def test(word):
assert_equal(split_ascii(word), split_word)
assert_equal(
[
convert_ascii_reverse(letter) for letter in
split_unicode(convert_ascii(word))
],
split_word
)
def test(unicode_word):
assert_equal(convert_ascii_reverse(unicode_word), ascii_word)
def test(word):
assert_equal(split_ascii(word), split_word)
assert_equal([
convert_ascii_reverse(letter)
for letter in split_unicode(convert_ascii(word))
], split_word)
print(text)
if len(roots) > 0:
add_definitions(text)
return sorted(definitions)
if __name__ == "__main__":
filename = os.path.abspath(os.path.join(os.path.basename(__file__), "../source_material/lexicon.htm"))
html = open(filename).read()
definitions = extract_definitions(html)
check_correctness = False
if check_correctness:
roots = extract_roots(html)
print("found {} roots".format(len(roots)))
print("found {} definitions".format(len(definitions)))
defined_roots = set(r for (r, d) in definitions)
assert len(defined_roots) == len(definitions)
undefined_roots = set(r for r in roots if r not in defined_roots)
print("defined roots", len(defined_roots))
print("undefined roots", len(undefined_roots))
for r in sorted(undefined_roots):
print(r)
print(set(defined_roots) - set(roots))
for root, definition in definitions:
root = convert_ascii_reverse(root)
print("{}|{}".format(root, definition))
# This should be run from the root, i.e:
# python extract/convert_tables_yaml.py
from ithkuil.grammar import *
from ithkuil.phonology import convert_ascii_reverse
keys = list(itertools.product(*ca_order))
with open('data/ca.yaml', 'w') as f:
for key in keys:
affixes = ca_table[frozenset(key)]
affixes = [convert_ascii_reverse(a) for a in affixes]
affix_lines = [
' - {}'.format(a)
for a in affixes
]
line = '{}:\n{}\n'.format(
'/'.join(key),
'\n'.join(affix_lines),
)
f.write(line)
def test(unicode_word):
assert_equal(convert_ascii_reverse(unicode_word), ascii_word)
add_definitions(text)
return sorted(definitions)
if __name__ == '__main__':
filename = os.path.abspath(
os.path.join(os.path.basename(__file__),
'../source_material/lexicon.htm'))
html = open(filename).read()
definitions = extract_definitions(html)
check_correctness = False
if check_correctness:
roots = extract_roots(html)
print('found {} roots'.format(len(roots)))
print('found {} definitions'.format(len(definitions)))
defined_roots = set(r for (r, d) in definitions)
assert len(defined_roots) == len(definitions)
undefined_roots = set(r for r in roots if r not in defined_roots)
print('defined roots', len(defined_roots))
print('undefined roots', len(undefined_roots))
for r in sorted(undefined_roots):
print(r)
print(set(defined_roots) - set(roots))
for root, definition in definitions:
root = convert_ascii_reverse(root)
print('{}|{}'.format(root, definition))
