Code Duplication    Length = 19-19 lines in 2 locations

abydos/stemmer.py 2 locations

@@ 1444-1462 (lines=19) @@
1441
1442
    # Step 1
1443
    _r1 = word[r1_start:]
1444
    if _r1[-7:] == 'erendes':
1445
        word = word[:-7]
1446
    elif _r1[-6:] in {'erende', 'hedens'}:
1447
        word = word[:-6]
1448
    elif _r1[-5:] in {'ethed', 'erede', 'heden', 'heder', 'endes', 'ernes',
1449
                      'erens', 'erets'}:
1450
        word = word[:-5]
1451
    elif _r1[-4:] in {'ered', 'ende', 'erne', 'eren', 'erer', 'heds', 'enes',
1452
                      'eres', 'eret'}:
1453
        word = word[:-4]
1454
    elif _r1[-3:] in {'hed', 'ene', 'ere', 'ens', 'ers', 'ets'}:
1455
        word = word[:-3]
1456
    elif _r1[-2:] in {'en', 'er', 'es', 'et'}:
1457
        word = word[:-2]
1458
    elif _r1[-1:] == 'e':
1459
        word = word[:-1]
1460
    elif _r1[-1:] == 's':
1461
        if len(word) > 1 and word[-2] in _s_endings:
1462
            word = word[:-1]
1463
1464
    # Step 2
1465
    if word[r1_start:][-2:] in {'gd', 'dt', 'gt', 'kt'}:
@@ 1378-1396 (lines=19) @@
1375
1376
    # Step 1
1377
    _r1 = word[r1_start:]
1378
    if _r1[-7:] == 'heterna':
1379
        word = word[:-7]
1380
    elif _r1[-6:] == 'hetens':
1381
        word = word[:-6]
1382
    elif _r1[-5:] in {'anden', 'heten', 'heter', 'arnas', 'ernas', 'ornas',
1383
                      'andes', 'arens', 'andet'}:
1384
        word = word[:-5]
1385
    elif _r1[-4:] in {'arna', 'erna', 'orna', 'ande', 'arne', 'aste', 'aren',
1386
                      'ades', 'erns'}:
1387
        word = word[:-4]
1388
    elif _r1[-3:] in {'ade', 'are', 'ern', 'ens', 'het', 'ast'}:
1389
        word = word[:-3]
1390
    elif _r1[-2:] in {'ad', 'en', 'ar', 'er', 'or', 'as', 'es', 'at'}:
1391
        word = word[:-2]
1392
    elif _r1[-1:] in {'a', 'e'}:
1393
        word = word[:-1]
1394
    elif _r1[-1:] == 's':
1395
        if len(word) > 1 and word[-2] in _s_endings:
1396
            word = word[:-1]
1397
1398
    # Step 2
1399
    if word[r1_start:][-2:] in {'dd', 'gd', 'nn', 'dt', 'gt', 'kt', 'tt'}: