summaryrefslogtreecommitdiff
path: root/macros/latex/contrib/l3kernel/l3expan.dtx
blob: 163e867881d742a5e552eb009220a9ded5d092fc (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
% \iffalse meta-comment
%
%% File: l3expan.dtx
%
% Copyright (C) 1990-2021 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{l3expan} package\\ Argument expansion^^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 2021-05-11}
%
% \maketitle
%
% \begin{documentation}
%
% This module provides generic methods for expanding \TeX{} arguments in a
% systematic manner. The functions in this module all have prefix |exp|.
%
% Not all possible variations are implemented for every base
% function. Instead only those that are used within the \LaTeX3 kernel
% or otherwise seem to be of general interest are implemented.
% Consult the module description to find out which functions are
% actually defined. The next section explains how to define missing
% variants.
%
% \section{Defining new variants}
% \label{sec:l3expan:defining-variants}
%
% The definition of variant forms for base functions may be necessary
% when writing new functions or when applying a kernel function in a
% situation that we haven't thought of before.
%
% Internally preprocessing of arguments is done with functions of the form
% \cs[no-index]{exp_\ldots{}}.  They all look alike, an example would be
% \cs{exp_args:NNo}. This function has three arguments, the first and the
% second are a single tokens, while the third argument should be given
% in braces. Applying \cs{exp_args:NNo} expands the content of third
% argument once before any expansion of the first and second arguments.
% If \cs{seq_gpush:No} was not defined it could be coded in the following way:
% \begin{verbatim}
%   \exp_args:NNo \seq_gpush:Nn
%      \g_file_name_stack
%      { \l_tmpa_tl }
% \end{verbatim}
% In other words, the first argument to \cs{exp_args:NNo} is the base
% function and the other arguments are preprocessed and then passed to
% this base function. In the example the first argument to the base
% function should be a single token which is left unchanged while the
% second argument is expanded once. From this example we can also see
% how the variants are defined. They just expand into the appropriate
% |\exp_| function followed by the desired base function, \emph{e.g.}
% \begin{quote}
%   |\cs_generate_variant:Nn \seq_gpush:Nn { No } |
% \end{quote}
% results in the definition of |\seq_gpush:No|
% \begin{quote}
%   |\cs_new:Npn \seq_gpush:No { \exp_args:NNo \seq_gpush:Nn }|
% \end{quote}
% Providing variants in this way in style files is safe as the
% \cs{cs_generate_variant:Nn} function will only create new definitions if
% there is not already one available. Therefore adding
% such definition to later releases of the kernel will not make such
% style files obsolete.
%
% The steps above may be automated by using the function
% \cs{cs_generate_variant:Nn}, described next.
%
% \section{Methods for defining variants}
% \label{sec:l3expan:variants-method}
%
% We recall the set of available argument specifiers.
% \begin{itemize}
% \item |N|~is used for single-token arguments while |c|~constructs a
%   control sequence from its name and passes it to a parent function as
%   an |N|-type argument.
% \item Many argument types extract or expand some tokens and provide it
%   as an |n|-type argument, namely a braced multiple-token argument:
%   |V|~extracts the value of a variable, |v|~extracts the value from
%   the name of a variable, |n|~uses the argument as it is, |o|~expands
%   once, |f|~expands fully the front of the token list, |e| and
%   |x|~expand fully all tokens (differences are explained later).
% \item A few odd argument types remain: |T|~and |F|~for conditional
%   processing, otherwise identical to |n|-type arguments,
%   |p|~for the parameter text
%   in definitions, |w|~for arguments with a specific syntax, and |D|~to
%   denote primitives that should not be used directly.
% \end{itemize}
%
% \begin{function}[updated = 2017-11-28]
%   {\cs_generate_variant:Nn, \cs_generate_variant:cn}
%   \begin{syntax}
%     \cs{cs_generate_variant:Nn} \meta{parent control sequence} \Arg{variant argument specifiers}
%   \end{syntax}
%   This function is used to define argument-specifier variants of the
%   \meta{parent control sequence} for \LaTeX3 code-level macros. The
%   \meta{parent control sequence} is first separated into the
%   \meta{base name} and \meta{original argument specifier}. The
%   comma-separated list of \meta{variant argument specifiers} is
%   then used to define variants of the
%   \meta{original argument specifier} if these are not already
%   defined. For each \meta{variant} given, a function is created
%   that expands its arguments as detailed and passes them
%   to the \meta{parent control sequence}. So for example
%   \begin{verbatim}
%     \cs_set:Npn \foo:Nn #1#2 { code here }
%     \cs_generate_variant:Nn \foo:Nn { c }
%   \end{verbatim}
%   creates a new function |\foo:cn| which expands its first
%   argument into a control sequence name and passes the result to
%   |\foo:Nn|. Similarly
%   \begin{verbatim}
%     \cs_generate_variant:Nn \foo:Nn { NV , cV }
%   \end{verbatim}
%   generates the functions |\foo:NV| and |\foo:cV| in the same
%   way. The \cs{cs_generate_variant:Nn} function can only be applied if
%   the \meta{parent control sequence} is already defined. If the \meta{parent
%   control sequence} is protected or if the \meta{variant} involves any
%   |x|~argument, then the \meta{variant control sequence} is also
%   protected.  The \meta{variant} is created globally, as is any
%   \cs[no-index]{exp_args:N\meta{variant}} function needed to carry out the
%   expansion.
%
%   Only |n|~and |N| arguments can be changed to other types.  The only
%   allowed changes are
%   \begin{itemize}
%   \item |c|~variant of an |N|~parent;
%   \item |o|, |V|, |v|, |f|, |e|, or~|x| variant of an |n|~parent;
%   \item |N|, |n|, |T|, |F|, or |p| argument unchanged.
%   \end{itemize}
%   This means the \meta{parent} of a \meta{variant} form is always
%   unambiguous, even in cases where both an |n|-type parent and an
%   |N|-type parent exist, such as for \cs{tl_count:n} and
%   \cs{tl_count:N}.
%
%   For backward compatibility it is currently possible to make |n|,
%   |o|, |V|, |v|, |f|, |e|, or |x|-type variants of an |N|-type argument or
%   |N| or |c|-type variants of an |n|-type argument.  Both are
%   deprecated.  The first because passing more than one token to an
%   |N|-type argument will typically break the parent function's code.
%   The second because programmers who use that most often want to
%   access the value of a variable given its name, hence should use a
%   |V|-type or |v|-type variant instead of |c|-type.  In those cases,
%   using the lower-level \cs{exp_args:No} or \cs{exp_args:Nc}
%   functions explicitly is preferred to defining confusing variants.
% \end{function}
%
% \section{Introducing the variants}
%
% The |V| type returns the value of a register, which can be one of
% |tl|, |clist|, |int|, |skip|, |dim|, |muskip|, or built-in \TeX{}
% registers. The |v| type is the same except it first creates a
% control sequence out of its argument before returning the
% value.
%
% In general, the programmer should not need to be concerned with
% expansion control. When simply using the content of a variable,
% functions with a |V| specifier should be used. For those referred to by
% (cs)name, the |v| specifier is available for the same purpose. Only when
% specific expansion steps are needed, such as when using delimited
% arguments, should the lower-level functions with |o| specifiers be employed.
%
% The |e| type expands all tokens fully, starting from the first.  More
% precisely the expansion is identical to that of \TeX{}'s \tn{message}
% (in particular |#| needs not be doubled).  It was added in May 2018.
% In recent enough engines (starting around 2019) it relies on the
% primitive \tn{expanded} hence is fast.  In older engines it is very
% much slower.  As a result it should only be used in performance
% critical code if typical users will have a recent installation of the
% \TeX{} ecosystem.
%
% The |x| type expands all tokens fully, starting from the first.  In
% contrast to |e|, all macro parameter characters |#| must be doubled,
% and omitting this leads to low-level errors.  In addition this type of
% expansion is not expandable, namely functions that have |x| in their
% signature do not themselves expand when appearing inside |x| or |e|
% expansion.
%
% The |f| type is so special that it deserves an example.  It is
% typically used in contexts where only expandable commands are allowed.
% Then |x|-expansion cannot be used, and |f|-expansion provides an
% alternative that expands the front of the token list
% as much as can be done in such contexts.  For
% instance, say that we want to evaluate the integer expression $3 + 4$
% and pass the result $7$ as an argument to an expandable function
% |\example:n|.  For this, one should define a variant using
% \cs{cs_generate_variant:Nn} |\example:n| |{| |f| |}|, then do
% \begin{quote}
%   |\example:f { \int_eval:n { 3 + 4 } }|
% \end{quote}
% Note that |x|-expansion would also expand \cs{int_eval:n} fully to its
% result~$7$, but the variant |\example:x| cannot be expandable.  Note
% also that |o|-expansion would not expand \cs{int_eval:n} fully to its
% result since that function requires several expansions.  Besides the
% fact that |x|-expansion is protected rather than expandable, another
% difference between |f|-expansion and |x|-expansion is that
% |f|-expansion expands tokens from the beginning and stops as soon as a
% non-expandable token is encountered, while |x|-expansion continues
% expanding further tokens.  Thus, for instance
% \begin{quote}
%   |\example:f { \int_eval:n { 1 + 2 } , \int_eval:n { 3 + 4 } }|
% \end{quote}
% results in the call
% \begin{quote}
%   |\example:n { 3 , \int_eval:n { 3 + 4 } }|
% \end{quote}
% while using |\example:x| or |\example:e| instead results in
% \begin{quote}
%   |\example:n { 3 , 7 }|
% \end{quote}
% at the cost of being protected (for |x| type) or very much slower in
% old engines (for |e| type).
% If you use |f| type expansion in conditional processing then
% you should stick to using |TF|  type functions only as the expansion
% does not finish any |\if... \fi:| itself!
%
% It is important to note that both \texttt{f}- and \texttt{o}-type
% expansion are concerned with the expansion of tokens from left to
% right in their arguments. In particular, \texttt{o}-type expansion
% applies to the first \emph{token} in the argument it receives: it
% is conceptually similar to
% \begin{verbatim}
%   \exp_after:wN <base function> \exp_after:wN { <argument> }
% \end{verbatim}
% At the same time, \texttt{f}-type expansion stops at the \emph{first}
% non-expandable token. This means for example that both
% \begin{verbatim}
%    \tl_set:No \l_tmpa_tl { { \g_tmpb_tl } }
% \end{verbatim}
% and
% \begin{verbatim}
%    \tl_set:Nf \l_tmpa_tl { { \g_tmpb_tl } }
% \end{verbatim}
% leave |\g_tmpb_tl| unchanged: |{| is the first token in the
% argument and is non-expandable.
%
% It is usually best to keep the following in mind when using variant
% forms.
% \begin{itemize}
%   \item
%     Variants with |x|-type arguments (that are fully expanded before
%     being passed to the |n|-type base function) are never expandable
%     even when the base function is.  Such variants cannot work
%     correctly in arguments that are themselves subject to expansion.
%     Consider using |f| or |e| expansion.
%   \item
%     In contrast, |e|~expansion (full expansion, almost like~|x| except
%     for the treatment of~|#|) does not prevent variants from being
%     expandable (if the base function is).  The drawback is that
%     |e|~expansion is very much slower in old engines (before 2019).
%     Consider using |f|~expansion if that type of expansion is
%     sufficient to perform the required expansion, or |x|~expansion if
%     the variant will not itself need to be expandable.
%   \item
%     Finally |f|~expansion only expands the front of the token list,
%     stopping at the first non-expandable token.  This may fail to
%     fully expand the argument.
% \end{itemize}
%
% When speed is essential (for functions that do very little work and
% whose variants are used numerous times in a document) the following
% considerations apply because internal functions for argument expansion
% come in two flavours, some faster than others.
% \begin{itemize}
%   \item
%     Arguments that might need expansion should come first in the list
%     of arguments.
%   \item
%     Arguments that should consist of single tokens |N|, |c|, |V|, or
%     |v| should come first among these.
%   \item
%     Arguments that appear after the first multi-token argument |n|,
%     |f|, |e|, or |o| require slightly slower special processing to be
%     expanded.  Therefore it is best to use the optimized functions,
%     namely those that contain only |N|, |c|, |V|, and |v|, and, in the
%     last position, |o|, |f|, |e|, with possible trailing |N| or |n| or
%     |T| or |F|, which are not expanded.  Any |x|-type argument causes
%     slightly slower processing.
% \end{itemize}
%
% \section{Manipulating the first argument}
%
% These functions are described in detail: expansion of multiple tokens follows
% the same rules but is described in a shorter fashion.
%
% \begin{function}[EXP]{\exp_args:Nc, \exp_args:cc}
%   \begin{syntax}
%     \cs{exp_args:Nc} \meta{function} \Arg{tokens}
%   \end{syntax}
%   This function absorbs two arguments (the \meta{function} name and
%   the \meta{tokens}). The \meta{tokens} are expanded until only characters
%   remain, and are then turned into a control sequence.
%   The result is inserted into the input stream \emph{after} reinsertion
%   of the \meta{function}. Thus the \meta{function} may take more than
%   one argument: all others are left unchanged.
%
%   The |:cc| variant constructs the \meta{function} name in the same
%   manner as described for the \meta{tokens}.
%   \begin{texnote}
%     Protected macros that appear in a \texttt{c}-type argument are
%     expanded despite being protected; \cs{exp_not:n} also has no
%     effect.  An internal error occurs if non-characters or active
%     characters remain after full expansion, as the conversion to a
%     control sequence is not possible.
%   \end{texnote}
% \end{function}
%
% \begin{function}[EXP]{\exp_args:No}
%   \begin{syntax}
%     \cs{exp_args:No} \meta{function} \Arg{tokens} ...
%   \end{syntax}
%   This function absorbs two arguments (the \meta{function} name and
%   the \meta{tokens}). The \meta{tokens} are expanded once, and the result
%   is inserted in braces into the input stream \emph{after} reinsertion
%   of the \meta{function}. Thus the \meta{function} may take more than
%   one argument: all others are left unchanged.
% \end{function}
%
% \begin{function}[EXP]{\exp_args:NV}
%   \begin{syntax}
%     \cs{exp_args:NV} \meta{function} \meta{variable}
%   \end{syntax}
%   This function absorbs two arguments (the names of the \meta{function} and
%   the \meta{variable}). The content of the \meta{variable} are recovered
%   and placed inside braces into the input stream \emph{after} reinsertion
%   of the \meta{function}. Thus the \meta{function} may take more than
%   one argument: all others are left unchanged.
% \end{function}
%
% \begin{function}[EXP]{\exp_args:Nv}
%   \begin{syntax}
%     \cs{exp_args:Nv} \meta{function} \Arg{tokens}
%   \end{syntax}
%   This function absorbs two arguments (the \meta{function} name and
%   the \meta{tokens}). The \meta{tokens} are expanded until only characters
%   remain, and are then turned into a control sequence.
%   This control sequence should
%   be the name of a \meta{variable}.   The content of the \meta{variable} are
%   recovered and placed inside braces into the input stream \emph{after}
%   reinsertion of the \meta{function}. Thus the \meta{function} may take more
%   than one argument: all others are left unchanged.
%   \begin{texnote}
%     Protected macros that appear in a \texttt{v}-type argument are
%     expanded despite being protected; \cs{exp_not:n} also has no
%     effect.  An internal error occurs if non-characters or active
%     characters remain after full expansion, as the conversion to a
%     control sequence is not possible.
%   \end{texnote}
% \end{function}
%
% \begin{function}[EXP, added = 2018-05-15]{\exp_args:Ne}
%   \begin{syntax}
%     \cs{exp_args:Ne} \meta{function} \Arg{tokens}
%   \end{syntax}
%   This function absorbs two arguments (the \meta{function} name and
%   the \meta{tokens}) and exhaustively expands the \meta{tokens}.
%   The result is inserted in braces into the input stream
%   \emph{after} reinsertion of the \meta{function}.
%   Thus the \meta{function} may take more
%   than one argument: all others are left unchanged.
%   \begin{texnote}
%     This relies on the \tn{expanded} primitive when available (in
%     \LuaTeX{} and starting around 2019 in other engines).  Otherwise
%     it uses some fall-back code that is very much slower.  As a result
%     it should only be used in performance-critical code if typical
%     users have a recent installation of the \TeX{} ecosystem.
%   \end{texnote}
% \end{function}
%
% \begin{function}[EXP]{\exp_args:Nf}
%   \begin{syntax}
%     \cs{exp_args:Nf} \meta{function} \Arg{tokens}
%   \end{syntax}
%   This function absorbs two arguments (the \meta{function} name and
%   the \meta{tokens}). The \meta{tokens} are fully expanded until the
%   first non-expandable token is found (if that is a space it is
%   removed), and the result
%   is inserted in braces into the input stream \emph{after} reinsertion
%   of the \meta{function}. Thus the \meta{function} may take more than
%   one argument: all others are left unchanged.
% \end{function}
%
% \begin{function}{\exp_args:Nx}
%   \begin{syntax}
%     \cs{exp_args:Nx} \meta{function} \Arg{tokens}
%   \end{syntax}
%   This function absorbs two arguments (the \meta{function} name and
%   the \meta{tokens}) and exhaustively expands the \meta{tokens}.
%   The result is inserted in braces into the input stream
%   \emph{after} reinsertion of the \meta{function}.
%   Thus the \meta{function} may take more
%   than one argument: all others are left unchanged.
% \end{function}
%
% \section{Manipulating two arguments}
%
% \begin{function}[EXP, updated = 2018-05-15]
%   {
%     \exp_args:NNc,
%     \exp_args:NNo,
%     \exp_args:NNV,
%     \exp_args:NNv,
%     \exp_args:NNe,
%     \exp_args:NNf,
%     \exp_args:Ncc,
%     \exp_args:Nco,
%     \exp_args:NcV,
%     \exp_args:Ncv,
%     \exp_args:Ncf,
%     \exp_args:NVV
%   }
%   \begin{syntax}
%     \cs{exp_args:NNc} \meta{token_1} \meta{token_2} \Arg{tokens}
%   \end{syntax}
%   These optimized functions absorb three arguments and expand the second and
%   third as detailed by their argument specifier. The first argument
%   of the function is then the next item on the input stream, followed
%   by the expansion of the second and third arguments.
% \end{function}
%
% \begin{function}[EXP, updated = 2018-05-15]
%   {
%     \exp_args:Nnc,
%     \exp_args:Nno,
%     \exp_args:NnV,
%     \exp_args:Nnv,
%     \exp_args:Nne,
%     \exp_args:Nnf,
%     \exp_args:Noc,
%     \exp_args:Noo,
%     \exp_args:Nof,
%     \exp_args:NVo,
%     \exp_args:Nfo,
%     \exp_args:Nff,
%     \exp_args:Nee,
%   }
%   \begin{syntax}
%     \cs{exp_args:Noo} \meta{token} \Arg{tokens_1} \Arg{tokens_2}
%   \end{syntax}
%   These functions absorb three arguments and expand the second and
%   third as detailed by their argument specifier. The first argument
%   of the function is then the next item on the input stream, followed
%   by the expansion of the second and third arguments.
%   These functions need slower processing.
% \end{function}
%
% \begin{function}
%   {
%     \exp_args:NNx,
%     \exp_args:Ncx,
%     \exp_args:Nnx,
%     \exp_args:Nox,
%     \exp_args:Nxo,
%     \exp_args:Nxx
%   }
%   \begin{syntax}
%     \cs{exp_args:NNx} \meta{token_1} \meta{token_2} \Arg{tokens}
%   \end{syntax}
%   These functions absorb three arguments and expand the second and
%   third as detailed by their argument specifier. The first argument
%   of the function is then the next item on the input stream, followed
%   by the expansion of the second and third arguments. These functions
%   are not expandable due to their |x|-type argument.
% \end{function}
%
% \section{Manipulating three arguments}
%
% \begin{function}[EXP]
%   {
%     \exp_args:NNNo,
%     \exp_args:NNNV,
%     \exp_args:NNNv,
%     \exp_args:Nccc,
%     \exp_args:NcNc,
%     \exp_args:NcNo,
%     \exp_args:Ncco
%   }
%   \begin{syntax}
%     \cs{exp_args:NNNo} \meta{token_1} \meta{token_2} \meta{token_3} \Arg{tokens}
%   \end{syntax}
%   These optimized functions absorb four arguments and expand the second, third
%   and fourth as detailed by their argument specifier. The first
%   argument of the function is then the next item on the input stream,
%   followed by the expansion of the second argument, \emph{etc}.
% \end{function}
%
% \begin{function}[EXP]
%   {
%     \exp_args:NNcf,
%     \exp_args:NNno,
%     \exp_args:NNnV,
%     \exp_args:NNoo,
%     \exp_args:NNVV,
%     \exp_args:Ncno,
%     \exp_args:NcnV,
%     \exp_args:Ncoo,
%     \exp_args:NcVV,
%     \exp_args:Nnnc,
%     \exp_args:Nnno,
%     \exp_args:Nnnf,
%     \exp_args:Nnff,
%     \exp_args:Nooo,
%     \exp_args:Noof,
%     \exp_args:Nffo,
%     \exp_args:Neee
%   }
%   \begin{syntax}
%     \cs{exp_args:NNoo} \meta{token_1} \meta{token_2} \Arg{token_3} \Arg{tokens}
%   \end{syntax}
%   These functions absorb four arguments and expand the second, third
%   and fourth as detailed by their argument specifier. The first
%   argument of the function is then the next item on the input stream,
%   followed by the expansion of the second argument, \emph{etc}.
%   These functions need slower processing.
% \end{function}
%
% \begin{function}[added = 2015-08-12]
%   {
%     \exp_args:NNNx,
%     \exp_args:NNnx,
%     \exp_args:NNox,
%     \exp_args:Nccx,
%     \exp_args:Ncnx,
%     \exp_args:Nnnx,
%     \exp_args:Nnox,
%     \exp_args:Noox,
%   }
%   \begin{syntax}
%     \cs{exp_args:NNnx} \meta{token_1} \meta{token_2} \Arg{tokens_1} \Arg{tokens_2}
%   \end{syntax}
%   These functions absorb four arguments and expand the second, third
%   and fourth as detailed by their argument specifier. The first
%   argument of the function is then the next item on the input stream,
%   followed by the expansion of the second argument, \emph{etc.}
% \end{function}
%
% \section{Unbraced expansion}
%
% \begin{function}[EXP, updated = 2018-05-15]
%   {
%     \exp_last_unbraced:No,
%     \exp_last_unbraced:NV,
%     \exp_last_unbraced:Nv,
%     \exp_last_unbraced:Ne,
%     \exp_last_unbraced:Nf,
%     \exp_last_unbraced:NNo,
%     \exp_last_unbraced:NNV,
%     \exp_last_unbraced:NNf,
%     \exp_last_unbraced:Nco,
%     \exp_last_unbraced:NcV,
%     \exp_last_unbraced:Nno,
%     \exp_last_unbraced:Noo,
%     \exp_last_unbraced:Nfo,
%     \exp_last_unbraced:NNNo,
%     \exp_last_unbraced:NNNV,
%     \exp_last_unbraced:NNNf,
%     \exp_last_unbraced:NnNo,
%     \exp_last_unbraced:NNNNo,
%     \exp_last_unbraced:NNNNf,
%   }
%   \begin{syntax}
%     \cs{exp_last_unbraced:Nno} \meta{token} \Arg{tokens_1} \Arg{tokens_2}
%   \end{syntax}
%   These functions absorb the number of arguments given by their
%   specification, carry out the expansion
%   indicated and leave the results in the input stream, with the
%   last argument not surrounded by the usual braces.
%   Of these, the |:Nno|, |:Noo|, |:Nfo| and |:NnNo|
%   variants need slower processing.
%   \begin{texnote}
%     As an optimization, the last argument is unbraced by some
%     of those functions before expansion. This can cause problems
%     if the argument is empty: for instance,
%     \cs{exp_last_unbraced:Nf} |\foo_bar:w| |{ }| \cs{q_stop}
%     leads to an infinite loop, as the quark is \texttt{f}-expanded.
%   \end{texnote}
% \end{function}
%
% \begin{function}{\exp_last_unbraced:Nx}
%   \begin{syntax}
%     \cs{exp_last_unbraced:Nx} \meta{function} \Arg{tokens}
%   \end{syntax}
%   This function fully expands the \meta{tokens} and leaves the result
%   in the input stream after reinsertion of the \meta{function}.
%   This function is not expandable.
% \end{function}
%
% \begin{function}[EXP]{\exp_last_two_unbraced:Noo}
%   \begin{syntax}
%     \cs{exp_last_two_unbraced:Noo} \meta{token} \Arg{tokens_1} \Arg{tokens_2}
%   \end{syntax}
%   This function absorbs three arguments and expands the second and third
%   once. The first argument of the function is then the next item on the
%   input stream, followed by the expansion of the second and third arguments,
%   which are not wrapped in braces.
%   This function needs special (slower) processing.
% \end{function}
%
% \begin{function}[EXP]{\exp_after:wN}
%   \begin{syntax}
%     \cs{exp_after:wN} \meta{token_1} \meta{token_2}
%   \end{syntax}
%   Carries out a single expansion of \meta{token_2} (which may consume
%   arguments) prior to the expansion of \meta{token_1}. If \meta{token_2} has
%   no expansion (for example, if it is a character) then it is left
%   unchanged. It is important to notice that \meta{token_1} may be
%   \emph{any} single token, including group-opening and -closing
%   tokens (|{| or |}| assuming normal \TeX{} category codes). Unless
%   specifically required this should be avoided: expansion should be carried out using an
%   appropriate argument specifier variant or the appropriate
%   \cs[no-index]{exp_arg:N} function.
%   \begin{texnote}
%     This is the \TeX{} primitive \tn{expandafter} renamed.
%   \end{texnote}
% \end{function}
%
%
% \section{Preventing expansion}
%
% Despite the fact that the following functions are all about preventing
% expansion, they're designed to be used in an expandable context and hence
% are all marked as being `expandable' since they themselves disappear
% after the expansion has completed.
%
% \begin{function}[EXP]{\exp_not:N}
%   \begin{syntax}
%     \cs{exp_not:N} \meta{token}
%   \end{syntax}
%   Prevents expansion of the \meta{token} in a context where it would
%   otherwise be expanded, for example an |x|-type argument or the first
%   token in an |o| or |e| or |f| argument.
%   \begin{texnote}
%     This is the \TeX{} \tn{noexpand} primitive.  It only prevents
%     expansion.  At the beginning of an |f|-type argument, a space
%     \meta{token} is removed even if it appears as \cs{exp_not:N}
%     \cs{c_space_token}.  In an |x|-expanding definition
%     (\cs{cs_new:Npx}), a macro parameter introduces an argument even
%     if it appears as \cs{exp_not:N} |#| |1|.  This differs from
%     \cs{exp_not:n}.
%   \end{texnote}
% \end{function}
%
% \begin{function}[EXP]{\exp_not:c}
%   \begin{syntax}
%     \cs{exp_not:c} \Arg{tokens}
%   \end{syntax}
%   Expands the \meta{tokens} until only characters remain, and then
%   converts this into a control sequence.
%   Further expansion of this control sequence is then inhibited using
%   \cs{exp_not:N}.
%   \begin{texnote}
%     Protected macros that appear in a \texttt{c}-type argument are
%     expanded despite being protected; \cs{exp_not:n} also has no
%     effect.  An internal error occurs if non-characters or active
%     characters remain after full expansion, as the conversion to a
%     control sequence is not possible.
%   \end{texnote}
% \end{function}
%
% \begin{function}[EXP]{\exp_not:n}
%   \begin{syntax}
%     \cs{exp_not:n} \Arg{tokens}
%   \end{syntax}
%   Prevents expansion of the \meta{tokens} in an |e| or |x|-type argument.  In
%   all other cases the \meta{tokens} continue to be expanded, for
%   example in the input stream or in other types of arguments such as
%   \texttt{c}, \texttt{f}, \texttt{v}.  The argument of \cs{exp_not:n}
%   \emph{must} be surrounded by braces.
%   \begin{texnote}
%     This is the \eTeX{} \tn{unexpanded} primitive.  In an
%     |x|-expanding definition (\cs{cs_new:Npx}), \cs{exp_not:n}~|{#1}|
%     is equivalent to |##1| rather than to~|#1|, namely it inserts the
%     two characters |#| and~|1|.  In an |e|-type argument
%     \cs{exp_not:n}~|{#}| is equivalent to |#|, namely it inserts the
%     character~|#|.
%   \end{texnote}
% \end{function}
%
% \begin{function}[EXP]{\exp_not:o}
%   \begin{syntax}
%     \cs{exp_not:o} \Arg{tokens}
%   \end{syntax}
%   Expands the \meta{tokens} once, then prevents any further expansion
%   in |x|-type or \texttt{e}-type arguments using \cs{exp_not:n}.
% \end{function}
%
% \begin{function}[EXP]{\exp_not:V}
%   \begin{syntax}
%     \cs{exp_not:V} \meta{variable}
%   \end{syntax}
%   Recovers the content of the \meta{variable}, then prevents expansion
%   of this material in |x|-type or \texttt{e}-type arguments using \cs{exp_not:n}.
% \end{function}
%
% \begin{function}[EXP]{\exp_not:v}
%   \begin{syntax}
%     \cs{exp_not:v} \Arg{tokens}
%   \end{syntax}
%   Expands the \meta{tokens} until only characters remains, and then
%   converts this into a control sequence which should be a \meta{variable}
%   name.
%   The content of the \meta{variable} is recovered, and further
%   expansion in |x|-type or \texttt{e}-type arguments is prevented using \cs{exp_not:n}.
%   \begin{texnote}
%     Protected macros that appear in a \texttt{v}-type argument are
%     expanded despite being protected; \cs{exp_not:n} also has no
%     effect.  An internal error occurs if non-characters or active
%     characters remain after full expansion, as the conversion to a
%     control sequence is not possible.
%   \end{texnote}
% \end{function}
%
% \begin{function}[EXP]{\exp_not:e}
%   \begin{syntax}
%     \cs{exp_not:e} \Arg{tokens}
%   \end{syntax}
%   Expands \meta{tokens} exhaustively, then protects the result of the
%   expansion (including any tokens which were not expanded) from
%   further expansion in |e| or |x|-type arguments using \cs{exp_not:n}.
%   This is very rarely useful but is provided for consistency.
% \end{function}
%
% \begin{function}[EXP]{\exp_not:f}
%   \begin{syntax}
%     \cs{exp_not:f} \Arg{tokens}
%   \end{syntax}
%   Expands \meta{tokens} fully until the first unexpandable token is
%   found (if it is a space it is removed). Expansion then stops, and
%   the result of the expansion (including any tokens which were not
%   expanded) is protected from further expansion in |x|-type or \texttt{e}-type arguments
%   using \cs{exp_not:n}.
% \end{function}
%
% \begin{function}[updated = 2011-06-03, EXP]{\exp_stop_f:}
%   \begin{syntax}
%     |\foo_bar:f| \{ \meta{tokens} \cs{exp_stop_f:} \meta{more tokens} \}
%   \end{syntax}
%   This function terminates an \texttt{f}-type expansion. Thus if
%   a function |\foo_bar:f| starts an \texttt{f}-type expansion
%   and all of \meta{tokens} are expandable \cs{exp_stop_f:}
%   terminates the expansion of tokens even if \meta{more tokens}
%   are also expandable. The function itself is an implicit space
%   token. Inside an \texttt{x}-type expansion, it retains its
%   form, but when typeset it produces the underlying space (\verb*| |).
% \end{function}
%
%
% \section{Controlled expansion}
%
% The \pkg{expl3} language makes all efforts to hide the complexity of
% \TeX{} expansion from the programmer by providing concepts that
% evaluate/expand arguments of functions prior to calling the \enquote{base}
% functions. Thus, instead of using many \tn{expandafter} calls and
% other trickery it is usually a matter of choosing the right variant
% of a function to achieve a desired result.
%
% Of course, deep down \TeX{} is using expansion as always and there
% are cases where a programmer needs to control that expansion
% directly; typical situations are basic data manipulation tools. This
% section documents the functions for that level. These
% commands are used throughout the kernel code, but we hope that outside
% the kernel there will be little need to resort to them. Instead the
% argument manipulation methods document above should usually be sufficient.
%
% While \cs{exp_after:wN} expands one token (out of order) it is
% sometimes necessary to expand several tokens in one go. The next set
% of commands provide this functionality. Be aware that it is
% absolutely required that the programmer has full control over the
% tokens to be expanded, i.e., it is not possible to use these
% functions to expand unknown input as part of
% \meta{expandable-tokens} as that will break badly if unexpandable
% tokens are encountered in that place!
%
% \begin{function}[added=2015-08-23,EXP]
%       {
%         \exp:w ,
%         \exp_end:
%       }
%   \begin{syntax}
%     \cs{exp:w} \meta{expandable tokens} \cs{exp_end:} \\
%   \end{syntax}
%   Expands \meta{expandable-tokens} until reaching \cs{exp_end:} at
%   which point expansion stops.
%   The full expansion of \meta{expandable tokens} has to be empty.
%   If any token in \meta{expandable tokens} or any token generated by
%   expanding the tokens therein is not expandable the expansion will end
%   prematurely and as a result \cs{exp_end:} will be misinterpreted
%   later on.\footnotemark
%
%   In typical use cases the \cs{exp_end:} is hidden somewhere
%   in the replacement text of \meta{expandable-tokens} rather than
%   being on the same expansion level than \cs{exp:w}, e.g., you may
%   see code such as
%\begin{verbatim}
%    \exp:w \@@_case:NnTF #1 {#2} { } { }
%\end{verbatim}
%   where somewhere during the expansion of |\@@_case:NnTF| the
%   \cs{exp_end:} gets generated.
%   \begin{texnote}
%     The current implementation uses \tn{romannumeral} hence ignores
%     space tokens and explicit signs |+| and |-| in the expansion of the
%     \meta{expandable tokens}, but this should not be relied upon.
%   \end{texnote}
% \end{function}
% \footnotetext{Due to the implementation you might get the character
%   in position 0 in the current font (typically \enquote{\texttt{`}})
%   in the output without any error message!}
%
% \begin{function}[added=2015-08-23, EXP, label = \exp_end_continue_f:w]
%       {
%         \exp:w ,
%         \exp_end_continue_f:w
%       }
%   \begin{syntax}
%     \cs{exp:w} \meta{expandable-tokens} \cs{exp_end_continue_f:w} \meta{further-tokens}
%   \end{syntax}
%   Expands \meta{expandable-tokens} until reaching \cs{exp_end_continue_f:w} at
%   which point expansion continues as an \texttt{f}-type expansion expanding
%   \meta{further-tokens} until an unexpandable token is encountered (or
%   the \texttt{f}-type expansion is explicitly terminated by
%   \cs{exp_stop_f:}). As with all \texttt{f}-type expansions a space ending
%   the expansion gets removed.
%
%   The full expansion of \meta{expandable-tokens} has to be empty.
%   If any token in \meta{expandable-tokens} or any token generated by
%   expanding the tokens therein is not expandable the expansion will end
%   prematurely and as a result \cs{exp_end_continue_f:w} will be misinterpreted
%   later on.\footnotemark
%
%
%   In typical use cases  \meta{expandable-tokens} contains no tokens at all,
%   e.g., you will see code such as
%\begin{verbatim}
%    \exp_after:wN { \exp:w \exp_end_continue_f:w #2 }
%\end{verbatim}
%   where the \cs{exp_after:wN} triggers an \texttt{f}-expansion of the tokens
%   in |#2|. For technical reasons this has to happen using two tokens
%   (if they would be hidden inside another command \cs{exp_after:wN}
%   would only expand the command but not trigger any additional
%   |f|-expansion).
%
%   You might wonder why there are two different approaches available,
%   after all the effect of
%   \begin{quote}
%     \cs{exp:w} \meta{expandable-tokens} \cs{exp_end:}
%   \end{quote}
%   can be alternatively achieved through an \texttt{f}-type expansion by using
%   \cs{exp_stop_f:}, i.e.
%   \begin{quote}
%     \cs{exp:w} \cs{exp_end_continue_f:w} \meta{expandable-tokens} \cs{exp_stop_f:}
%   \end{quote}
%   The reason is simply that the first approach is slightly faster
%   (one less token to parse and less expansion internally)
%   so in places where such performance really matters and where we
%   want to explicitly stop the expansion at a defined point the first
%   form is preferable.
% \end{function}
% \footnotetext{In this particular case you may get a character into
%    the output as well as an error message.}
%
% \begin{function}[added=2015-08-23, EXP, label = \exp_end_continue_f:nw]
%       {
%         \exp:w ,
%         \exp_end_continue_f:nw
%       }
%   \begin{syntax}
%     \cs{exp:w} \meta{expandable-tokens} \cs{exp_end_continue_f:nw} \meta{further-tokens}
%   \end{syntax}
%   The difference to \cs{exp_end_continue_f:w} is that we first we pick
%   up an argument which is then returned to the input stream.  If
%   \meta{further-tokens} starts with space tokens then these space
%   tokens are removed while searching for the argument.  If it starts
%   with a brace group then the braces are removed. Thus such spaces or
%   braces will not terminate the \texttt{f}-type expansion.
% \end{function}
%
% \section{Internal functions}
%
% \begin{function}{\::n, \::N, \::p, \::c, \::o, \::e, \::f, \::x, \::v, \::V, \:::}
%   \begin{syntax}
%     |\cs_new:Npn \exp_args:Ncof { \::c \::o \::f \::: }|
%   \end{syntax}
%   Internal forms for the base expansion types. These names do \emph{not}
%   conform to the general \LaTeX3 approach as this makes them more readily
%   visible in the log and so forth.  They should not be used outside this module.
% \end{function}
%
% \begin{function}
%   {\::o_unbraced, \::e_unbraced, \::f_unbraced, \::x_unbraced, \::v_unbraced, \::V_unbraced}
%   \begin{syntax}
%     |\cs_new:Npn \exp_last_unbraced:Nno { \::n \::o_unbraced \::: }|
%   \end{syntax}
%   Internal forms for the expansion types which leave the terminal argument
%   unbraced. These names do \emph{not}
%   conform to the general \LaTeX3 approach as this makes them more readily
%   visible in the log and so forth.  They should not be used outside this module.
% \end{function}
%
% \end{documentation}
%
% \begin{implementation}
%
% \section{\pkg{l3expan} implementation}
%
%    \begin{macrocode}
%<*package>
%    \end{macrocode}
%
%    \begin{macrocode}
%<@@=exp>
%    \end{macrocode}
%
% \begin{variable}{\l_@@_internal_tl}
%   The |\exp_| module has its private variable to temporarily store the
%   result of |x|-type argument expansion. This is done to avoid interference
%   with other functions using temporary variables.
% \end{variable}
%
% \begin{macro}{\exp_after:wN}
% \begin{macro}{\exp_not:N}
% \begin{macro}{\exp_not:n}
%   These are defined in \pkg{l3basics}, as they are needed
%   \enquote{early}.  This is just a reminder of that fact!
% \end{macro}
% \end{macro}
% \end{macro}
%
% \subsection{General expansion}
%
% In this section a general mechanism for defining functions that handle
% arguments is defined.  These general expansion functions are
% expandable unless |x| is used.  (Any version of |x| is going to have
% to use one of the \LaTeX3 names for \cs{cs_set:Npx} at some
% point, and so is never going to be expandable.)
%
% The definition of expansion functions with this technique happens
% in section~\ref{sec:l3expan:gendef}.
% In section~\ref{sec:l3expan:handtune} some common cases are coded by a more direct
% method for efficiency, typically using calls to \cs{exp_after:wN}.
%
% \begin{variable}{\l_@@_internal_tl}
%   This scratch token list variable is defined in \pkg{l3basics}.
% \end{variable}
%
% This code uses internal functions with names that start with |\::| to
% perform the expansions. All macros are |long| since the tokens
% undergoing expansion may be arbitrary user input.
%
% An argument manipulator |\::|\meta{Z} always has signature |#1\:::#2#3|
% where |#1| holds the remaining argument manipulations to be performed,
% \cs{:::} serves as an end marker for the list of manipulations, |#2|
% is the carried over result of the previous expansion steps and |#3| is
% the argument about to be processed.
% One exception to this rule is \cs{::p}, which has to grab an argument
% delimited by a left brace.
%
% \begin{macro}[EXP]{\@@_arg_next:nnn}
% \begin{macro}[EXP]{\@@_arg_next:Nnn}
%   |#1| is the result of an expansion step, |#2| is the remaining
%   argument manipulations and |#3| is the current result of the
%   expansion chain.  This auxiliary function moves |#1| back after
%   |#3| in the input stream and checks if any expansion is left to
%   be done by calling |#2|. In by far the most cases we need
%   to add a set of braces to the result of an argument manipulation
%   so it is more effective to do it directly here. Actually, so far
%   only the |c| of the final argument manipulation variants does not
%   require a set of braces.
%    \begin{macrocode}
\cs_new:Npn \@@_arg_next:nnn #1#2#3 { #2 \::: { #3 {#1} } }
\cs_new:Npn \@@_arg_next:Nnn #1#2#3 { #2 \::: { #3 #1 } }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\:::}
%   The end marker is just another name for the identity function.
%    \begin{macrocode}
\cs_new:Npn \::: #1 {#1}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\::n}
%   This function is used to skip an argument that doesn't need to
%   be expanded.
%    \begin{macrocode}
\cs_new:Npn \::n #1 \::: #2#3 { #1 \::: { #2 {#3} } }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\::N}
%   This function is used to skip an argument that consists of a
%   single token and doesn't need to be expanded.
%    \begin{macrocode}
\cs_new:Npn \::N #1 \::: #2#3 { #1 \::: {#2#3} }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\::p}
%   This function is used to skip an argument that is delimited by a
%   left brace and doesn't need to be expanded.  It is not
%   wrapped in braces in the result.
%    \begin{macrocode}
\cs_new:Npn \::p #1 \::: #2#3# { #1 \::: {#2#3} }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\::c}
%   This function is used to skip an argument that is turned into
%   a control sequence without expansion.
%    \begin{macrocode}
\cs_new:Npn \::c #1 \::: #2#3
  { \exp_after:wN \@@_arg_next:Nnn \cs:w #3 \cs_end: {#1} {#2} }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\::o}
%   This function is used to expand an argument once.
%    \begin{macrocode}
\cs_new:Npn \::o #1 \::: #2#3
  { \exp_after:wN \@@_arg_next:nnn \exp_after:wN {#3} {#1} {#2} }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\::e}
%   With the \tn{expanded} primitive available, just expand.  Otherwise
%   defer to \cs{exp_args:Ne} implemented later.
%    \begin{macrocode}
\cs_if_exist:NTF \tex_expanded:D
  {
    \cs_new:Npn \::e #1 \::: #2#3
      { \tex_expanded:D { \exp_not:n { #1 \::: } { \exp_not:n {#2} {#3} } } }
  }
  {
    \cs_new:Npn \::e #1 \::: #2#3
      { \exp_args:Ne \@@_arg_next:nnn {#3} {#1} {#2} }
  }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\::f}
% \begin{macro}{\exp_stop_f:}
%   This function is used to expand a token list until the first
%   unexpandable token is found. This is achieved through \cs{exp:w}
%   \cs{exp_end_continue_f:w} that expands everything in its way
%   following it. This scanning procedure is terminated once the
%   expansion hits something non-expandable (if that is a space it is
%   removed).  We introduce \cs{exp_stop_f:} to mark such an
%   end-of-expansion marker.  For example, |f|-expanding
%   |\cs_set_eq:Nc \aaa { b \l_tmpa_tl b }| where |\l_tmpa_tl| contains
%   the characters |lur| gives |\tex_let:D \aaa = \blurb| which then
%   turns out to start with the non-expandable token |\tex_let:D|.
%   Since the expansion of \cs{exp:w} \cs{exp_end_continue_f:w} is
%   empty, we wind up with a fully expanded list, only \TeX{} has not
%   tried to execute any of the non-expandable tokens. This is what
%   differentiates this function from the |x| argument type.
%    \begin{macrocode}
\cs_new:Npn \::f #1 \::: #2#3
  {
    \exp_after:wN \@@_arg_next:nnn
      \exp_after:wN { \exp:w \exp_end_continue_f:w #3 }
      {#1} {#2}
  }
\use:nn { \cs_new_eq:NN \exp_stop_f: } { ~ }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\::x}
%   This function is used to expand an argument fully.
%   We build in the expansion of \cs{@@_arg_next:nnn}.
%    \begin{macrocode}
\cs_new_protected:Npn \::x #1 \::: #2#3
  {
    \cs_set_nopar:Npx \l_@@_internal_tl
      { \exp_not:n { #1 \::: } { \exp_not:n {#2} {#3} } }
    \l_@@_internal_tl
  }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\::v}
% \begin{macro}[EXP]{\::V}
%   These functions return the value of a register, i.e., one of
%   |tl|, |clist|, |int|, |skip|, |dim|, |muskip|, or built-in
%   \TeX{} register. The |V| version
%   expects a single token whereas |v| like |c| creates a csname from
%   its argument given in braces and then evaluates it as if it was a
%   |V|. The \cs{exp:w} sets off an expansion
%   similar to an |f|-type expansion, which we terminate using
%   \cs{exp_end:}. The argument is returned in braces.
%    \begin{macrocode}
\cs_new:Npn \::V #1 \::: #2#3
  {
    \exp_after:wN \@@_arg_next:nnn
      \exp_after:wN { \exp:w \@@_eval_register:N #3 }
      {#1} {#2}
}
\cs_new:Npn \::v #1 \::: #2#3
  {
    \exp_after:wN \@@_arg_next:nnn
      \exp_after:wN { \exp:w \@@_eval_register:c {#3} }
      {#1} {#2}
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}[EXP]{\@@_eval_register:N, \@@_eval_register:c}
% \begin{macro}[EXP]{\@@_eval_error_msg:w}
%   This function evaluates a register. Now a register might exist as
%   one of two things: A parameter-less macro or a built-in \TeX{}
%   register such as |\count|. For the \TeX{} registers we have to
%   utilize a \tn{the} whereas for the macros we merely have to
%   expand them once. The trick is to find out when to use
%   \tn{the} and when not to. What we want here is to find out
%   whether the token expands to something else when hit with
%   \cs{exp_after:wN}. The technique is to compare the meaning of the
%   token in question when it has been prefixed with \cs{exp_not:N}
%   and the token itself. If it is a macro, the prefixed
%   \cs{exp_not:N} temporarily turns it into the primitive
%   \cs{scan_stop:}.
%    \begin{macrocode}
\cs_new:Npn \@@_eval_register:N #1
  {
    \exp_after:wN \if_meaning:w \exp_not:N #1 #1
%    \end{macrocode}
%   If the token was not a macro it may be a malformed variable from a
%   |c| expansion in which case it is equal to the primitive
%   \cs{scan_stop:}. In that case we throw an error. We could let \TeX{}
%   do it for us but that would result in the rather obscure
%   \begin{quote}
%     |! You can't use `\relax' after \the.|
%   \end{quote}
%   which while quite true doesn't give many hints as to what actually
%   went wrong. We provide something more sensible.
%    \begin{macrocode}
      \if_meaning:w \scan_stop: #1
        \@@_eval_error_msg:w
      \fi:
%    \end{macrocode}
%   The next bit requires some explanation. The function must be
%   initiated by \cs{exp:w} and we want to
%   terminate this expansion chain by inserting the \cs{exp_end:}
%    token.
%   However, we have to expand the register |#1| before we do
%   that. If it is a \TeX{} register, we need to execute the sequence
%   |\exp_after:wN \exp_end: \tex_the:D #1| and if it is a macro we
%   need to execute |\exp_after:wN \exp_end: #1|. We therefore issue
%   the longer of the two sequences and if the register is a macro, we
%   remove the \cs{tex_the:D}.
%    \begin{macrocode}
    \else:
      \exp_after:wN \use_i_ii:nnn
    \fi:
    \exp_after:wN \exp_end: \tex_the:D #1
  }
\cs_new:Npn \@@_eval_register:c #1
  { \exp_after:wN \@@_eval_register:N \cs:w #1 \cs_end: }
%    \end{macrocode}
%   Clean up nicely, then call the undefined control sequence. The
%   result is an error message looking like this:
%   \begin{verbatim}
%     ! Undefined control sequence.
%     <argument> \LaTeX3 error:
%                               Erroneous variable used!
%     l.55 \tl_set:Nv \l_tmpa_tl {undefined_tl}
%   \end{verbatim}
%    \begin{macrocode}
\cs_new:Npn \@@_eval_error_msg:w #1 \tex_the:D #2
  {
      \fi:
    \fi:
    \__kernel_msg_expandable_error:nnn { kernel } { bad-variable } {#2}
    \exp_end:
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \subsection{Hand-tuned definitions}
% \label{sec:l3expan:handtune}
%
% One of the most important features of these functions is that they
% are fully expandable.
%
% \begin{macro}[EXP]{\exp_args:Nc, \exp_args:cc}
%   In \pkg{l3basics}.
% \end{macro}
%
% \begin{macro}[EXP]{\exp_args:NNc, \exp_args:Ncc, \exp_args:Nccc}
%   Here are the functions that turn their argument into csnames but are
%   expandable.
%    \begin{macrocode}
\cs_new:Npn \exp_args:NNc #1#2#3
  { \exp_after:wN #1 \exp_after:wN #2 \cs:w # 3\cs_end: }
\cs_new:Npn \exp_args:Ncc #1#2#3
  { \exp_after:wN #1 \cs:w #2 \exp_after:wN \cs_end: \cs:w #3 \cs_end: }
\cs_new:Npn \exp_args:Nccc #1#2#3#4
  {
    \exp_after:wN #1
      \cs:w #2 \exp_after:wN \cs_end:
      \cs:w #3 \exp_after:wN \cs_end:
      \cs:w #4 \cs_end:
  }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\exp_args:No}
% \begin{macro}[EXP]{\exp_args:NNo}
% \begin{macro}[EXP]{\exp_args:NNNo}
%   Those lovely runs of expansion!
%    \begin{macrocode}
\cs_new:Npn \exp_args:No #1#2 { \exp_after:wN #1 \exp_after:wN {#2} }
\cs_new:Npn \exp_args:NNo #1#2#3
  { \exp_after:wN #1 \exp_after:wN #2 \exp_after:wN {#3} }
\cs_new:Npn \exp_args:NNNo #1#2#3#4
  { \exp_after:wN #1 \exp_after:wN#2 \exp_after:wN #3 \exp_after:wN {#4} }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}[EXP]{\exp_args:Ne}
%   When the \tn{expanded} primitive is available, use it.  Otherwise
%   use \cs{@@_e:nn}, defined later, to fully expand tokens.
%    \begin{macrocode}
\cs_if_exist:NTF \tex_expanded:D
  {
    \cs_new:Npn \exp_args:Ne #1#2
      { \exp_after:wN #1 \tex_expanded:D { {#2} } }
  }
  {
    \cs_new:Npn \exp_args:Ne #1#2
      {
        \exp_after:wN #1 \exp_after:wN
          { \exp:w \@@_e:nn {#2} { } }
      }
  }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\exp_args:Nf, \exp_args:NV, \exp_args:Nv}
%    \begin{macrocode}
\cs_new:Npn \exp_args:Nf #1#2
  { \exp_after:wN #1 \exp_after:wN { \exp:w \exp_end_continue_f:w #2 } }
\cs_new:Npn \exp_args:Nv #1#2
  {
    \exp_after:wN #1 \exp_after:wN
      { \exp:w \@@_eval_register:c {#2} }
  }
\cs_new:Npn \exp_args:NV #1#2
  {
    \exp_after:wN #1 \exp_after:wN
      { \exp:w \@@_eval_register:N #2 }
  }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]
%   {
%     \exp_args:NNV, \exp_args:NNv, \exp_args:NNe, \exp_args:NNf,
%     \exp_args:Nco, \exp_args:NcV, \exp_args:Ncv, \exp_args:Ncf,
%     \exp_args:NVV,
%   }
%   Some more hand-tuned function with three arguments.
%   If we forced that an |o| argument always has braces,
%   we could implement \cs{exp_args:Nco} with less tokens
%   and only two arguments.
%    \begin{macrocode}
\cs_new:Npn \exp_args:NNV #1#2#3
  {
    \exp_after:wN #1
    \exp_after:wN #2
    \exp_after:wN { \exp:w \@@_eval_register:N #3 }
  }
\cs_new:Npn \exp_args:NNv #1#2#3
  {
    \exp_after:wN #1
    \exp_after:wN #2
    \exp_after:wN { \exp:w \@@_eval_register:c {#3} }
  }
\cs_if_exist:NTF \tex_expanded:D
  {
    \cs_new:Npn \exp_args:NNe #1#2#3
      {
        \exp_after:wN #1
        \exp_after:wN #2
        \tex_expanded:D { {#3} }
      }
  }
  { \cs_new:Npn \exp_args:NNe { \::N \::e \::: } }
\cs_new:Npn \exp_args:NNf #1#2#3
  {
    \exp_after:wN #1
    \exp_after:wN #2
    \exp_after:wN { \exp:w \exp_end_continue_f:w #3 }
  }
\cs_new:Npn \exp_args:Nco #1#2#3
  {
    \exp_after:wN #1
    \cs:w #2 \exp_after:wN \cs_end:
    \exp_after:wN {#3}
  }
\cs_new:Npn \exp_args:NcV #1#2#3
  {
    \exp_after:wN #1
    \cs:w #2 \exp_after:wN \cs_end:
    \exp_after:wN { \exp:w \@@_eval_register:N #3 }
  }
\cs_new:Npn \exp_args:Ncv #1#2#3
  {
    \exp_after:wN #1
    \cs:w #2 \exp_after:wN \cs_end:
    \exp_after:wN { \exp:w \@@_eval_register:c {#3} }
  }
\cs_new:Npn \exp_args:Ncf #1#2#3
  {
    \exp_after:wN #1
    \cs:w #2 \exp_after:wN \cs_end:
    \exp_after:wN { \exp:w \exp_end_continue_f:w #3 }
  }
\cs_new:Npn \exp_args:NVV #1#2#3
  {
    \exp_after:wN #1
    \exp_after:wN { \exp:w \exp_after:wN
      \@@_eval_register:N \exp_after:wN #2 \exp_after:wN }
    \exp_after:wN { \exp:w \@@_eval_register:N #3 }
  }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]
%   {
%     \exp_args:NNNV, \exp_args:NNNv,
%     \exp_args:NcNc, \exp_args:NcNo,
%     \exp_args:Ncco,
%   }
%   A few more that we can hand-tune.
%    \begin{macrocode}
\cs_new:Npn \exp_args:NNNV #1#2#3#4
  {
    \exp_after:wN #1
    \exp_after:wN #2
    \exp_after:wN #3
    \exp_after:wN { \exp:w \@@_eval_register:N #4 }
  }
\cs_new:Npn \exp_args:NNNv #1#2#3#4
  {
    \exp_after:wN #1
    \exp_after:wN #2
    \exp_after:wN #3
    \exp_after:wN { \exp:w \@@_eval_register:c {#4} }
  }
\cs_new:Npn \exp_args:NcNc #1#2#3#4
  {
    \exp_after:wN #1
    \cs:w #2 \exp_after:wN \cs_end:
    \exp_after:wN #3
    \cs:w #4 \cs_end:
  }
\cs_new:Npn \exp_args:NcNo #1#2#3#4
  {
    \exp_after:wN #1
    \cs:w #2 \exp_after:wN \cs_end:
    \exp_after:wN #3
    \exp_after:wN {#4}
  }
\cs_new:Npn \exp_args:Ncco #1#2#3#4
  {
    \exp_after:wN #1
    \cs:w #2 \exp_after:wN \cs_end:
    \cs:w #3 \exp_after:wN \cs_end:
    \exp_after:wN {#4}
  }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\exp_args:Nx}
%    \begin{macrocode}
\cs_new_protected:Npn \exp_args:Nx #1#2
  { \use:x { \exp_not:N #1 {#2} } }
%    \end{macrocode}
% \end{macro}
%
% \subsection{Last-unbraced versions}
%
% \begin{macro}[EXP]{\@@_arg_last_unbraced:nn}
% \begin{macro}[EXP]{\::o_unbraced}
% \begin{macro}[EXP]{\::V_unbraced}
% \begin{macro}[EXP]{\::v_unbraced}
% \begin{macro}[EXP]{\::e_unbraced}
% \begin{macro}[EXP]{\::f_unbraced}
% \begin{macro}[EXP]{\::x_unbraced}
%   There are a few places where the last argument needs to be available
%   unbraced. First some helper macros.
%    \begin{macrocode}
\cs_new:Npn \@@_arg_last_unbraced:nn #1#2 { #2#1 }
\cs_new:Npn \::o_unbraced \::: #1#2
  { \exp_after:wN \@@_arg_last_unbraced:nn \exp_after:wN {#2} {#1} }
\cs_new:Npn \::V_unbraced \::: #1#2
  {
    \exp_after:wN \@@_arg_last_unbraced:nn
      \exp_after:wN { \exp:w \@@_eval_register:N #2 } {#1}
  }
\cs_new:Npn \::v_unbraced \::: #1#2
  {
    \exp_after:wN \@@_arg_last_unbraced:nn
      \exp_after:wN { \exp:w \@@_eval_register:c {#2} } {#1}
  }
\cs_if_exist:NTF \tex_expanded:D
  {
    \cs_new:Npn \::e_unbraced \::: #1#2
      { \tex_expanded:D { \exp_not:n {#1} #2 } }
  }
  {
    \cs_new:Npn \::e_unbraced \::: #1#2
      { \exp:w \@@_e:nn {#2} {#1} }
  }
\cs_new:Npn \::f_unbraced \::: #1#2
  {
    \exp_after:wN \@@_arg_last_unbraced:nn
      \exp_after:wN { \exp:w \exp_end_continue_f:w #2 } {#1}
  }
\cs_new_protected:Npn \::x_unbraced \::: #1#2
  {
    \cs_set_nopar:Npx \l_@@_internal_tl { \exp_not:n {#1} #2 }
    \l_@@_internal_tl
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}[EXP]
%   {
%     \exp_last_unbraced:No,
%     \exp_last_unbraced:NV,
%     \exp_last_unbraced:Nv,
%     \exp_last_unbraced:Ne,
%     \exp_last_unbraced:Nf,
%     \exp_last_unbraced:NNo,
%     \exp_last_unbraced:NNV,
%     \exp_last_unbraced:NNf,
%     \exp_last_unbraced:Nco,
%     \exp_last_unbraced:NcV,
%     \exp_last_unbraced:NNNo,
%     \exp_last_unbraced:NNNV,
%     \exp_last_unbraced:NNNf,
%     \exp_last_unbraced:Nno,
%     \exp_last_unbraced:Noo,
%     \exp_last_unbraced:Nfo,
%     \exp_last_unbraced:NnNo,
%     \exp_last_unbraced:NNNNo,
%     \exp_last_unbraced:NNNNf,
%   }
% \begin{macro}{\exp_last_unbraced:Nx}
%   Now the business end: most of these are hand-tuned for speed, but the
%   general system is in place.
%    \begin{macrocode}
\cs_new:Npn \exp_last_unbraced:No #1#2 { \exp_after:wN #1 #2 }
\cs_new:Npn \exp_last_unbraced:NV #1#2
  { \exp_after:wN #1 \exp:w \@@_eval_register:N #2 }
\cs_new:Npn \exp_last_unbraced:Nv #1#2
  { \exp_after:wN #1 \exp:w \@@_eval_register:c {#2} }
\cs_if_exist:NTF \tex_expanded:D
  {
    \cs_new:Npn \exp_last_unbraced:Ne #1#2
      { \exp_after:wN #1 \tex_expanded:D {#2} }
  }
  { \cs_new:Npn \exp_last_unbraced:Ne { \::e_unbraced \::: } }
\cs_new:Npn \exp_last_unbraced:Nf #1#2
  { \exp_after:wN #1 \exp:w \exp_end_continue_f:w #2 }
\cs_new:Npn \exp_last_unbraced:NNo #1#2#3
  { \exp_after:wN #1 \exp_after:wN #2 #3 }
\cs_new:Npn \exp_last_unbraced:NNV #1#2#3
  {
    \exp_after:wN #1
    \exp_after:wN #2
    \exp:w \@@_eval_register:N #3
  }
\cs_new:Npn \exp_last_unbraced:NNf #1#2#3
  {
    \exp_after:wN #1
    \exp_after:wN #2
    \exp:w \exp_end_continue_f:w #3
  }
\cs_new:Npn \exp_last_unbraced:Nco #1#2#3
  { \exp_after:wN #1 \cs:w #2 \exp_after:wN \cs_end: #3 }
\cs_new:Npn \exp_last_unbraced:NcV #1#2#3
  {
    \exp_after:wN #1
    \cs:w #2 \exp_after:wN \cs_end:
    \exp:w \@@_eval_register:N #3
  }
\cs_new:Npn \exp_last_unbraced:NNNo #1#2#3#4
  { \exp_after:wN #1 \exp_after:wN #2 \exp_after:wN #3 #4 }
\cs_new:Npn \exp_last_unbraced:NNNV #1#2#3#4
  {
    \exp_after:wN #1
    \exp_after:wN #2
    \exp_after:wN #3
    \exp:w \@@_eval_register:N #4
  }
\cs_new:Npn \exp_last_unbraced:NNNf #1#2#3#4
  {
    \exp_after:wN #1
    \exp_after:wN #2
    \exp_after:wN #3
    \exp:w \exp_end_continue_f:w #4
  }
\cs_new:Npn \exp_last_unbraced:Nno { \::n \::o_unbraced \::: }
\cs_new:Npn \exp_last_unbraced:Noo { \::o \::o_unbraced \::: }
\cs_new:Npn \exp_last_unbraced:Nfo { \::f \::o_unbraced \::: }
\cs_new:Npn \exp_last_unbraced:NnNo { \::n \::N \::o_unbraced \::: }
\cs_new:Npn \exp_last_unbraced:NNNNo #1#2#3#4#5
  { \exp_after:wN #1 \exp_after:wN #2 \exp_after:wN #3 \exp_after:wN #4 #5 }
\cs_new:Npn \exp_last_unbraced:NNNNf #1#2#3#4#5
  {
    \exp_after:wN #1
    \exp_after:wN #2
    \exp_after:wN #3
    \exp_after:wN #4
    \exp:w \exp_end_continue_f:w #5
  }
\cs_new_protected:Npn \exp_last_unbraced:Nx { \::x_unbraced \::: }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}[EXP]{\exp_last_two_unbraced:Noo}
% \begin{macro}[EXP]{\@@_last_two_unbraced:noN}
%   If |#2| is a single token then this can be implemented as
%   \begin{verbatim}
%     \cs_new:Npn \exp_last_two_unbraced:Noo #1 #2 #3
%      { \exp_after:wN \exp_after:wN \exp_after:wN #1 \exp_after:wN #2 #3 }
%   \end{verbatim}
%   However, for robustness this is not suitable. Instead, a bit of a
%   shuffle is used to ensure that |#2| can be multiple tokens.
%    \begin{macrocode}
\cs_new:Npn \exp_last_two_unbraced:Noo #1#2#3
  { \exp_after:wN \@@_last_two_unbraced:noN \exp_after:wN {#3} {#2} #1 }
\cs_new:Npn \@@_last_two_unbraced:noN #1#2#3
   { \exp_after:wN #3 #2 #1 }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \subsection{Preventing expansion}
%
% \begin{macro}{\__kernel_exp_not:w}
%   At the kernel level, we need the primitive behaviour to allow expansion
%   \emph{before} the brace group.
%    \begin{macrocode}
\cs_new_eq:NN \__kernel_exp_not:w \tex_unexpanded:D
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\exp_not:c}
% \begin{macro}[EXP]{\exp_not:o}
% \begin{macro}[EXP]{\exp_not:e}
% \begin{macro}[EXP]{\exp_not:f}
% \begin{macro}[EXP]{\exp_not:V}
% \begin{macro}[EXP]{\exp_not:v}
%   All these except \cs{exp_not:c} call the kernel-internal
%   \cs{__kernel_exp_not:w} namely \cs{tex_unexpanded:D}.
%    \begin{macrocode}
\cs_new:Npn \exp_not:c #1 { \exp_after:wN \exp_not:N \cs:w #1 \cs_end: }
\cs_new:Npn \exp_not:o #1 { \__kernel_exp_not:w \exp_after:wN {#1} }
\cs_if_exist:NTF \tex_expanded:D
  {
    \cs_new:Npn \exp_not:e #1
      { \__kernel_exp_not:w \tex_expanded:D { {#1} } }
  }
  {
    \cs_new:Npn \exp_not:e
      { \__kernel_exp_not:w \exp_args:Ne \prg_do_nothing: }
  }
\cs_new:Npn \exp_not:f #1
  { \__kernel_exp_not:w \exp_after:wN { \exp:w \exp_end_continue_f:w #1 } }
\cs_new:Npn \exp_not:V #1
  {
    \__kernel_exp_not:w \exp_after:wN
      { \exp:w \@@_eval_register:N #1 }
  }
\cs_new:Npn \exp_not:v #1
  {
    \__kernel_exp_not:w \exp_after:wN
      { \exp:w \@@_eval_register:c {#1} }
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \subsection{Controlled expansion}
%
% \begin{macro}{\exp:w}
% \begin{macro}{\exp_end:}
% \begin{macro}{\exp_end_continue_f:w}
% \begin{macro}{\exp_end_continue_f:nw}
%   To trigger a sequence of \enquote{arbitrarily} many expansions we
%   need a method to invoke \TeX's expansion mechanism in such a way
%   that (a) we are able to stop it in a controlled manner and (b) the
%   result of what triggered the expansion in the first place is null,
%   i.e.\@, that we do not get any unwanted side effects. There aren't
%   that many possibilities in \TeX{}; in fact the one explained below
%   might well be the only one (as normally the result of expansion is
%   not null).
%
%   The trick here is to make use of the fact that
%   \cs{tex_romannumeral:D} expands the tokens following it when looking
%   for a number and that its expansion is null if that number turns out
%   to be zero or negative. So we use that to start the expansion
%   sequence: \cs{exp:w} is set equal to \cs{tex_romannumeral:D} in
%   \pkg{l3basics}.  To stop the expansion sequence in a controlled way
%   all we need to provide is a constant integer zero as part of
%   expanded tokens. As this is an integer constant it immediately stops
%   \cs{tex_romannumeral:D}'s search for a number.  Again, the
%   definition of \cs{exp_end:} as the integer constant zero is in
%   \pkg{l3basics}.  (Note that according to our specification all
%   tokens we expand initiated by \cs{exp:w} are supposed to be
%   expandable (as well as their replacement text in the expansion) so
%   we will not encounter a \enquote{number} that actually result in a
%   roman numeral being generated. Or if we do then the programmer made
%   a mistake.)
%
%   If on the other hand we want to stop the initial expansion sequence
%   but continue with an \texttt{f}-type expansion we provide the
%   alphabetic constant |`^^@| that also represents |0| but this time
%   \TeX's syntax for a \meta{number} continues searching for an
%   optional space (and it continues expansion doing that) --- see
%   \TeX{}book page~269 for details.
%    \begin{macrocode}
\group_begin:
  \tex_catcode:D `\^^@ = 13
  \cs_new_protected:Npn \exp_end_continue_f:w { `^^@ }
%    \end{macrocode}
%    If the above definition ever appears outside its proper context
%    the active character |^^@| will be executed so we turn this into an
%    error. The test for existence covers the (unlikely) case that some
%    other code has already defined |^^@|: this is true for example for
%    \texttt{xmltex.tex}.
%    \begin{macrocode}
  \if_cs_exist:N ^^@
  \else:
    \cs_new:Npn ^^@
      { \__kernel_msg_expandable_error:nn { kernel } { bad-exp-end-f } }
  \fi:
%    \end{macrocode}
%   The same but grabbing an argument to remove spaces and braces.
%    \begin{macrocode}
  \cs_new:Npn \exp_end_continue_f:nw #1 { `^^@ #1 }
\group_end:
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
%
% \subsection{Emulating \texttt{e}-type expansion}
%
% When the \tn{expanded} primitive is available it is used to implement
% \texttt{e}-type expansion; otherwise we emulate it.
%
%    \begin{macrocode}
\cs_if_exist:NF \tex_expanded:D
  {
%    \end{macrocode}
%
% \begin{macro}[EXP]{\@@_e:nn, \@@_e_end:nn}
%   Repeatedly expand tokens, keeping track of fully-expanded tokens in
%   the second argument to \cs{@@_e:nn}; this function eventually
%   calls \cs{@@_e_end:nn} to leave \cs{exp_end:} in the input
%   stream, followed by the result of the expansion.  There are many
%   special cases: spaces, brace groups, \tn{noexpand}, \tn{unexpanded},
%   \tn{the}, \tn{primitive}.
%   While we use brace tricks \cs{if_false:} |{| \cs{fi:}, the expansion
%   of this function is always triggered by \cs{exp:w} so brace balance
%   is eventually restored after that is hit with a single step of
%   expansion.  Otherwise we could not nest \texttt{e}-type expansions
%   within each other.
%    \begin{macrocode}
    \cs_new:Npn \@@_e:nn #1
      {
        \if_false: { \fi:
          \tl_if_head_is_N_type:nTF {#1}
            { \@@_e:N }
            {
              \tl_if_head_is_group:nTF {#1}
                { \@@_e_group:n }
                {
                  \tl_if_empty:nTF {#1}
                    { \exp_after:wN \@@_e_end:nn }
                    { \exp_after:wN \@@_e_space:nn }
                  \exp_after:wN { \if_false: } \fi:
                }
            }
          #1
        }
      }
    \cs_new:Npn \@@_e_end:nn #1#2 { \exp_end: #2 }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\@@_e_space:nn}
%   For an explicit space character, remove it by \texttt{f}-expansion
%   and put it in the (future) output.
%    \begin{macrocode}
    \cs_new:Npn \@@_e_space:nn #1#2
      { \exp_args:Nf \@@_e:nn {#1} { #2 ~ } }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\@@_e_group:n, \@@_e_put:nn, \@@_e_put:nnn}
%   For a group, expand its contents, wrap it in two pairs of braces,
%   and call \cs{@@_e_put:nn}.  This function places the first item (the
%   double-brace wrapped result) into the output.  Importantly,
%   \cs{tl_head:n} works even if the input contains quarks.
%    \begin{macrocode}
    \cs_new:Npn \@@_e_group:n #1
      {
        \exp_after:wN \@@_e_put:nn
        \exp_after:wN { \exp_after:wN { \exp_after:wN {
              \exp:w \if_false: } \fi: \@@_e:nn {#1} { } } }
      }
    \cs_new:Npn \@@_e_put:nn #1
      {
        \exp_args:NNo \exp_args:No \@@_e_put:nnn
          { \tl_head:n {#1} } {#1}
      }
    \cs_new:Npn \@@_e_put:nnn #1#2#3
      { \exp_args:No \@@_e:nn { \use_none:n #2 } { #3 #1 } }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]
%   {\@@_e:N, \@@_e:Nnn, \@@_e_protected:Nnn, \@@_e_expandable:Nnn}
%   For an \texttt{N}-type token, call \cs{@@_e:Nnn} with arguments the
%   \meta{first token}, the remaining tokens to expand and what's
%   already been expanded.  If the \meta{first token} is non-expandable,
%   including \tn{protected} (\tn{long} or not) macros, it is put in the
%   result by \cs{@@_e_protected:Nnn}.  The four special primitives
%   \tn{unexpanded}, \tn{noexpand}, \tn{the}, \tn{primitive} are
%   detected; otherwise the token is expanded by
%   \cs{@@_e_expandable:Nnn}.
%    \begin{macrocode}
    \cs_new:Npn \@@_e:N #1
      {
        \exp_after:wN \@@_e:Nnn
        \exp_after:wN #1
        \exp_after:wN { \if_false: } \fi:
      }
    \cs_new:Npn \@@_e:Nnn #1
      {
        \if_case:w
          \exp_after:wN \if_meaning:w \exp_not:N #1 #1 1 ~ \fi:
          \token_if_protected_macro:NT #1 { 1 ~ }
          \token_if_protected_long_macro:NT #1 { 1 ~ }
          \if_meaning:w \exp_not:n #1 2 ~ \fi:
          \if_meaning:w \exp_not:N #1 3 ~ \fi:
          \if_meaning:w \tex_the:D #1 4 ~ \fi:
          \if_meaning:w \tex_primitive:D #1 5 ~ \fi:
          0 ~
          \exp_after:wN \@@_e_expandable:Nnn
        \or: \exp_after:wN \@@_e_protected:Nnn
        \or: \exp_after:wN \@@_e_unexpanded:Nnn
        \or: \exp_after:wN \@@_e_noexpand:Nnn
        \or: \exp_after:wN \@@_e_the:Nnn
        \or: \exp_after:wN \@@_e_primitive:Nnn
        \fi:
        #1
      }
    \cs_new:Npn \@@_e_protected:Nnn #1#2#3
      { \@@_e:nn {#2} { #3 #1 } }
    \cs_new:Npn \@@_e_expandable:Nnn #1#2
      { \exp_args:No \@@_e:nn { #1 #2 } }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]
%   {
%     \@@_e_primitive:Nnn,
%     \@@_e_primitive_aux:NNw,
%     \@@_e_primitive_aux:NNnn,
%     \@@_e_primitive_other:NNnn,
%     \@@_e_primitive_other_aux:nNNnn
%   }
%   We don't try hard to make sensible error recovery since the error
%   recovery of \cs{tex_primitive:D} when followed by something else
%   than a primitive depends on the engine.  The only valid case is when
%   what follows is \texttt{N}-type.  Then distinguish special
%   primitives \tn{unexpanded}, \tn{noexpand}, \tn{the}, \tn{primitive}
%   from other primitives.  In the \enquote{other} case, the only
%   reasonable way to check if the primitive that follows
%   \cs{tex_primitive:D} is expandable is to expand and compare the
%   before-expansion and after-expansion results.  If they coincide then
%   probably the primitive is non-expandable and should be put in the
%   output together with \cs{tex_primitive:D} (one can cook up contrived
%   counter-examples where the true \tn{expanded} would have an infinite
%   loop), and otherwise one should continue expanding.
%    \begin{macrocode}
    \cs_new:Npn \@@_e_primitive:Nnn #1#2
      {
        \if_false: { \fi:
          \tl_if_head_is_N_type:nTF {#2}
            { \@@_e_primitive_aux:NNw #1 }
            {
              \__kernel_msg_expandable_error:nnn { kernel } { e-type }
                { Missing~primitive~name }
              \@@_e_primitive_aux:NNw #1 \c_empty_tl
            }
          #2
        }
      }
    \cs_new:Npn \@@_e_primitive_aux:NNw #1#2
      {
        \exp_after:wN \@@_e_primitive_aux:NNnn
        \exp_after:wN #1
        \exp_after:wN #2
        \exp_after:wN { \if_false: } \fi:
      }
    \cs_new:Npn \@@_e_primitive_aux:NNnn #1#2
      {
        \exp_args:Nf \str_case_e:nnTF { \cs_to_str:N #2 }
          {
            { unexpanded } { \@@_e_unexpanded:Nnn \exp_not:n }
            { noexpand } { \@@_e_noexpand:Nnn \exp_not:N }
            { the } { \@@_e_the:Nnn \tex_the:D }
            {
              \sys_if_engine_xetex:T { pdf }
              \sys_if_engine_luatex:T { pdf }
              primitive
            } { \@@_e_primitive:Nnn #1 }
          }
          { \@@_e_primitive_other:NNnn #1 #2 }
      }
    \cs_new:Npn \@@_e_primitive_other:NNnn #1#2#3
      {
        \exp_args:No \@@_e_primitive_other_aux:nNNnn
          { #1 #2 #3 }
          #1 #2 {#3}
      }
    \cs_new:Npn \@@_e_primitive_other_aux:nNNnn #1#2#3#4#5
      {
        \str_if_eq:nnTF {#1} { #2 #3 #4 }
          { \@@_e:nn {#4} { #5 #2 #3 } }
          { \@@_e:nn {#1} {#5} }
      }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\@@_e_noexpand:Nnn}
%   The \tn{noexpand} primitive has no effect when followed by a token
%   that is not \texttt{N}-type; otherwise \cs{@@_e_put:nn} can grab the
%   next token and put it in the result unchanged.
%    \begin{macrocode}
    \cs_new:Npn \@@_e_noexpand:Nnn #1#2
      {
        \tl_if_head_is_N_type:nTF {#2}
          { \@@_e_put:nn } { \@@_e:nn } {#2}
      }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]
%   {
%     \@@_e_unexpanded:Nnn, \@@_e_unexpanded:nn,
%     \@@_e_unexpanded:nN, \@@_e_unexpanded:N
%   }
%   The \tn{unexpanded} primitive expands and ignores any space,
%   \cs{scan_stop:}, or token affected by \cs{exp_not:N}, then expects a
%   brace group.  Since we only support brace-balanced token lists it is
%   impossible to support the case where the argument of \tn{unexpanded}
%   starts with an implicit brace.  Even though we want to expand and
%   ignore spaces we cannot blindly \texttt{f}-expand because tokens
%   affected by \cs{exp_not:N} should discarded without being expanded
%   further.
%
%   As usual distinguish four cases: brace group (the normal case, where
%   we just put the item in the result), space (just \texttt{f}-expand
%   to remove the space), empty (an error), or \texttt{N}-type
%   \meta{token}.  In the last case call \cs{@@_e_unexpanded:nN}
%   triggered by an \texttt{f}-expansion.  Having a non-expandable
%   \meta{token} after \tn{unexpanded} is an error (we recover by
%   passing |{}| to \tn{unexpanded}; this is different from \TeX{}
%   because the error recovery of \tn{unexpanded} changes the balance of
%   braces), unless that \meta{token} is \cs{scan_stop:} or a space
%   (recall that we don't implement the case of an implicit begin-group
%   token).  An expandable \meta{token} is instead expanded, unless it
%   is \tn{noexpand}.  The latter primitive can be followed by an expandable
%   \texttt{N}-type token (removed), by a non-expandable one (kept
%   and later causing an error), by a space (removed by
%   \texttt{f}-expansion), or by a brace group or nothing (later causing
%   an error).
%    \begin{macrocode}
    \cs_new:Npn \@@_e_unexpanded:Nnn #1 { \@@_e_unexpanded:nn }
    \cs_new:Npn \@@_e_unexpanded:nn #1
      {
        \tl_if_head_is_N_type:nTF {#1}
          {
            \exp_args:Nf \@@_e_unexpanded:nn
              { \@@_e_unexpanded:nN {#1} #1 }
          }
          {
            \tl_if_head_is_group:nTF {#1}
              { \@@_e_put:nn }
              {
                \tl_if_empty:nTF {#1}
                  {
                    \__kernel_msg_expandable_error:nnn
                      { kernel } { e-type }
                      { \unexpanded missing~brace }
                    \@@_e_end:nn
                  }
                  { \exp_args:Nf \@@_e_unexpanded:nn }
              }
            {#1}
          }
      }
    \cs_new:Npn \@@_e_unexpanded:nN #1#2
      {
        \exp_after:wN \if_meaning:w \exp_not:N #2 #2
          \exp_after:wN \use_i:nn
        \else:
          \exp_after:wN \use_ii:nn
        \fi:
        {
          \token_if_eq_catcode:NNTF #2 \c_space_token
            { \exp_stop_f: }
            {
              \token_if_eq_meaning:NNTF #2 \scan_stop:
                { \exp_stop_f: }
                {
                  \__kernel_msg_expandable_error:nnn
                    { kernel } { e-type }
                    { \unexpanded missing~brace }
                  { }
                }
            }
        }
        {
          \token_if_eq_meaning:NNTF #2 \exp_not:N
            {
              \exp_args:No \tl_if_head_is_N_type:nT { \use_none:n #1 }
                { \@@_e_unexpanded:N }
            }
            { \exp_after:wN \exp_stop_f: #2 }
        }
      }
    \cs_new:Npn \@@_e_unexpanded:N #1
      {
        \exp_after:wN \if_meaning:w \exp_not:N #1 #1 \else:
          \exp_after:wN \use_i:nn
        \fi:
        \exp_stop_f: #1
      }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\@@_e_the:Nnn, \@@_e_the:N, \@@_e_the_toks_reg:N}
%   Finally implement \tn{the}.  Followed by anything other than an
%   \texttt{N}-type \meta{token} this causes an error (we just let
%   \TeX{} make one), otherwise we test the \meta{token}.  If the
%   \meta{token} is expandable, expand it.  Otherwise it could be any
%   kind of register, or things like \tn{numexpr}, so there is no way to
%   deal with all cases.  Thankfully, only \tn{toks} data needs to be
%   protected from expansion since everything else gives a string of
%   characters.  If the \meta{token} is \tn{toks} we find a number and
%   unpack using the |the_toks| functions.  If it is a token register we
%   unpack it in a brace group and call \cs{@@_e_put:nn} to move it to
%   the result.  Otherwise we unpack and continue expanding (useless but
%   safe) since it is basically impossible to have a handle on where the
%   result of \tn{the} ends.
%    \begin{macrocode}
    \cs_new:Npn \@@_e_the:Nnn #1#2
      {
        \tl_if_head_is_N_type:nTF {#2}
          { \if_false: { \fi: \@@_e_the:N #2 } }
          { \exp_args:No \@@_e:nn { \tex_the:D #2 } }
      }
    \cs_new:Npn \@@_e_the:N #1
      {
        \exp_after:wN \if_meaning:w \exp_not:N #1 #1
          \exp_after:wN \use_i:nn
        \else:
          \exp_after:wN \use_ii:nn
        \fi:
        {
          \if_meaning:w \tex_toks:D #1
            \exp_after:wN \@@_e_the_toks:wnn \int_value:w
            \exp_after:wN \@@_e_the_toks:n
            \exp_after:wN { \int_value:w \if_false: } \fi:
          \else:
            \@@_e_if_toks_register:NTF #1
              { \exp_after:wN \@@_e_the_toks_reg:N }
              {
                \exp_after:wN \@@_e:nn \exp_after:wN {
                  \tex_the:D \if_false: } \fi:
              }
            \exp_after:wN #1
          \fi:
        }
        {
          \exp_after:wN \@@_e_the:Nnn \exp_after:wN ?
          \exp_after:wN { \exp:w \if_false: } \fi:
          \exp_after:wN \exp_end: #1
        }
      }
    \cs_new:Npn \@@_e_the_toks_reg:N #1
      {
        \exp_after:wN \@@_e_put:nn \exp_after:wN {
          \exp_after:wN {
            \tex_the:D \if_false: } \fi: #1 }
      }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\@@_e_the_toks:wnn, \@@_e_the_toks:n, \@@_e_the_toks:N}
%   The calling function has applied \cs{int_value:w} so we collect
%   digits with \cs{@@_e_the_toks:n} (which gets the token list as an
%   argument) and \cs{@@_e_the_toks:N} (which gets the first token in
%   case it is \texttt{N}-type).  The digits are themselves collected
%   into an \cs{int_value:w} argument to \cs{@@_e_the_toks:wnn}.  Then
%   that function unpacks the \tn{toks}\meta{number} into the result.
%   We include |?| because \cs{@@_e_put:nnn} removes one item from its
%   second argument.  Note that our approach is rather crude: in cases
%   like |\the\toks12~34| the first \cs{int_value:w} removes the space
%   and we will incorrectly unpack the |\the\toks1234|.
%    \begin{macrocode}
    \cs_new:Npn \@@_e_the_toks:wnn #1; #2
      {
        \exp_args:No \@@_e_put:nnn
          { \tex_the:D \tex_toks:D #1 } { ? #2 }
      }
    \cs_new:Npn \@@_e_the_toks:n #1
      {
        \tl_if_head_is_N_type:nTF {#1}
          { \exp_after:wN \@@_e_the_toks:N \if_false: { \fi: #1 } }
          { ; {#1} }
      }
    \cs_new:Npn \@@_e_the_toks:N #1
      {
        \if_int_compare:w 10 < 9 \token_to_str:N #1 \exp_stop_f:
          \exp_after:wN \use_i:nn
        \else:
          \exp_after:wN \use_ii:nn
        \fi:
        {
          #1
          \exp_after:wN \@@_e_the_toks:n
          \exp_after:wN { \if_false: } \fi:
        }
        {
          \exp_after:wN ;
          \exp_after:wN { \if_false: } \fi: #1
        }
      }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[EXP]{\@@_e_if_toks_register:NTF}
% \begin{macro}[EXP]
%   {
%     \@@_e_the_XeTeXinterchartoks:,
%     \@@_e_the_errhelp:,
%     \@@_e_the_everycr:,
%     \@@_e_the_everydisplay:,
%     \@@_e_the_everyeof:,
%     \@@_e_the_everyhbox:,
%     \@@_e_the_everyjob:,
%     \@@_e_the_everymath:,
%     \@@_e_the_everypar:,
%     \@@_e_the_everyvbox:,
%     \@@_e_the_output:,
%     \@@_e_the_pdfpageattr:,
%     \@@_e_the_pdfpageresources:,
%     \@@_e_the_pdfpagesattr:,
%     \@@_e_the_pdfpkmode:
%   }
%   We need to detect both \tn{toks} registers like \tn{toks@} in
%   \LaTeXe{} and parameters such as \tn{everypar}, as the result of
%   unpacking the register should not expand further.  Registers are
%   found by \cs{token_if_toks_register:NTF} by inspecting the meaning.
%   The list of parameters is finite so we just use a
%   \cs{cs_if_exist:cTF} test to look up in a table.  We abuse
%   \cs{cs_to_str:N}'s ability to remove a leading escape character
%   whatever it is.
%    \begin{macrocode}
    \prg_new_conditional:Npnn \@@_e_if_toks_register:N #1 { TF }
      {
        \token_if_toks_register:NTF #1 { \prg_return_true: }
          {
            \cs_if_exist:cTF
              {
                @@_e_the_
                \exp_after:wN \cs_to_str:N
                \token_to_meaning:N #1
                :
              } { \prg_return_true: } { \prg_return_false: }
          }
      }
    \cs_new_eq:NN \@@_e_the_XeTeXinterchartoks: ?
    \cs_new_eq:NN \@@_e_the_errhelp: ?
    \cs_new_eq:NN \@@_e_the_everycr: ?
    \cs_new_eq:NN \@@_e_the_everydisplay: ?
    \cs_new_eq:NN \@@_e_the_everyeof: ?
    \cs_new_eq:NN \@@_e_the_everyhbox: ?
    \cs_new_eq:NN \@@_e_the_everyjob: ?
    \cs_new_eq:NN \@@_e_the_everymath: ?
    \cs_new_eq:NN \@@_e_the_everypar: ?
    \cs_new_eq:NN \@@_e_the_everyvbox: ?
    \cs_new_eq:NN \@@_e_the_output: ?
    \cs_new_eq:NN \@@_e_the_pdfpageattr: ?
    \cs_new_eq:NN \@@_e_the_pdfpageresources: ?
    \cs_new_eq:NN \@@_e_the_pdfpagesattr: ?
    \cs_new_eq:NN \@@_e_the_pdfpkmode: ?
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% We are done emulating \texttt{e}-type argument expansion when
% \tn{expanded} is unavailable.
%    \begin{macrocode}
  }
%    \end{macrocode}
%
%
% \subsection{Defining function variants}
%
%    \begin{macrocode}
%<@@=cs>
%    \end{macrocode}
%
% \begin{variable}{\s_@@_mark,\s_@@_stop}
%   Internal scan marks. No \pkg{l3quark} yet, so do things by hand.
%    \begin{macrocode}
\cs_new_eq:NN \s_@@_mark \scan_stop:
\cs_new_eq:NN \s_@@_stop \scan_stop:
%    \end{macrocode}
% \end{variable}
%
% \begin{variable}{\q_@@_recursion_stop}
%   Internal recursion quarks. No \pkg{l3quark} yet, so do things by hand.
%    \begin{macrocode}
\cs_new:Npn \q_@@_recursion_stop { \q_@@_recursion_stop }
%    \end{macrocode}
% \end{variable}
%
% \begin{macro}[EXP]{
%     \@@_use_none_delimit_by_s_stop:w,
%     \@@_use_i_delimit_by_s_stop:nw,
%     \@@_use_none_delimit_by_q_recursion_stop:w
%   }
%   Internal scan marks.
%    \begin{macrocode}
\cs_new:Npn \@@_use_none_delimit_by_s_stop:w #1 \s_@@_stop { }
\cs_new:Npn \@@_use_i_delimit_by_s_stop:nw #1 #2 \s_@@_stop {#1}
\cs_new:Npn \@@_use_none_delimit_by_q_recursion_stop:w
  #1 \q_@@_recursion_stop { }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\cs_generate_variant:Nn, \cs_generate_variant:cn}
%   \begin{arguments}
%     \item Base form of a function; \emph{e.g.},~\cs{tl_set:Nn}
%     \item One or more variant argument specifiers; e.g., |{Nx,c,cx}|
%   \end{arguments}
%   After making sure that the base form exists, test whether it is
%   protected or not and define \cs{@@_tmp:w} as either
%   \cs{cs_new:Npx} or \cs{cs_new_protected:Npx}, which is
%   then used to define all the variants (except those involving
%   \texttt{x}-expansion, always protected).  Split up the original base
%   function only once, to grab its name and signature.  Then we wish to
%   iterate through the comma list of variant argument specifiers, which
%   we first convert to a string: the reason is explained later.
%    \begin{macrocode}
\cs_new_protected:Npn \cs_generate_variant:Nn #1#2
  {
    \@@_generate_variant:N #1
    \use:x
      {
        \@@_generate_variant:nnNN
          \cs_split_function:N #1
          \exp_not:N #1
          \tl_to_str:n {#2} ,
            \exp_not:N \scan_stop: ,
            \exp_not:N \q_@@_recursion_stop
      }
  }
\cs_new_protected:Npn \cs_generate_variant:cn
  { \exp_args:Nc \cs_generate_variant:Nn }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\@@_generate_variant:N}
% \begin{macro}{\@@_generate_variant:ww, \@@_generate_variant:wwNw}
%   The goal here is to pick up protected parent functions.  There are
%   four cases: the parent function can be a primitive or a macro, and
%   can be expandable or not.  For non-expandable primitives, all
%   variants should be protected; skipping the \cs{else:} branch is safe
%   because non-expandable primitives cannot be \TeX{} conditionals.
%
%   The other case where variants should be protected is when the parent
%   function is a protected macro: then |protected| appears in the
%   meaning before the fist occurrence of |macro|.  The |ww| auxiliary
%   removes everything in the meaning string after the first |ma|.  We
%   use |ma| rather than the full |macro| because the meaning of the
%   \tn{firstmark} primitive (and four others) can contain an arbitrary
%   string after a leading |firstmark:|.  Then, look for |pr| in the
%   part we extracted: no need to look for anything longer: the only
%   strings we can have are an empty string, \verb*|\long |,
%   \verb*|\protected |, \verb*|\protected\long |, |\first|, |\top|,
%   |\bot|, |\splittop|, or |\splitbot|, with |\| replaced by the
%   appropriate escape character.  If |pr| appears in the part before
%   |ma|, the first \cs{s_@@_mark} is taken as an argument of the |wwNw|
%   auxiliary, and |#3| is \cs{cs_new_protected:Npx}, otherwise it
%   is \cs{cs_new:Npx}.
%    \begin{macrocode}
\cs_new_protected:Npx \@@_generate_variant:N #1
  {
    \exp_not:N \exp_after:wN \exp_not:N \if_meaning:w
      \exp_not:N \exp_not:N #1 #1
      \cs_set_eq:NN \exp_not:N \@@_tmp:w \cs_new_protected:Npx
    \exp_not:N \else:
      \exp_not:N \exp_after:wN \exp_not:N \@@_generate_variant:ww
        \exp_not:N \token_to_meaning:N #1 \tl_to_str:n { ma }
          \s_@@_mark
        \s_@@_mark \cs_new_protected:Npx
        \tl_to_str:n { pr }
        \s_@@_mark \cs_new:Npx
        \s_@@_stop
    \exp_not:N \fi:
  }
\exp_last_unbraced:NNNNo
  \cs_new_protected:Npn \@@_generate_variant:ww
    #1 { \tl_to_str:n { ma } } #2 \s_@@_mark
    { \@@_generate_variant:wwNw #1 }
\exp_last_unbraced:NNNNo
  \cs_new_protected:Npn \@@_generate_variant:wwNw
    #1 { \tl_to_str:n { pr } } #2 \s_@@_mark #3 #4 \s_@@_stop
    { \cs_set_eq:NN \@@_tmp:w #3 }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\@@_generate_variant:nnNN}
%   \begin{arguments}
%     \item Base name.
%     \item Base signature.
%     \item Boolean.
%     \item Base function.
%   \end{arguments}
%   If the boolean is \cs{c_false_bool}, the base function has no colon
%   and we abort with an error; otherwise, set off a loop through the
%   desired variant forms. The original function is retained as |#4| for
%   efficiency.
%    \begin{macrocode}
\cs_new_protected:Npn \@@_generate_variant:nnNN #1#2#3#4
  {
    \if_meaning:w \c_false_bool #3
      \__kernel_msg_error:nnx { kernel } { missing-colon }
        { \token_to_str:c {#1} }
      \exp_after:wN \@@_use_none_delimit_by_q_recursion_stop:w
    \fi:
    \@@_generate_variant:Nnnw #4 {#1}{#2}
  }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\@@_generate_variant:Nnnw}
%   \begin{arguments}
%     \item Base function.
%     \item Base name.
%     \item Base signature.
%     \item Beginning of variant signature.
%   \end{arguments}
%   First check whether to terminate the loop over variant forms.  Then,
%   for each variant form, construct a new function name using the
%   original base name, the variant signature consisting of $l$ letters
%   and the last $k-l$ letters of the base signature (of length $k$).
%   For example, for a base function \cs{prop_put:Nnn} which needs a
%   |cV| variant form, we want the new signature to be |cVn|.
%
%   There are further subtleties:
%   \begin{itemize}
%     \item In \cs{cs_generate_variant:Nn} |\foo:nnTF| |{xxTF}|, we must define
%       |\foo:xxTF| using |\exp_args:Nxx|,
%       rather than a hypothetical |\exp_args:NxxTF|.  Thus, we wish to
%       trim a common trailing part from the base signature and the
%       variant signature.
%     \item In \cs{cs_generate_variant:Nn} |\foo:on| |{ox}|, the
%       function |\foo:ox| must be defined using |\exp_args:Nnx|, not
%       |\exp_args:Nox|, to avoid double |o| expansion.
%     \item Lastly, \cs{cs_generate_variant:Nn} |\foo:on| |{xn}| must
%       trigger an error, because we do not have a means to replace
%       |o|-expansion by |x|-expansion.
%       More generally, we can only convert |N| to |c|, or convert |n|
%       to |V|, |v|, |o|, |f|, |x|.
%     \end{itemize}
%     All this boils down to a few rules.  Only |n| and |N|-type
%     arguments can be replaced by \cs{cs_generate_variant:Nn}.  Other
%     argument types are allowed to be passed unchanged from the base
%     form to the variant: in the process they are changed to |n|
%     except for |N| and |p|-type arguments.  A common trailing
%     part is ignored.
%
%     We compare the base and variant signatures one character at a time
%     within |x|-expansion.  The result is given to
%     \cs{@@_generate_variant:wwNN} (defined later) in the form
%     \meta{processed variant signature} \cs{s_@@_mark} \meta{errors}
%     \cs{s_@@_stop} \meta{base function} \meta{new function}.  If all went
%     well, \meta{errors} is empty; otherwise, it is a kernel error
%     message and some clean-up code.
%
%     Note the space after |#3| and after the following brace group.
%     Those are ignored by \TeX{} when fetching the last argument for
%     \cs{@@_generate_variant_loop:nNwN}, but can be used as a delimiter
%     for \cs{@@_generate_variant_loop_end:nwwwNNnn}.
%    \begin{macrocode}
\cs_new_protected:Npn \@@_generate_variant:Nnnw #1#2#3#4 ,
  {
    \if_meaning:w \scan_stop: #4
      \exp_after:wN \@@_use_none_delimit_by_q_recursion_stop:w
    \fi:
    \use:x
      {
        \exp_not:N \@@_generate_variant:wwNN
        \@@_generate_variant_loop:nNwN { }
          #4
          \@@_generate_variant_loop_end:nwwwNNnn
          \s_@@_mark
          #3 ~
          { ~ { } \fi: \@@_generate_variant_loop_long:wNNnn } ~
          { }
          \s_@@_stop
        \exp_not:N #1 {#2} {#4}
      }
    \@@_generate_variant:Nnnw #1 {#2} {#3}
  }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[rEXP]
%   {
%     \@@_generate_variant_loop:nNwN,
%     \@@_generate_variant_loop_base:N,
%     \@@_generate_variant_loop_same:w,
%     \@@_generate_variant_loop_end:nwwwNNnn,
%     \@@_generate_variant_loop_long:wNNnn,
%     \@@_generate_variant_loop_invalid:NNwNNnn,
%     \@@_generate_variant_loop_special:NNwNNnn
%   }
%   \begin{arguments}
%     \item Last few consecutive letters common between the base and
%       variant (more precisely, \cs{@@_generate_variant_same:N}
%       \meta{letter} for each letter).
%     \item Next variant letter.
%     \item Remainder of variant form.
%     \item Next base letter.
%   \end{arguments}
%   The first argument is populated by
%   \cs{@@_generate_variant_loop_same:w} when a variant letter and a
%   base letter match.  It is flushed into the input stream whenever the
%   two letters are different: if the loop ends before, the argument is
%   dropped, which means that trailing common letters are ignored.
%
%   The case where the two letters are different is only allowed if the
%   base is |N| and the variant is |c|, or when the base is |n| and the
%   variant is |o|, |V|, |v|, |f| or |x|.  Otherwise, call
%   \cs{@@_generate_variant_loop_invalid:NNwNNnn} to remove the end of
%   the loop, get arguments at the end of the loop, and place an
%   appropriate error message as a second argument of
%   \cs{@@_generate_variant:wwNN}.  If the letters are distinct and the
%   base letter is indeed |n| or |N|, leave in the input stream whatever
%   argument |#1| was collected, and the next variant letter |#2|, then
%   loop by calling \cs{@@_generate_variant_loop:nNwN}.
%
%   The loop can stop in three ways.
%   \begin{itemize}
%     \item If the end of the variant form is encountered first, |#2| is
%       \cs{@@_generate_variant_loop_end:nwwwNNnn} (expanded by the
%       conditional \cs{if:w}), which inserts some tokens to end the
%       conditional; grabs the \meta{base name} as |#7|, the
%       \meta{variant signature} |#8|, the \meta{next base letter} |#1|
%       and the part |#3| of the base signature that wasn't read yet;
%       and combines those into the \meta{new function} to be defined.
%     \item If the end of the base form is encountered first, |#4| is
%       |~{}\fi:| which ends the conditional (with an empty expansion),
%       followed by \cs{@@_generate_variant_loop_long:wNNnn}, which
%       places an error as the second argument of
%       \cs{@@_generate_variant:wwNN}.
%     \item The loop can be interrupted early if the requested expansion
%       is unavailable, namely when the variant and base letters differ
%       and the base is not the right one (|n| or |N| to support the
%       variant).  In that case too an error is placed as the second
%       argument of \cs{@@_generate_variant:wwNN}.
%   \end{itemize}
%   Note that if the variant form has the same length as the base form,
%   |#2| is as described in the first point, and |#4| as described in
%   the second point above.  The \cs{@@_generate_variant_loop_end:nwwwNNnn}
%   breaking function takes the empty brace group in |#4| as its first
%   argument: this empty brace group produces the correct signature for
%   the full variant.
%    \begin{macrocode}
\cs_new:Npn \@@_generate_variant_loop:nNwN #1#2#3 \s_@@_mark #4
  {
    \if:w #2 #4
      \exp_after:wN \@@_generate_variant_loop_same:w
    \else:
      \if:w #4 \@@_generate_variant_loop_base:N #2 \else:
        \if:w 0
          \if:w N #4 \else: \if:w n #4 \else: 1 \fi: \fi:
          \if:w \scan_stop: \@@_generate_variant_loop_base:N #2 1 \fi:
          0
          \@@_generate_variant_loop_special:NNwNNnn #4#2
        \else:
          \@@_generate_variant_loop_invalid:NNwNNnn #4#2
        \fi:
      \fi:
    \fi:
    #1
    \prg_do_nothing:
    #2
    \@@_generate_variant_loop:nNwN { } #3 \s_@@_mark
  }
\cs_new:Npn \@@_generate_variant_loop_base:N #1
  {
    \if:w c #1 N \else:
      \if:w o #1 n \else:
        \if:w V #1 n \else:
          \if:w v #1 n \else:
            \if:w f #1 n \else:
              \if:w e #1 n \else:
                \if:w x #1 n \else:
                  \if:w n #1 n \else:
                    \if:w N #1 N \else:
                      \scan_stop:
                    \fi:
                  \fi:
                \fi:
              \fi:
            \fi:
          \fi:
        \fi:
      \fi:
    \fi:
  }
\cs_new:Npn \@@_generate_variant_loop_same:w
    #1 \prg_do_nothing: #2#3#4
  { #3 { #1 \@@_generate_variant_same:N #2 } }
\cs_new:Npn \@@_generate_variant_loop_end:nwwwNNnn
    #1#2 \s_@@_mark #3 ~ #4 \s_@@_stop #5#6#7#8
  {
    \scan_stop: \scan_stop: \fi:
    \s_@@_mark \s_@@_stop
    \exp_not:N #6
    \exp_not:c { #7 : #8 #1 #3 }
  }
\cs_new:Npn \@@_generate_variant_loop_long:wNNnn #1 \s_@@_stop #2#3#4#5
  {
    \exp_not:n
      {
        \s_@@_mark
        \__kernel_msg_error:nnxx { kernel } { variant-too-long }
          {#5} { \token_to_str:N #3 }
        \use_none:nnn
        \s_@@_stop
        #3
        #3
      }
  }
\cs_new:Npn \@@_generate_variant_loop_invalid:NNwNNnn
    #1#2 \fi: \fi: \fi: #3 \s_@@_stop #4#5#6#7
  {
    \fi: \fi: \fi:
    \exp_not:n
      {
        \s_@@_mark
        \__kernel_msg_error:nnxxxx { kernel } { invalid-variant }
          {#7} { \token_to_str:N #5 } {#1} {#2}
        \use_none:nnn
        \s_@@_stop
        #5
        #5
      }
  }
\cs_new:Npn \@@_generate_variant_loop_special:NNwNNnn
  #1#2#3 \s_@@_stop #4#5#6#7
  {
    #3 \s_@@_stop #4 #5 {#6} {#7}
    \exp_not:n
      {
        \__kernel_msg_error:nnxxxx
          { kernel } { deprecated-variant }
          {#7} { \token_to_str:N #5 } {#1} {#2}
      }
  }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}[rEXP]{\@@_generate_variant_same:N}
%   When the base and variant letters are identical, don't do any
%   expansion.  For most argument types, we can use the |n|-type
%   no-expansion, but the |N| and |p| types require a slightly different
%   behaviour with respect to braces.  For |V|-type this function could
%   output |N| to avoid adding useless braces but that is not a problem.
%    \begin{macrocode}
\cs_new:Npn \@@_generate_variant_same:N #1
  {
    \if:w N #1 #1 \else:
      \if:w p #1 #1 \else:
        \token_to_str:N n
        \if:w n #1 \else:
          \@@_generate_variant_loop_special:NNwNNnn #1#1
        \fi:
      \fi:
    \fi:
  }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\@@_generate_variant:wwNN}
%   If the variant form has already been defined, log its existence
%   (provided \texttt{log-functions} is active).
