Code Duplication    Length = 19-19 lines in 2 locations

abydos/stemmer.py 2 locations

@@ 1447-1465 (lines=19) @@
1444
1445
    # Step 1
1446
    _r1 = word[r1_start:]
1447
    if _r1[-7:] == 'erendes':
1448
        word = word[:-7]
1449
    elif _r1[-6:] in {'erende', 'hedens'}:
1450
        word = word[:-6]
1451
    elif _r1[-5:] in {'ethed', 'erede', 'heden', 'heder', 'endes', 'ernes',
1452
                      'erens', 'erets'}:
1453
        word = word[:-5]
1454
    elif _r1[-4:] in {'ered', 'ende', 'erne', 'eren', 'erer', 'heds', 'enes',
1455
                      'eres', 'eret'}:
1456
        word = word[:-4]
1457
    elif _r1[-3:] in {'hed', 'ene', 'ere', 'ens', 'ers', 'ets'}:
1458
        word = word[:-3]
1459
    elif _r1[-2:] in {'en', 'er', 'es', 'et'}:
1460
        word = word[:-2]
1461
    elif _r1[-1:] == 'e':
1462
        word = word[:-1]
1463
    elif _r1[-1:] == 's':
1464
        if len(word) > 1 and word[-2] in _s_endings:
1465
            word = word[:-1]
1466
1467
    # Step 2
1468
    if word[r1_start:][-2:] in {'gd', 'dt', 'gt', 'kt'}:
@@ 1381-1399 (lines=19) @@
1378
1379
    # Step 1
1380
    _r1 = word[r1_start:]
1381
    if _r1[-7:] == 'heterna':
1382
        word = word[:-7]
1383
    elif _r1[-6:] == 'hetens':
1384
        word = word[:-6]
1385
    elif _r1[-5:] in {'anden', 'heten', 'heter', 'arnas', 'ernas', 'ornas',
1386
                      'andes', 'arens', 'andet'}:
1387
        word = word[:-5]
1388
    elif _r1[-4:] in {'arna', 'erna', 'orna', 'ande', 'arne', 'aste', 'aren',
1389
                      'ades', 'erns'}:
1390
        word = word[:-4]
1391
    elif _r1[-3:] in {'ade', 'are', 'ern', 'ens', 'het', 'ast'}:
1392
        word = word[:-3]
1393
    elif _r1[-2:] in {'ad', 'en', 'ar', 'er', 'or', 'as', 'es', 'at'}:
1394
        word = word[:-2]
1395
    elif _r1[-1:] in {'a', 'e'}:
1396
        word = word[:-1]
1397
    elif _r1[-1:] == 's':
1398
        if len(word) > 1 and word[-2] in _s_endings:
1399
            word = word[:-1]
1400
1401
    # Step 2
1402
    if word[r1_start:][-2:] in {'dd', 'gd', 'nn', 'dt', 'gt', 'kt', 'tt'}: