Menu
▾
▴
[Woc-commit] CVS: woc/src/woc/src/cas CAS2NSC.txt,NONE,1.1 CAS2SMILES.txt,NONE,1.1 CAS2UN.txt,NONE,1.1 Makefile.am,NONE,1.1 add_nsc_number.pl,NONE,1.1 add_nsc_number.pl.in,NONE,1.1 add_numbers.pl,NONE,1.1 add_numbers.pl.in,NONE,1.1 add_smiles.pl,NONE,1.1 add_smiles.pl.in,NONE,1.1 add_un_nummers.pl,NONE,1.1 add_un_nummers.pl.in,NONE,1.1 check_cas_nummers.pl,NONE,1.1 check_cas_nummers.pl.in,NONE,1.1 check_doublenumbers.pl,NONE,1.1 check_doublenumbers.pl.in,NONE,1.1 zoek-UNs-zonder-wml.pl,NONE,1.1 zoek-UNs-zonder-wml.pl.in,NONE,1.1
Update of /cvsroot/woc/woc/src/woc/src/cas
In directory usw-pr-cvs1:/tmp/cvs-serv23171/woc/src/cas
Added Files:
CAS2NSC.txt CAS2SMILES.txt CAS2UN.txt Makefile.am
add_nsc_number.pl add_nsc_number.pl.in add_numbers.pl
add_numbers.pl.in add_smiles.pl add_smiles.pl.in
add_un_nummers.pl add_un_nummers.pl.in check_cas_nummers.pl
check_cas_nummers.pl.in check_doublenumbers.pl
check_doublenumbers.pl.in zoek-UNs-zonder-wml.pl
zoek-UNs-zonder-wml.pl.in
Log Message:
--- NEW FILE: CAS2NSC.txt ---
100-00-5 9792
100-01-6 9797
100-02-7 1317
100-06-1 5601
100-06-1 209523
100-07-2 86125
100-09-4 7926
100-09-4 32742
100-10-7 5517
100-11-8 4609
100-12-9 858
100-14-1 9803
100-15-2 5390
100-16-3 4079
100-17-4 5507
100-18-5 84198
100-19-6 41590
100-20-9 41885
100-21-0 36973
[...126750 lines suppressed...]
998-29-8 96837
998-30-1 124134
998-40-3 91700
998-41-4 111642
998-91-4 2321
998-91-4 224119
998-98-1 79255
999-21-3 4799
999-23-5 6406
999-29-1 58406
999-30-4 16588
999-33-7 223056
999-37-1 509380
999-47-3 34835
999-50-8 58807
999-55-3 20945
999-68-8 93208
999-79-1 525131
999-81-5 34858
999-97-3 93895
--- NEW FILE: CAS2SMILES.txt ---
100-00-5 [N+]([O-])(=O)C1=CC=C(Cl)C=C1
100-01-6 [N+]([O-])(=O)C1=CC=C(N)C=C1
100-02-7 [N+]([O-])(=O)C1=CC=C(O)C=C1
100-06-1 C(C)(=O)C1=CC=C(OC)C=C1
100-09-4 C(=O)(O)C1=CC=C(OC)C=C1
100-10-7 N(C)(C)C1=CC=C(C=O)C=C1
100-11-8 [N+]([O-])(=O)C1=CC=C(CBr)C=C1
100-12-9 [N+]([O-])(=O)C1=CC=C(CC)C=C1
100-14-1 [N+]([O-])(=O)C1=CC=C(CCl)C=C1
100-15-2 [N+]([O-])(=O)C1=CC=C(NC)C=C1
100-16-3 [N+]([O-])(=O)C1=CC=C(NN)C=C1
100-17-4 [N+]([O-])(=O)C1=CC=C(OC)C=C1
100-23-2 N(C)(C)C1=CC=C([N+](=O)[O-])C=C1
100-25-4 [N+]([O-])(=O)C1=CC=C([N+](=O)[O-])C=C1
100-26-5 C(=O)(O)C1=CC=C(C(=O)O)N=C1
100-28-7 [N+]([O-])(=O)C1=CC=C(N=C=O)C=C1
100-29-8 [N+]([O-])(=O)C1=CC=C(OCC)C=C1
100-32-3 [N+]([O-])(=O)C2=CC=C(SSC1=CC=C([N+](=O)[O-])C=C1)C=C2
100-33-4 C(=N)(N)C2=CC=C(OCCCCCOC1=CC=C(C(=N)N)C=C1)C=C2
[...9537 lines suppressed...]
99-87-6 C(C)(C)C1=CC=C(C)C=C1
99-88-7 C(C)(C)C1=CC=C(N)C=C1
99-89-8 C(C)(C)C1=CC=C(O)C=C1
99-91-2 C(C)(=O)C1=CC=C(Cl)C=C1
99-92-3 C(C)(=O)C1=CC=C(N)C=C1
99-93-4 C(C)(=O)C1=CC=C(O)C=C1
99-94-5 C(=O)(O)C1=CC=C(C)C=C1
99-96-7 C(=O)(O)C1=CC=C(O)C=C1
99-97-8 N(C)(C)C1=CC=C(C)C=C1
99-98-9 N(C)(C)C1=CC=C(N)C=C1
99-99-0 [N+]([O-])(=O)C1=CC=C(C)C=C1
992-59-6 S(=O)(=O)(O)C1=C6C(=C(N)C(=C1)N=NC5=C(C)C=C(C4=CC(=C(N=NC3=C(N)C2=C(C=CC=C2)C(=C3)S(=O)(=O)O)C=C4)C)C=C5)C=CC=C6
994-31-0 [Sn](CC)(CC)(CC)Cl
996-19-0 S(=O)(=O)(O)O.C(NN)(=N)N
996-97-4 N(C(CCCCCCC)=O)(CC)CC
996-98-5 C(C(NN)=O)(NN)=O
997-95-5 N(CC(C)C)(CC(C)C)N=O
998-91-4 C(=CC(OCC)=O)(OCC)C
999-21-3 C(C=CC(OCC=C)=O)(OCC=C)=O
999-23-5 N(CCN(CCC#N)C)(CCC#N)C
--- NEW FILE: CAS2UN.txt ---
#CAS-nummer <=> UN-nummer
100-00-5 1578
100-01-6 1661
100-02-7 1663
100-25-4 1597
100-37-8 2686
100-39-0 1737
100-40-3 1993
100-41-4 1175
100-42-5 2055
100-44-7 1738
100-47-0 2224
100-51-6 2810
100-52-7 1990
100-61-8 2294
100-63-0 2572
100-80-1 2618
100-97-0 1328
10024-97-2 1070
10025-67-9 1828
10025-69-1 2923
10025-78-2 1295
10025-87-3 1810
10025-91-9 1733
10026-04-7 1818
10026-13-8 1806
10031-13-7 1618
10035-10-6 1048
10045-94-0 1625
10048-95-0 1685
101-02-0 2811
101-68-8 2489
101-72-4 2811
101-77-9 2651
101-84-8 3077
10102-18-8 2630
10102-44-0 1067
10103-50-1 1622
10103-61-4 1557
10108-64-2 2570
10112-91-1 2025
10124-50-2 1678
10192-30-0 2693
102-82-9 2542
10265-92-6 2783
10290-12-7 1586
10294-33-4 2692
10294-34-5 1741
10326-27-9 1564
10361-37-2 1564
104-15-4 2585
104-94-9 02431
105-46-4 1123
105-58-8 2366
105-74-8 2124,
106-35-4 1224
106-42-3 1307
106-44-5 2076
106-46-7 2811
106-47-8 2018
106-48-9 2020
106-49-0 1708
106-50-3 1673
106-51-4 2587
106-87-6 2810
106-88-7 3022
106-89-8 2023
106-92-3 2219
106-93-4 1605
106-97-8 1011
106-98-9 1012
106-99-0 1010
107-01-7 1012
107-02-8 1092
107-05-1 1100
107-06-2 1184
107-07-3 1135
107-10-8 1277
107-11-9 2334
107-12-0 2404
107-13-1 1093
107-14-2 2668
107-15-3 1604
107-18-6 1098
107-19-7 2929
107-20-0 2232
107-30-2 1239
107-31-3 1243
107-39-1 2050
107-40-4 2050
107-49-3 3018
107-83-5 1208
107-87-9 1249
107-98-2 3092
108-01-0 2051
108-03-2 2608
108-05-4 1301
108-10-1 1245
108-11-2 2053
108-18-9 1158
108-20-3 1159
108-21-4 1220
108-23-6 2407
108-24-7 1715
108-31-6 2215
108-38-3 1307
108-39-4 2076
108-42-9 2019
108-43-0 2020
108-44-1 1708
108-46-3 2876
108-65-6 1993
108-67-8 2325
108-70-3 2321
108-77-0 2670
108-83-8 1157
108-87-2 2296
108-88-3 1294
108-89-4 2313
108-90-7 1134
108-91-8 2357
108-94-1 1915
108-95-2 1671
108-98-5 2337
108-99-6 2313
109-06-8 2313
109-55-7 2734
109-60-4 1276
109-66-0 1265
109-69-3 1127
109-73-9 1125
109-79-5 2347
109-86-4 1188
109-87-5 1234
109-89-7 1154
109-92-2 1302
109-94-4 1190
109-99-9 2056
110-00-9 2389
110-01-0 2412
110-02-1 2414
110-12-3 2302
110-16-7 1759
110-19-0 1213
110-43-0 1110
110-49-6 1189
110-54-3 1208
110-63-4 2810
110-80-5 1171
110-82-7 1145
110-83-8 2256
110-86-1 1282
110-89-4 2401
110-91-8 2054
11097-69-1 2315
111-14-8 3265
111-15-9 1172
111-30-8 2810
111-40-0 2079
111-65-9 1262
111-66-0 1993
111-69-3 2205
111-76-2 2369
111-84-2 1920
1113-38-8 2811
112-24-3 2259
112-55-0 1228
114-26-1 2757
115-11-7 1055
115-29-7 2761
115-32-2 2761
116-06-3 2757
116-29-0 2761
118-74-1 2729
118-75-2 2588
118-96-7 0209
119-93-7 2811
120-36-5 2765
120-51-4 2810
120-71-8 2431
120-80-9 2811
120-82-1 2321
120-83-2 2020
120-92-3 2245
12001-29-5 2590
12057-74-8 2011
12079-65-1 2811
121-14-2 2038
121-43-7 2416
121-44-8 1296
121-69-7 2253
121-75-5 3082
12108-13-3 2810
12122-67-7 2771
12125-01-8 2505
12136-45-7 2033
122-14-5 3018
122-34-9 2763
122-39-4 2811
122-52-1 2323
122-79-2 2810
123-05-7 1191
123-31-9 2662
123-38-6 1275
123-51-3 1105
123-54-6 2310
123-62-6 2496
123-72-8 1129
123-77-3 3242
123-86-4 1123
123-91-1 1165
123-92-2 1104
124-02-7 2359
124-09-4 2280
124-18-5 2247
124-38-9 1013
124-40-3 1032
124-41-4 1431
124-58-3 1557
124-63-0 3246
12427-38-2 2210
126-30-7 1325
126-75-0 3018
126-98-7 3079
126-99-8 1991
127-18-4 1897
1300-73-8 1711
1302-42-7 2812
1303-28-2 1559
1304-28-5 1884
1304-29-6 1449
1305-62-0 1759
1305-78-8 1910
1306-19-0 2570
1309-64-4 1549
131-17-9 2810
131-52-2 2567
1310-58-3 1813
1310-65-2 2680
1310-66-3 2680
1310-73-2 1823
1312-73-8 1382
1313-99-1 2811
1314-34-7 3285
1314-56-3 1807
1314-62-1 2862
1314-84-7 1714
1317-36-8 3288
13171-21-6 3018
1319-77-3 2076,
1321-65-9 2811
1327-53-3 1561
133-06-2 2773
1333-74-0 1049
1333-82-0 1463
1333-86-4 1361
1336-21-6 2672
1338-02-9 3009
13463-39-3 1259
13463-40-6 1994
13477-00-4 1445
13494-80-9 2811
137-26-8 2771
137-30-4 2771
137-32-6 1105
13768-86-0 2928
13952-84-6 1992
140-88-5 1917
141-32-2 2348
141-43-5 2491
141-66-2 3018
141-78-6 1173
141-79-7 1229
142-04-1 1548
142-28-9 1992
142-62-1 2829
142-82-5 1206
142-96-1 1149
143-10-2 1228
143-33-9 1689
144-49-0 2642
144-62-7 2923
14484-64-1 2771
14977-61-8 1758
151-50-8 1680
151-56-4 1185
151-67-7 1610
1563-66-2 2757
1569-69-3 3054
15950-66-0 2020
15972-60-8 2588
1600-27-7 1629
16219-75-3 1993
1634-04-4 2398
16752-77-5 2757
16871-90-2 2655
16872-11-0 1775
1689-83-4 2588
16893-85-9 2674
16961-83-4 1778
17702-41-9 1868
17804-35-2 2757
1836-75-5 2779
1897-45-6 2588
1910-42-5 2781
1912-24-9 2763
1912-26-1 2763
1918-00-9 2769
19287-45-7 1911
19624-22-7 1380
19750-95-9 2588
2032-59-9 2757
2050-92-2 2841
20816-12-0 2471
20859-73-8 1397
2104-64-5 2783
21908-53-2 1641
22224-92-6 2783
2243-62-1 None
2303-17-5 2757
2310-17-0 2783
2425-06-1 2773
25013-15-4 2618
25057-89-0 2588
25154-52-3 3082
25154-54-5 1597
2524-04-1 2751
25321-14-6 2038
25340-17-4 2049
2551-62-4 1080
26628-22-8 1687
2807-30-9 1993
2813-95-8 3014
28434-00-6 2902
2855-13-2 2289
287-92-3 1146
2893-78-9 2465
2921-88-2 2783
298-00-0 2783
299-86-5 2783
300-76-5 3018
301-04-2 1616
302-01-2 2029
302-17-0 2811
30525-89-4 2213
309-00-2 2761
3173-53-3 2488
3209-22-1 1578
3268-49-3 2785
333-41-5 3018
3333-67-3 2811
3383-96-8 2783
3452-97-9 n.o.s
353-50-4 2417
353-59-3 1974
35400-43-2 3018
3689-24-5 1704
3710-30-3 2309
3724-65-0 2823
3811-04-9 1485
39300-45-3 3013
3982-91-0 1837
4098-71-9 2290
4109-96-0 2189
4170-30-3 1143
420-04-2 3276
431-03-8 2346
479-45-8 0208
485-31-4 2779
50-00-0 1198
50-29-3 2761
50-78-2 2811
504-29-0 2671
506-68-3 1889
506-77-4 1589
506-78-5 1588
506-93-4 1467
506-96-7 1716
507-09-5 2436
51-28-5 1320
513-36-0 1127
51630-58-1 2902
52-51-7 3241
52-68-6 2783
52315-07-8 2902
52645-53-1 2902
528-29-0 1597
52918-63-5 2588
532-27-4 1697
5329-14-6 2967
533-74-4 2588
54-11-5 1654
540-59-0 1150
540-84-1 1262
541-41-3 1182
542-88-1 2249
542-92-7 1993
55-38-9 3018
55-63-0 0143
55-68-5 1895
554-00-7 1590
554-13-2 2811
556-52-5 2810
558-13-4 2516
56-23-5 1846
56-35-9 3020
56-38-2 3018
56-72-4 3027
563-12-2 3018
563-80-4 2397
57-06-7 1545
57-14-7 1163
57-24-9 1692
57-57-8 2810
57-74-9 2996
58-08-2 1544
58-89-9 2761
583-59-5 2617
583-78-8 2020
584-02-1 1105
584-79-2 2902
584-84-9 2078
59-50-7 2669
590-01-2 1914
590-18-1 1012
591-35-5 2020
591-78-6 1224
592-01-8 1575
592-27-8 1262
592-41-6 2370
592-57-4 1993
592-84-7 1128
594-42-3 1670
60-24-2 2966
60-29-7 1155
60-34-4 1244
60-35-5 2811
60-41-3 1692
60-51-5 2783
60-57-1 2761
602-01-7 2038
606-20-2 2038
608-27-5 1590
608-31-1 1590
608-73-1 2761
61-82-5 2588
610-39-9 2038
610-40-2 1577
611-06-3 1578
611-15-4 2618
6164-98-3 2588
61789-51-3 2001
61790-14-5 2810
62-38-4 1674
62-53-3 1547
62-55-5 2811
62-56-6 2810
62-73-7 3018
62-74-8 2629
62-75-9 2810
622-45-7 2243
622-97-9 2618
624-64-6 1012
624-83-9 2480
626-38-0 1104
626-93-7 2282
628-63-7 1104
63-25-2 2757
630-08-0 1016
63989-69-5 1607
64-17-5 1170
64-18-6 1779
64-19-7 2789
64-67-5 1594
640-15-3 3018
6484-52-2 1942
65-30-5 1658
65-31-6 1659
66-81-9 2588
67-56-1 1230
67-63-0 1219
67-64-1 1090
67-66-3 1888
674-82-8 2521
68-12-2 2265
68085-85-8 2902
681-84-5 2606
6834-92-0 1759
68694-11-1 2588
6923-22-4 2783
70-30-4 2875
71-23-8 1274
71-36-3 1120
71-41-0 1105
71-43-2 1114
71-55-6 2831
732-11-6 2783
74-82-8 1971
74-83-9 1062
74-84-0 1035
74-85-1 1962
74-86-2 1001
74-87-3 1063
74-89-5 1061
74-90-8 1051
74-93-1 1064
74-95-3 2664
74-97-5 1887
74-98-6 1978
74-99-7 1954
7439-90-9 1970
7439-92-1 3288
7439-93-2 1415
7439-95-4 1418
7439-95-4 1869
7439-97-6 2809
7440-01-9 1065
7440-16-6 3089
7440-28-0 1707
7440-29-1 2975
7440-36-0 2871
7440-37-1 1951
7440-38-2 1558
7440-39-3 1400
7440-41-7 1567
7440-43-9 2570
7440-44-0 1361
7440-48-4 3178
7440-58-6 2545
7440-59-7 1963
7440-63-3 2036
7440-66-6 1436
7440-70-2 1401
7446-08-4 2811
7446-09-5 1079
7446-18-6 1707
7487-94-7 1624
75-00-3 1037
75-01-4 1086
75-02-5 1860
75-04-7 1036
75-05-8 1648
75-07-0 1089
75-08-1 2363
75-09-2 1593
75-15-0 1131
75-18-3 1164
75-20-7 1402
75-21-8 1040
75-25-2 2515
75-31-0 1221
75-34-3 2362
75-35-4 1303
75-36-5 1717
75-38-7 1959
75-43-4 1029
75-44-5 1076
75-45-6 1018
75-46-7 1984
75-50-3 1083
75-52-5 1261
75-54-7 1242
75-55-8 1921
75-56-9 1280
75-63-8 1009
75-65-0 1120
75-69-4 1017
75-71-8 1028
75-72-9 1022
75-73-0 1982
75-74-1 1649
75-77-4 1298
75-79-6 1250
75-86-5 1541
75-91-2 3109
75-98-9 1759
7550-45-0 1838
7580-67-8 1414
759-94-4 2992
76-15-3 1020
76-44-8 2761
7601-89-0 1502
7616-94-6 3083
7631-99-4 1498
7632-00-0 1500
7637-07-2 1008
7646-78-8 1827
7647-01-0 1050
7664-38-2 1805
7664-39-3 1052
7664-41-7 1005
7664-93-9 1830
768-52-5 2810
7681-52-9 1791
7697-37-2 2031
77-58-7 2788
77-78-1 1595
7704-34-9 1350
7719-12-2 1809
7722-84-1 2015
7723-14-0 1381
7726-95-6 1744
7727-37-9 1066
7727-37-9 1977
7727-43-7 1564
7757-79-1 1486
7758-01-2 1484
7758-09-0 1488
7761-88-8 1493
7772-99-8 2923
7775-09-9 1495
7778-43-0 1685
7778-44-1 1573
7778-74-7 1489
7779-88-6 1514
7782-41-4 1045
7782-44-7 1072
7782-44-7 1073
7782-49-2 2658
7782-50-5 1017
7782-65-2 2192
7783-00-8 2811
7783-06-4 1053
7783-07-5 2202
7783-35-9 1645
7783-41-7 2190
7783-47-3 3288
7783-54-2 2451
7783-61-1 1859
7783-70-2 1732
7783-79-1 2194
7783-81-5 2978
7784-34-1 1560
7784-40-9 1617
7784-41-0 1677
7784-42-1 2188
7784-44-3 1546
7786-34-7 3018
7789-06-2 2811
7789-30-2 1745
7789-38-0 1494
7790-91-2 1749
7791-23-3 2879
7791-25-5 1834
78-00-2 1649
78-10-4 1292
78-30-8 2574
78-34-2 3018
78-78-4 1265
78-79-5 1218
78-81-9 1214
78-83-1 1212
78-84-2 2045
78-85-3 2396
78-87-5 1279
78-90-0 2258
78-92-2 1120
78-93-3 1193
78-95-5 1695
78-96-6 2735
78-99-9 1992
7803-51-2 2199
7803-52-3 2676
7803-62-5 2203
786-19-6 3018
79-00-5 3082
79-01-6 1710
79-04-9 1752
79-06-1 2074
79-07-2 2811
79-09-4 1848
79-10-7 2218
79-11-8 1751
79-20-9 1231
79-22-1 1238
79-24-3 2842
79-27-6 2504
79-34-5 1702
79-36-7 1765
79-41-4 2531
79-43-6 1764
79-46-9 2608
80-15-9 3109
80-62-6 1247
8001-35-2 2761
8008-20-6 1223
8018-01-7 2771
8052-41-3 1268
81-81-2 3027
811-97-2 3159
818-08-6 3146
822-06-0 2281
83-79-4 2588
836-30-6 1325
84-74-2 3082
85-44-9 2214
86-50-0 2783
86-88-4 1651
867-27-6 3018
87-61-6 2811
87-66-1 2811
87-68-3 2279
87-86-5 3155
873-66-5 2618
88-06-2 2020
88-72-2 1664
88-73-3 1578
88-74-4 1661
88-85-7 2779
88-89-1 0154
89-98-5 1760
90-04-0 2431
90-30-2 2811
90035-08-8 3027
91-22-5 2656
91-64-5 2811
91465-08-6 2588
919-86-8 3018
92-67-1 2811
92-87-5 1885
93-58-3 2938
93-65-2 2765
93-76-5 2765
933-75-5 2020
933-78-8 2020
935-95-5 2020
94-36-0 3102
94-74-6 2765
94-75-7 2765
944-22-9 3018
95-47-6 1307
95-48-7 2076
95-50-1 1591
95-51-2 2019
95-53-4 1708
95-55-6 2512
95-57-8 2021
95-76-1 1590
95-80-7 1709
95-82-9 1590
95-95-4 2020
95266-40-3 2588
957-51-7 2588
96-12-8 2872
96-14-0 1208
96-18-4 2810
96-22-0 1156
96-33-3 1919
97-00-7 1577
97-02-9 1596
97-63-2 2277
98-00-0 2874
98-01-1 1199
98-05-5 1557
98-07-7 2226
98-82-8 1918
98-83-9 2303
98-87-3 1886
98-88-4 1736
98-95-3 1662
99-09-2 1661
99-54-7 1578
99-65-0 1597
99-87-6 2046
99-97-8 2810
99-99-0 1664
999-61-1 1760
--- NEW FILE: Makefile.am ---
bindir=${prefix}/bin/cas
bin_SCRIPTS = \
add_nsc_number.pl \
add_numbers.pl \
add_smiles.pl \
add_un_nummers.pl \
check_cas_nummers.pl \
check_doublenumbers.pl \
zoek-UNs-zonder-wml.pl
CLEANFILES = $(bin_SCRIPTS)
--- NEW FILE: add_nsc_number.pl ---
#! /usr/bin/perl -w
use strict;
use diagnostics;
# Check command line options
if (@ARGV == 0) {
print "Usage: $0 <xml-files>$/";
exit;
}
# Global variables
my $CAS_NSC_table_file = "CAS2NSC.txt";
my @CAS_NSC_table;
my $nr_NSC_numbers = 0;
my $nr_added_NSC_numbers = 0;
my $nr_unable = 0;
# Read CAS<=>NSC table
if (open (FILE, $CAS_NSC_table_file)) {
while (<FILE>) {
chomp;
push (@CAS_NSC_table, $_);
}
close (FILE) || die "$CAS_NSC_table_file: $!$/";
} else {
die "$CAS_NSC_table_file: $!$/";
}
# Loop over all files
foreach my $arg (@ARGV) {
&add_NSC_number_to_file ($arg);
}
# Print statistics
print " ------------------$/";
if ($nr_NSC_numbers) {
print "NSC numbers found : $nr_NSC_numbers$/";
} else {
print "No NSC numbers found.$/";
}
if ($nr_unable) {
print "Unable to changes: $nr_unable$/" if ($nr_unable);
}
if ($nr_added_NSC_numbers) {
print "NSC numbers added : $nr_added_NSC_numbers$/";
print "NSC numbers total : ",$nr_NSC_numbers+$nr_added_NSC_numbers,$/;
} else {
print "No NSC numbers added.$/";
}
sub add_NSC_number_to_file {
my $file = shift;
my @XML_data;
my $changed = "";
print "Processing $file. ";
if (open (FILE, $file)) {
my $cas_found = "";
my $chemical = "";
while (<FILE>) {
if (/CHEMICAL/i) {
push (@XML_data, $_);
$chemical = "True";
} elsif (/CAS-NUMBER/i && /\>(.+?)\</) {
$cas_found = "True";
my $CAS_line = $_;
my $NSC_line = $CAS_line;
my $CAS_number = $1;
push (@XML_data, $CAS_line);
my $next_line = <FILE> if (!eof(FILE));
if ($next_line =~ /NSC-NUMBER/) {
push (@XML_data, $next_line);
$nr_NSC_numbers++;
} else {
my $NSC_number;
if ($NSC_number = &get_NSC_number($CAS_number)) {
$NSC_line =~ s/CAS-NUMBER/NSC-NUMBER/;
$NSC_line =~ s/\>.*?\</\>$NSC_number\</;
push (@XML_data, $NSC_line);
$changed = "True";
}
push (@XML_data, $next_line);
}
} elsif (/WOC-NUMBER/ && /\>(.+?)\</) {
push (@XML_data, $_);
$cas_found = "True";
} else {
push (@XML_data, $_);
}
}
print "No CAS number found!! " if ($chemical && !$cas_found);
close (FILE) || warn "$file: $!$/";
} else {
warn "$file: $!$/";
}
if ($changed) {
if (open (FILE, ">$file")) {
foreach my $line (@XML_data) {
print FILE $line;
}
close (FILE) || die $!;
print "NSC number added...$/";
$nr_added_NSC_numbers++;
} else {
warn "$file: $!$/";
print "Unable to change...$/";
$nr_unable++;
}
} else {
print "No changes...$/";
}
}
sub get_NSC_number {
my $CAS_number = shift;
my $NSC_number = "";
foreach my $line (@CAS_NSC_table) {
if ($line =~ /^$CAS_number\s*(.*)/) {
$NSC_number = $1;
}
}
return $NSC_number;
}
--- NEW FILE: add_nsc_number.pl.in ---
#! @PATHTOPERL@ -w
use strict;
use diagnostics;
# Check command line options
if (@ARGV == 0) {
print "Usage: $0 <xml-files>$/";
exit;
}
# Global variables
my $CAS_NSC_table_file = "CAS2NSC.txt";
my @CAS_NSC_table;
my $nr_NSC_numbers = 0;
my $nr_added_NSC_numbers = 0;
my $nr_unable = 0;
# Read CAS<=>NSC table
if (open (FILE, $CAS_NSC_table_file)) {
while (<FILE>) {
chomp;
push (@CAS_NSC_table, $_);
}
close (FILE) || die "$CAS_NSC_table_file: $!$/";
} else {
die "$CAS_NSC_table_file: $!$/";
}
# Loop over all files
foreach my $arg (@ARGV) {
&add_NSC_number_to_file ($arg);
}
# Print statistics
print " ------------------$/";
if ($nr_NSC_numbers) {
print "NSC numbers found : $nr_NSC_numbers$/";
} else {
print "No NSC numbers found.$/";
}
if ($nr_unable) {
print "Unable to changes: $nr_unable$/" if ($nr_unable);
}
if ($nr_added_NSC_numbers) {
print "NSC numbers added : $nr_added_NSC_numbers$/";
print "NSC numbers total : ",$nr_NSC_numbers+$nr_added_NSC_numbers,$/;
} else {
print "No NSC numbers added.$/";
}
sub add_NSC_number_to_file {
my $file = shift;
my @XML_data;
my $changed = "";
print "Processing $file. ";
if (open (FILE, $file)) {
my $cas_found = "";
my $chemical = "";
while (<FILE>) {
if (/CHEMICAL/i) {
push (@XML_data, $_);
$chemical = "True";
} elsif (/CAS-NUMBER/i && /\>(.+?)\</) {
$cas_found = "True";
my $CAS_line = $_;
my $NSC_line = $CAS_line;
my $CAS_number = $1;
push (@XML_data, $CAS_line);
my $next_line = <FILE> if (!eof(FILE));
if ($next_line =~ /NSC-NUMBER/) {
push (@XML_data, $next_line);
$nr_NSC_numbers++;
} else {
my $NSC_number;
if ($NSC_number = &get_NSC_number($CAS_number)) {
$NSC_line =~ s/CAS-NUMBER/NSC-NUMBER/;
$NSC_line =~ s/\>.*?\</\>$NSC_number\</;
push (@XML_data, $NSC_line);
$changed = "True";
}
push (@XML_data, $next_line);
}
} elsif (/WOC-NUMBER/ && /\>(.+?)\</) {
push (@XML_data, $_);
$cas_found = "True";
} else {
push (@XML_data, $_);
}
}
print "No CAS number found!! " if ($chemical && !$cas_found);
close (FILE) || warn "$file: $!$/";
} else {
warn "$file: $!$/";
}
if ($changed) {
if (open (FILE, ">$file")) {
foreach my $line (@XML_data) {
print FILE $line;
}
close (FILE) || die $!;
print "NSC number added...$/";
$nr_added_NSC_numbers++;
} else {
warn "$file: $!$/";
print "Unable to change...$/";
$nr_unable++;
}
} else {
print "No changes...$/";
}
}
sub get_NSC_number {
my $CAS_number = shift;
my $NSC_number = "";
foreach my $line (@CAS_NSC_table) {
if ($line =~ /^$CAS_number\s*(.*)/) {
$NSC_number = $1;
}
}
return $NSC_number;
}
--- NEW FILE: add_numbers.pl ---
#! /usr/bin/perl -w
use strict;
use diagnostics;
my $number_type = "SMILES";
# Check command line options
if (@ARGV < 1) {
print "Usage: $0 [-type <TYPE>] <xml-files>$/";
print "\t<TYPE> = (SMILES | UN[-NUMBER] | NSC[-NUMBER])$/";
exit;
} elsif ((@ARGV >= 3) && ($ARGV[0] =~ /^-t/i)) {
shift @ARGV;
$number_type = shift @ARGV;
}
# Global variables
my $CAS_Lookup_table_file = "CAS2SMILES.txt";
my @CAS_Lookup_table;
my $nr_numbers = 0;
my $nr_added_numbers = 0;
my $nr_unable = 0;
my $nr_double = 0;
if ($number_type =~ /^SMILES/i) {
$number_type = "SMILES";
$CAS_Lookup_table_file = "CAS2SMILES.txt";
} elsif ($number_type =~ /^NSC/i) {
$number_type = "NSC-NUMBER";
$CAS_Lookup_table_file = "CAS2NSC.txt";
} elsif ($number_type =~ /^UN/i) {
$number_type = "UN-NUMBER";
$CAS_Lookup_table_file = "CAS2UN.txt";
} else {
print "Error: unrecognized type $number_type!$/";
print "Usage: $0 [-type <TYPE>] <xml-files>$/";
print "\t<TYPE> = (SMILES | UN[-NUMBER] | NSC[-NUMBER])$/";
exit;
}
# Read CAS<=><TYPE> table
if (open (FILE, $CAS_Lookup_table_file)) {
while (<FILE>) {
chomp;
push (@CAS_Lookup_table, $_);
}
close (FILE) || die "$CAS_Lookup_table_file: $!$/";
} else {
die "$CAS_Lookup_table_file: $!$/";
}
# Loop over all files
foreach my $arg (@ARGV) {
&add_number_to_file ($arg);
}
# Print statistics
print " ------------------$/";
if ($nr_numbers) {
print "$number_type strings found : $nr_numbers$/";
} else {
print "No $number_type strings found.$/";
}
if ($nr_unable) {
print "Unable to changes: $nr_unable$/" if ($nr_unable);
}
if ($nr_double) {
print "Double $number_type strings: $nr_double$/" if ($nr_double);
}
if ($nr_added_numbers) {
print "$number_type strings added : $nr_added_numbers$/";
print "$number_type strings total : ",$nr_numbers+$nr_added_numbers,$/;
} else {
print "No $number_type strings added.$/";
}
sub add_number_to_file {
my $file = shift;
my @XML_data;
my $changed = "";
my $number_found = "";
my $double_number = "";
my @numbers = ();
print "Processing $file. ";
if (open (FILE, $file)) {
my $cas_found = "";
my $chemical = "";
while (<FILE>) {
if (/CHEMICAL/i) {
push (@XML_data, $_);
$chemical = "True";
} elsif (/CAS-NUMBER/i && /\>(.+?)\</) {
$cas_found = "True";
if (!$number_found) {
my $CAS_line = $_;
my $number_line = $CAS_line;
my $CAS_number = $1;
push (@XML_data, $CAS_line);
my $next_line = <FILE> if (!eof(FILE));
if ($next_line =~ /$number_type/) {
push (@XML_data, $next_line);
$number_found = "True";
} else {
my $number;
if ($number = &get_number($CAS_number)) {
$number_line =~ s/CAS-NUMBER/$number_type/;
$number_line =~ s/\>.*?\</\>$number\</;
push (@numbers, $number);
push (@XML_data, $number_line);
$changed = "True";
}
push (@XML_data, $next_line);
}
}
} elsif (/$number_type/ && /\>(.+?)\</) {
push (@numbers, $1);
if (!$number_found) {
push (@XML_data, $_);
$number_found = "True";
} else {
$double_number = "True";
}
} elsif (/WOC-NUMBER/ && /\>(.+?)\</) {
push (@XML_data, $_);
$cas_found = "True";
} else {
push (@XML_data, $_);
}
}
print "No CAS number found!! " if ($chemical && !$cas_found);
close (FILE) || warn "$file: $!$/";
} else {
warn "$file: $!$/";
}
if ($double_number) {
print "Double $number_type string found!";
$nr_numbers++;
$nr_double++;
$changed = "True";
} elsif ($number_found) {
print "No changes needed...$/";
$nr_numbers++;
$changed = "";
}
if ($changed) {
if (open (FILE, ">$file")) {
foreach my $line (@XML_data) {
print FILE $line;
}
close (FILE) || die $!;
if ($double_number) {
print " Attempting to remove...$/";
} else {
print "$number_type number added...$/";
$nr_added_numbers++;
}
} else {
warn "$file: $!$/";
print "Unable to change...$/";
$nr_unable++;
}
} else {
print "No changes...$/";
}
}
sub get_number {
my $CAS_number = shift;
my $number = "";
foreach my $line (@CAS_Lookup_table) {
if ($line =~ /^$CAS_number\s*(.*)/) {
$number = $1;
}
}
return $number;
}
--- NEW FILE: add_numbers.pl.in ---
#! @PATHTOPERL@ -w
use strict;
use diagnostics;
my $number_type = "SMILES";
# Check command line options
if (@ARGV < 1) {
print "Usage: $0 [-type <TYPE>] <xml-files>$/";
print "\t<TYPE> = (SMILES | UN[-NUMBER] | NSC[-NUMBER])$/";
exit;
} elsif ((@ARGV >= 3) && ($ARGV[0] =~ /^-t/i)) {
shift @ARGV;
$number_type = shift @ARGV;
}
# Global variables
my $CAS_Lookup_table_file = "CAS2SMILES.txt";
my @CAS_Lookup_table;
my $nr_numbers = 0;
my $nr_added_numbers = 0;
my $nr_unable = 0;
my $nr_double = 0;
if ($number_type =~ /^SMILES/i) {
$number_type = "SMILES";
$CAS_Lookup_table_file = "CAS2SMILES.txt";
} elsif ($number_type =~ /^NSC/i) {
$number_type = "NSC-NUMBER";
$CAS_Lookup_table_file = "CAS2NSC.txt";
} elsif ($number_type =~ /^UN/i) {
$number_type = "UN-NUMBER";
$CAS_Lookup_table_file = "CAS2UN.txt";
} else {
print "Error: unrecognized type $number_type!$/";
print "Usage: $0 [-type <TYPE>] <xml-files>$/";
print "\t<TYPE> = (SMILES | UN[-NUMBER] | NSC[-NUMBER])$/";
exit;
}
# Read CAS<=><TYPE> table
if (open (FILE, $CAS_Lookup_table_file)) {
while (<FILE>) {
chomp;
push (@CAS_Lookup_table, $_);
}
close (FILE) || die "$CAS_Lookup_table_file: $!$/";
} else {
die "$CAS_Lookup_table_file: $!$/";
}
# Loop over all files
foreach my $arg (@ARGV) {
&add_number_to_file ($arg);
}
# Print statistics
print " ------------------$/";
if ($nr_numbers) {
print "$number_type strings found : $nr_numbers$/";
} else {
print "No $number_type strings found.$/";
}
if ($nr_unable) {
print "Unable to changes: $nr_unable$/" if ($nr_unable);
}
if ($nr_double) {
print "Double $number_type strings: $nr_double$/" if ($nr_double);
}
if ($nr_added_numbers) {
print "$number_type strings added : $nr_added_numbers$/";
print "$number_type strings total : ",$nr_numbers+$nr_added_numbers,$/;
} else {
print "No $number_type strings added.$/";
}
sub add_number_to_file {
my $file = shift;
my @XML_data;
my $changed = "";
my $number_found = "";
my $double_number = "";
my @numbers = ();
print "Processing $file. ";
if (open (FILE, $file)) {
my $cas_found = "";
my $chemical = "";
while (<FILE>) {
if (/CHEMICAL/i) {
push (@XML_data, $_);
$chemical = "True";
} elsif (/CAS-NUMBER/i && /\>(.+?)\</) {
$cas_found = "True";
if (!$number_found) {
my $CAS_line = $_;
my $number_line = $CAS_line;
my $CAS_number = $1;
push (@XML_data, $CAS_line);
my $next_line = <FILE> if (!eof(FILE));
if ($next_line =~ /$number_type/) {
push (@XML_data, $next_line);
$number_found = "True";
} else {
my $number;
if ($number = &get_number($CAS_number)) {
$number_line =~ s/CAS-NUMBER/$number_type/;
$number_line =~ s/\>.*?\</\>$number\</;
push (@numbers, $number);
push (@XML_data, $number_line);
$changed = "True";
}
push (@XML_data, $next_line);
}
}
} elsif (/$number_type/ && /\>(.+?)\</) {
push (@numbers, $1);
if (!$number_found) {
push (@XML_data, $_);
$number_found = "True";
} else {
$double_number = "True";
}
} elsif (/WOC-NUMBER/ && /\>(.+?)\</) {
push (@XML_data, $_);
$cas_found = "True";
} else {
push (@XML_data, $_);
}
}
print "No CAS number found!! " if ($chemical && !$cas_found);
close (FILE) || warn "$file: $!$/";
} else {
warn "$file: $!$/";
}
if ($double_number) {
print "Double $number_type string found!";
$nr_numbers++;
$nr_double++;
$changed = "True";
} elsif ($number_found) {
print "No changes needed...$/";
$nr_numbers++;
$changed = "";
}
if ($changed) {
if (open (FILE, ">$file")) {
foreach my $line (@XML_data) {
print FILE $line;
}
close (FILE) || die $!;
if ($double_number) {
print " Attempting to remove...$/";
} else {
print "$number_type number added...$/";
$nr_added_numbers++;
}
} else {
warn "$file: $!$/";
print "Unable to change...$/";
$nr_unable++;
}
} else {
print "No changes...$/";
}
}
sub get_number {
my $CAS_number = shift;
my $number = "";
foreach my $line (@CAS_Lookup_table) {
if ($line =~ /^$CAS_number\s*(.*)/) {
$number = $1;
}
}
return $number;
}
--- NEW FILE: add_smiles.pl ---
#! /usr/bin/perl -w
use strict;
use diagnostics;
# Check command line options
if (@ARGV == 0) {
print "Usage: $0 <xml-files>$/";
exit;
}
# Global variables
my $CAS_SMILES_table_file = "CAS2SMILES.txt";
my @CAS_SMILES_table;
my $nr_SMILES_numbers = 0;
my $nr_added_SMILES_numbers = 0;
my $nr_unable = 0;
my $nr_double = 0;
# Read CAS<=>SMILES table
if (open (FILE, $CAS_SMILES_table_file)) {
while (<FILE>) {
chomp;
push (@CAS_SMILES_table, $_);
}
close (FILE) || die "$CAS_SMILES_table_file: $!$/";
} else {
die "$CAS_SMILES_table_file: $!$/";
}
# Loop over all files
foreach my $arg (@ARGV) {
&add_SMILES_number_to_file ($arg);
}
# Print statistics
print " ------------------$/";
if ($nr_SMILES_numbers) {
print "SMILES strings found : $nr_SMILES_numbers$/";
} else {
print "No SMILES strings found.$/";
}
if ($nr_unable) {
print "Unable to changes: $nr_unable$/" if ($nr_unable);
}
if ($nr_double) {
print "Double SMILES strings: $nr_double$/" if ($nr_double);
}
if ($nr_added_SMILES_numbers) {
print "SMILES strings added : $nr_added_SMILES_numbers$/";
print "SMILES strings total : ",$nr_SMILES_numbers+$nr_added_SMILES_numbers,$/;
} else {
print "No SMILES strings added.$/";
}
sub add_SMILES_number_to_file {
my $file = shift;
my @XML_data;
my $changed = "";
my $smiles_found = "";
my $double_smiles = "";
my @smiles = ();
print "Processing $file. ";
if (open (FILE, $file)) {
my $cas_found = "";
my $chemical = "";
while (<FILE>) {
if (/CHEMICAL/i) {
push (@XML_data, $_);
$chemical = "True";
} elsif (/CAS-NUMBER/i && /\>(.+?)\</) {
$cas_found = "True";
if (!$smiles_found) {
my $CAS_line = $_;
my $SMILES_line = $CAS_line;
my $CAS_number = $1;
push (@XML_data, $CAS_line);
my $next_line = <FILE> if (!eof(FILE));
if ($next_line =~ /SMILES/) {
push (@XML_data, $next_line);
$smiles_found = "True";
} else {
my $SMILES_number;
if ($SMILES_number = &get_SMILES_number($CAS_number)) {
$SMILES_line =~ s/CAS-NUMBER/SMILES/;
$SMILES_line =~ s/\>.*?\</\>$SMILES_number\</;
push (@XML_data, $SMILES_line);
$changed = "True";
}
push (@XML_data, $next_line);
}
}
} elsif (/SMILES/ && /\>(.+?)\</) {
if (!$smiles_found) {
push (@XML_data, $_);
$smiles_found = "True";
} else {
$double_smiles = "True";
}
} elsif (/WOC-NUMBER/ && /\>(.+?)\</) {
push (@XML_data, $_);
$cas_found = "True";
} else {
push (@XML_data, $_);
}
}
print "No CAS number found!! " if ($chemical && !$cas_found);
close (FILE) || warn "$file: $!$/";
} else {
warn "$file: $!$/";
}
if ($double_smiles) {
print "Double SMILES string found!";
$nr_SMILES_numbers++;
$nr_double++;
$changed = "True";
} elsif ($smiles_found) {
print "No changes needed...$/";
$nr_SMILES_numbers++;
$changed = "";
}
if ($changed) {
if (open (FILE, ">$file")) {
foreach my $line (@XML_data) {
print FILE $line;
}
close (FILE) || die $!;
if ($double_smiles) {
print " Attempting to remove...$/";
} else {
print "SMILES number added...$/";
$nr_added_SMILES_numbers++;
}
} else {
warn "$file: $!$/";
print "Unable to change...$/";
$nr_unable++;
}
} else {
print "No changes...$/";
}
}
sub get_SMILES_number {
my $CAS_number = shift;
my $SMILES_number = "";
foreach my $line (@CAS_SMILES_table) {
if ($line =~ /^$CAS_number\s*(.*)/) {
$SMILES_number = $1;
}
}
return $SMILES_number;
}
--- NEW FILE: add_smiles.pl.in ---
#! @PATHTOPERL@ -w
use strict;
use diagnostics;
# Check command line options
if (@ARGV == 0) {
print "Usage: $0 <xml-files>$/";
exit;
}
# Global variables
my $CAS_SMILES_table_file = "CAS2SMILES.txt";
my @CAS_SMILES_table;
my $nr_SMILES_numbers = 0;
my $nr_added_SMILES_numbers = 0;
my $nr_unable = 0;
my $nr_double = 0;
# Read CAS<=>SMILES table
if (open (FILE, $CAS_SMILES_table_file)) {
while (<FILE>) {
chomp;
push (@CAS_SMILES_table, $_);
}
close (FILE) || die "$CAS_SMILES_table_file: $!$/";
} else {
die "$CAS_SMILES_table_file: $!$/";
}
# Loop over all files
foreach my $arg (@ARGV) {
&add_SMILES_number_to_file ($arg);
}
# Print statistics
print " ------------------$/";
if ($nr_SMILES_numbers) {
print "SMILES strings found : $nr_SMILES_numbers$/";
} else {
print "No SMILES strings found.$/";
}
if ($nr_unable) {
print "Unable to changes: $nr_unable$/" if ($nr_unable);
}
if ($nr_double) {
print "Double SMILES strings: $nr_double$/" if ($nr_double);
}
if ($nr_added_SMILES_numbers) {
print "SMILES strings added : $nr_added_SMILES_numbers$/";
print "SMILES strings total : ",$nr_SMILES_numbers+$nr_added_SMILES_numbers,$/;
} else {
print "No SMILES strings added.$/";
}
sub add_SMILES_number_to_file {
my $file = shift;
my @XML_data;
my $changed = "";
my $smiles_found = "";
my $double_smiles = "";
my @smiles = ();
print "Processing $file. ";
if (open (FILE, $file)) {
my $cas_found = "";
my $chemical = "";
while (<FILE>) {
if (/CHEMICAL/i) {
push (@XML_data, $_);
$chemical = "True";
} elsif (/CAS-NUMBER/i && /\>(.+?)\</) {
$cas_found = "True";
if (!$smiles_found) {
my $CAS_line = $_;
my $SMILES_line = $CAS_line;
my $CAS_number = $1;
push (@XML_data, $CAS_line);
my $next_line = <FILE> if (!eof(FILE));
if ($next_line =~ /SMILES/) {
push (@XML_data, $next_line);
$smiles_found = "True";
} else {
my $SMILES_number;
if ($SMILES_number = &get_SMILES_number($CAS_number)) {
$SMILES_line =~ s/CAS-NUMBER/SMILES/;
$SMILES_line =~ s/\>.*?\</\>$SMILES_number\</;
push (@XML_data, $SMILES_line);
$changed = "True";
}
push (@XML_data, $next_line);
}
}
} elsif (/SMILES/ && /\>(.+?)\</) {
if (!$smiles_found) {
push (@XML_data, $_);
$smiles_found = "True";
} else {
$double_smiles = "True";
}
} elsif (/WOC-NUMBER/ && /\>(.+?)\</) {
push (@XML_data, $_);
$cas_found = "True";
} else {
push (@XML_data, $_);
}
}
print "No CAS number found!! " if ($chemical && !$cas_found);
close (FILE) || warn "$file: $!$/";
} else {
warn "$file: $!$/";
}
if ($double_smiles) {
print "Double SMILES string found!";
$nr_SMILES_numbers++;
$nr_double++;
$changed = "True";
} elsif ($smiles_found) {
print "No changes needed...$/";
$nr_SMILES_numbers++;
$changed = "";
}
if ($changed) {
if (open (FILE, ">$file")) {
foreach my $line (@XML_data) {
print FILE $line;
}
close (FILE) || die $!;
if ($double_smiles) {
print " Attempting to remove...$/";
} else {
print "SMILES number added...$/";
$nr_added_SMILES_numbers++;
}
} else {
warn "$file: $!$/";
print "Unable to change...$/";
$nr_unable++;
}
} else {
print "No changes...$/";
}
}
sub get_SMILES_number {
my $CAS_number = shift;
my $SMILES_number = "";
foreach my $line (@CAS_SMILES_table) {
if ($line =~ /^$CAS_number\s*(.*)/) {
$SMILES_number = $1;
}
}
return $SMILES_number;
}
--- NEW FILE: add_un_nummers.pl ---
#! /usr/bin/perl -w
use strict;
use diagnostics;
# Check command line options
if (@ARGV == 0) {
print "Usage: $0 <xml-files>$/";
exit;
}
# Global variables
my $CAS_UN_table_file = "CAS2UN.txt";
my @CAS_UN_table;
my $nr_UN_numbers = 0;
my $nr_added_UN_numbers = 0;
my $nr_unable = 0;
# Read CAS<=>UN table
if (open (FILE, $CAS_UN_table_file)) {
while (<FILE>) {
chomp;
push (@CAS_UN_table, $_);
}
close (FILE) || die "$CAS_UN_table_file: $!$/";
} else {
die "$CAS_UN_table_file: $!$/";
}
# Loop over all files
foreach my $arg (@ARGV) {
&add_UN_number_to_file ($arg);
}
# Print statistics
print " ------------------$/";
if ($nr_UN_numbers) {
print "UN numbers found : $nr_UN_numbers$/";
} else {
print "No UN numbers found.$/";
}
if ($nr_unable) {
print "Unable to changes: $nr_unable$/" if ($nr_unable);
}
if ($nr_added_UN_numbers) {
print "UN numbers added : $nr_added_UN_numbers$/";
print "UN numbers total : ",$nr_UN_numbers+$nr_added_UN_numbers,$/;
} else {
print "No UN numbers added.$/";
}
sub add_UN_number_to_file {
my $file = shift;
my @XML_data;
my $changed = "";
print "Processing $file. ";
if (open (FILE, $file)) {
my $cas_found = "";
my $chemical = "";
while (<FILE>) {
if (/CHEMICAL/i) {
push (@XML_data, $_);
$chemical = "True";
} elsif (/CAS-NUMBER/i && /\>(.+?)\</) {
$cas_found = "True";
my $CAS_line = $_;
my $UN_line = $CAS_line;
my $CAS_number = $1;
push (@XML_data, $CAS_line);
my $next_line = <FILE> if (!eof(FILE));
if ($next_line =~ /UN-NUMBER/) {
push (@XML_data, $next_line);
$nr_UN_numbers++;
} else {
my $UN_number;
if ($UN_number = &get_UN_number($CAS_number)) {
$UN_line =~ s/CAS-NUMBER/UN-NUMBER/;
$UN_line =~ s/\>.*?\</\>$UN_number\</;
push (@XML_data, $UN_line);
$changed = "True";
}
push (@XML_data, $next_line);
}
} elsif (/WOC-NUMBER/ && /\>(.+?)\</) {
push (@XML_data, $_);
$cas_found = "True";
} else {
push (@XML_data, $_);
}
}
print "No CAS number found!! " if ($chemical && !$cas_found);
close (FILE) || warn "$file: $!$/";
} else {
warn "$file: $!$/";
}
if ($changed) {
if (open (FILE, ">$file")) {
foreach my $line (@XML_data) {
print FILE $line;
}
close (FILE) || die $!;
print "UN number added...$/";
$nr_added_UN_numbers++;
} else {
warn "$file: $!$/";
print "Unable to change...$/";
$nr_unable++;
}
} else {
print "No changes...$/";
}
}
sub get_UN_number {
my $CAS_number = shift;
my $UN_number = "";
foreach my $line (@CAS_UN_table) {
if ($line =~ /^$CAS_number\s*\<\=\>\s*(.*)/) {
$UN_number = $1;
}
}
return $UN_number;
}
--- NEW FILE: add_un_nummers.pl.in ---
#! @PATHTOPERL@ -w
use strict;
use diagnostics;
# Check command line options
if (@ARGV == 0) {
print "Usage: $0 <xml-files>$/";
exit;
}
# Global variables
my $CAS_UN_table_file = "CAS2UN.txt";
my @CAS_UN_table;
my $nr_UN_numbers = 0;
my $nr_added_UN_numbers = 0;
my $nr_unable = 0;
# Read CAS<=>UN table
if (open (FILE, $CAS_UN_table_file)) {
while (<FILE>) {
chomp;
push (@CAS_UN_table, $_);
}
close (FILE) || die "$CAS_UN_table_file: $!$/";
} else {
die "$CAS_UN_table_file: $!$/";
}
# Loop over all files
foreach my $arg (@ARGV) {
&add_UN_number_to_file ($arg);
}
# Print statistics
print " ------------------$/";
if ($nr_UN_numbers) {
print "UN numbers found : $nr_UN_numbers$/";
} else {
print "No UN numbers found.$/";
}
if ($nr_unable) {
print "Unable to changes: $nr_unable$/" if ($nr_unable);
}
if ($nr_added_UN_numbers) {
print "UN numbers added : $nr_added_UN_numbers$/";
print "UN numbers total : ",$nr_UN_numbers+$nr_added_UN_numbers,$/;
} else {
print "No UN numbers added.$/";
}
sub add_UN_number_to_file {
my $file = shift;
my @XML_data;
my $changed = "";
print "Processing $file. ";
if (open (FILE, $file)) {
my $cas_found = "";
my $chemical = "";
while (<FILE>) {
if (/CHEMICAL/i) {
push (@XML_data, $_);
$chemical = "True";
} elsif (/CAS-NUMBER/i && /\>(.+?)\</) {
$cas_found = "True";
my $CAS_line = $_;
my $UN_line = $CAS_line;
my $CAS_number = $1;
push (@XML_data, $CAS_line);
my $next_line = <FILE> if (!eof(FILE));
if ($next_line =~ /UN-NUMBER/) {
push (@XML_data, $next_line);
$nr_UN_numbers++;
} else {
my $UN_number;
if ($UN_number = &get_UN_number($CAS_number)) {
$UN_line =~ s/CAS-NUMBER/UN-NUMBER/;
$UN_line =~ s/\>.*?\</\>$UN_number\</;
push (@XML_data, $UN_line);
$changed = "True";
}
push (@XML_data, $next_line);
}
} elsif (/WOC-NUMBER/ && /\>(.+?)\</) {
push (@XML_data, $_);
$cas_found = "True";
} else {
push (@XML_data, $_);
}
}
print "No CAS number found!! " if ($chemical && !$cas_found);
close (FILE) || warn "$file: $!$/";
} else {
warn "$file: $!$/";
}
if ($changed) {
if (open (FILE, ">$file")) {
foreach my $line (@XML_data) {
print FILE $line;
}
close (FILE) || die $!;
print "UN number added...$/";
$nr_added_UN_numbers++;
} else {
warn "$file: $!$/";
print "Unable to change...$/";
$nr_unable++;
}
} else {
print "No changes...$/";
}
}
sub get_UN_number {
my $CAS_number = shift;
my $UN_number = "";
foreach my $line (@CAS_UN_table) {
if ($line =~ /^$CAS_number\s*\<\=\>\s*(.*)/) {
$UN_number = $1;
}
}
return $UN_number;
}
--- NEW FILE: check_cas_nummers.pl ---
#! /usr/bin/perl -w
use strict;
use diagnostics;
# Check argument line
if (@ARGV == 0) {
print "Usage: $0 <XML-files>$/";
exit;
}
# Global variables
my $nr_ignored = 0;
my $nr_inserted = 0;
my $nr_unable = 0;
# Loop over all files
foreach my $arg (@ARGV) {
&check_CAS_and_WOC_numbers_in_file ($arg);
}
# Print statistics
print " ------------------------$/";
if ($nr_ignored + $nr_inserted == 0) {
print "All files passed check.$/";
} else {
print "Errors ignored : $nr_ignored$/" if ($nr_ignored);
print "Unable to changes : $nr_unable$/" if ($nr_unable);
print "Check digits inserted: $nr_inserted$/" if ($nr_inserted);
}
sub check_CAS_and_WOC_numbers_in_file {
my $file = shift;
my @XML_data;
my $changed = "";
print "Processing $file, ";
if (open (FILE, $file)) {
while (<FILE>) {
if (/(CAS|WOC)-NUMBER.*?\>(.+?)\</) {
my $type = $1;
my $line = $_;
my $number = $2;
my $check = &check_digit ($number);
if (!$check) {
print "$/ Wrong check digit or typing error in $type number in $file! Ignoring error...$/ ";
$nr_ignored++;
#chop ($number);
#chop ($number);
#$number = &check_digit ($number);
#$line =~ s/\>.*?\</\>$number\</;
#print "Corrected...$/";
#$nr_corrected++;
#$changed = "True";
} elsif ($check =~ /Error/i) {
print "$/ $check in $file! Ignoring error...$/ ";
$nr_ignored++;
} elsif ($check !~ /Correct/i) {
print "$/ Missing check digit of $type number in $file!$/ ";
$number = &check_digit ($number);
$line =~ s/\>.*?\</\>$number\</;
$changed = "True";
}
push (@XML_data, $line);
} else {
push (@XML_data, $_);
}
}
close (FILE) || warn "$file: $!$/";
} else {
warn "$file: $!$/";
}
if ($changed) {
if (open (FILE, ">$file")) {
foreach my $line (@XML_data) {
print FILE $line;
}
close (FILE) || die "$file: $!$/";
print "Check digit inserted...$/";
$nr_inserted++;
} else {
warn "$file: $!$/";
print "Unable to change...$/";
$nr_unable++;
}
} else {
print "No changes...$/";
}
}
--- NEW FILE: check_cas_nummers.pl.in ---
#! @PATHTOPERL@ -w
use strict;
use diagnostics;
# Check argument line
if (@ARGV == 0) {
print "Usage: $0 <XML-files>$/";
exit;
}
# Global variables
my $nr_ignored = 0;
my $nr_inserted = 0;
my $nr_unable = 0;
# Loop over all files
foreach my $arg (@ARGV) {
&check_CAS_and_WOC_numbers_in_file ($arg);
}
# Print statistics
print " ------------------------$/";
if ($nr_ignored + $nr_inserted == 0) {
print "All files passed check.$/";
} else {
print "Errors ignored : $nr_ignored$/" if ($nr_ignored);
print "Unable to changes : $nr_unable$/" if ($nr_unable);
print "Check digits inserted: $nr_inserted$/" if ($nr_inserted);
}
sub check_CAS_and_WOC_numbers_in_file {
my $file = shift;
my @XML_data;
my $changed = "";
print "Processing $file, ";
if (open (FILE, $file)) {
while (<FILE>) {
if (/(CAS|WOC)-NUMBER.*?\>(.+?)\</) {
my $type = $1;
my $line = $_;
my $number = $2;
my $check = &check_digit ($number);
if (!$check) {
print "$/ Wrong check digit or typing error in $type number in $file! Ignoring error...$/ ";
$nr_ignored++;
#chop ($number);
#chop ($number);
#$number = &check_digit ($number);
#$line =~ s/\>.*?\</\>$number\</;
#print "Corrected...$/";
#$nr_corrected++;
#$changed = "True";
} elsif ($check =~ /Error/i) {
print "$/ $check in $file! Ignoring error...$/ ";
$nr_ignored++;
} elsif ($check !~ /Correct/i) {
print "$/ Missing check digit of $type number in $file!$/ ";
$number = &check_digit ($number);
$line =~ s/\>.*?\</\>$number\</;
$changed = "True";
}
push (@XML_data, $line);
} else {
push (@XML_data, $_);
}
}
close (FILE) || warn "$file: $!$/";
} else {
warn "$file: $!$/";
}
if ($changed) {
if (open (FILE, ">$file")) {
foreach my $line (@XML_data) {
print FILE $line;
}
close (FILE) || die "$file: $!$/";
print "Check digit inserted...$/";
$nr_inserted++;
} else {
warn "$file: $!$/";
print "Unable to change...$/";
$nr_unable++;
}
} else {
print "No changes...$/";
}
}
--- NEW FILE: check_doublenumbers.pl ---
#! /usr/bin/perl -w
use strict;
use diagnostics;
# Check command line options
if (@ARGV == 0) {
print "Usage: $0 <xml-files>$/";
exit;
}
# Global variables
my @NUMBER_count = ();
my @NUMBER_doublecount = ();
# Loop over all files
foreach my $arg (@ARGV) {
&check_numbers_in_file ($arg);
}
# Print statistics
print " ------------------$/";
if ($nr_SMILES_numbers) {
print "SMILES strings found : $nr_SMILES_numbers$/";
} else {
print "No SMILES strings found.$/";
}
if ($nr_unable) {
print "Unable to changes: $nr_unable$/" if ($nr_unable);
}
if ($nr_double) {
print "Double SMILES strings: $nr_double$/" if ($nr_double);
}
if ($nr_added_SMILES_numbers) {
print "SMILES strings added : $nr_added_SMILES_numbers$/";
print "SMILES strings total : ",$nr_SMILES_numbers+$nr_added_SMILES_numbers,$/;
} else {
print "No SMILES strings added.$/";
}
sub check_numbers_in_file {
my $file = shift;
my @XML_data;
my $changed = "";
my $doubles_found = "";
my $mismatching_doubles_found = "";
my @NUMBERS = ();
print "Processing $file. ";
if (open (FILE, $file)) {
my $cas_found = "";
my $chemical = "";
while (<FILE>) {
if (/CHEMICAL/i) {
push (@XML_data, $_);
$chemical = "True";
} elsif (/INDEX CLASS/i && /\>(.+?)\</) {
$number_found = "True";
my $CAS_line = $_;
my $SMILES_line = $CAS_line;
my $CAS_number = $1;
push (@XML_data, $CAS_line);
my $next_line = <FILE> if (!eof(FILE));
if ($next_line =~ /SMILES/) {
push (@XML_data, $next_line);
$smiles_found = "True";
} else {
my $SMILES_number;
if ($SMILES_number = &get_SMILES_number($CAS_number)) {
$SMILES_line =~ s/CAS-NUMBER/SMILES/;
$SMILES_line =~ s/\>.*?\</\>$SMILES_number\</;
push (@XML_data, $SMILES_line);
$changed = "True";
}
push (@XML_data, $next_line);
}
}
} elsif (/SMILES/ && /\>(.+?)\</) {
if (!$smiles_found) {
push (@XML_data, $_);
$smiles_found = "True";
} else {
$double_smiles = "True";
}
} elsif (/WOC-NUMBER/ && /\>(.+?)\</) {
push (@XML_data, $_);
$cas_found = "True";
} else {
push (@XML_data, $_);
}
}
print "No CAS number found!! " if ($chemical && !$cas_found);
close (FILE) || warn "$file: $!$/";
} else {
warn "$file: $!$/";
}
if ($double_smiles) {
print "Double SMILES string found...$/";
$nr_SMILES_numbers++;
$nr_double++;
} elsif ($changed && $smiles_found) {
print "No changes needed...$/";
$nr_SMILES_numbers++;
} elsif ($changed) {
if (open (FILE, ">$file")) {
foreach my $line (@XML_data) {
print FILE $line;
}
close (FILE) || die $!;
print "SMILES number added...$/";
$nr_added_SMILES_numbers++;
} else {
warn "$file: $!$/";
print "Unable to change...$/";
$nr_unable++;
}
} else {
print "No changes...$/";
}
}
sub get_SMILES_number {
my $CAS_number = shift;
my $SMILES_number = "";
foreach my $line (@CAS_SMILES_table) {
if ($line =~ /^$CAS_number\s*(.*)/) {
$SMILES_number = $1;
}
}
return $SMILES_number;
}
--- NEW FILE: check_doublenumbers.pl.in ---
#! @PATHTOPERL@ -w
use strict;
use diagnostics;
# Check command line options
if (@ARGV == 0) {
print "Usage: $0 <xml-files>$/";
exit;
}
# Global variables
my @NUMBER_count = ();
my @NUMBER_doublecount = ();
# Loop over all files
foreach my $arg (@ARGV) {
&check_numbers_in_file ($arg);
}
# Print statistics
print " ------------------$/";
if ($nr_SMILES_numbers) {
print "SMILES strings found : $nr_SMILES_numbers$/";
} else {
print "No SMILES strings found.$/";
}
if ($nr_unable) {
print "Unable to changes: $nr_unable$/" if ($nr_unable);
}
if ($nr_double) {
print "Double SMILES strings: $nr_double$/" if ($nr_double);
}
if ($nr_added_SMILES_numbers) {
print "SMILES strings added : $nr_added_SMILES_numbers$/";
print "SMILES strings total : ",$nr_SMILES_numbers+$nr_added_SMILES_numbers,$/;
} else {
print "No SMILES strings added.$/";
}
sub check_numbers_in_file {
my $file = shift;
my @XML_data;
my $changed = "";
my $doubles_found = "";
my $mismatching_doubles_found = "";
my @NUMBERS = ();
print "Processing $file. ";
if (open (FILE, $file)) {
my $cas_found = "";
my $chemical = "";
while (<FILE>) {
if (/CHEMICAL/i) {
push (@XML_data, $_);
$chemical = "True";
} elsif (/INDEX CLASS/i && /\>(.+?)\</) {
$number_found = "True";
my $CAS_line = $_;
my $SMILES_line = $CAS_line;
my $CAS_number = $1;
push (@XML_data, $CAS_line);
my $next_line = <FILE> if (!eof(FILE));
if ($next_line =~ /SMILES/) {
push (@XML_data, $next_line);
$smiles_found = "True";
} else {
my $SMILES_number;
if ($SMILES_number = &get_SMILES_number($CAS_number)) {
$SMILES_line =~ s/CAS-NUMBER/SMILES/;
$SMILES_line =~ s/\>.*?\</\>$SMILES_number\</;
push (@XML_data, $SMILES_line);
$changed = "True";
}
push (@XML_data, $next_line);
}
}
} elsif (/SMILES/ && /\>(.+?)\</) {
if (!$smiles_found) {
push (@XML_data, $_);
$smiles_found = "True";
} else {
$double_smiles = "True";
}
} elsif (/WOC-NUMBER/ && /\>(.+?)\</) {
push (@XML_data, $_);
$cas_found = "True";
} else {
push (@XML_data, $_);
}
}
print "No CAS number found!! " if ($chemical && !$cas_found);
close (FILE) || warn "$file: $!$/";
} else {
warn "$file: $!$/";
}
if ($double_smiles) {
print "Double SMILES string found...$/";
$nr_SMILES_numbers++;
$nr_double++;
} elsif ($changed && $smiles_found) {
print "No changes needed...$/";
$nr_SMILES_numbers++;
} elsif ($changed) {
if (open (FILE, ">$file")) {
foreach my $line (@XML_data) {
print FILE $line;
}
close (FILE) || die $!;
print "SMILES number added...$/";
$nr_added_SMILES_numbers++;
} else {
warn "$file: $!$/";
print "Unable to change...$/";
$nr_unable++;
}
} else {
print "No changes...$/";
}
}
sub get_SMILES_number {
my $CAS_number = shift;
my $SMILES_number = "";
foreach my $line (@CAS_SMILES_table) {
if ($line =~ /^$CAS_number\s*(.*)/) {
$SMILES_number = $1;
}
}
return $SMILES_number;
}
--- NEW FILE: zoek-UNs-zonder-wml.pl ---
#! /usr/bin/perl -w
use strict;
use diagnostics;
my $debug = "";
my $wmldir = "../../web-docs/data/wml";
my $existsfile = "tmp-nowml-exists";
# Check command line options
#if (@ARGV == 0) {
# print "Usage: $0 <xml-files>$/";
# exit;
#}
# Global variables
my $CAS_UN_table_file = "CAS-UN.table";
my @CAS_UN_table;
my $nr_UN_numbers = 0;
my $nr_added_UN_numbers = 0;
my $nr_unable = 0;
# Read CAS<=>UN table
if (open (FILE, $CAS_UN_table_file)) {
<FILE>; # skip first line (comment) #
while (<FILE>) {
chomp;
push (@CAS_UN_table, $_);
}
close (FILE) || die "$CAS_UN_table_file: $!$/";
} else {
die "$CAS_UN_table_file: $!$/";
}
&create_existskeyfile ("NUMBER");
foreach my $line (@CAS_UN_table) {
my ($cas) = split (/\s*<=>\s*/, $line);
my $nummatch = &exists (("NUMBER" => $cas));
if ($nummatch) {
print "NUMBER found:\t$nummatch$/" if ($debug);
} else {
print "Dangling CAS/UN pair: $line$/";
}
}
&remove_existskeyfile ("NUMBER");
sub create_existskeyfile {
my $key = shift;
`grep -i $key $wmldir/*.xml > $existsfile$key`;
}
sub remove_existskeyfile {
my $key = shift;
`rm $existsfile$key`;
}
sub exists {
my %filters = @_;
foreach my $key (keys %filters) {
my $out = `grep -i $filters{$key} $existsfile$key`;
if ($out) {
chomp ($out);
$out =~ s/$wmldir\///ig;
$out =~ s/:\s*/:\t/ig;
return "$out";
}
}
return "";
}
--- NEW FILE: zoek-UNs-zonder-wml.pl.in ---
#! @PATHTOPERL@ -w
use strict;
use diagnostics;
my $debug = "";
my $wmldir = "../../web-docs/data/wml";
my $existsfile = "tmp-nowml-exists";
# Check command line options
#if (@ARGV == 0) {
# print "Usage: $0 <xml-files>$/";
# exit;
#}
# Global variables
my $CAS_UN_table_file = "CAS-UN.table";
my @CAS_UN_table;
my $nr_UN_numbers = 0;
my $nr_added_UN_numbers = 0;
my $nr_unable = 0;
# Read CAS<=>UN table
if (open (FILE, $CAS_UN_table_file)) {
<FILE>; # skip first line (comment) #
while (<FILE>) {
chomp;
push (@CAS_UN_table, $_);
}
close (FILE) || die "$CAS_UN_table_file: $!$/";
} else {
die "$CAS_UN_table_file: $!$/";
}
&create_existskeyfile ("NUMBER");
foreach my $line (@CAS_UN_table) {
my ($cas) = split (/\s*<=>\s*/, $line);
my $nummatch = &exists (("NUMBER" => $cas));
if ($nummatch) {
print "NUMBER found:\t$nummatch$/" if ($debug);
} else {
print "Dangling CAS/UN pair: $line$/";
}
}
&remove_existskeyfile ("NUMBER");
sub create_existskeyfile {
my $key = shift;
`grep -i $key $wmldir/*.xml > $existsfile$key`;
}
sub remove_existskeyfile {
my $key = shift;
`rm $existsfile$key`;
}
sub exists {
my %filters = @_;
foreach my $key (keys %filters) {
my $out = `grep -i $filters{$key} $existsfile$key`;
if ($out) {
chomp ($out);
$out =~ s/$wmldir\///ig;
$out =~ s/:\s*/:\t/ig;
return "$out";
}
}
return "";
}
|
