Code Duplication    Length = 19-19 lines in 2 locations

abydos/stemmer.py 2 locations

@@ 1581-1599 (lines=19) @@
1578
1579
    # Step 1
1580
    _r1 = word[r1_start:]
1581
    if _r1[-7:] == 'erendes':
1582
        word = word[:-7]
1583
    elif _r1[-6:] in {'erende', 'hedens'}:
1584
        word = word[:-6]
1585
    elif _r1[-5:] in {'ethed', 'erede', 'heden', 'heder', 'endes', 'ernes',
1586
                      'erens', 'erets'}:
1587
        word = word[:-5]
1588
    elif _r1[-4:] in {'ered', 'ende', 'erne', 'eren', 'erer', 'heds', 'enes',
1589
                      'eres', 'eret'}:
1590
        word = word[:-4]
1591
    elif _r1[-3:] in {'hed', 'ene', 'ere', 'ens', 'ers', 'ets'}:
1592
        word = word[:-3]
1593
    elif _r1[-2:] in {'en', 'er', 'es', 'et'}:
1594
        word = word[:-2]
1595
    elif _r1[-1:] == 'e':
1596
        word = word[:-1]
1597
    elif _r1[-1:] == 's':
1598
        if len(word) > 1 and word[-2] in _s_endings:
1599
            word = word[:-1]
1600
1601
    # Step 2
1602
    if word[r1_start:][-2:] in {'gd', 'dt', 'gt', 'kt'}:
@@ 1515-1533 (lines=19) @@
1512
1513
    # Step 1
1514
    _r1 = word[r1_start:]
1515
    if _r1[-7:] == 'heterna':
1516
        word = word[:-7]
1517
    elif _r1[-6:] == 'hetens':
1518
        word = word[:-6]
1519
    elif _r1[-5:] in {'anden', 'heten', 'heter', 'arnas', 'ernas', 'ornas',
1520
                      'andes', 'arens', 'andet'}:
1521
        word = word[:-5]
1522
    elif _r1[-4:] in {'arna', 'erna', 'orna', 'ande', 'arne', 'aste', 'aren',
1523
                      'ades', 'erns'}:
1524
        word = word[:-4]
1525
    elif _r1[-3:] in {'ade', 'are', 'ern', 'ens', 'het', 'ast'}:
1526
        word = word[:-3]
1527
    elif _r1[-2:] in {'ad', 'en', 'ar', 'er', 'or', 'as', 'es', 'at'}:
1528
        word = word[:-2]
1529
    elif _r1[-1:] in {'a', 'e'}:
1530
        word = word[:-1]
1531
    elif _r1[-1:] == 's':
1532
        if len(word) > 1 and word[-2] in _s_endings:
1533
            word = word[:-1]
1534
1535
    # Step 2
1536
    if word[r1_start:][-2:] in {'dd', 'gd', 'nn', 'dt', 'gt', 'kt', 'tt'}: