
        { \int_eval:n { \dt_rows:N #2 - \c_one } }
        \dt_remove_row_aux:nw #3 \q_stop
        \dt_remove_row_loop:nw #5 \q_recursion_tail \q_@@_row
          \__prg_break_point:Nn \dt_map_break: { }
      }
  }
\cs_new_eq:NN \dt_remove_row_aux:nw \@@_add_row:nw
\cs_new:Npn \dt_remove_row_loop:nw #1#2 \q_@@_row
  {
    \__quark_if_recursion_tail_break:nN {#1} \dt_map_break:
    \int_eval:n { #1 - \c_one }
    \exp_not:n { #2 \q_@@_row }
    \dt_remove_row_loop:nw
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \subsection{Accessing data in data tables}
%
% \begin{macro}{\dt_get:NnnN}
% \begin{macro}[aux]{\dt_get_aux:nNnnn}
% \begin{macro}[aux]{\dt_get_aux:nNnnn}
%   Recovering a value from a row means doing two splits: first find the row,
%   then find the key. Nothing exciting, just a question of tracking the
%   returned items.
%    \begin{macrocode}
\cs_new_protected:Npn \dt_get:NnnN #1#2#3#4
  {
    \@@_split_row:NnTF #1 {#2}
      { \dt_get_aux:nNnnn {#3} #4 }
      { \tl_set:Nn #4 { \q_no_value } }
  }
\cs_new_protected:Npn \dt_get_aux:nNnnn #1#2#3#4#5
  {
    \@@_split_key:nnTF {#4} {#1}
      { \dt_get_aux:Nnnn #2 }
      { \tl_set:Nn #2 { \q_no_value } }
  }
\cs_new_protected:Npn \dt_get_aux:Nnnn #1#2#3#4 { \tl_set:Nn #1 {#3} }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}[TF]{\dt_get:NnnN}
% \begin{macro}[aux]{\@@_get_true:nNnnn}
% \begin{macro}[aux]{\@@_get_true:Nnnn}
%   The same idea as the standard method, but built as a conditional.
%    \begin{macrocode}
\prg_new_protected_conditional:Npnn \dt_get:NnnN #1#2#3#4 { T , F , TF }
  {
    \@@_split_row:NnTF #1 {#2}
      { \@@_get_true:nNnnn {#3} #4 }
      { \prg_return_false: }
  }
\cs_new_protected:Npn \@@_get_true:nNnnn #1#2#3#4#5
  {
    \@@_split_key:nnTF {#4} {#1}
      { \@@_get_true:Nnnn #2 }
      { \prg_return_false: }
  }
\cs_new_protected:Npn \@@_get_true:Nnnn #1#2#3#4
  {
    \tl_set:Nn #1 {#3}
    \prg_return_true:
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\dt_get:NnN}
% \begin{macro}[TF]{\dt_get:NnN}
%   Simple wrappers.
%    \begin{macrocode}
\cs_new_protected:Npn \dt_get:NnN #1
  { \dt_get:NnnN  #1 { \dt_rows:N #1 } }
\cs_new_protected:Npn \dt_get:NnNT #1
  { \dt_get:NnnNF  #1 { \dt_rows:N #1 } }
\cs_new_protected:Npn \dt_get:NnNF #1
  { \dt_get:NnnNF  #1 { \dt_rows:N #1 } }
\cs_new_protected:Npn \dt_get:NnNTF #1
  { \dt_get:NnnNTF #1 { \dt_rows:N #1 } }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \subsection{Mapping to data tables}
%
% \begin{variable}{\g_dt_map_level_int}
%   Unlike other mappings, the mapping level here has to be available
%   and so linked to the module.
%    \begin{macrocode}
\int_new:N \g_dt_map_level_int
%    \end{macrocode}
% \end{variable}
%
% \begin{macro}{\dt_map_variables:Nnn}
% \begin{macro}[aux]{\@@_map_variables_key:nn}
% \begin{macro}[aux]{\@@_map_variables:nnn}
% \begin{macro}[aux]{\@@_map_variables:nNNw}
% \begin{macro}[aux]{\@@_map_variables:nnw}
%   Mapping across a data table is more complex than other cases as there
%   are two \enquote{dimensions} to worry about: the rows and the keys.
%   The first stage of the mapping is to convert the key--variable mapping
%   into a sequence that can be used later. This is done with the assumption
%   that any key without a variable can simply be dropped entirely. The
%   header of the table is then split from the body.
%    \begin{macrocode}
\cs_new_protected:Npn \dt_map_variables:Nnn #1#2#3
  {
    \int_gincr:N \g_dt_map_level_int
    \seq_gclear_new:c { g_dt_map_ \int_use:N \g_dt_map_level_int _seq }
    \keyval_parse:NNn \use_none:n \@@_map_variables_key:nn {#2}
    \@@_split_header:NT #1 { \@@_map_variables:nnn {#3} }
  }
\cs_new_protected:Npn \@@_map_variables_key:nn #1#2
  {
    \seq_gput_right:cn { g_dt_map_ \int_use:N \g_dt_map_level_int _seq }
      { {#1} #2 }
  }
%    \end{macrocode}
%   As \cs{@@_split_header:NT} will leave a couple of tokens at the front
%   of the body part of the split, there is a quick piece of tidying up
%   to remove them.
%    \begin{macrocode}
\cs_new_protected:Npn \@@_map_variables:nnn #1#2#3
  { \@@_map_variables:nNNw {#1} #3 \q_stop }
\cs_new_protected:Npn \@@_map_variables:nNNw
  #1 \q_@@_header \q_@@_row #2 \q_stop
  {
    \int_zero_new:c { l_dt_map_ \int_use:N \g_dt_map_level_int _row_int }
    \@@_map_variables:nnw {#1} #2 { } \q_recursion_tail \q_@@_row
      \__prg_break_point:Nn \dt_map_break:
        { \int_gdecr:N \g_dt_map_level_int }
  }
\cs_new_protected:Npn \@@_map_variables:nnw #1#2#3#4 \q_@@_row
  {
    \__quark_if_recursion_tail_break:nN {#3} \dt_map_break:
    \seq_map_inline:cn { g_dt_map_ \int_use:N \g_dt_map_level_int _seq }
      { \dt_get_aux:nNnnn ##1 { } {#3#4} { } }
    #1
    \int_incr:c { l_dt_map_ \int_use:N \g_dt_map_level_int _row_int }
    \@@_map_variables:nnw {#1}
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}[rEXP]{\dt_map_break:}
% \begin{macro}[rEXP]{\dt_map_break:n}
%   The break statements use the general \cs{__prg_map_break:Nn}.
%    \begin{macrocode}
\cs_new_nopar:Npn \dt_map_break:
  { \__prg_map_break:Nn \dt_map_break: { } }
\cs_new_nopar:Npn \dt_map_break:n
  { \__prg_map_break:Nn \dt_map_break: }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \subsection{Data table conditionals}
%
% \begin{macro}[pTF, EXP]{\dt_if_empty:N}
%   An empty data table has not only no rows but also no keys. (The number of
%   rows can be tested using \cs{dt_rows:N} and an \texttt{int} test.)
%    \begin{macrocode}
\prg_new_conditional:Npnn \dt_if_empty:N #1 { T , F , TF , p }
  {
    \if_meaning:w #1 \c_empty_dt
      \prg_return_true:
    \else:
      \prg_return_false:
    \fi:
  }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[pTF, EXP]{\dt_if_in:Nn}
% \begin{macro}[aux, EXP]{\@@_if_in:nnn}
% \begin{macro}[aux, EXP]{\@@_if_in:nwN}
% \begin{macro}[aux, EXP]{\@@_if_in:n}
%   Expandably checking for the presence of a key in the table as a whole
%   requires a mapping to the header. The idea is the usual recursion set
%   up with a string-based comparison only after checking for the end of
%   the loop.
%    \begin{macrocode}
\prg_new_conditional:Npnn \dt_if_in:Nn #1#2 { p , T , F , TF }
  { \@@_split_header:NT #1 { \@@_if_in:nnn {#2} } }
\cs_new:Npn \@@_if_in:nnn #1#2#3
  {
    \exp_last_unbraced:Nno \@@_if_in:nwN {#1} { \use_none:nn #2 }
      \q_recursion_tail \q_@@
      \__prg_break_point:
  }
\cs_new:Npn \@@_if_in:nwN #1#2 \q_@@
  {
    \if_meaning:w \q_recursion_tail #2
      \exp_after:wN \__prg_break:n
    \else:
      \exp_after:wN \use_none:n
    \fi:
      { \prg_return_false: }
    \str_if_eq:nnTF {#1} {#2}
      { \__prg_break:n { \prg_return_true: } }
      { \@@_if_in:nwN {#1} }
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}[pTF, EXP]{\dt_if_in_row:Nnn}
% \begin{macro}[aux, EXP]{\@@_if_in_row:nw}
% \begin{macro}[aux, EXP]{\@@_if_in_row:nn}
% \begin{macro}[aux, EXP]{\@@_if_in_row:nwn}
% \begin{macro}[aux, EXP]{\@@_if_in_row:N}
%   Finding a key in a single row in an expandable way requires two mappings.
%   To start of with, there is a search for the row. This uses for termination
%   the fact that each row starts \cs{q_@@_row} and ends \cs{q_nil}, and
%   always contains at least the row number as the first \meta{balanced
%   text}. That can be replaced by the tail marker to terminate iteration:
%   all that is then needed is the correct placement of the clean-up code.
%    \begin{macrocode}
\prg_new_conditional:Npnn \dt_if_in_row:Nnn #1#2#3 { p , T , F , TF }
  {
    \exp_last_unbraced:Nno \@@_if_in_row:nw {#2} #1
      \q_recursion_tail \q_nil
      \__prg_break_point:
      { \tl_to_str:n {#3} }
  }
%    \end{macrocode}
%   The row iteration does a numerical comparison to see if the target row has
%   been found. That means that the row argument does not need to be converted
%   to a number earlier.
%    \begin{macrocode}
\cs_new:Npn \@@_if_in_row:nw #1#2 \q_@@_row #3#4 \q_nil
  {
    \if_meaning:w \q_recursion_tail #3
      \exp_after:wN \__prg_break:n
    \else:
      \exp_after:wN \use_none:n
    \fi:
      {
        \use_i:nn
        \prg_return_false:
      }
    \int_compare:nNnTF {#1} = {#3}
      { \__prg_break:n { \exp_args:Nno \@@_if_in_row:nn {#4} } }
      { \@@_if_in_row:nw {#1} }
  }
%    \end{macrocode}
%   The second iteration is along the row. This is basically the same as
%   \cs{prop_if_in:NnTF} with the \cs{q_@@} in place of \cs{q__prop}.
%    \begin{macrocode}
\cs_new:Npn \@@_if_in_row:nn #1#2
  {
    \@@_if_in_row:nwn {#2} #1 {#2} \q_@@ { } \q_recursion_tail
      \__prg_break_point:
  }
\cs_new:Npn \@@_if_in_row:nwn #1 \q_@@ #2 \q_@@ #3
  {
    \str_if_eq_x:nnTF {#1} {#2}
      { \@@_if_in_row:N }
      { \@@_if_in_row:nwn {#1} }
  }
\cs_new:Npn \@@_if_in_row:N #1
  {
    \if_meaning:w \q_@@ #1
      \prg_return_true:
    \else:
      \prg_return_false:
    \fi:
    \__prg_break:
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}[pTF, EXP]{\dt_if_in_row:Nn}
%   Simple wrappers.
%    \begin{macrocode}
\cs_new:Npn \dt_if_in_row_p:Nn #1
  { \dt_if_in_row_p:Nnn #1 { \dt_rows:N #1 } }
\cs_new:Npn \dt_if_in_row:NnT  #1
  { \dt_if_in_row:NnnT  #1 { \dt_rows:N #1 } }
\cs_new:Npn \dt_if_in_row:NnF  #1
  { \dt_if_in_row:NnnF  #1 { \dt_rows:N #1 } }
\cs_new:Npn \dt_if_in_row:NnTF #1
  { \dt_if_in_row:NnnTF #1 { \dt_rows:N #1 } }
%    \end{macrocode}
% \end{macro}
%
% \subsection{Messages}
%
%    \begin{macrocode}
\__msg_kernel_new:nnnn { dt } { unknown-row }
  { Data~table~#1~does~not~contain~a~row~'#2'. }
  {
    Data~table~#1~contains~#3~rows.~These~must~be~accessed~by~number:~row~
    #2~is~not~present~in~the~table.
  }
%    \end{macrocode}
%
%    \begin{macrocode}
%</initex|package>
%    \end{macrocode}
%
% \end{implementation}
%
% \PrintIndex