1 parent c2ce6f6 commit 2d71b4cCopy full SHA for 2d71b4c
1 file changed
Source/Data/scripts/analyze_data.py
@@ -12,9 +12,15 @@
12
unigram_1char_count = 0
13
unigram_multichar_count = 0
14
emoji_count = 0
15
+macro_count = 0
16
17
for (r, v, s) in data:
18
# Skip emojis.
19
+
20
+ if v.startswith('MACRO@'):
21
+ macro_count += 1
22
+ continue
23
24
if s == -8:
25
emoji_count += 1
26
key = '-'.join(r)
@@ -95,7 +101,7 @@
95
101
print('%6d unigrams with one character' % unigram_1char_count)
96
102
print('%6d unigrams with multiple characters' % unigram_multichar_count)
97
103
print("%6d emojis" % emoji_count)
98
-
104
+print("%6d macros" % macro_count)
99
105
print(separator)
100
106
print('summary for unigrams with scores lower than their competing characters:')
107
print('%6d unigrams that are indifferent since the characters are the same' %
0 commit comments