Code Duplication    Length = 19-19 lines in 2 locations

abydos/stemmer.py 2 locations

@@ 1446-1464 (lines=19) @@
1443
1444
    # Step 1
1445
    _r1 = word[r1_start:]
1446
    if _r1[-7:] == 'erendes':
1447
        word = word[:-7]
1448
    elif _r1[-6:] in {'erende', 'hedens'}:
1449
        word = word[:-6]
1450
    elif _r1[-5:] in {'ethed', 'erede', 'heden', 'heder', 'endes', 'ernes',
1451
                      'erens', 'erets'}:
1452
        word = word[:-5]
1453
    elif _r1[-4:] in {'ered', 'ende', 'erne', 'eren', 'erer', 'heds', 'enes',
1454
                      'eres', 'eret'}:
1455
        word = word[:-4]
1456
    elif _r1[-3:] in {'hed', 'ene', 'ere', 'ens', 'ers', 'ets'}:
1457
        word = word[:-3]
1458
    elif _r1[-2:] in {'en', 'er', 'es', 'et'}:
1459
        word = word[:-2]
1460
    elif _r1[-1:] == 'e':
1461
        word = word[:-1]
1462
    elif _r1[-1:] == 's':
1463
        if len(word) > 1 and word[-2] in _s_endings:
1464
            word = word[:-1]
1465
1466
    # Step 2
1467
    if word[r1_start:][-2:] in {'gd', 'dt', 'gt', 'kt'}:
@@ 1380-1398 (lines=19) @@
1377
1378
    # Step 1
1379
    _r1 = word[r1_start:]
1380
    if _r1[-7:] == 'heterna':
1381
        word = word[:-7]
1382
    elif _r1[-6:] == 'hetens':
1383
        word = word[:-6]
1384
    elif _r1[-5:] in {'anden', 'heten', 'heter', 'arnas', 'ernas', 'ornas',
1385
                      'andes', 'arens', 'andet'}:
1386
        word = word[:-5]
1387
    elif _r1[-4:] in {'arna', 'erna', 'orna', 'ande', 'arne', 'aste', 'aren',
1388
                      'ades', 'erns'}:
1389
        word = word[:-4]
1390
    elif _r1[-3:] in {'ade', 'are', 'ern', 'ens', 'het', 'ast'}:
1391
        word = word[:-3]
1392
    elif _r1[-2:] in {'ad', 'en', 'ar', 'er', 'or', 'as', 'es', 'at'}:
1393
        word = word[:-2]
1394
    elif _r1[-1:] in {'a', 'e'}:
1395
        word = word[:-1]
1396
    elif _r1[-1:] == 's':
1397
        if len(word) > 1 and word[-2] in _s_endings:
1398
            word = word[:-1]
1399
1400
    # Step 2
1401
    if word[r1_start:][-2:] in {'dd', 'gd', 'nn', 'dt', 'gt', 'kt', 'tt'}: