def generate_stroke():
abc = abcNLPChar('C')
db = abcSql("abcNLP.db")
db.recreate_stroke_order()
abc.getStrokeOrderDict(db=db)
db.commit()
db.close()
def generate_stroke():
abc = abcNLPChar("C")
db = abcSql("abcNLP.db")
db.recreate_stroke_order()
abc.getStrokeOrderDict(db=db)
db.commit()
db.close()
def generate_variant():
abc = abcNLPChar("C")
db = abcSql("abcNLP.db")
db.recreate_char_variant()
dic = db.fetch_stroke_orders()
stk = abc.getStrokeOrderSimilar(dic, db=db)
db.commit()
db.close
def generate_variant():
abc = abcNLPChar('C')
db = abcSql("abcNLP.db")
db.recreate_char_variant()
dic = db.fetch_stroke_orders()
stk = abc.getStrokeOrderSimilar(dic, db=db)
db.commit()
db.close
def generate_tongyin_variant(maxnum=500000):
abc = abcNLPChar('C')
db = abcSql("abcNLP.db")
db.recreate_tongyin_variant()
count = 0
for src in abc.getDomainCharacterIterator():
if (count >= maxnum):
print "Max number %d reached!" % count
break
try:
sc = abc.getStrokeCount(src)
py = abc.getReadingForCharacter(src, 'Pinyin')
# sorry, only process one Pinyin of each of char
tongyins = abc.getCharactersForReading(py[0], 'Pinyin')
except:
continue
# default, the Pinyin is a variant (score set to 0)
db.insert_tongyin_char_variant(src, py[0], 0)
# print "%s --> %s (0)" % (src, py[0])
distances = [ (500, ""), (500, ""), (500, "")]
for ty in tongyins:
if ty == src:
continue
try:
# distance = abs(abc.getStrokeCount(ty) - sc) ; use difference
distance = abc.getStrokeCount(ty) # the simple character is better
except:
continue
distances.pop()
distances.append((distance, ty))
insertionSort(distances)
for distance in distances:
if distance[0] == 500:
break
#print "%s --> %s (%d)" % (src, distance[1], distance[0])
db.insert_tongyin_char_variant(src, distance[1], distance[0])
count = count + 1
db.commit()
db.close()
def generate_default_variant(maxnum=500000):
abc = abcNLPChar('C')
db = abcSql("abcNLP.db")
db.recreate_default_variant()
count = 0
for src in abc.getDomainCharacterIterator():
if (count >= maxnum):
print "Max number %d reached!" % count
break
try:
exts = get_abc_extend_char(abc, src)
except:
continue
for ext in exts:
if ext == src:
continue
#print "%s --> %s " % (src, ext)
db.insert_default_char_variant(src, ext, 0)
count = count + 1
db.commit()
db.close()
def clear_db():
db = abcSql("abcNLP.db")
db.recreate_stroke_order()
db.recreate_char_variant()
db.close
def clear_db():
db = abcSql("abcNLP.db")
db.recreate_stroke_order()
db.recreate_char_variant()
db.close
def remove_other_variant():
abc = abcNLPChar('C')
db = abcSql("abcNLP.db")
db.remove_old_tables()
db.close()
def generate_bigone_variant():
abc = abcNLPChar('C')
db = abcSql("abcNLP.db")
db.recreate_allinone_variant()
db.merge_to_one_variant()
db.close()
def generate_decomp_variant(maxnum=500000, depth=1):
abc = abcNLPChar('C')
db = abcSql("abcNLP.db")
if depth == 0:
db.recreate_decomp_variant()
elif depth > 0:
db.recreate_decompext_variant()
count = 0
for src in abc.getDomainCharacterIterator():
# [u'待', u'法', u'⾽']:
if (count >= maxnum):
print "Max number %d reached!" % count
break
try:
decomps = abc.getDecompositionEntries(src)
except:
continue
level = depth
decomp = []
if decomps:
decomp = decomps[0]
if decomp:
idc = decomp[0]
if not abc.isBinaryIDSOperator(idc) and not abc.isTrinaryIDSOperator(idc):
raise exception.NoInformationError("IDC of char %s is error: %s" % (char, idc))
if idc == u'⿰'and type(decomp[1]) == type(()) and type(decomp[2]) == type(()):
a = decomp[1][0];
b = decomp[2][0];
if (a == u'？' or b == u'？'):
continue
ao, bo = a, b
if level > 1:
a = get_abc_extend_char(abc, a).pop()
if ao != a:
level = level - 1
if level > 1:
b = get_abc_extend_char(abc, b).pop()
if bo != b:
level = level - 1
# for the reason of level, b may hasn't been changed
if ao == bo and ao != a:
b = a
chg = not (ao == a and bo == b)
#print "%s --> <%s%s> %d" % (src, a, b, chg)
if depth == 0:
db.insert_decomp_char_variant(src, a+b, 0)
elif depth > 0:
db.insert_decompext_char_variant(src, a+b, 0)
count = count + 1
elif idc == u'⿲' and type(decomp[1]) == type(()) and type(decomp[2]) == type(()) \
and type(decomp[3]) == type(()):
a = decomp[1][0];
b = decomp[2][0];
c = decomp[3][0];
if (a == u'？' or b == u'？' or c == u'？'):
continue
ao, bo, co = a, b, c
if level > 1:
a = get_abc_extend_char(abc, a).pop()
if ao != a:
level = level - 1
if level > 1:
b = get_abc_extend_char(abc, b).pop()
if bo != b:
level = level - 1
if level > 1:
b = get_abc_extend_char(abc, b).pop()
if bo != b:
level = level - 1
# for the reason of level, b, c may hasn't been changed
if ao == bo and ao != a:
b = a
if ao == co and ao != a:
c = a
if bo == co and bo != b:
c = b
chg = not (ao == a and bo == b and co == c)
#print "%s --> <%s%s%s> %d" % (src, a, b, c, chg)
if depth == 0:
db.insert_decomp_char_variant(src, a+b+c, 0)
elif depth > 0:
db.insert_decompext_char_variant(src, a+b+c, 0)
count = count + 1
db.commit()
db.close()