%   Otherwise, make sure that the |\exp_args:N #3| form is defined, and
%   if it contains |x|, change \cs{@@_tmp:w} locally to
%   \cs{cs_new_protected:Npx}.  Then define the variant by
%   combining the |\exp_args:N #3| variant and the base function.
%    \begin{macrocode}
\cs_new_protected:Npn \@@_generate_variant:wwNN
    #1 \s_@@_mark #2 \s_@@_stop #3#4
  {
    #2
    \cs_if_free:NT #4
      {
        \group_begin:
          \@@_generate_internal_variant:n {#1}
          \@@_tmp:w #4 { \exp_not:c { exp_args:N #1 } \exp_not:N #3 }
        \group_end:
      }
  }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\@@_generate_internal_variant:n}
% \begin{macro}[rEXP]{\@@_generate_internal_variant_loop:n}
%   First test for the presence of |x| (this is where working with
%   strings makes our lives easier), as the result should be protected,
%   and the next variant to be defined using that internal variant
%   should be protected (done by setting \cs{@@_tmp:w}).  Then call
%   \cs{@@_generate_internal_variant:NNn} with arguments
%   \cs{cs_new_protected:cpn} \cs{use:x} (for protected) or
%   \cs{cs_new:cpn} \cs{tex_expanded:D} (expandable) and the signature.  If |p|
%   appears in the signature, or if the function to be defined is
%   expandable and the primitive \tn{expanded} is not available, or if there
%   are more than $8$ arguments, call
%   some fall-back code that just puts the appropriate |\::| commands.
%   Otherwise, call \cs{@@_generate_internal_one_go:NNn} to construct
%   the \cs[no-index]{exp_args:N\dots} function as a macro taking up to
%   $9$~arguments and expanding them using \cs{use:x} or \cs{tex_expanded:D}.
%    \begin{macrocode}
\cs_new_protected:Npx \@@_generate_internal_variant:n #1
  {
    \exp_not:N \@@_generate_internal_variant:wwnNwn
      #1 \s_@@_mark
        { \cs_set_eq:NN \exp_not:N \@@_tmp:w \cs_new_protected:Npx }
        \cs_new_protected:cpn
        \use:x
      \token_to_str:N x \s_@@_mark
        { }
        \cs_new:cpn
        \exp_not:N \tex_expanded:D
    \s_@@_stop
      {#1}
  }
\exp_last_unbraced:NNNNo
  \cs_new_protected:Npn \@@_generate_internal_variant:wwnNwn #1
    { \token_to_str:N x } #2 \s_@@_mark #3#4#5#6 \s_@@_stop #7
  {
    #3
    \cs_if_free:cT { exp_args:N #7 }
      { \@@_generate_internal_variant:NNn #4 #5 {#7} }
  }
\cs_set_protected:Npn \@@_tmp:w #1
  {
    \cs_new_protected:Npn \@@_generate_internal_variant:NNn ##1##2##3
      {
        \if_catcode:w X \use_none:nnnnnnnn ##3
            \prg_do_nothing: \prg_do_nothing: \prg_do_nothing:
            \prg_do_nothing: \prg_do_nothing: \prg_do_nothing:
            \prg_do_nothing: \prg_do_nothing: X
          \exp_after:wN \@@_generate_internal_test:Nw \exp_after:wN ##2
        \else:
          \exp_after:wN \@@_generate_internal_test_aux:w \exp_after:wN #1
        \fi:
        ##3
        \s_@@_mark
        {
          \use:x
            {
              ##1 { exp_args:N ##3 }
                { \@@_generate_internal_variant_loop:n ##3 { : \use_i:nn } }
            }
        }
        #1
        \s_@@_mark
        { \exp_not:n { \@@_generate_internal_one_go:NNn ##1 ##2 {##3} } }
        \s_@@_stop
      }
    \cs_new_protected:Npn \@@_generate_internal_test_aux:w
        ##1 #1 ##2 \s_@@_mark ##3 ##4 \s_@@_stop {##3}
    \cs_if_exist:NTF \tex_expanded:D
      {
        \cs_new_eq:NN \@@_generate_internal_test:Nw
          \@@_generate_internal_test_aux:w
      }
      {
        \cs_new_protected:Npn \@@_generate_internal_test:Nw ##1
          {
            \if_meaning:w \tex_expanded:D ##1
              \exp_after:wN \@@_generate_internal_test_aux:w
              \exp_after:wN #1
            \else:
              \exp_after:wN \@@_generate_internal_test_aux:w
            \fi:
          }
      }
  }
\exp_args:No \@@_tmp:w { \token_to_str:N p }
\cs_new_protected:Npn \@@_generate_internal_one_go:NNn #1#2#3
  {
    \@@_generate_internal_loop:nwnnw
      { \exp_not:N ##1 } 1 . { } { }
      #3 { ? \@@_generate_internal_end:w } X ;
      23456789 { ? \@@_generate_internal_long:w } ;
    #1 #2 {#3}
  }
\cs_new_protected:Npn \@@_generate_internal_loop:nwnnw #1#2 . #3#4#5#6 ; #7
  {
    \use_none:n #5
    \use_none:n #7
    \cs_if_exist_use:cF { @@_generate_internal_#5:NN }
      { \@@_generate_internal_other:NN }
        #5 #7
    #7 .
    { #3 #1 } { #4 ## #2 }
    #6 ;
  }
\cs_new_protected:Npn \@@_generate_internal_N:NN #1#2
  { \@@_generate_internal_loop:nwnnw { \exp_not:N ###2 } }
\cs_new_protected:Npn \@@_generate_internal_c:NN #1#2
  { \exp_args:No \@@_generate_internal_loop:nwnnw { \exp_not:c {###2} } }
\cs_new_protected:Npn \@@_generate_internal_n:NN #1#2
  { \@@_generate_internal_loop:nwnnw { { \exp_not:n {###2} } } }
\cs_new_protected:Npn \@@_generate_internal_x:NN #1#2
  { \@@_generate_internal_loop:nwnnw { {###2} } }
\cs_new_protected:Npn \@@_generate_internal_other:NN #1#2
  {
    \exp_args:No \@@_generate_internal_loop:nwnnw
      {
        \exp_after:wN
        {
          \exp:w \exp_args:NNc \exp_after:wN \exp_end:
          { exp_not:#1 } {###2}
        }
      }
  }
\cs_new_protected:Npn \@@_generate_internal_end:w #1 . #2#3#4 ; #5 ; #6#7#8
  { #6 { exp_args:N #8 } #3 { #7 {#2} } }
\cs_new_protected:Npn \@@_generate_internal_long:w #1 N #2#3 . #4#5#6#
  {
    \exp_args:Nx \@@_generate_internal_long:nnnNNn
      { \@@_generate_internal_variant_loop:n #2 #6 { : \use_i:nn } }
      {#4} {#5}
  }
\cs_new:Npn \@@_generate_internal_long:nnnNNn #1#2#3#4 ; ; #5#6#7
  { #5 { exp_args:N #7 } #3 { #6 { \exp_not:n {#1} {#2} } } }
%    \end{macrocode}
%   This command grabs char by char outputting |\::#1| (not expanded
%   further).  We avoid tests by putting a trailing |: \use_i:nn|, which
%   leaves \cs{cs_end:} and removes the looping macro.  The colon is in
%   fact also turned into \cs{:::} so that the required structure for
%   |\exp_args:N...| commands is correctly terminated.
%    \begin{macrocode}
\cs_new:Npn \@@_generate_internal_variant_loop:n #1
  {
    \exp_after:wN \exp_not:N \cs:w :: #1 \cs_end:
    \@@_generate_internal_variant_loop:n
  }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}
%   {
%     \prg_generate_conditional_variant:Nnn,
%     \@@_generate_variant:nnNnn,
%     \@@_generate_variant:w,
%     \@@_generate_variant:n,
%     \@@_generate_variant_p_form:nnn,
%     \@@_generate_variant_T_form:nnn,
%     \@@_generate_variant_F_form:nnn,
%     \@@_generate_variant_TF_form:nnn,
%   }
%    \begin{macrocode}
\cs_new_protected:Npn \prg_generate_conditional_variant:Nnn #1
  {
    \use:x
      {
        \@@_generate_variant:nnNnn
          \cs_split_function:N #1
      }
  }
\cs_new_protected:Npn \@@_generate_variant:nnNnn #1#2#3#4#5
  {
    \if_meaning:w \c_false_bool #3
      \__kernel_msg_error:nnx { kernel } { missing-colon }
        { \token_to_str:c {#1} }
      \@@_use_i_delimit_by_s_stop:nw
    \fi:
    \exp_after:wN \@@_generate_variant:w
    \tl_to_str:n {#5} , \scan_stop: , \q_@@_recursion_stop
    \@@_use_none_delimit_by_s_stop:w \s_@@_mark {#1} {#2} {#4} \s_@@_stop
  }
\cs_new_protected:Npn \@@_generate_variant:w
    #1 , #2 \s_@@_mark #3#4#5
  {
    \if_meaning:w \scan_stop: #1 \scan_stop:
      \if_meaning:w \q_@@_nil #1 \q_@@_nil
        \use_i:nnn
      \fi:
      \exp_after:wN \@@_use_none_delimit_by_q_recursion_stop:w
    \else:
      \cs_if_exist_use:cTF { @@_generate_variant_#1_form:nnn }
        { {#3} {#4} {#5} }
        {
          \__kernel_msg_error:nnxx
            { kernel } { conditional-form-unknown }
            {#1} { \token_to_str:c { #3 : #4 } }
        }
    \fi:
    \@@_generate_variant:w #2 \s_@@_mark {#3} {#4} {#5}
  }
\cs_new_protected:Npn \@@_generate_variant_p_form:nnn #1#2
  { \cs_generate_variant:cn { #1 _p : #2 } }
\cs_new_protected:Npn \@@_generate_variant_T_form:nnn #1#2
  { \cs_generate_variant:cn { #1 : #2 T } }
\cs_new_protected:Npn \@@_generate_variant_F_form:nnn #1#2
  { \cs_generate_variant:cn { #1 : #2 F } }
\cs_new_protected:Npn \@@_generate_variant_TF_form:nnn #1#2
  { \cs_generate_variant:cn { #1 : #2 TF } }
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\exp_args_generate:n}
%   This function is not used in the kernel hence we can use functions
%   that are defined in later modules.  It also does not need to be fast
%   so use inline mappings.  For each requested variant we check that
%   there are no characters besides |NnpcofVvx|, in particular that
%   there are no spaces.  Then we just call the internal function.
%    \begin{macrocode}
\cs_new_protected:Npn \exp_args_generate:n #1
  {
    \exp_args:No \clist_map_inline:nn { \tl_to_str:n {#1} }
      {
        \str_map_inline:nn {##1}
          {
            \str_if_in:nnF { NnpcofeVvx } {####1}
              {
                \__kernel_msg_error:nnnn { kernel } { invalid-exp-args }
                  {####1} {##1}
                \str_map_break:n { \use_none:nn }
              }
          }
        \@@_generate_internal_variant:n {##1}
      }
  }
%    \end{macrocode}
% \end{macro}
%
% \subsection{Definitions with the automated technique}
% \label{sec:l3expan:gendef}
%
% Some of these could be done more efficiently, but the complexity of
% coding then becomes an issue. Notice that the auto-generated functions
% actually take no arguments themselves.
%
% \begin{macro}[EXP]
%   {
%     \exp_args:Nnc, \exp_args:Nno, \exp_args:NnV, \exp_args:Nnv, \exp_args:Nne, \exp_args:Nnf,
%     \exp_args:Noc, \exp_args:Noo, \exp_args:Nof,
%     \exp_args:NVo, \exp_args:Nfo, \exp_args:Nff, \exp_args:Nee
%   }
% \begin{macro}
%   {
%     \exp_args:NNx, \exp_args:Ncx, \exp_args:Nnx,
%     \exp_args:Nox, \exp_args:Nxo, \exp_args:Nxx,
%   }
%   Here are the actual function definitions, using the helper
%   functions above.  The group is used because
%   \cs{@@_generate_internal_variant:n} redefines \cs{@@_tmp:w} locally.
%    \begin{macrocode}
\cs_set_protected:Npn \@@_tmp:w #1
  {
    \group_begin:
      \exp_args:No \@@_generate_internal_variant:n
        { \tl_to_str:n {#1} }
    \group_end:
  }
\@@_tmp:w { nc }
\@@_tmp:w { no }
\@@_tmp:w { nV }
\@@_tmp:w { nv }
\@@_tmp:w { ne }
\@@_tmp:w { nf }
\@@_tmp:w { oc }
\@@_tmp:w { oo }
\@@_tmp:w { of }
\@@_tmp:w { Vo }
\@@_tmp:w { fo }
\@@_tmp:w { ff }
\@@_tmp:w { ee }
\@@_tmp:w { Nx }
\@@_tmp:w { cx }
\@@_tmp:w { nx }
\@@_tmp:w { ox }
\@@_tmp:w { xo }
\@@_tmp:w { xx }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}[EXP]
%   {
%     \exp_args:NNcf,
%     \exp_args:NNno, \exp_args:NNnV, \exp_args:NNoo, \exp_args:NNVV,
%     \exp_args:Ncno, \exp_args:NcnV, \exp_args:Ncoo, \exp_args:NcVV,
%     \exp_args:Nnnc, \exp_args:Nnno, \exp_args:Nnnf, \exp_args:Nnff,
%     \exp_args:Nooo, \exp_args:Noof, \exp_args:Nffo, \exp_args:Neee
%   }
% \begin{macro}
%   {
%     \exp_args:NNNx, \exp_args:NNnx, \exp_args:NNox,
%     \exp_args:Nccx, \exp_args:Ncnx,
%     \exp_args:Nnnx, \exp_args:Nnox, \exp_args:Noox,
%   }
%    \begin{macrocode}
\@@_tmp:w { Ncf }
\@@_tmp:w { Nno }
\@@_tmp:w { NnV }
\@@_tmp:w { Noo }
\@@_tmp:w { NVV }
\@@_tmp:w { cno }
\@@_tmp:w { cnV }
\@@_tmp:w { coo }
\@@_tmp:w { cVV }
\@@_tmp:w { nnc }
\@@_tmp:w { nno }
\@@_tmp:w { nnf }
\@@_tmp:w { nff }
\@@_tmp:w { ooo }
\@@_tmp:w { oof }
\@@_tmp:w { ffo }
\@@_tmp:w { eee }
\@@_tmp:w { NNx }
\@@_tmp:w { Nnx }
\@@_tmp:w { Nox }
\@@_tmp:w { nnx }
\@@_tmp:w { nox }
\@@_tmp:w { ccx }
\@@_tmp:w { cnx }
\@@_tmp:w { oox }
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
%    \begin{macrocode}
%</package>
%    \end{macrocode}
%
% \end{implementation}
%
% \PrintIndex