Fix NST lexicon accents

_VOWELS = [
    "}:",
    "2:",
    "9",
    "a",
    "a*U",
    "A:",
    "e",
    "E",
    "E*U",
    "e:",
    "E:",
    "I",
    "i:",
    "O",
    "o:",
    "U",
    "u:",
    "u0",
    "Y",
    "y:"
]

_SAMPLE = """
AFTENPOSTEN	"a f t e n %p O s t e n
AFTONBLADET	"a f t O n %b l A: d e t
AFTONBLADETS	"a f t O n %b l A: d e t s
AFTONBRISVÄGEN	"a f t O n b r i: s %v E: g e n
AFTONGATAN	"a f t O N %g A: t a n
AFTONVÄGEN	"a f t O n %v E: g e n
AFZELIIVÄGEN	a f "s e: l I %v E: g e n
AFZELIUS	a f "s e: l I u0 s
AGADIR	a g a "d i: r
AGAMEMNON	a g a "m E m n O n
AGARD	"A: g a d`
AGARDH	"A: g a d`
AGARDHSGATAN	"A: g a d` s` %g A: t a n
AGARDSSON	"A: g a d` s` O n
AGASSI	a "g a s I
AGASSIS	a "g a s I s
AGATA	a "g A: t a
AGATAS	a "g A: t a s
"""

def split_phone(inphone):
    _STRESSMARKS = ['""', '"', '%']
    outmark = ''
    outphone = inphone
    for sm in _STRESSMARKS:
        if inphone.startswith(sm):
            outmark = sm
            outphone = inphone.replace(sm, '')
    return (outmark, outphone)

out_words = []
for line in _SAMPLE.split('\n'):
    if line == '':
        continue
    phones_out = []
    parts = line.split('\t')
    assert len(parts) == 2
    current_mark = ''
    for phone in parts[1].split(' '):
        tmp_mark, actual_phone = split_phone(phone)
        if tmp_mark != '':
            current_mark = tmp_mark
        if actual_phone in _VOWELS and current_mark != '':
            phones_out.append(current_mark + actual_phone)
            current_mark = ''
        else:
            phones_out.append(actual_phone)
    new_phones = ' '.join(phones_out)
    out_words.append(f"{parts[0]}\t{new_phones}")

out_words

['AFTENPOSTEN\t"a f t e n p %O s t e n',
 'AFTONBLADET\t"a f t O n b l %A: d e t',
 'AFTONBLADETS\t"a f t O n b l %A: d e t s',
 'AFTONBRISVÄGEN\t"a f t O n b r i: s v %E: g e n',
 'AFTONGATAN\t"a f t O N g %A: t a n',
 'AFTONVÄGEN\t"a f t O n v %E: g e n',
 'AFZELIIVÄGEN\ta f s "e: l I v %E: g e n',
 'AFZELIUS\ta f s "e: l I u0 s',
 'AGADIR\ta g a d "i: r',
 'AGAMEMNON\ta g a m "E m n O n',
 'AGARD\t"A: g a d`',
 'AGARDH\t"A: g a d`',
 'AGARDHSGATAN\t"A: g a d` s` g %A: t a n',
 'AGARDSSON\t"A: g a d` s` O n',
 'AGASSI\ta g "a s I',
 'AGASSIS\ta g "a s I s',
 'AGATA\ta g "A: t a',
 'AGATAS\ta g "A: t a s']