aboutsummaryrefslogblamecommitdiffstats
path: root/VideoPreview/src/lib/JEDI-SDL/SDL/Pas/libxmlparser.pas
blob: 63e7b7fb53f842bad036833db55b1d1a50bf3439 (plain) (tree)
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































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































                                                                                                                                      
(**
===============================================================================================
Name    : LibXmlParser
===============================================================================================
Project : All Projects
===============================================================================================
Subject : Progressive XML Parser for all types of XML Files
===============================================================================================
Author  : Stefan Heymann
          Eschenweg 3
          72076 T�bingen
          GERMANY

E-Mail:   stefan@destructor.de
URL:      www.destructor.de
===============================================================================================
Source, Legals ("Licence")
--------------------------
The official site to get this parser is http://www.destructor.de/

Usage and Distribution of this Source Code is ruled by the
"Destructor.de Source code Licence" (DSL) which comes with this file or
can be downloaded at http://www.destructor.de/

IN SHORT: Usage and distribution of this source code is free.
          You use it completely on your own risk.

Postcardware
------------
If you like this code, please send a postcard of your city to my above address.
===============================================================================================
!!!  All parts of this code which are not finished or not conforming exactly to
     the XmlSpec are marked with three exclamation marks

-!-  Parts where the parser may be able to detect errors in the document's syntax are
     marked with the dash-exlamation mark-dash sequence.
===============================================================================================
Terminology:
------------
- Start:   Start of a buffer part
- Final:   End (last character) of a buffer part
- DTD:     Document Type Definition
- DTDc:    Document Type Declaration
- XMLSpec: The current W3C XML Recommendation (version 1.0 as of 1998-02-10), Chapter No.
- Cur*:    Fields concerning the "Current" part passed back by the "Scan" method
===============================================================================================
Scanning the XML document
-------------------------
- Create TXmlParser Instance                     MyXml := TXmlParser.Create;
- Load XML Document                              MyXml.LoadFromFile (Filename);
- Start Scanning                                 MyXml.StartScan;
- Scan Loop                                      WHILE MyXml.Scan DO
- Test for Part Type                               CASE MyXml.CurPartType OF
- Handle Parts                                       ... : ;;;
- Handle Parts                                       ... : ;;;
- Handle Parts                                       ... : ;;;
                                                     END;
- Destroy                                        MyXml.Free;
===============================================================================================
Loading the XML document
------------------------
You can load the XML document from a file with the "LoadFromFile" method.
It is beyond the scope of this parser to perform HTTP or FTP accesses. If you want your
application to handle such requests (URLs), you can load the XML via HTTP or FTP or whatever
protocol and hand over the data buffer using the "LoadFromBuffer" or "SetBuffer" method.
"LoadFromBuffer" loads the internal buffer of TXmlParser with the given null-terminated
string, thereby creating a copy of that buffer.
"SetBuffer" just takes the pointer to another buffer, which means that the given
buffer pointer must be valid while the document is accessed via TXmlParser.
===============================================================================================
Encodings:
----------
This XML parser kind of "understands" the following encodings:
- UTF-8
- ISO-8859-1
- Windows-1252

Any flavor of multi-byte characters (and this includes UTF-16) is not supported. Sorry.

Every string which has to be passed to the application passes the virtual method
"TranslateEncoding" which translates the string from the current encoding (stored in
"CurEncoding") into the encoding the application wishes to receive.
The "TranslateEncoding" method that is built into TXmlParser assumes that the application
wants to receive Windows ANSI (Windows-1252, about the same as ISO-8859-1) and is able
to convert UTF-8 and ISO-8859-1 encodings.
For other source and target encodings, you will have to override "TranslateEncoding".
===============================================================================================
Buffer Handling
---------------
- The document must be loaded completely into a piece of RAM
- All character positions are referenced by PChar pointers
- The TXmlParser instance can either "own" the buffer itself (then, FBufferSize is > 0)
  or reference the buffer of another instance or object (then, FBuffersize is 0 and
  FBuffer is not NIL)
- The Property DocBuffer passes back a pointer to the first byte of the document. If there
  is no document stored (FBuffer is NIL), the DocBuffer returns a pointer to a NULL character.
===============================================================================================
Whitespace Handling
-------------------
The TXmlParser property "PackSpaces" determines how Whitespace is returned in Text Content:
While PackSpaces is true, all leading and trailing whitespace characters are trimmed of, all
Whitespace is converted to Space #x20 characters and contiguous Whitespace characters are
compressed to one.
If the "Scan" method reports a ptContent part, the application can get the original text
with all whitespace characters by extracting the characters from "CurStart" to "CurFinal".
If the application detects an xml:space attribute, it can set "PackSpaces" accordingly or
use CurStart/CurFinal.
Please note that TXmlParser does _not_ normalize Line Breaks to single LineFeed characters
as the XmlSpec requires (XmlSpec 2.11).
The xml:space attribute is not handled by TXmlParser. This is on behalf of the application.
===============================================================================================
Non-XML-Conforming
------------------
TXmlParser does not conform 100 % exactly to the XmlSpec:
- UTF-16 is not supported (XmlSpec 2.2)
  (Workaround: Convert UTF-16 to UTF-8 and hand the buffer over to TXmlParser)
- As the parser only works with single byte strings, all Unicode characters > 255
  can currently not be handled correctly.
- Line breaks are not normalized to single Linefeed #x0A characters (XmlSpec 2.11)
  (Workaround: The Application can access the text contents on its own [CurStart, CurFinal],
  thereby applying every normalization it wishes to)
- The attribute value normalization does not work exactly as defined in the
  Second Edition of the XML 1.0 specification.
- See also the code parts marked with three consecutive exclamation marks. These are
  parts which are not finished in the current code release.

This list may be incomplete, so it may grow if I get to know any other points.
As work on the parser proceeds, this list may also shrink.
===============================================================================================
Things Todo
-----------
- Introduce a new event/callback which is called when there is an unresolvable
  entity or character reference
- Support Unicode
- Use Streams instead of reading the whole XML into memory
===============================================================================================
Change History, Version numbers
-------------------------------
The Date is given in ISO Year-Month-Day (YYYY-MM-DD) order.
Versions are counted from 1.0.0 beginning with the version from 2000-03-16.
Unreleased versions don't get a version number.

Date        Author Version Changes
-----------------------------------------------------------------------------------------------
2000-03-16  HeySt  1.0.0   Start
2000-03-28  HeySt  1.0.1   Initial Publishing of TXmlParser on the destructor.de Web Site
2000-03-30  HeySt  1.0.2   TXmlParser.AnalyzeCData: Call "TranslateEncoding" for CurContent
2000-03-31  HeySt  1.0.3   Deleted the StrPosE function (was not needed anyway)
2000-04-04  HeySt  1.0.4   TDtdElementRec modified: Start/Final for all Elements;
                           Should be backwards compatible.
                           AnalyzeDtdc: Set CurPartType to ptDtdc
2000-04-23  HeySt  1.0.5   New class TObjectList. Eliminated reference to the Delphi 5
                           "Contnrs" unit so LibXmlParser is Delphi 4 compatible.
2000-07-03  HeySt  1.0.6   TNvpNode: Added Constructor
2000-07-11  HeySt  1.0.7   Removed "Windows" from USES clause
                           Added three-exclamation-mark comments for Utf8ToAnsi/AnsiToUtf8
                           Added three-exclamation-mark comments for CHR function calls
2000-07-23  HeySt  1.0.8   TXmlParser.Clear: CurAttr.Clear; EntityStack.Clear;
                           (This was not a bug; just defensive programming)
2000-07-29  HeySt  1.0.9   TNvpList: Added methods: Node(Index), Value(Index), Name(Index);
2000-10-07  HeySt          Introduced Conditional Defines
                           Uses Contnrs unit and its TObjectList class again for
                           Delphi 5 and newer versions
2001-01-30  HeySt          Introduced Version Numbering
                           Made LoadFromFile and LoadFromBuffer BOOLEAN functions
                           Introduced FileMode parameter for LoadFromFile
                           BugFix: TAttrList.Analyze: Must add CWhitespace to ExtractName call
                           Comments worked over
2001-02-28  HeySt  1.0.10  Completely worked over and tested the UTF-8 functions
                           Fixed a bug in TXmlParser.Scan which caused it to start over when it
                           was called after the end of scanning, resulting in an endless loop
                           TEntityStack is now a TObjectList instead of TList
2001-07-03  HeySt  1.0.11  Updated Compiler Version IFDEFs for Kylix
2001-07-11  HeySt  1.0.12  New TCustomXmlScanner component (taken over from LibXmlComps.pas)
2001-07-14  HeySt  1.0.13  Bugfix TCustomXmlScanner.FOnTranslateEncoding
2001-10-22  HeySt          Don't clear CurName anymore when the parser finds a CDATA section.
2001-12-03  HeySt  1.0.14  TObjectList.Clear: Make call to INHERITED method (fixes a memory leak)
2001-12-05  HeySt  1.0.15  TObjectList.Clear: removed call to INHERITED method
                           TObjectList.Destroy: Inserted SetCapacity call.
                           Reduces need for frequent re-allocation of pointer buffer
                           Dedicated to my father, Theodor Heymann
2002-06-26  HeySt  1.0.16  TXmlParser.Scan: Fixed a bug with PIs whose name is beginning
                           with 'xml'. Thanks to Uwe Kamm for submitting this bug.
                           The CurEncoding property is now always in uppercase letters (the XML
                           spec wants it to be treated case independently so when it's uppercase
                           comparisons are faster)
2002-03-04  HeySt  1.0.17  Included an IFDEF for Delphi 7 (VER150) and Kylix
                           There is a new symbol HAS_CONTNRS_UNIT which is used now to
                           distinguish between IDEs which come with the Contnrs unit and
                           those that don't.
*)

UNIT libxmlparser;

{$I jedi-sdl.inc}

INTERFACE

USES
  SysUtils, Classes,
  (*$IFDEF HAS_CONTNRS_UNIT *)  // The Contnrs Unit was introduced in Delphi 5
  Contnrs,
  (*$ENDIF*)
  Math;

CONST
  CVersion = '1.0.17';  // This variable will be updated for every release
                        // (I hope, I won't forget to do it everytime ...)

TYPE
  TPartType    = // --- Document Part Types
                 (ptNone,            // Nothing
                  ptXmlProlog,       // XML Prolog                  XmlSpec 2.8 / 4.3.1
                  ptComment,         // Comment                     XmlSpec 2.5
                  ptPI,              // Processing Instruction      XmlSpec 2.6
                  ptDtdc,            // Document Type Declaration   XmlSpec 2.8
                  ptStartTag,        // Start Tag                   XmlSpec 3.1
                  ptEmptyTag,        // Empty-Element Tag           XmlSpec 3.1
                  ptEndTag,          // End Tag                     XmlSpec 3.1
                  ptContent,         // Text Content between Tags
                  ptCData);          // CDATA Section               XmlSpec 2.7

  TDtdElemType = // --- DTD Elements
                 (deElement,         // !ELEMENT declaration
                  deAttList,         // !ATTLIST declaration
                  deEntity,          // !ENTITY declaration
                  deNotation,        // !NOTATION declaration
                  dePI,              // PI in DTD
                  deComment,         // Comment in DTD
                  deError);          // Error found in the DTD

TYPE
  TAttrList    = CLASS;
  TEntityStack = CLASS;
  TNvpList     = CLASS;
  TElemDef     = CLASS;
  TElemList    = CLASS;
  TEntityDef   = CLASS;
  TNotationDef = CLASS;

  TDtdElementRec = RECORD    // --- This Record is returned by the DTD parser callback function
                     Start, Final : PChar;                             // Start/End of the Element's Declaration
                     CASE ElementType : TDtdElemType OF                // Type of the Element
                       deElement,                                      // <!ELEMENT>
                       deAttList  : (ElemDef      : TElemDef);         // <!ATTLIST>
                       deEntity   : (EntityDef    : TEntityDef);       // <!ENTITY>
                       deNotation : (NotationDef  : TNotationDef);     // <!NOTATION>
                       dePI       : (Target       : PChar;             // <?PI ?>
                                     Content      : PChar;
                                     AttrList     : TAttrList);
                       deError    : (Pos          : PChar);            // Error
                       // deComment : ((No additional fields here));   // <!-- Comment -->
                   END;

  TXmlParser = CLASS                             // --- Internal Properties and Methods
               PROTECTED
                 FBuffer      : PChar;           // NIL if there is no buffer available
                 FBufferSize  : INTEGER;         // 0 if the buffer is not owned by the Document instance
                 FSource      : STRING;          // Name of Source of document. Filename for Documents loaded with LoadFromFile

                 FXmlVersion  : STRING;          // XML version from Document header. Default is '1.0'
                 FEncoding    : STRING;          // Encoding from Document header. Default is 'UTF-8'
                 FStandalone  : BOOLEAN;         // Standalone declaration from Document header. Default is 'yes'
                 FRootName    : STRING;          // Name of the Root Element (= DTD name)
                 FDtdcFinal   : PChar;           // Pointer to the '>' character terminating the DTD declaration

                 FNormalize   : BOOLEAN;         // If true: Pack Whitespace and don't return empty contents
                 EntityStack  : TEntityStack;    // Entity Stack for Parameter and General Entities
                 FCurEncoding : STRING;          // Current Encoding during parsing (always uppercase)

                 PROCEDURE AnalyzeProlog;                                         // Analyze XML Prolog or Text Declaration
                 PROCEDURE AnalyzeComment (Start : PChar; VAR Final : PChar);     // Analyze Comments
                 PROCEDURE AnalyzePI      (Start : PChar; VAR Final : PChar);     // Analyze Processing Instructions (PI)
                 PROCEDURE AnalyzeDtdc;                                           // Analyze Document Type Declaration
                 PROCEDURE AnalyzeDtdElements (Start : PChar; VAR Final : PChar); // Analyze DTD declarations
                 PROCEDURE AnalyzeTag;                                            // Analyze Start/End/Empty-Element Tags
                 PROCEDURE AnalyzeCData;                                          // Analyze CDATA Sections
                 PROCEDURE AnalyzeText (VAR IsDone : BOOLEAN);                    // Analyze Text Content between Tags
                 PROCEDURE AnalyzeElementDecl  (Start : PChar; VAR Final : PChar);
                 PROCEDURE AnalyzeAttListDecl  (Start : PChar; VAR Final : PChar);
                 PROCEDURE AnalyzeEntityDecl   (Start : PChar; VAR Final : PChar);
                 PROCEDURE AnalyzeNotationDecl (Start : PChar; VAR Final : PChar);

                 PROCEDURE PushPE (VAR Start : PChar);
                 PROCEDURE ReplaceCharacterEntities (VAR Str : STRING);
                 PROCEDURE ReplaceParameterEntities (VAR Str : STRING);
                 PROCEDURE ReplaceGeneralEntities   (VAR Str : STRING);

                 FUNCTION GetDocBuffer : PChar;  // Returns FBuffer or a pointer to a NUL char if Buffer is empty

               PUBLIC                         // --- Document Properties
                 PROPERTY XmlVersion : STRING  READ FXmlVersion;                 // XML version from the Document Prolog
                 PROPERTY Encoding   : STRING  READ FEncoding;                   // Document Encoding from Prolog
                 PROPERTY Standalone : BOOLEAN READ FStandalone;                 // Standalone Declaration from Prolog
                 PROPERTY RootName   : STRING  READ FRootName;                   // Name of the Root Element
                 PROPERTY Normalize  : BOOLEAN READ FNormalize WRITE FNormalize; // True if Content is to be normalized
                 PROPERTY Source     : STRING  READ FSource;                     // Name of Document Source (Filename)
                 PROPERTY DocBuffer  : PChar   READ GetDocBuffer;                // Returns document buffer
               PUBLIC                         // --- DTD Objects
                 Elements    : TElemList;     // Elements: List of TElemDef (contains Attribute Definitions)
                 Entities    : TNvpList;      // General Entities: List of TEntityDef
                 ParEntities : TNvpList;      // Parameter Entities: List of TEntityDef
                 Notations   : TNvpList;      // Notations: List of TNotationDef
               PUBLIC
                 CONSTRUCTOR Create;
                 DESTRUCTOR Destroy;                                      OVERRIDE;

                 // --- Document Handling
                 FUNCTION  LoadFromFile   (Filename : STRING;
                                           FileMode : INTEGER = fmOpenRead OR fmShareDenyNone) : BOOLEAN;
                                                                          // Loads Document from given file
                 FUNCTION  LoadFromBuffer (Buffer : PChar) : BOOLEAN;     // Loads Document from another buffer
                 PROCEDURE SetBuffer      (Buffer : PChar);               // References another buffer
                 PROCEDURE Clear;                                         // Clear Document

               PUBLIC
                 // --- Scanning through the document
                 CurPartType : TPartType;                         // Current Type
                 CurName     : STRING;                            // Current Name
                 CurContent  : STRING;                            // Current Normalized Content
                 CurStart    : PChar;                             // Current First character
                 CurFinal    : PChar;                             // Current Last character
                 CurAttr     : TAttrList;                         // Current Attribute List
                 PROPERTY CurEncoding : STRING READ FCurEncoding; // Current Encoding
                 PROCEDURE StartScan;
                 FUNCTION  Scan : BOOLEAN;

                 // --- Events / Callbacks
                 FUNCTION  LoadExternalEntity (SystemId, PublicId,
                                               Notation : STRING) : TXmlParser;     VIRTUAL;
                 FUNCTION  TranslateEncoding  (CONST Source : STRING) : STRING;     VIRTUAL;
                 PROCEDURE DtdElementFound (DtdElementRec : TDtdElementRec);        VIRTUAL;
               END;

  TValueType   = // --- Attribute Value Type
                 (vtNormal,       // Normal specified Attribute
                  vtImplied,      // #IMPLIED attribute value
                  vtFixed,        // #FIXED attribute value
                  vtDefault);     // Attribute value from default value in !ATTLIST declaration

  TAttrDefault = // --- Attribute Default Type
                 (adDefault,      // Normal default value
                  adRequired,     // #REQUIRED attribute
                  adImplied,      // #IMPLIED attribute
                  adFixed);       // #FIXED attribute

  TAttrType    = // --- Type of attribute
                 (atUnknown,      // Unknown type
                  atCData,        // Character data only
                  atID,           // ID
                  atIdRef,        // ID Reference
                  atIdRefs,       // Several ID References, separated by Whitespace
                  atEntity,       // Name of an unparsed Entity
                  atEntities,     // Several unparsed Entity names, separated by Whitespace
                  atNmToken,      // Name Token
                  atNmTokens,     // Several Name Tokens, separated by Whitespace
                  atNotation,     // A selection of Notation names (Unparsed Entity)
                  atEnumeration); // Enumeration

  TElemType    = // --- Element content type
                 (etEmpty,        // Element is always empty
                  etAny,          // Element can have any mixture of PCDATA and any elements
                  etChildren,     // Element must contain only elements
                  etMixed);       // Mixed PCDATA and elements

  (*$IFDEF HAS_CONTNRS_UNIT *)
  TObjectList = Contnrs.TObjectList;    // Re-Export this identifier
  (*$ELSE *)
  TObjectList = CLASS (TList)
                  DESTRUCTOR Destroy; OVERRIDE;
                  PROCEDURE Delete (Index : INTEGER);
                  PROCEDURE Clear; OVERRIDE;
                END;
  (*$ENDIF *)

  TNvpNode  = CLASS                     // Name-Value Pair Node
                 Name  : STRING;
                 Value : STRING;
                 CONSTRUCTOR Create (TheName : STRING = ''; TheValue : STRING = '');
              END;

  TNvpList  = CLASS (TObjectList)       // Name-Value Pair List
                PROCEDURE Add   (Node  : TNvpNode);
                FUNCTION  Node  (Name  : STRING)  : TNvpNode;  OVERLOAD;
                FUNCTION  Node  (Index : INTEGER) : TNvpNode;  OVERLOAD;
                FUNCTION  Value (Name  : STRING)  : STRING;    OVERLOAD;
                FUNCTION  Value (Index : INTEGER) : STRING;    OVERLOAD;
                FUNCTION  Name  (Index : INTEGER) : STRING;
              END;

  TAttr     = CLASS (TNvpNode)          // Attribute of a Start-Tag or Empty-Element-Tag
                 ValueType : TValueType;
                 AttrType  : TAttrType;
               END;

  TAttrList = CLASS (TNvpList)          // List of Attributes
                PROCEDURE Analyze (Start : PChar; VAR Final : PChar);
              END;

  TEntityStack = CLASS (TObjectList)    // Stack where current position is stored before parsing entities
                 PROTECTED
                   Owner : TXmlParser;
                 PUBLIC
                   CONSTRUCTOR Create (TheOwner : TXmlParser);
                   PROCEDURE Push (LastPos : PChar);                      OVERLOAD;
                   PROCEDURE Push (Instance : TObject; LastPos : PChar);  OVERLOAD;
                   FUNCTION  Pop : PChar;         // Returns next char or NIL if EOF is reached. Frees Instance.
                 END;

  TAttrDef    = CLASS (TNvpNode)        // Represents a <!ATTLIST Definition. "Value" is the default value
                  TypeDef     : STRING;           // Type definition from the DTD
                  Notations   : STRING;           // Notation List, separated by pipe symbols '|'
                  AttrType    : TAttrType;        // Attribute Type
                  DefaultType : TAttrDefault;     // Default Type
                END;

  TElemDef    = CLASS (TNvpList)       // Represents a <!ELEMENT Definition. Is a list of TAttrDef-Nodes
                  Name       : STRING;            // Element name
                  ElemType   : TElemType;         // Element type
                  Definition : STRING;            // Element definition from DTD
                END;

  TElemList   = CLASS (TObjectList)    // List of TElemDef nodes
                  FUNCTION  Node (Name : STRING) : TElemDef;
                  PROCEDURE Add (Node : TElemDef);
                END;

  TEntityDef  = CLASS (TNvpNode)       // Represents a <!ENTITY Definition.
                  SystemId     : STRING;
                  PublicId     : STRING;
                  NotationName : STRING;
                END;

  TNotationDef = CLASS (TNvpNode)      // Represents a <!NOTATION Definition. Value is the System ID
                   PublicId : STRING;
                 END;

  TCharset = SET OF CHAR;


CONST
  CWhitespace   = [#32, #9, #13, #10];                // Whitespace characters (XmlSpec 2.3)
  CLetter       = [#$41..#$5A, #$61..#$7A, #$C0..#$D6, #$D8..#$F6, #$F8..#$FF];
  CDigit        = [#$30..#$39];
  CNameChar     = CLetter + CDigit + ['.', '-', '_', ':', #$B7];
  CNameStart    = CLetter + ['_', ':'];
  CQuoteChar    = ['"', ''''];
  CPubidChar    = [#32, ^M, ^J, #9, 'a'..'z', 'A'..'Z', '0'..'9',
                   '-', '''', '(', ')', '+', ',', '.', '/', ':',
                   '=', '?', ';', '!', '*', '#', '@', '$', '_', '%'];

  CDStart       = '<![CDATA[';
  CDEnd         = ']]>';

  // --- Name Constants for the above enumeration types
  CPartType_Name : ARRAY [TPartType] OF STRING =
                   ('', 'XML Prolog', 'Comment', 'PI',
                    'DTD Declaration', 'Start Tag', 'Empty Tag', 'End Tag',
                    'Text', 'CDATA');
  CValueType_Name   : ARRAY [TValueType]    OF STRING = ('Normal', 'Implied', 'Fixed', 'Default');
  CAttrDefault_Name : ARRAY [TAttrDefault]  OF STRING = ('Default', 'Required', 'Implied', 'Fixed');
  CElemType_Name    : ARRAY [TElemType]     OF STRING = ('Empty', 'Any', 'Childs only', 'Mixed');
  CAttrType_Name    : ARRAY [TAttrType]     OF STRING = ('Unknown', 'CDATA',
                                                         'ID', 'IDREF', 'IDREFS',
                                                         'ENTITY', 'ENTITIES',
                                                         'NMTOKEN', 'NMTOKENS',
                                                         'Notation', 'Enumeration');

FUNCTION  ConvertWs   (Source: STRING; PackWs: BOOLEAN) : STRING;          // Convert WS to spaces #x20
PROCEDURE SetStringSF (VAR S : STRING; BufferStart, BufferFinal : PChar);  // SetString by Start/Final of buffer
FUNCTION  StrSFPas    (Start, Finish : PChar) : STRING;                    // Convert buffer part to Pascal string
FUNCTION  TrimWs      (Source : STRING) : STRING;                          // Trim Whitespace

FUNCTION  AnsiToUtf8  (Source : ANSISTRING) : STRING;                            // Convert Win-1252 to UTF-8
FUNCTION  Utf8ToAnsi  (Source : STRING; UnknownChar : CHAR = '�') : ANSISTRING;  // Convert UTF-8 to Win-1252


(*
===============================================================================================
TCustomXmlScanner event based component wrapper for TXmlParser
===============================================================================================
*)

TYPE
  TCustomXmlScanner = CLASS;
  TXmlPrologEvent   = PROCEDURE (Sender : TObject; XmlVersion, Encoding: STRING; Standalone : BOOLEAN) OF OBJECT;
  TCommentEvent     = PROCEDURE (Sender : TObject; Comment : STRING)                                   OF OBJECT;
  TPIEvent          = PROCEDURE (Sender : TObject; Target, Content: STRING; Attributes : TAttrList)    OF OBJECT;
  TDtdEvent         = PROCEDURE (Sender : TObject; RootElementName : STRING)                           OF OBJECT;
  TStartTagEvent    = PROCEDURE (Sender : TObject; TagName : STRING; Attributes : TAttrList)           OF OBJECT;
  TEndTagEvent      = PROCEDURE (Sender : TObject; TagName : STRING)                                   OF OBJECT;
  TContentEvent     = PROCEDURE (Sender : TObject; Content : STRING)                                   OF OBJECT;
  TElementEvent     = PROCEDURE (Sender : TObject; ElemDef : TElemDef)                                 OF OBJECT;
  TEntityEvent      = PROCEDURE (Sender : TObject; EntityDef : TEntityDef)                             OF OBJECT;
  TNotationEvent    = PROCEDURE (Sender : TObject; NotationDef : TNotationDef)                         OF OBJECT;
  TErrorEvent       = PROCEDURE (Sender : TObject; ErrorPos : PChar)                                   OF OBJECT;
  TExternalEvent    = PROCEDURE (Sender : TObject; SystemId, PublicId, NotationId : STRING;
                                 VAR Result : TXmlParser)                                              OF OBJECT;
  TEncodingEvent    = FUNCTION  (Sender : TObject; CurrentEncoding, Source : STRING) : STRING          OF OBJECT;


  TCustomXmlScanner = CLASS (TComponent)
    PROTECTED
      FXmlParser           : TXmlParser;
      FOnXmlProlog         : TXmlPrologEvent;
      FOnComment           : TCommentEvent;
      FOnPI                : TPIEvent;
      FOnDtdRead           : TDtdEvent;
      FOnStartTag          : TStartTagEvent;
      FOnEmptyTag          : TStartTagEvent;
      FOnEndTag            : TEndTagEvent;
      FOnContent           : TContentEvent;
      FOnCData             : TContentEvent;
      FOnElement           : TElementEvent;
      FOnAttList           : TElementEvent;
      FOnEntity            : TEntityEvent;
      FOnNotation          : TNotationEvent;
      FOnDtdError          : TErrorEvent;
      FOnLoadExternal      : TExternalEvent;
      FOnTranslateEncoding : TEncodingEvent;
      FStopParser          : BOOLEAN;
      FUNCTION  GetNormalize : BOOLEAN;
      PROCEDURE SetNormalize (Value : BOOLEAN);

      PROCEDURE WhenXmlProlog(XmlVersion, Encoding: STRING; Standalone : BOOLEAN); VIRTUAL;
      PROCEDURE WhenComment  (Comment : STRING);                                   VIRTUAL;
      PROCEDURE WhenPI       (Target, Content: STRING; Attributes : TAttrList);    VIRTUAL;
      PROCEDURE WhenDtdRead  (RootElementName : STRING);                           VIRTUAL;
      PROCEDURE WhenStartTag (TagName : STRING; Attributes : TAttrList);           VIRTUAL;
      PROCEDURE WhenEmptyTag (TagName : STRING; Attributes : TAttrList);           VIRTUAL;
      PROCEDURE WhenEndTag   (TagName : STRING);                                   VIRTUAL;
      PROCEDURE WhenContent  (Content : STRING);                                   VIRTUAL;
      PROCEDURE WhenCData    (Content : STRING);                                   VIRTUAL;
      PROCEDURE WhenElement  (ElemDef : TElemDef);                                 VIRTUAL;
      PROCEDURE WhenAttList  (ElemDef : TElemDef);                                 VIRTUAL;
      PROCEDURE WhenEntity   (EntityDef : TEntityDef);                             VIRTUAL;
      PROCEDURE WhenNotation (NotationDef : TNotationDef);                         VIRTUAL;
      PROCEDURE WhenDtdError (ErrorPos : PChar);                                   VIRTUAL;

    PUBLIC
      CONSTRUCTOR Create (AOwner: TComponent); OVERRIDE;
      DESTRUCTOR Destroy;                      OVERRIDE;

      PROCEDURE LoadFromFile   (Filename : TFilename);   // Load XML Document from file
      PROCEDURE LoadFromBuffer (Buffer : PChar);         // Load XML Document from buffer
      PROCEDURE SetBuffer      (Buffer : PChar);         // Refer to Buffer
      FUNCTION  GetFilename : TFilename;

      PROCEDURE Execute;                                 // Perform scanning

    PROTECTED
      PROPERTY XmlParser           : TXmlParser        READ FXmlParser;
      PROPERTY StopParser          : BOOLEAN           READ FStopParser          WRITE FStopParser;
      PROPERTY Filename            : TFilename         READ GetFilename          WRITE LoadFromFile;
      PROPERTY Normalize           : BOOLEAN           READ GetNormalize         WRITE SetNormalize;
      PROPERTY OnXmlProlog         : TXmlPrologEvent   READ FOnXmlProlog         WRITE FOnXmlProlog;
      PROPERTY OnComment           : TCommentEvent     READ FOnComment           WRITE FOnComment;
      PROPERTY OnPI                : TPIEvent          READ FOnPI                WRITE FOnPI;
      PROPERTY OnDtdRead           : TDtdEvent         READ FOnDtdRead           WRITE FOnDtdRead;
      PROPERTY OnStartTag          : TStartTagEvent    READ FOnStartTag          WRITE FOnStartTag;
      PROPERTY OnEmptyTag          : TStartTagEvent    READ FOnEmptyTag          WRITE FOnEmptyTag;
      PROPERTY OnEndTag            : TEndTagEvent      READ FOnEndTag            WRITE FOnEndTag;
      PROPERTY OnContent           : TContentEvent     READ FOnContent           WRITE FOnContent;
      PROPERTY OnCData             : TContentEvent     READ FOnCData             WRITE FOnCData;
      PROPERTY OnElement           : TElementEvent     READ FOnElement           WRITE FOnElement;
      PROPERTY OnAttList           : TElementEvent     READ FOnAttList           WRITE FOnAttList;
      PROPERTY OnEntity            : TEntityEvent      READ FOnEntity            WRITE FOnEntity;
      PROPERTY OnNotation          : TNotationEvent    READ FOnNotation          WRITE FOnNotation;
      PROPERTY OnDtdError          : TErrorEvent       READ FOnDtdError          WRITE FOnDtdError;
      PROPERTY OnLoadExternal      : TExternalEvent    READ FOnLoadExternal      WRITE FOnLoadExternal;
      PROPERTY OnTranslateEncoding : TEncodingEvent    READ FOnTranslateEncoding WRITE FOnTranslateEncoding;
    END;

(*
===============================================================================================
IMPLEMENTATION
===============================================================================================
*)

IMPLEMENTATION


(*
===============================================================================================
Unicode and UTF-8 stuff
===============================================================================================
*)

CONST
  // --- Character Translation Table for Unicode <-> Win-1252
  WIN1252_UNICODE : ARRAY [$00..$FF] OF WORD = (
                    $0000, $0001, $0002, $0003, $0004, $0005, $0006, $0007, $0008, $0009,
                    $000A, $000B, $000C, $000D, $000E, $000F, $0010, $0011, $0012, $0013,
                    $0014, $0015, $0016, $0017, $0018, $0019, $001A, $001B, $001C, $001D,
                    $001E, $001F, $0020, $0021, $0022, $0023, $0024, $0025, $0026, $0027,
                    $0028, $0029, $002A, $002B, $002C, $002D, $002E, $002F, $0030, $0031,
                    $0032, $0033, $0034, $0035, $0036, $0037, $0038, $0039, $003A, $003B,
                    $003C, $003D, $003E, $003F, $0040, $0041, $0042, $0043, $0044, $0045,
                    $0046, $0047, $0048, $0049, $004A, $004B, $004C, $004D, $004E, $004F,
                    $0050, $0051, $0052, $0053, $0054, $0055, $0056, $0057, $0058, $0059,
                    $005A, $005B, $005C, $005D, $005E, $005F, $0060, $0061, $0062, $0063,
                    $0064, $0065, $0066, $0067, $0068, $0069, $006A, $006B, $006C, $006D,
                    $006E, $006F, $0070, $0071, $0072, $0073, $0074, $0075, $0076, $0077,
                    $0078, $0079, $007A, $007B, $007C, $007D, $007E, $007F,

                    $20AC, $0081, $201A, $0192, $201E, $2026, $2020, $2021, $02C6, $2030,
                    $0160, $2039, $0152, $008D, $017D, $008F, $0090, $2018, $2019, $201C,
                    $201D, $2022, $2013, $2014, $02DC, $2122, $0161, $203A, $0153, $009D,
                    $017E, $0178, $00A0, $00A1, $00A2, $00A3, $00A4, $00A5, $00A6, $00A7,
                    $00A8, $00A9, $00AA, $00AB, $00AC, $00AD, $00AE, $00AF, $00B0, $00B1,
                    $00B2, $00B3, $00B4, $00B5, $00B6, $00B7, $00B8, $00B9, $00BA, $00BB,
                    $00BC, $00BD, $00BE, $00BF, $00C0, $00C1, $00C2, $00C3, $00C4, $00C5,
                    $00C6, $00C7, $00C8, $00C9, $00CA, $00CB, $00CC, $00CD, $00CE, $00CF,
                    $00D0, $00D1, $00D2, $00D3, $00D4, $00D5, $00D6, $00D7, $00D8, $00D9,
                    $00DA, $00DB, $00DC, $00DD, $00DE, $00DF, $00E0, $00E1, $00E2, $00E3,
                    $00E4, $00E5, $00E6, $00E7, $00E8, $00E9, $00EA, $00EB, $00EC, $00ED,
                    $00EE, $00EF, $00F0, $00F1, $00F2, $00F3, $00F4, $00F5, $00F6, $00F7,
                    $00F8, $00F9, $00FA, $00FB, $00FC, $00FD, $00FE, $00FF);

(* UTF-8  (somewhat simplified)
   -----
   Character Range    Byte sequence
   ---------------    --------------------------     (x=Bits from original character)
   $0000..$007F       0xxxxxxx
   $0080..$07FF       110xxxxx 10xxxxxx
   $8000..$FFFF       1110xxxx 10xxxxxx 10xxxxxx

   Example
   --------
   Transforming the Unicode character U+00E4 LATIN SMALL LETTER A WITH DIAERESIS  ("�"):

         ISO-8859-1,           Decimal  228
         Win1252,              Hex      $E4
         ANSI                  Bin      1110 0100
                                        abcd efgh

         UTF-8                 Binary   1100xxab 10cdefgh
                               Binary   11000011 10100100
                               Hex      $C3      $A4
                               Decimal  195      164
                               ANSI     �        �         *)


FUNCTION  AnsiToUtf8 (Source : ANSISTRING) : STRING;
          (* Converts the given Windows ANSI (Win1252) String to UTF-8. *)
VAR
  I   : INTEGER;  // Loop counter
  U   : WORD;     // Current Unicode value
  Len : INTEGER;  // Current real length of "Result" string
BEGIN
  SetLength (Result, Length (Source) * 3);   // Worst case
  Len := 0;
  FOR I := 1 TO Length (Source) DO BEGIN
    U := WIN1252_UNICODE [ORD (Source [I])];
    CASE U OF
      $0000..$007F : BEGIN
                       INC (Len);
                       Result [Len] := CHR (U);
                     END;
      $0080..$07FF : BEGIN
                       INC (Len);
                       Result [Len] := CHR ($C0 OR (U SHR 6));
                       INC (Len);
                       Result [Len] := CHR ($80 OR (U AND $3F));
                     END;
      $0800..$FFFF : BEGIN
                       INC (Len);
                       Result [Len] := CHR ($E0 OR (U SHR 12));
                       INC (Len);
                       Result [Len] := CHR ($80 OR ((U SHR 6) AND $3F));
                       INC (Len);
                       Result [Len] := CHR ($80 OR (U AND $3F));
                     END;
      END;
    END;
  SetLength (Result, Len);
END;


FUNCTION  Utf8ToAnsi (Source : STRING; UnknownChar : CHAR = '�') : ANSISTRING;
          (* Converts the given UTF-8 String to Windows ANSI (Win-1252).
             If a character can not be converted, the "UnknownChar" is inserted. *)
VAR
  SourceLen : INTEGER;  // Length of Source string
  I, K      : INTEGER;
  A         : BYTE;     // Current ANSI character value
  U         : WORD;
  Ch        : CHAR;     // Dest char
  Len       : INTEGER;  // Current real length of "Result" string
BEGIN
  SourceLen := Length (Source);
  SetLength (Result, SourceLen);   // Enough room to live
  Len := 0;
  I   := 1;
  WHILE I <= SourceLen DO BEGIN
    A := ORD (Source [I]);
    IF A < $80 THEN BEGIN                                               // Range $0000..$007F
      INC (Len);
      Result [Len] := Source [I];
      INC (I);
      END
    ELSE BEGIN                                                          // Determine U, Inc I
      IF (A AND $E0 = $C0) AND (I < SourceLen) THEN BEGIN               // Range $0080..$07FF
        U := (WORD (A AND $1F) SHL 6) OR (ORD (Source [I+1]) AND $3F);
        INC (I, 2);
        END
      ELSE IF (A AND $F0 = $E0) AND (I < SourceLen-1) THEN BEGIN        // Range $0800..$FFFF
        U := (WORD (A AND $0F) SHL 12) OR
             (WORD (ORD (Source [I+1]) AND $3F) SHL 6) OR
             (      ORD (Source [I+2]) AND $3F);
        INC (I, 3);
        END
      ELSE BEGIN                                                        // Unknown/unsupported
        INC (I);
        FOR K := 7 DOWNTO 0 DO
          IF A AND (1 SHL K) = 0 THEN BEGIN
            INC (I, (A SHR (K+1))-1);
            BREAK;
            END;
        U := WIN1252_UNICODE [ORD (UnknownChar)];
        END;
      Ch := UnknownChar;                                                // Retrieve ANSI char
      FOR A := $00 TO $FF DO
        IF WIN1252_UNICODE [A] = U THEN BEGIN
          Ch := CHR (A);
          BREAK;
          END;
      INC (Len);
      Result [Len] := Ch;
      END;
    END;
  SetLength (Result, Len);
END;


(*
===============================================================================================
"Special" Helper Functions

Don't ask me why. But including these functions makes the parser *DRAMATICALLY* faster
on my K6-233 machine. You can test it yourself just by commenting them out.
They do exactly the same as the Assembler routines defined in SysUtils.
(This is where you can see how great the Delphi compiler really is. The compiled code is
faster than hand-coded assembler!)
===============================================================================================
--> Just move this line below the StrScan function -->  *)


FUNCTION StrPos (CONST Str, SearchStr : PChar) : PChar;
         // Same functionality as SysUtils.StrPos
VAR
  First : CHAR;
  Len   : INTEGER;
BEGIN
  First  := SearchStr^;
  Len    := StrLen (SearchStr);
  Result := Str;
  REPEAT
    IF Result^ = First THEN
      IF StrLComp (Result, SearchStr, Len) = 0 THEN BREAK;
    IF Result^ = #0 THEN BEGIN
      Result := NIL;
      BREAK;
      END;
    INC (Result);
  UNTIL FALSE;
END;


FUNCTION StrScan (CONST Start : PChar; CONST Ch : CHAR) : PChar;
         // Same functionality as SysUtils.StrScan
BEGIN
  Result := Start;
  WHILE Result^ <> Ch DO BEGIN
    IF Result^ = #0 THEN BEGIN
      Result := NIL;
      EXIT;
      END;
    INC (Result);
    END;
END;


(*
===============================================================================================
Helper Functions
===============================================================================================
*)

FUNCTION  DelChars (Source : STRING; CharsToDelete : TCharset) : STRING;
          // Delete all "CharsToDelete" from the string
VAR
  I : INTEGER;
BEGIN
  Result := Source;
  FOR I := Length (Result) DOWNTO 1 DO
    IF Result [I] IN CharsToDelete THEN
      Delete (Result, I, 1);
END;


FUNCTION  TrimWs (Source : STRING) : STRING;
          // Trimms off Whitespace characters from both ends of the string
VAR
  I : INTEGER;
BEGIN
  // --- Trim Left
  I := 1;
  WHILE (I <= Length (Source)) AND (Source [I] IN CWhitespace) DO
    INC (I);
  Result := Copy (Source, I, MaxInt);

  // --- Trim Right
  I := Length (Result);
  WHILE (I > 1) AND (Result [I] IN CWhitespace) DO
    DEC (I);
  Delete (Result, I+1, Length (Result)-I);
END;


FUNCTION  ConvertWs (Source: STRING; PackWs: BOOLEAN) : STRING;
          // Converts all Whitespace characters to the Space #x20 character
          // If "PackWs" is true, contiguous Whitespace characters are packed to one
VAR
  I : INTEGER;
BEGIN
  Result := Source;
  FOR I := Length (Result) DOWNTO 1 DO
    IF (Result [I] IN CWhitespace) THEN
      IF PackWs AND (I > 1) AND (Result [I-1] IN CWhitespace)
        THEN Delete (Result, I, 1)
        ELSE Result [I] := #32;
END;


PROCEDURE SetStringSF (VAR S : STRING; BufferStart, BufferFinal : PChar);
BEGIN
  SetString (S, BufferStart, BufferFinal-BufferStart+1);
END;


FUNCTION  StrLPas  (Start : PChar; Len : INTEGER) : STRING;
BEGIN
  SetString (Result, Start, Len);
END;


FUNCTION  StrSFPas (Start, Finish : PChar) : STRING;
BEGIN
  SetString (Result, Start, Finish-Start+1);
END;


FUNCTION  StrScanE (CONST Source : PChar; CONST CharToScanFor : CHAR) : PChar;
          // If "CharToScanFor" is not found, StrScanE returns the last char of the
          // buffer instead of NIL
BEGIN
  Result := StrScan (Source, CharToScanFor);
  IF Result = NIL THEN
    Result := StrEnd (Source)-1;
END;


PROCEDURE ExtractName (Start : PChar; Terminators : TCharset; VAR Final : PChar);
          (* Extracts the complete Name beginning at "Start".
             It is assumed that the name is contained in Markup, so the '>' character is
             always a Termination.
             Start:       IN  Pointer to first char of name. Is always considered to be valid
             Terminators: IN  Characters which terminate the name
             Final:       OUT Pointer to last char of name *)
BEGIN
  Final := Start+1;
  Include (Terminators, #0);
  Include (Terminators, '>');
  WHILE NOT (Final^ IN Terminators) DO
    INC (Final);
  DEC (Final);
END;


PROCEDURE ExtractQuote (Start : PChar; VAR Content : STRING; VAR Final : PChar);
          (* Extract a string which is contained in single or double Quotes.
             Start:    IN   Pointer to opening quote
             Content:  OUT  The quoted string
             Final:    OUT  Pointer to closing quote *)
BEGIN
  Final := StrScan (Start+1, Start^);
  IF Final = NIL THEN BEGIN
    Final := StrEnd (Start+1)-1;
    SetString (Content, Start+1, Final-Start);
    END
  ELSE
    SetString (Content, Start+1, Final-1-Start);
END;


(*
===============================================================================================
TEntityStackNode
This Node is pushed to the "Entity Stack" whenever the parser parses entity replacement text.
The "Instance" field holds the Instance pointer of an External Entity buffer. When it is
popped, the Instance is freed.
The "Encoding" field holds the name of the Encoding. External Parsed Entities may have
another encoding as the document entity (XmlSpec 4.3.3). So when there is an "<?xml" PI
found in the stream (= Text Declaration at the beginning of external parsed entities), the
Encoding found there is used for the External Entity (is assigned to TXmlParser.CurEncoding)
Default Encoding is for the Document Entity is UTF-8. It is assumed that External Entities
have the same Encoding as the Document Entity, unless they carry a Text Declaration.
===============================================================================================
*)

TYPE
  TEntityStackNode = CLASS
                       Instance : TObject;
                       Encoding : STRING;
                       LastPos  : PChar;
                     END;

(*
===============================================================================================
TEntityStack
For nesting of Entities.
When there is an entity reference found in the data stream, the corresponding entity
definition is searched and the current position is pushed to this stack.
From then on, the program scans the entitiy replacement text as if it were normal content.
When the parser reaches the end of an entity, the current position is popped off the
stack again.
===============================================================================================
*)

CONSTRUCTOR TEntityStack.Create (TheOwner : TXmlParser);
BEGIN
  INHERITED Create;
  Owner := TheOwner;
END;


PROCEDURE TEntityStack.Push (LastPos : PChar);
BEGIN
  Push (NIL, LastPos);
END;


PROCEDURE TEntityStack.Push (Instance : TObject; LastPos : PChar);
VAR
  ESN : TEntityStackNode;
BEGIN
  ESN := TEntityStackNode.Create;
  ESN.Instance := Instance;
  ESN.Encoding := Owner.FCurEncoding;  // Save current Encoding
  ESN.LastPos  := LastPos;
  Add (ESN);
END;


FUNCTION  TEntityStack.Pop : PChar;
VAR
  ESN : TEntityStackNode;
BEGIN
  IF Count > 0 THEN BEGIN
    ESN := TEntityStackNode (Items [Count-1]);
    Result := ESN.LastPos;
    IF ESN.Instance <> NIL THEN
      ESN.Instance.Free;
    IF ESN.Encoding <> '' THEN
      Owner.FCurEncoding := ESN.Encoding;   // Restore current Encoding
    Delete (Count-1);
    END
  ELSE
    Result := NIL;
END;


(*
===============================================================================================
TExternalID
-----------
XmlSpec 4.2.2:  ExternalID ::= 'SYSTEM' S SystemLiteral |
                               'PUBLIC' S PubidLiteral S SystemLiteral
XmlSpec 4.7:    PublicID   ::= 'PUBLIC' S PubidLiteral
SystemLiteral and PubidLiteral are quoted
===============================================================================================
*)

TYPE
  TExternalID = CLASS
                  PublicId : STRING;
                  SystemId : STRING;
                  Final    : PChar;
                  CONSTRUCTOR Create (Start : PChar);
                END;

CONSTRUCTOR TExternalID.Create (Start : PChar);
BEGIN
  INHERITED Create;
  Final := Start;
  IF StrLComp (Start, 'SYSTEM', 6) = 0 THEN BEGIN
    WHILE NOT (Final^ IN (CQuoteChar + [#0, '>', '['])) DO INC (Final);
    IF NOT (Final^ IN CQuoteChar) THEN EXIT;
    ExtractQuote (Final, SystemID, Final);
    END
  ELSE IF StrLComp (Start, 'PUBLIC', 6) = 0 THEN BEGIN
    WHILE NOT (Final^ IN (CQuoteChar + [#0, '>', '['])) DO INC (Final);
    IF NOT (Final^ IN CQuoteChar) THEN EXIT;
    ExtractQuote (Final, PublicID, Final);
    INC (Final);
    WHILE NOT (Final^ IN (CQuoteChar + [#0, '>', '['])) DO INC (Final);
    IF NOT (Final^ IN CQuoteChar) THEN EXIT;
    ExtractQuote (Final, SystemID, Final);
    END;
END;


(*
===============================================================================================
TXmlParser
===============================================================================================
*)

CONSTRUCTOR TXmlParser.Create;
BEGIN
  INHERITED Create;
  FBuffer     := NIL;
  FBufferSize := 0;
  Elements    := TElemList.Create;
  Entities    := TNvpList.Create;
  ParEntities := TNvpList.Create;
  Notations   := TNvpList.Create;
  CurAttr     := TAttrList.Create;
  EntityStack := TEntityStack.Create (Self);
  Clear;
END;


DESTRUCTOR TXmlParser.Destroy;
BEGIN
  Clear;
  Elements.Free;
  Entities.Free;
  ParEntities.Free;
  Notations.Free;
  CurAttr.Free;
  EntityStack.Free;
  INHERITED Destroy;
END;


PROCEDURE TXmlParser.Clear;
          // Free Buffer and clear all object attributes
BEGIN
  IF (FBufferSize > 0) AND (FBuffer <> NIL) THEN
    FreeMem (FBuffer);
  FBuffer     := NIL;
  FBufferSize := 0;
  FSource     := '';
  FXmlVersion := '';
  FEncoding   := '';
  FStandalone := FALSE;
  FRootName   := '';
  FDtdcFinal  := NIL;
  FNormalize  := TRUE;
  Elements.Clear;
  Entities.Clear;
  ParEntities.Clear;
  Notations.Clear;
  CurAttr.Clear;
  EntityStack.Clear;
END;


FUNCTION  TXmlParser.LoadFromFile (Filename : STRING; FileMode : INTEGER = fmOpenRead OR fmShareDenyNone) : BOOLEAN;
          // Loads Document from given file
          // Returns TRUE if successful
VAR
  f           : FILE;
  ReadIn      : INTEGER;
  OldFileMode : INTEGER;
BEGIN
  Result := FALSE;
  Clear;

  // --- Open File
  OldFileMode := SYSTEM.FileMode;
  TRY
    SYSTEM.FileMode := FileMode;
    TRY
      AssignFile (f, Filename);
      Reset (f, 1);
    EXCEPT
      EXIT;
      END;

    TRY
      // --- Allocate Memory
      TRY
        FBufferSize := Filesize (f) + 1;
        GetMem (FBuffer, FBufferSize);
      EXCEPT
        Clear;
        EXIT;
        END;

      // --- Read File
      TRY
        BlockRead (f, FBuffer^, FBufferSize, ReadIn);
        (FBuffer+ReadIn)^ := #0;  // NULL termination
      EXCEPT
        Clear;
        EXIT;
        END;
    FINALLY
      CloseFile (f);
      END;

    FSource := Filename;
    Result  := TRUE;

  FINALLY
    SYSTEM.FileMode := OldFileMode;
    END;
END;


FUNCTION  TXmlParser.LoadFromBuffer (Buffer : PChar) : BOOLEAN;
          // Loads Document from another buffer
          // Returns TRUE if successful
          // The "Source" property becomes '<MEM>' if successful
BEGIN
  Result := FALSE;
  Clear;
  FBufferSize := StrLen (Buffer) + 1;
  TRY
    GetMem (FBuffer, FBufferSize);
  EXCEPT
    Clear;
    EXIT;
    END;
  StrCopy (FBuffer, Buffer);
  FSource := '<MEM>';
  Result := TRUE;
END;


PROCEDURE TXmlParser.SetBuffer (Buffer : PChar);      // References another buffer
BEGIN
  Clear;
  FBuffer     := Buffer;
  FBufferSize := 0;
  FSource := '<REFERENCE>';
END;


//-----------------------------------------------------------------------------------------------
// Scanning through the document
//-----------------------------------------------------------------------------------------------

PROCEDURE TXmlParser.StartScan;
BEGIN
  CurPartType := ptNone;
  CurName     := '';
  CurContent  := '';
  CurStart    := NIL;
  CurFinal    := NIL;
  CurAttr.Clear;
  EntityStack.Clear;
END;


FUNCTION  TXmlParser.Scan : BOOLEAN;
          // Scans the next Part
          // Returns TRUE if a part could be found, FALSE if there is no part any more
          //
          // "IsDone" can be set to FALSE by AnalyzeText in order to go to the next part
          // if there is no Content due to normalization
VAR
  IsDone : BOOLEAN;
BEGIN
  REPEAT
    IsDone := TRUE;

    // --- Start of next Part
    IF CurStart = NIL
      THEN CurStart := DocBuffer
      ELSE CurStart := CurFinal+1;
    CurFinal := CurStart;

    // --- End of Document of Pop off a new part from the Entity stack?
    IF CurStart^ = #0 THEN
      CurStart := EntityStack.Pop;

    // --- No Document or End Of Document: Terminate Scan
    IF (CurStart = NIL) OR (CurStart^ = #0) THEN BEGIN
      CurStart := StrEnd (DocBuffer);
      CurFinal := CurStart-1;
      EntityStack.Clear;
      Result   := FALSE;
      EXIT;
      END;

    IF (StrLComp (CurStart, '<?xml', 5) = 0) AND
       ((CurStart+5)^ IN CWhitespace) THEN AnalyzeProlog                                      // XML Declaration, Text Declaration
    ELSE IF StrLComp (CurStart, '<?',        2) = 0 THEN AnalyzePI (CurStart, CurFinal)       // PI
    ELSE IF StrLComp (CurStart, '<!--',      4) = 0 THEN AnalyzeComment (CurStart, CurFinal)  // Comment
    ELSE IF StrLComp (CurStart, '<!DOCTYPE', 9) = 0 THEN AnalyzeDtdc                          // DTDc
    ELSE IF StrLComp (CurStart, CDStart, Length (CDStart)) = 0 THEN AnalyzeCdata              // CDATA Section
    ELSE IF StrLComp (CurStart, '<',         1) = 0 THEN AnalyzeTag                           // Start-Tag, End-Tag, Empty-Element-Tag
    ELSE AnalyzeText (IsDone);                                                                // Text Content
  UNTIL IsDone;
  Result := TRUE;
END;


PROCEDURE TXmlParser.AnalyzeProlog;
          // Analyze XML Prolog or Text Declaration
VAR
  F : PChar;
BEGIN
  CurAttr.Analyze (CurStart+5, F);
  IF EntityStack.Count = 0 THEN BEGIN
    FXmlVersion := CurAttr.Value ('version');
    FEncoding   := CurAttr.Value ('encoding');
    FStandalone := CurAttr.Value ('standalone') = 'yes';
    END;
  CurFinal := StrPos (F, '?>');
  IF CurFinal <> NIL
    THEN INC (CurFinal)
    ELSE CurFinal := StrEnd (CurStart)-1;
  FCurEncoding := AnsiUpperCase (CurAttr.Value ('encoding'));
  IF FCurEncoding = '' THEN
    FCurEncoding := 'UTF-8';   // Default XML Encoding is UTF-8
  CurPartType  := ptXmlProlog;
  CurName      := '';
  CurContent   := '';
END;


PROCEDURE TXmlParser.AnalyzeComment (Start : PChar; VAR Final : PChar);
          // Analyze Comments
BEGIN
  Final := StrPos (Start+4, '-->');
  IF Final = NIL
    THEN Final := StrEnd (Start)-1
    ELSE INC (Final, 2);
  CurPartType := ptComment;
END;


PROCEDURE TXmlParser.AnalyzePI (Start : PChar; VAR Final : PChar);
          // Analyze Processing Instructions (PI)
          // This is also called for Character
VAR
  F : PChar;
BEGIN
  CurPartType := ptPI;
  Final := StrPos (Start+2, '?>');
  IF Final = NIL
    THEN Final := StrEnd (Start)-1
    ELSE INC (Final);
  ExtractName (Start+2, CWhitespace + ['?', '>'], F);
  SetStringSF (CurName, Start+2, F);
  SetStringSF (CurContent, F+1, Final-2);
  CurAttr.Analyze (F+1, F);
END;


PROCEDURE TXmlParser.AnalyzeDtdc;
          (* Analyze Document Type Declaration
                 doctypedecl  ::= '<!DOCTYPE' S Name (S ExternalID)? S? ('[' (markupdecl | PEReference | S)* ']' S?)? '>'
                 markupdecl   ::= elementdecl | AttlistDecl | EntityDecl | NotationDecl | PI | Comment
                 PEReference  ::= '%' Name ';'

                 elementdecl  ::= '<!ELEMENT' S Name S contentspec S?                    '>'
                 AttlistDecl  ::= '<!ATTLIST' S Name AttDef* S?                          '>'
                 EntityDecl   ::= '<!ENTITY' S Name S EntityDef S?                       '>' |
                                  '<!ENTITY' S '%' S Name S PEDef S?                     '>'
                 NotationDecl ::= '<!NOTATION' S Name S (ExternalID |  PublicID) S?      '>'
                 PI           ::=  '<?' PITarget (S (Char* - (Char* '?>' Char* )))?     '?>'
                 Comment      ::= '<!--' ((Char - '-') | ('-' (Char - '-')))*          '-->'  *)
TYPE
  TPhase = (phName, phDtd, phInternal, phFinishing);
VAR
  Phase       : TPhase;
  F           : PChar;
  ExternalID  : TExternalID;
  ExternalDTD : TXmlParser;
  DER         : TDtdElementRec;
BEGIN
  DER.Start := CurStart;
  EntityStack.Clear;    // Clear stack for Parameter Entities
  CurPartType := ptDtdc;

  // --- Don't read DTDc twice
  IF FDtdcFinal <> NIL THEN BEGIN
    CurFinal := FDtdcFinal;
    EXIT;
    END;

  // --- Scan DTDc
  CurFinal := CurStart + 9;    // First char after '<!DOCTYPE'
  Phase    := phName;
  REPEAT
    CASE CurFinal^ OF
      '%' : BEGIN
              PushPE (CurFinal);
              CONTINUE;
            END;
      #0  : IF EntityStack.Count = 0 THEN
              BREAK
            ELSE BEGIN
              CurFinal := EntityStack.Pop;
              CONTINUE;
              END;
      '[' : BEGIN
              Phase := phInternal;
              AnalyzeDtdElements (CurFinal+1, CurFinal);
              CONTINUE;
            END;
      ']' : Phase := phFinishing;
      '>' : BREAK;
      ELSE  IF NOT (CurFinal^ IN CWhitespace) THEN BEGIN
              CASE Phase OF
                phName : IF (CurFinal^ IN CNameStart)  THEN BEGIN
                           ExtractName (CurFinal, CWhitespace + ['[', '>'], F);
                           SetStringSF (FRootName, CurFinal, F);
                           CurFinal := F;
                           Phase := phDtd;
                           END;
                phDtd  : IF (StrLComp (CurFinal, 'SYSTEM', 6) = 0) OR
                            (StrLComp (CurFinal, 'PUBLIC', 6) = 0) THEN BEGIN
                           ExternalID  := TExternalID.Create (CurFinal);
                           ExternalDTD := LoadExternalEntity (ExternalId.SystemId, ExternalID.PublicId, '');
                           F := StrPos (ExternalDtd.DocBuffer, '<!');
                           IF F <> NIL THEN
                             AnalyzeDtdElements (F, F);
                           ExternalDTD.Free;
                           CurFinal := ExternalID.Final;
                           ExternalID.Free;
                           END;
                ELSE     BEGIN
                           DER.ElementType := deError;
                           DER.Pos         := CurFinal;
                           DER.Final       := CurFinal;
                           DtdElementFound (DER);
                         END;
                END;

              END;
      END;
    INC (CurFinal);
  UNTIL FALSE;

  CurPartType := ptDtdc;
  CurName     := '';
  CurContent  := '';

  // It is an error in the document if "EntityStack" is not empty now
  IF EntityStack.Count > 0 THEN BEGIN
    DER.ElementType := deError;
    DER.Final       := CurFinal;
    DER.Pos         := CurFinal;
    DtdElementFound (DER);
    END;

  EntityStack.Clear;    // Clear stack for General Entities
  FDtdcFinal := CurFinal;
END;


PROCEDURE TXmlParser.AnalyzeDtdElements (Start : PChar; VAR Final : PChar);
          // Analyze the "Elements" of a DTD contained in the external or
          // internal DTD subset.
VAR
  DER : TDtdElementRec;
BEGIN
  Final := Start;
  REPEAT
    CASE Final^ OF
      '%' : BEGIN
              PushPE (Final);
              CONTINUE;
            END;
      #0  : IF EntityStack.Count = 0 THEN
              BREAK
            ELSE BEGIN
              CurFinal := EntityStack.Pop;
              CONTINUE;
              END;
      ']',
      '>' : BREAK;
      '<' : IF      StrLComp (Final, '<!ELEMENT',   9) = 0 THEN AnalyzeElementDecl  (Final, Final)
            ELSE IF StrLComp (Final, '<!ATTLIST',   9) = 0 THEN AnalyzeAttListDecl  (Final, Final)
            ELSE IF StrLComp (Final, '<!ENTITY',    8) = 0 THEN AnalyzeEntityDecl   (Final, Final)
            ELSE IF StrLComp (Final, '<!NOTATION', 10) = 0 THEN AnalyzeNotationDecl (Final, Final)
            ELSE IF StrLComp (Final, '<?',          2) = 0 THEN BEGIN   // PI in DTD
              DER.ElementType := dePI;
              DER.Start       := Final;
              AnalyzePI (Final, Final);
              DER.Target      := PChar (CurName);
              DER.Content     := PChar (CurContent);
              DER.AttrList    := CurAttr;
              DER.Final       := Final;
              DtdElementFound (DER);
              END
            ELSE IF StrLComp (Final, '<!--', 4) = 0 THEN BEGIN   // Comment in DTD
              DER.ElementType := deComment;
              DER.Start       := Final;
              AnalyzeComment  (Final, Final);
              DER.Final       := Final;
              DtdElementFound (DER);
              END
            ELSE BEGIN
              DER.ElementType := deError;
              DER.Start       := Final;
              DER.Pos         := Final;
              DER.Final       := Final;
              DtdElementFound (DER);
              END;

      END;
    INC (Final);
  UNTIL FALSE;
END;


PROCEDURE TXmlParser.AnalyzeTag;
          // Analyze Tags
VAR
  S, F    : PChar;
  Attr    : TAttr;
  ElemDef : TElemDef;
  AttrDef : TAttrDef;
  I       : INTEGER;
BEGIN
  CurPartType := ptStartTag;
  S := CurStart+1;
  IF S^ = '/' THEN BEGIN
    CurPartType := ptEndTag;
    INC (S);
    END;
  ExtractName (S, CWhitespace + ['/'], F);
  SetStringSF (CurName, S, F);
  CurAttr.Analyze (F+1, CurFinal);
  IF CurFinal^ = '/' THEN BEGIN
    CurPartType := ptEmptyTag;
    END;
  CurFinal := StrScanE (CurFinal, '>');

  // --- Set Default Attribute values for nonexistent attributes
  IF (CurPartType = ptStartTag) OR (CurPartType = ptEmptyTag) THEN BEGIN
    ElemDef := Elements.Node (CurName);
    IF ElemDef <> NIL THEN BEGIN
      FOR I := 0 TO ElemDef.Count-1 DO BEGIN
        AttrDef := TAttrDef (ElemDef [I]);
        Attr := TAttr (CurAttr.Node (AttrDef.Name));
        IF (Attr = NIL) AND (AttrDef.Value <> '') THEN BEGIN
          Attr           := TAttr.Create (AttrDef.Name, AttrDef.Value);
          Attr.ValueType := vtDefault;
          CurAttr.Add (Attr);
          END;
        IF Attr <> NIL THEN BEGIN
          CASE AttrDef.DefaultType OF
            adDefault  : ;
            adRequired : ; // -!- It is an error in the document if "Attr.Value" is an empty string
            adImplied  : Attr.ValueType := vtImplied;
            adFixed    : BEGIN
                           Attr.ValueType := vtFixed;
                           Attr.Value     := AttrDef.Value;
                         END;
            END;
          Attr.AttrType := AttrDef.AttrType;
          END;
        END;
      END;

    // --- Normalize Attribute Values. XmlSpec:
           // - a character reference is processed by appending the referenced character to the attribute value
           // - an entity reference is processed by recursively processing the replacement text of the entity
           // - a whitespace character (#x20, #xD, #xA, #x9) is processed by appending #x20 to the normalized value,
           //   except that only a single #x20 is appended for a "#xD#xA" sequence that is part of an external
           //   parsed entity or the literal entity value of an internal parsed entity
           // - other characters are processed by appending them to the normalized value
           // If the declared value is not CDATA, then the XML processor must further process the
           // normalized attribute value by discarding any leading and trailing space (#x20) characters,
           // and by replacing sequences of space (#x20) characters by a single space (#x20) character.
           // All attributes for which no declaration has been read should be treated by a
           // non-validating parser as if declared CDATA.
           // !!! The XML 1.0 SE specification is somewhat different here
           //     This code does not conform exactly to this specification
    FOR I := 0 TO CurAttr.Count-1 DO
      WITH TAttr (CurAttr [I]) DO BEGIN
        ReplaceGeneralEntities   (Value);
        ReplaceCharacterEntities (Value);
        IF (AttrType <> atCData) AND (AttrType <> atUnknown)
          THEN Value := TranslateEncoding (TrimWs (ConvertWs (Value, TRUE)))
          ELSE Value := TranslateEncoding (ConvertWs (Value, FALSE));
        END;
    END;
END;


PROCEDURE TXmlParser.AnalyzeCData;
          // Analyze CDATA Sections
BEGIN
  CurPartType := ptCData;
  CurFinal := StrPos (CurStart, CDEnd);
  IF CurFinal = NIL THEN BEGIN
    CurFinal   := StrEnd (CurStart)-1;
    CurContent := TranslateEncoding (StrPas (CurStart+Length (CDStart)));
    END
  ELSE BEGIN
    SetStringSF (CurContent, CurStart+Length (CDStart), CurFinal-1);
    INC (CurFinal, Length (CDEnd)-1);
    CurContent := TranslateEncoding (CurContent);
    END;
END;


PROCEDURE TXmlParser.AnalyzeText (VAR IsDone : BOOLEAN);
          (* Analyzes Text Content between Tags. CurFinal will point to the last content character.
             Content ends at a '<' character or at the end of the document.
             Entity References and Character Entity references are resolved.
             If PackSpaces is TRUE, contiguous Whitespace Characters will be compressed to
             one Space #x20 character, Whitespace at the beginning and end of content will
             be trimmed off and content which is or becomes empty is not returned to
             the application (in this case, "IsDone" is set to FALSE which causes the
             Scan method to proceed directly to the next part. *)

  PROCEDURE ProcessEntity;
            (* Is called if there is an ampsersand '&' character found in the document.
               IN  "CurFinal" points to the ampersand
               OUT "CurFinal" points to the first character after the semi-colon ';' *)
  VAR
    P              : PChar;
    Name           : STRING;
    EntityDef      : TEntityDef;
    ExternalEntity : TXmlParser;
  BEGIN
    P := StrScan (CurFinal , ';');
    IF P <> NIL THEN BEGIN
      SetStringSF (Name, CurFinal+1, P-1);

      // Is it a Character Entity?
      IF (CurFinal+1)^ = '#' THEN BEGIN
        IF UpCase ((CurFinal+2)^) = 'X'       // !!! Can't use "CHR" for Unicode characters > 255:
          THEN CurContent := CurContent + CHR (StrToIntDef ('$'+Copy (Name, 3, MaxInt), 32))
          ELSE CurContent := CurContent + CHR (StrToIntDef (Copy (Name, 2, MaxInt), 32));
        CurFinal := P+1;
        EXIT;
        END

      // Is it a Predefined Entity?
      ELSE IF Name = 'lt'   THEN BEGIN CurContent := CurContent + '<';  CurFinal := P+1; EXIT; END
      ELSE IF Name = 'gt'   THEN BEGIN CurContent := CurContent + '>';  CurFinal := P+1; EXIT; END
      ELSE IF Name = 'amp'  THEN BEGIN CurContent := CurContent + '&';  CurFinal := P+1; EXIT; END
      ELSE IF Name = 'apos' THEN BEGIN CurContent := CurContent + ''''; CurFinal := P+1; EXIT; END
      ELSE IF Name = 'quot' THEN BEGIN CurContent := CurContent + '"';  CurFinal := P+1; EXIT; END;

      // Replace with Entity from DTD
      EntityDef := TEntityDef (Entities.Node (Name));
      IF EntityDef <> NIL THEN BEGIN
        IF EntityDef.Value <> '' THEN BEGIN
          EntityStack.Push (P+1);
          CurFinal := PChar (EntityDef.Value);
          END
        ELSE BEGIN
          ExternalEntity := LoadExternalEntity (EntityDef.SystemId, EntityDef.PublicId, EntityDef.NotationName);
          EntityStack.Push (ExternalEntity, P+1);
          CurFinal := ExternalEntity.DocBuffer;
          END;
        END
      ELSE BEGIN
        CurContent := CurContent + Name;
        CurFinal   := P+1;
        END;
      END
    ELSE BEGIN
      INC (CurFinal);
      END;
  END;

VAR
  C  : INTEGER;
BEGIN
  CurFinal    := CurStart;
  CurPartType := ptContent;
  CurContent  := '';
  C           := 0;
  REPEAT
    CASE CurFinal^ OF
      '&' : BEGIN
              CurContent := CurContent + TranslateEncoding (StrLPas (CurFinal-C, C));
              C := 0;
              ProcessEntity;
              CONTINUE;
            END;
      #0  : BEGIN
              IF EntityStack.Count = 0 THEN
                BREAK
              ELSE BEGIN
                CurContent := CurContent + TranslateEncoding (StrLPas (CurFinal-C, C));
                C := 0;
                CurFinal := EntityStack.Pop;
                CONTINUE;
                END;
            END;
      '<' : BREAK;
      ELSE INC (C);
      END;
    INC (CurFinal);
  UNTIL FALSE;
  CurContent := CurContent + TranslateEncoding (StrLPas (CurFinal-C, C));
  DEC (CurFinal);

  IF FNormalize THEN BEGIN
    CurContent := ConvertWs (TrimWs (CurContent), TRUE);
    IsDone     := CurContent <> '';    // IsDone will only get FALSE if PackSpaces is TRUE
    END;
END;


PROCEDURE TXmlParser.AnalyzeElementDecl  (Start : PChar; VAR Final : PChar);
          (* Parse <!ELEMENT declaration starting at "Start"
             Final must point to the terminating '>' character
             XmlSpec 3.2:
                 elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>'
                 contentspec ::= 'EMPTY' | 'ANY' | Mixed | children
                 Mixed       ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*'   |
                                 '(' S? '#PCDATA' S? ')'
                 children    ::= (choice | seq) ('?' | '*' | '+')?
                 choice      ::= '(' S? cp ( S? '|' S? cp )* S? ')'
                 cp          ::= (Name | choice | seq) ('?' | '*' | '+')?
                 seq         ::= '(' S? cp ( S? ',' S? cp )* S? ')'

             More simply:
                 contentspec ::= EMPTY
                                 ANY
                                 '(#PCDATA)'
                                 '(#PCDATA | A | B)*'
                                 '(A, B, C)'
                                 '(A | B | C)'
                                 '(A?, B*, C+),
                                 '(A, (B | C | D)* )'                       *)
VAR
  Element : TElemDef;
  Elem2   : TElemDef;
  F       : PChar;
  DER     : TDtdElementRec;
BEGIN
  Element   := TElemDef.Create;
  Final     := Start + 9;
  DER.Start := Start;
  REPEAT
    IF Final^ = '>' THEN BREAK;
    IF (Final^ IN CNameStart) AND (Element.Name = '') THEN BEGIN
      ExtractName (Final, CWhitespace, F);
      SetStringSF (Element.Name, Final, F);
      Final := F;
      F := StrScan (Final+1, '>');
      IF F = NIL THEN BEGIN
        Element.Definition := STRING (Final);
        Final := StrEnd (Final);
        BREAK;
        END
      ELSE BEGIN
        SetStringSF (Element.Definition, Final+1, F-1);
        Final := F;
        BREAK;
        END;
      END;
    INC (Final);
  UNTIL FALSE;
  Element.Definition := DelChars (Element.Definition, CWhitespace);
  ReplaceParameterEntities (Element.Definition);
  IF      Element.Definition = 'EMPTY' THEN Element.ElemType := etEmpty
  ELSE IF Element.Definition = 'ANY'   THEN Element.ElemType := etAny
  ELSE IF Copy (Element.Definition, 1, 8) = '(#PCDATA' THEN Element.ElemType := etMixed
  ELSE IF Copy (Element.Definition, 1, 1) = '('        THEN Element.ElemType := etChildren
  ELSE Element.ElemType := etAny;

  Elem2 := Elements.Node (Element.Name);
  IF Elem2 <> NIL THEN
    Elements.Delete (Elements.IndexOf (Elem2));
  Elements.Add (Element);
  Final := StrScanE (Final, '>');
  DER.ElementType := deElement;
  DER.ElemDef  := Element;
  DER.Final    := Final;
  DtdElementFound (DER);
END;


PROCEDURE TXmlParser.AnalyzeAttListDecl  (Start : PChar; VAR Final : PChar);
          (* Parse <!ATTLIST declaration starting at "Start"
             Final must point to the terminating '>' character
             XmlSpec 3.3:
                 AttlistDecl    ::= '<!ATTLIST' S Name AttDef* S? '>'
                 AttDef         ::= S Name S AttType S DefaultDecl
                 AttType        ::= StringType | TokenizedType | EnumeratedType
                 StringType     ::= 'CDATA'
                 TokenizedType  ::= 'ID' | 'IDREF' | 'IDREFS' | 'ENTITY' | 'ENTITIES' | 'NMTOKEN' | 'NMTOKENS'
                 EnumeratedType ::= NotationType | Enumeration
                 NotationType   ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
                 Enumeration    ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')'
                 DefaultDecl    ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue)
                 AttValue       ::= '"' ([^<&"] | Reference)* '"' | "'" ([^<&'] | Reference)* "'"
            Examples:
                 <!ATTLIST address
                           A1 CDATA "Default"
                           A2 ID    #REQUIRED
                           A3 IDREF #IMPLIED
                           A4 IDREFS #IMPLIED
                           A5 ENTITY #FIXED "&at;&#252;"
                           A6 ENTITIES #REQUIRED
                           A7 NOTATION (WMF | DXF) "WMF"
                           A8 (A | B | C) #REQUIRED>                *)
TYPE
  TPhase = (phElementName, phName, phType, phNotationContent, phDefault);
VAR
  Phase       : TPhase;
  F           : PChar;
  ElementName : STRING;
  ElemDef     : TElemDef;
  AttrDef     : TAttrDef;
  AttrDef2    : TAttrDef;
  Strg        : STRING;
  DER         : TDtdElementRec;
BEGIN
  Final     := Start + 9;   // The character after <!ATTLIST
  Phase     := phElementName;
  DER.Start := Start;
  AttrDef   := NIL;
  ElemDef   := NIL;
  REPEAT
    IF NOT (Final^ IN CWhitespace) THEN
      CASE Final^ OF
        '%' : BEGIN
                PushPE (Final);
                CONTINUE;
              END;
        #0  : IF EntityStack.Count = 0 THEN
                BREAK
              ELSE BEGIN
                Final := EntityStack.Pop;
                CONTINUE;
                END;
        '>' : BREAK;
        ELSE  CASE Phase OF
                phElementName     : BEGIN
                                      ExtractName (Final, CWhitespace + CQuoteChar + ['#'], F);
                                      SetStringSF (ElementName, Final, F);
                                      Final := F;
                                      ElemDef := Elements.Node (ElementName);
                                      IF ElemDef = NIL THEN BEGIN
                                        ElemDef := TElemDef.Create;
                                        ElemDef.Name       := ElementName;
                                        ElemDef.Definition := 'ANY';
                                        ElemDef.ElemType   := etAny;
                                        Elements.Add (ElemDef);
                                        END;
                                      Phase := phName;
                                    END;
                phName            : BEGIN
                                      AttrDef := TAttrDef.Create;
                                      ExtractName (Final, CWhitespace + CQuoteChar + ['#'], F);
                                      SetStringSF (AttrDef.Name, Final, F);
                                      Final := F;
                                      AttrDef2 := TAttrDef (ElemDef.Node (AttrDef.Name));
                                      IF AttrDef2 <> NIL THEN
                                        ElemDef.Delete (ElemDef.IndexOf (AttrDef2));
                                      ElemDef.Add (AttrDef);
                                      Phase := phType;
                                    END;
                phType            : BEGIN
                                      IF Final^ = '(' THEN BEGIN
                                        F := StrScan (Final+1, ')');
                                        IF F <> NIL
                                          THEN SetStringSF (AttrDef.TypeDef, Final+1, F-1)
                                          ELSE AttrDef.TypeDef := STRING (Final+1);
                                        AttrDef.TypeDef := DelChars (AttrDef.TypeDef, CWhitespace);
                                        AttrDef.AttrType := atEnumeration;
                                        ReplaceParameterEntities (AttrDef.TypeDef);
                                        ReplaceCharacterEntities (AttrDef.TypeDef);
                                        Phase := phDefault;
                                        END
                                      ELSE IF StrLComp (Final, 'NOTATION', 8) = 0 THEN BEGIN
                                        INC (Final, 8);
                                        AttrDef.AttrType := atNotation;
                                        Phase := phNotationContent;
                                        END
                                      ELSE BEGIN
                                        ExtractName (Final, CWhitespace+CQuoteChar+['#'], F);
                                        SetStringSF (AttrDef.TypeDef, Final, F);
                                        IF      AttrDef.TypeDef = 'CDATA'    THEN AttrDef.AttrType := atCData
                                        ELSE IF AttrDef.TypeDef = 'ID'       THEN AttrDef.AttrType := atId
                                        ELSE IF AttrDef.TypeDef = 'IDREF'    THEN AttrDef.AttrType := atIdRef
                                        ELSE IF AttrDef.TypeDef = 'IDREFS'   THEN AttrDef.AttrType := atIdRefs
                                        ELSE IF AttrDef.TypeDef = 'ENTITY'   THEN AttrDef.AttrType := atEntity
                                        ELSE IF AttrDef.TypeDef = 'ENTITIES' THEN AttrDef.AttrType := atEntities
                                        ELSE IF AttrDef.TypeDef = 'NMTOKEN'  THEN AttrDef.AttrType := atNmToken
                                        ELSE IF AttrDef.TypeDef = 'NMTOKENS' THEN AttrDef.AttrType := atNmTokens;
                                        Phase := phDefault;
                                        END
                                    END;
                phNotationContent : BEGIN
                                      F := StrScan (Final, ')');
                                      IF F <> NIL THEN
                                        SetStringSF (AttrDef.Notations, Final+1, F-1)
                                      ELSE BEGIN
                                        AttrDef.Notations := STRING (Final+1);
                                        Final := StrEnd (Final);
                                        END;
                                      ReplaceParameterEntities (AttrDef.Notations);
                                      AttrDef.Notations := DelChars (AttrDef.Notations, CWhitespace);
                                      Phase := phDefault;
                                    END;
                phDefault :         BEGIN
                                      IF Final^ = '#' THEN BEGIN
                                        ExtractName (Final, CWhiteSpace + CQuoteChar, F);
                                        SetStringSF (Strg, Final, F);
                                        Final := F;
                                        ReplaceParameterEntities (Strg);
                                        IF      Strg = '#REQUIRED' THEN BEGIN AttrDef.DefaultType := adRequired; Phase := phName; END
                                        ELSE IF Strg = '#IMPLIED'  THEN BEGIN AttrDef.DefaultType := adImplied;  Phase := phName; END
                                        ELSE IF Strg = '#FIXED'    THEN       AttrDef.DefaultType := adFixed;
                                        END
                                      ELSE IF (Final^ IN CQuoteChar) THEN BEGIN
                                        ExtractQuote (Final, AttrDef.Value, Final);
                                        ReplaceParameterEntities (AttrDef.Value);
                                        ReplaceCharacterEntities (AttrDef.Value);
                                        Phase := phName;
                                        END;
                                      IF Phase = phName THEN BEGIN
                                        AttrDef := NIL;
                                        END;
                                    END;

                END;
        END;
    INC (Final);
  UNTIL FALSE;

  Final := StrScan (Final, '>');

  DER.ElementType := deAttList;
  DER.ElemDef  := ElemDef;
  DER.Final    := Final;
  DtdElementFound (DER);
END;


PROCEDURE TXmlParser.AnalyzeEntityDecl   (Start : PChar; VAR Final : PChar);
          (* Parse <!ENTITY declaration starting at "Start"
             Final must point to the terminating '>' character
             XmlSpec 4.2:
                 EntityDecl  ::= '<!ENTITY' S Name S EntityDef S? '>'   |
                                 '<!ENTITY' S '%' S Name S PEDef S? '>'
                 EntityDef   ::= EntityValue | (ExternalID NDataDecl?)
                 PEDef       ::= EntityValue | ExternalID
                 NDataDecl   ::= S 'NDATA' S Name
                 EntityValue ::= '"' ([^%&"] | PEReference | EntityRef | CharRef)* '"'   |
                                 "'" ([^%&'] | PEReference | EntityRef | CharRef)* "'"
                 PEReference ::= '%' Name ';'

             Examples
                 <!ENTITY test1 "Stefan Heymann">                   <!-- Internal, general, parsed              -->
                 <!ENTITY test2 SYSTEM "ent2.xml">                  <!-- External, general, parsed              -->
                 <!ENTITY test2 SYSTEM "ent3.gif" NDATA gif>        <!-- External, general, unparsed            -->
                 <!ENTITY % test3 "<!ELEMENT q ANY>">               <!-- Internal, parameter                    -->
                 <!ENTITY % test6 SYSTEM "ent6.xml">                <!-- External, parameter                    -->
                 <!ENTITY test4 "&test1; ist lieb">                 <!-- IGP, Replacement text <> literal value -->
                 <!ENTITY test5 "<p>Dies ist ein Test-Absatz</p>">  <!-- IGP, See XmlSpec 2.4                   -->
          *)
TYPE
  TPhase = (phName, phContent, phNData, phNotationName, phFinalGT);
VAR
  Phase         : TPhase;
  IsParamEntity : BOOLEAN;
  F             : PChar;
  ExternalID    : TExternalID;
  EntityDef     : TEntityDef;
  EntityDef2    : TEntityDef;
  DER           : TDtdElementRec;
BEGIN
  Final         := Start + 8;   // First char after <!ENTITY
  DER.Start     := Start;
  Phase         := phName;
  IsParamEntity := FALSE;
  EntityDef     := TEntityDef.Create;
  REPEAT
    IF NOT (Final^ IN CWhitespace) THEN
      CASE Final^ OF
        '%' : IsParamEntity := TRUE;
        '>' : BREAK;
        ELSE  CASE Phase OF
                phName         : IF Final^ IN CNameStart THEN BEGIN
                                   ExtractName (Final, CWhitespace + CQuoteChar, F);
                                   SetStringSF (EntityDef.Name, Final, F);
                                   Final := F;
                                   Phase := phContent;
                                   END;
                phContent      : IF Final^ IN CQuoteChar THEN BEGIN
                                   ExtractQuote (Final, EntityDef.Value, Final);
                                   Phase := phFinalGT;
                                   END
                                 ELSE IF (StrLComp (Final, 'SYSTEM', 6) = 0) OR
                                         (StrLComp (Final, 'PUBLIC', 6) = 0) THEN BEGIN
                                   ExternalID := TExternalID.Create (Final);
                                   EntityDef.SystemId := ExternalID.SystemId;
                                   EntityDef.PublicId := ExternalID.PublicId;
                                   Final      := ExternalID.Final;
                                   Phase      := phNData;
                                   ExternalID.Free;
                                   END;
                phNData        : IF StrLComp (Final, 'NDATA', 5) = 0 THEN BEGIN
                                   INC (Final, 4);
                                   Phase := phNotationName;
                                   END;
                phNotationName : IF Final^ IN CNameStart THEN BEGIN
                                   ExtractName (Final, CWhitespace + ['>'], F);
                                   SetStringSF (EntityDef.NotationName, Final, F);
                                   Final := F;
                                   Phase := phFinalGT;
                                   END;
                phFinalGT      : ; // -!- There is an error in the document if this branch is called
                END;
        END;
    INC (Final);
  UNTIL FALSE;
  IF IsParamEntity THEN BEGIN
    EntityDef2 := TEntityDef (ParEntities.Node (EntityDef.Name));
    IF EntityDef2 <> NIL THEN
      ParEntities.Delete (ParEntities.IndexOf (EntityDef2));
    ParEntities.Add (EntityDef);
    ReplaceCharacterEntities (EntityDef.Value);
    END
  ELSE BEGIN
    EntityDef2 := TEntityDef (Entities.Node (EntityDef.Name));
    IF EntityDef2 <> NIL THEN
      Entities.Delete (Entities.IndexOf (EntityDef2));
    Entities.Add (EntityDef);
    ReplaceParameterEntities (EntityDef.Value);  //  Create replacement texts (see XmlSpec 4.5)
    ReplaceCharacterEntities (EntityDef.Value);
    END;
  Final := StrScanE (Final, '>');

  DER.ElementType := deEntity;
  DER.EntityDef   := EntityDef;
  DER.Final       := Final;
  DtdElementFound (DER);
END;


PROCEDURE TXmlParser.AnalyzeNotationDecl (Start : PChar; VAR Final : PChar);
          // Parse <!NOTATION declaration starting at "Start"
          // Final must point to the terminating '>' character
          // XmlSpec 4.7: NotationDecl ::=  '<!NOTATION' S Name S (ExternalID |  PublicID) S? '>'
TYPE
  TPhase = (phName, phExtId, phEnd);
VAR
  ExternalID  : TExternalID;
  Phase       : TPhase;
  F           : PChar;
  NotationDef : TNotationDef;
  DER         : TDtdElementRec;
BEGIN
  Final       := Start + 10;   // Character after <!NOTATION
  DER.Start   := Start;
  Phase       := phName;
  NotationDef := TNotationDef.Create;
  REPEAT
    IF NOT (Final^ IN CWhitespace) THEN
      CASE Final^ OF
        '>',
        #0   : BREAK;
        ELSE   CASE Phase OF
                 phName  : BEGIN
                             ExtractName (Final, CWhitespace + ['>'], F);
                             SetStringSF (NotationDef.Name, Final, F);
                             Final := F;
                             Phase := phExtId;
                           END;
                 phExtId : BEGIN
                             ExternalID := TExternalID.Create (Final);
                             NotationDef.Value    := ExternalID.SystemId;
                             NotationDef.PublicId := ExternalID.PublicId;
                             Final := ExternalId.Final;
                             ExternalId.Free;
                             Phase := phEnd;
                           END;
                 phEnd   : ;   // -!- There is an error in the document if this branch is called
                 END;
        END;
    INC (Final);
  UNTIL FALSE;
  Notations.Add (NotationDef);
  Final := StrScanE (Final, '>');

  DER.ElementType := deNotation;
  DER.NotationDef := NotationDef;
  DER.Final       := Final;
  DtdElementFound (DER);
END;


PROCEDURE TXmlParser.PushPE (VAR Start : PChar);
          (* If there is a parameter entity reference found in the data stream,
             the current position will be pushed to the entity stack.
             Start:  IN  Pointer to the '%' character starting the PE reference
                     OUT Pointer to first character of PE replacement text *)
VAR
  P         : PChar;
  EntityDef : TEntityDef;
BEGIN
  P := StrScan (Start, ';');
  IF P <> NIL THEN BEGIN
    EntityDef := TEntityDef (ParEntities.Node (StrSFPas (Start+1, P-1)));
    IF EntityDef <> NIL THEN BEGIN
      EntityStack.Push (P+1);
      Start := PChar (EntityDef.Value);
      END
    ELSE
      Start := P+1;
    END;
END;


PROCEDURE TXmlParser.ReplaceCharacterEntities (VAR Str : STRING);
          // Replaces all Character Entity References in the String
VAR
  Start  : INTEGER;
  PAmp   : PChar;
  PSemi  : PChar;
  PosAmp : INTEGER;
  Len    : INTEGER;    // Length of Entity Reference
BEGIN
  IF Str = '' THEN EXIT;
  Start := 1;
  REPEAT
    PAmp := StrPos (PChar (Str) + Start-1, '&#');
    IF PAmp = NIL THEN BREAK;
    PSemi := StrScan (PAmp+2, ';');
    IF PSemi = NIL THEN BREAK;
    PosAmp := PAmp - PChar (Str) + 1;
    Len    := PSemi-PAmp+1;
    IF CompareText (Str [PosAmp+2], 'x') = 0          // !!! Can't use "CHR" for Unicode characters > 255
      THEN Str [PosAmp] := CHR (StrToIntDef ('$'+Copy (Str, PosAmp+3, Len-4), 0))
      ELSE Str [PosAmp] := CHR (StrToIntDef (Copy (Str, PosAmp+2, Len-3), 32));
    Delete (Str, PosAmp+1, Len-1);  
    Start := PosAmp + 1;
  UNTIL FALSE;
END;


PROCEDURE TXmlParser.ReplaceParameterEntities (VAR Str : STRING);
          // Recursively replaces all Parameter Entity References in the String
  PROCEDURE ReplaceEntities (VAR Str : STRING);
  VAR
    Start   : INTEGER;
    PAmp    : PChar;
    PSemi   : PChar;
    PosAmp  : INTEGER;
    Len     : INTEGER;
    Entity  : TEntityDef;
    Repl    : STRING;        // Replacement
  BEGIN
    IF Str = '' THEN EXIT;
    Start := 1;
    REPEAT
      PAmp := StrPos (PChar (Str)+Start-1, '%');
      IF PAmp = NIL THEN BREAK;
      PSemi := StrScan (PAmp+2, ';');
      IF PSemi = NIL THEN BREAK;
      PosAmp := PAmp - PChar (Str) + 1;
      Len    := PSemi-PAmp+1;
      Entity := TEntityDef (ParEntities.Node (Copy (Str, PosAmp+1, Len-2)));
      IF Entity <> NIL THEN BEGIN
        Repl := Entity.Value;
        ReplaceEntities (Repl);    // Recursion
        END
      ELSE
        Repl := Copy (Str, PosAmp, Len);
      Delete (Str, PosAmp, Len);  
      Insert (Repl, Str, PosAmp);
      Start := PosAmp + Length (Repl);
    UNTIL FALSE;
  END;
BEGIN
  ReplaceEntities (Str);
END;


PROCEDURE TXmlParser.ReplaceGeneralEntities (VAR Str : STRING);
          // Recursively replaces General Entity References in the String
  PROCEDURE ReplaceEntities (VAR Str : STRING);
  VAR
    Start          : INTEGER;
    PAmp           : PChar;
    PSemi          : PChar;
    PosAmp         : INTEGER;
    Len            : INTEGER;
    EntityDef      : TEntityDef;
    EntName        : STRING;
    Repl           : STRING;        // Replacement
    ExternalEntity : TXmlParser;
  BEGIN
    IF Str = '' THEN EXIT;
    Start := 1;
    REPEAT
      PAmp := StrPos (PChar (Str)+Start-1, '&');
      IF PAmp = NIL THEN BREAK;
      PSemi := StrScan (PAmp+2, ';');
      IF PSemi = NIL THEN BREAK;
      PosAmp := PAmp - PChar (Str) + 1;
      Len    := PSemi-PAmp+1;
      EntName := Copy (Str, PosAmp+1, Len-2);
      IF      EntName = 'lt'   THEN Repl := '<'
      ELSE IF EntName = 'gt'   THEN Repl := '>'
      ELSE IF EntName = 'amp'  THEN Repl := '&'
      ELSE IF EntName = 'apos' THEN Repl := ''''
      ELSE IF EntName = 'quot' THEN Repl := '"'
      ELSE BEGIN
        EntityDef := TEntityDef (Entities.Node (EntName));
        IF EntityDef <> NIL THEN BEGIN
          IF EntityDef.Value <> '' THEN   // Internal Entity
            Repl := EntityDef.Value
          ELSE BEGIN                      // External Entity
            ExternalEntity := LoadExternalEntity (EntityDef.SystemId, EntityDef.PublicId, EntityDef.NotationName);
            Repl := StrPas (ExternalEntity.DocBuffer);  // !!! What if it contains a Text Declaration?
            ExternalEntity.Free;
            END;
          ReplaceEntities (Repl);    // Recursion
          END
        ELSE
          Repl := Copy (Str, PosAmp, Len);
        END;
      Delete (Str, PosAmp, Len);  
      Insert (Repl, Str, PosAmp);
      Start := PosAmp + Length (Repl);
    UNTIL FALSE;
  END;
BEGIN
  ReplaceEntities (Str);
END;


FUNCTION  TXmlParser.LoadExternalEntity (SystemId, PublicId, Notation : STRING) : TXmlParser;
          // This will be called whenever there is a Parsed External Entity or
          // the DTD External Subset to be parsed.
          // It has to create a TXmlParser instance and load the desired Entity.
          // This instance of LoadExternalEntity assumes that "SystemId" is a valid
          // file name (relative to the Document source) and loads this file using
          // the LoadFromFile method.
VAR
  Filename : STRING;
BEGIN
  // --- Convert System ID to complete filename
  Filename := StringReplace (SystemId, '/', '\', [rfReplaceAll]);
  IF Copy (FSource, 1, 1) <> '<' THEN
    IF (Copy (Filename, 1, 2) = '\\') OR (Copy (Filename, 2, 1) = ':') THEN
      // Already has an absolute Path
    ELSE BEGIN
      Filename := ExtractFilePath (FSource) + Filename;
      END;

  // --- Load the File
  Result := TXmlParser.Create;
  Result.LoadFromFile (Filename);
END;


FUNCTION  TXmlParser.TranslateEncoding  (CONST Source : STRING) : STRING;
          // The member variable "CurEncoding" always holds the name of the current
          // encoding, e.g. 'UTF-8' or 'ISO-8859-1'.
          // This virtual method "TranslateEncoding" is responsible for translating
          // the content passed in the "Source" parameter to the Encoding which
          // is expected by the application.
          // This instance of "TranlateEncoding" assumes that the Application expects
          // Windows ANSI (Win1252) strings. It is able to transform UTF-8 or ISO-8859-1
          // encodings.
          // If you want your application to understand or create other encodings, you
          // override this function.
BEGIN
  IF CurEncoding = 'UTF-8'
    THEN Result := Utf8ToAnsi (Source)
    ELSE Result := Source;
END;


PROCEDURE TXmlParser.DtdElementFound (DtdElementRec : TDtdElementRec);
          // This method is called for every element which is found in the DTD
          // declaration. The variant record TDtdElementRec is passed which
          // holds informations about the element.
          // You can override this function to handle DTD declarations.
          // Note that when you parse the same Document instance a second time,
          // the DTD will not get parsed again.
BEGIN
END;


FUNCTION TXmlParser.GetDocBuffer: PChar;
         // Returns FBuffer or a pointer to a NUL char if Buffer is empty
BEGIN
  IF FBuffer = NIL
    THEN Result := #0
    ELSE Result := FBuffer;
END;


(*$IFNDEF HAS_CONTNRS_UNIT
===============================================================================================
TObjectList
===============================================================================================
*)

DESTRUCTOR TObjectList.Destroy;
BEGIN
  Clear;
  SetCapacity(0);
  INHERITED Destroy;
END;


PROCEDURE TObjectList.Delete (Index : INTEGER);
BEGIN
  IF (Index < 0) OR (Index >= Count) THEN EXIT;
  TObject (Items [Index]).Free;
  INHERITED Delete (Index);
END;


PROCEDURE TObjectList.Clear;
BEGIN
  WHILE Count > 0 DO
    Delete (Count-1);
END;

(*$ENDIF *)

(*
===============================================================================================
TNvpNode
--------
Node base class for the TNvpList
===============================================================================================
*)

CONSTRUCTOR TNvpNode.Create (TheName, TheValue : STRING);
BEGIN
  INHERITED Create;
  Name  := TheName;
  Value := TheValue;
END;


(*
===============================================================================================
TNvpList
--------
A generic List of Name-Value Pairs, based on the TObjectList introduced in Delphi 5
===============================================================================================
*)

PROCEDURE TNvpList.Add (Node : TNvpNode);
VAR
  I : INTEGER;
BEGIN
  FOR I := Count-1 DOWNTO 0 DO
    IF Node.Name > TNvpNode (Items [I]).Name THEN BEGIN
      Insert (I+1, Node);
      EXIT;
      END;
  Insert (0, Node);
END;



FUNCTION  TNvpList.Node (Name : STRING) : TNvpNode;
          // Binary search for Node
VAR
  L, H : INTEGER;    // Low, High Limit
  T, C : INTEGER;    // Test Index, Comparison result
  Last : INTEGER;    // Last Test Index
BEGIN
  IF Count=0 THEN BEGIN
    Result := NIL;
    EXIT;
    END;

  L    := 0;
  H    := Count;
  Last := -1;
  REPEAT
    T := (L+H) DIV 2;
    IF T=Last THEN BREAK;
    Result := TNvpNode (Items [T]);
    C := CompareStr (Result.Name, Name);
    IF      C = 0 THEN EXIT
    ELSE IF C < 0 THEN L := T
    ELSE               H := T;
    Last := T;
  UNTIL FALSE;
  Result := NIL;
END;


FUNCTION  TNvpList.Node (Index : INTEGER) : TNvpNode;
BEGIN
  IF (Index < 0) OR (Index >= Count)
    THEN Result := NIL
    ELSE Result := TNvpNode (Items [Index]);
END;


FUNCTION  TNvpList.Value (Name : STRING) : STRING;
VAR
  Nvp : TNvpNode;
BEGIN
  Nvp := TNvpNode (Node (Name));
  IF Nvp <> NIL
    THEN Result := Nvp.Value
    ELSE Result := '';
END;


FUNCTION  TNvpList.Value (Index : INTEGER) : STRING;
BEGIN
  IF (Index < 0) OR (Index >= Count)
    THEN Result := ''
    ELSE Result := TNvpNode (Items [Index]).Value;
END;


FUNCTION  TNvpList.Name (Index : INTEGER) : STRING;
BEGIN
  IF (Index < 0) OR (Index >= Count)
    THEN Result := ''
    ELSE Result := TNvpNode (Items [Index]).Name;
END;


(*
===============================================================================================
TAttrList
List of Attributes. The "Analyze" method extracts the Attributes from the given Buffer.
Is used for extraction of Attributes in Start-Tags, Empty-Element Tags and the "pseudo"
attributes in XML Prologs, Text Declarations and PIs.
===============================================================================================
*)

PROCEDURE TAttrList.Analyze (Start : PChar; VAR Final : PChar);
          // Analyze the Buffer for Attribute=Name pairs.
          // Terminates when there is a character which is not IN CNameStart
          // (e.g. '?>' or '>' or '/>')
TYPE
  TPhase = (phName, phEq, phValue);
VAR
  Phase : TPhase;
  F     : PChar;
  Name  : STRING;
  Value : STRING;
  Attr  : TAttr;
BEGIN
  Clear;
  Phase := phName;
  Final := Start;
  REPEAT
    IF (Final^ = #0) OR (Final^ = '>') THEN BREAK;
    IF NOT (Final^ IN CWhitespace) THEN
      CASE Phase OF
        phName  : BEGIN
                    IF NOT (Final^ IN CNameStart) THEN EXIT;
                    ExtractName (Final, CWhitespace + ['=', '/'], F);
                    SetStringSF (Name, Final, F);
                    Final := F;
                    Phase := phEq;
                  END;
        phEq    : BEGIN
                    IF Final^ = '=' THEN
                      Phase := phValue
                  END;
        phValue : BEGIN
                    IF Final^ IN CQuoteChar THEN BEGIN
                      ExtractQuote (Final, Value, F);
                      Attr := TAttr.Create;
                      Attr.Name      := Name;
                      Attr.Value     := Value;
                      Attr.ValueType := vtNormal;
                      Add (Attr);
                      Final := F;
                      Phase := phName;
                      END;
                  END;
        END;
    INC (Final);
  UNTIL FALSE;
END;


(*
===============================================================================================
TElemList
List of TElemDef nodes.
===============================================================================================
*)

FUNCTION  TElemList.Node (Name : STRING) : TElemDef;
          // Binary search for the Node with the given Name
VAR
  L, H : INTEGER;    // Low, High Limit
  T, C : INTEGER;    // Test Index, Comparison result
  Last : INTEGER;    // Last Test Index
BEGIN
  IF Count=0 THEN BEGIN
    Result := NIL;
    EXIT;
    END;

  L    := 0;
  H    := Count;
  Last := -1;
  REPEAT
    T := (L+H) DIV 2;
    IF T=Last THEN BREAK;
    Result := TElemDef (Items [T]);
    C := CompareStr (Result.Name, Name);
    IF C = 0 THEN EXIT
    ELSE IF C < 0 THEN L := T
    ELSE               H := T;
    Last := T;
  UNTIL FALSE;
  Result := NIL;
END;


PROCEDURE TElemList.Add (Node : TElemDef);
VAR
  I : INTEGER;
BEGIN
  FOR I := Count-1 DOWNTO 0 DO
    IF Node.Name > TElemDef (Items [I]).Name THEN BEGIN
      Insert (I+1, Node);
      EXIT;
      END;
  Insert (0, Node);
END;


(*
===============================================================================================
TScannerXmlParser
A TXmlParser descendant for the TCustomXmlScanner component
===============================================================================================
*)

TYPE
  TScannerXmlParser = CLASS (TXmlParser)
                       Scanner : TCustomXmlScanner;
                       CONSTRUCTOR Create (TheScanner : TCustomXmlScanner);
                       FUNCTION  LoadExternalEntity (SystemId, PublicId,
                                                     Notation : STRING) : TXmlParser;  OVERRIDE;
                       FUNCTION  TranslateEncoding  (CONST Source : STRING) : STRING;  OVERRIDE;
                       PROCEDURE DtdElementFound (DtdElementRec : TDtdElementRec);     OVERRIDE;
                      END;

CONSTRUCTOR TScannerXmlParser.Create (TheScanner : TCustomXmlScanner);
BEGIN
  INHERITED Create;
  Scanner := TheScanner;
END;


FUNCTION  TScannerXmlParser.LoadExternalEntity (SystemId, PublicId, Notation : STRING) : TXmlParser;
BEGIN
  IF Assigned (Scanner.FOnLoadExternal)
    THEN Scanner.FOnLoadExternal (Scanner, SystemId, PublicId, Notation, Result)
    ELSE Result :=  INHERITED LoadExternalEntity (SystemId, PublicId, Notation);
END;


FUNCTION  TScannerXmlParser.TranslateEncoding  (CONST Source : STRING) : STRING;
BEGIN
  IF Assigned (Scanner.FOnTranslateEncoding)
    THEN Result := Scanner.FOnTranslateEncoding (Scanner, CurEncoding, Source)
    ELSE Result := INHERITED TranslateEncoding (Source);
END;


PROCEDURE TScannerXmlParser.DtdElementFound (DtdElementRec : TDtdElementRec);
BEGIN
  WITH DtdElementRec DO
    CASE ElementType OF
      deElement  : Scanner.WhenElement  (ElemDef);
      deAttList  : Scanner.WhenAttList  (ElemDef);
      deEntity   : Scanner.WhenEntity   (EntityDef);
      deNotation : Scanner.WhenNotation (NotationDef);
      dePI       : Scanner.WhenPI       (STRING (Target), STRING (Content), AttrList);
      deComment  : Scanner.WhenComment  (StrSFPas (Start, Final));
      deError    : Scanner.WhenDtdError (Pos);
      END;
END;


(*
===============================================================================================
TCustomXmlScanner
===============================================================================================
*)

CONSTRUCTOR TCustomXmlScanner.Create (AOwner: TComponent);
BEGIN
  INHERITED;
  FXmlParser := TScannerXmlParser.Create (Self);
END;


DESTRUCTOR TCustomXmlScanner.Destroy;
BEGIN
  FXmlParser.Free;
  INHERITED;
END;


PROCEDURE TCustomXmlScanner.LoadFromFile (Filename : TFilename);
          // Load XML Document from file
BEGIN
  FXmlParser.LoadFromFile (Filename);
END;


PROCEDURE TCustomXmlScanner.LoadFromBuffer (Buffer : PChar);
          // Load XML Document from buffer
BEGIN
  FXmlParser.LoadFromBuffer (Buffer);
END;


PROCEDURE TCustomXmlScanner.SetBuffer (Buffer : PChar);
          // Refer to Buffer
BEGIN
  FXmlParser.SetBuffer (Buffer);
END;


FUNCTION  TCustomXmlScanner.GetFilename : TFilename;
BEGIN
  Result := FXmlParser.Source;
END;


FUNCTION  TCustomXmlScanner.GetNormalize : BOOLEAN;
BEGIN
  Result := FXmlParser.Normalize;
END;


PROCEDURE TCustomXmlScanner.SetNormalize (Value : BOOLEAN);
BEGIN
  FXmlParser.Normalize := Value;
END;


PROCEDURE TCustomXmlScanner.WhenXmlProlog(XmlVersion, Encoding: STRING; Standalone : BOOLEAN);
          // Is called when the parser has parsed the <? xml ?> declaration of the prolog
BEGIN
  IF Assigned (FOnXmlProlog) THEN FOnXmlProlog (Self, XmlVersion, Encoding, Standalone);
END;


PROCEDURE TCustomXmlScanner.WhenComment  (Comment : STRING);
          // Is called when the parser has parsed a <!-- comment -->
BEGIN
  IF Assigned (FOnComment) THEN FOnComment (Self, Comment);
END;


PROCEDURE TCustomXmlScanner.WhenPI (Target, Content: STRING; Attributes : TAttrList);
          // Is called when the parser has parsed a <?processing instruction ?>
BEGIN
  IF Assigned (FOnPI) THEN FOnPI (Self, Target, Content, Attributes);
END;


PROCEDURE TCustomXmlScanner.WhenDtdRead (RootElementName : STRING);
          // Is called when the parser has completely parsed the DTD
BEGIN
  IF Assigned (FOnDtdRead) THEN FOnDtdRead (Self, RootElementName);
END;


PROCEDURE TCustomXmlScanner.WhenStartTag (TagName : STRING; Attributes : TAttrList);
          // Is called when the parser has parsed a start tag like <p>
BEGIN
  IF Assigned (FOnStartTag) THEN FOnStartTag (Self, TagName, Attributes);
END;


PROCEDURE TCustomXmlScanner.WhenEmptyTag (TagName : STRING; Attributes : TAttrList);
          // Is called when the parser has parsed an Empty Element Tag like <br/>
BEGIN
  IF Assigned (FOnEmptyTag) THEN FOnEmptyTag (Self, TagName, Attributes);
END;


PROCEDURE TCustomXmlScanner.WhenEndTag (TagName : STRING);
          // Is called when the parser has parsed an End Tag like </p>
BEGIN
  IF Assigned (FOnEndTag) THEN FOnEndTag (Self, TagName);
END;


PROCEDURE TCustomXmlScanner.WhenContent (Content : STRING);
          // Is called when the parser has parsed an element's text content
BEGIN
  IF Assigned (FOnContent) THEN FOnContent (Self, Content);
END;


PROCEDURE TCustomXmlScanner.WhenCData (Content : STRING);
          // Is called when the parser has parsed a CDATA section
BEGIN
  IF Assigned (FOnCData) THEN FOnCData (Self, Content);
END;


PROCEDURE TCustomXmlScanner.WhenElement (ElemDef : TElemDef);
          // Is called when the parser has parsed an <!ELEMENT> definition
          // inside the DTD
BEGIN
  IF Assigned (FOnElement) THEN FOnElement (Self, ElemDef);
END;


PROCEDURE TCustomXmlScanner.WhenAttList (ElemDef : TElemDef);
          // Is called when the parser has parsed an <!ATTLIST> definition
          // inside the DTD
BEGIN
  IF Assigned (FOnAttList) THEN FOnAttList (Self, ElemDef);
END;


PROCEDURE TCustomXmlScanner.WhenEntity   (EntityDef : TEntityDef);
          // Is called when the parser has parsed an <!ENTITY> definition
          // inside the DTD
BEGIN
  IF Assigned (FOnEntity) THEN FOnEntity (Self, EntityDef);
END;


PROCEDURE TCustomXmlScanner.WhenNotation (NotationDef : TNotationDef);
          // Is called when the parser has parsed a <!NOTATION> definition
          // inside the DTD
BEGIN
  IF Assigned (FOnNotation) THEN FOnNotation (Self, NotationDef);
END;


PROCEDURE TCustomXmlScanner.WhenDtdError (ErrorPos : PChar);
          // Is called when the parser has found an Error in the DTD
BEGIN
  IF Assigned (FOnDtdError) THEN FOnDtdError (Self, ErrorPos);
END;


PROCEDURE TCustomXmlScanner.Execute;
          // Perform scanning
          // Scanning is done synchronously, i.e. you can expect events to be triggered
          // in the order of the XML data stream. Execute will finish when the whole XML
          // document has been scanned or when the StopParser property has been set to TRUE.
BEGIN
  FStopParser := FALSE;
  FXmlParser.StartScan;
  WHILE FXmlParser.Scan AND (NOT FStopParser) DO
    CASE FXmlParser.CurPartType OF
      ptNone      : ;
      ptXmlProlog : WhenXmlProlog (FXmlParser.XmlVersion, FXmlParser.Encoding, FXmlParser.Standalone);
      ptComment   : WhenComment   (StrSFPas (FXmlParser.CurStart, FXmlParser.CurFinal));
      ptPI        : WhenPI        (FXmlParser.CurName, FXmlParser.CurContent, FXmlParser.CurAttr);
      ptDtdc      : WhenDtdRead   (FXmlParser.RootName);
      ptStartTag  : WhenStartTag  (FXmlParser.CurName, FXmlParser.CurAttr);
      ptEmptyTag  : WhenEmptyTag  (FXmlParser.CurName, FXmlParser.CurAttr);
      ptEndTag    : WhenEndTag    (FXmlParser.CurName);
      ptContent   : WhenContent   (FXmlParser.CurContent);
      ptCData     : WhenCData     (FXmlParser.CurContent);
      END;
END;


END.