Skip to content

Commit 0fb5b04

Browse files
committed
fix: strip commas in numbers for Chinese TN Measure (#284)
Use cdrewrite to remove commas between digits before matching numbers in the Measure rule, so western-formatted numbers like 1,000,000米 correctly normalize to 一百万米. Note: numbers exceeding 万级 (>99,999,999) still need 亿级 support in Cardinal (#223), which is tracked separately.
1 parent 53156d2 commit 0fb5b04

1 file changed

Lines changed: 2 additions & 0 deletions

File tree

tn/chinese/rules/measure.py

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -36,6 +36,8 @@ def build_tagger(self):
3636
to = cross("-", "到") | cross("~", "到") | accep("到")
3737

3838
number = self.cardinal.number
39+
strip_comma = self.build_rule(delete(","), self.DIGIT, self.DIGIT)
40+
number = strip_comma @ number
3941
number @= self.build_rule(cross("二", "两"), "[BOS]", "[EOS]")
4042
# 1-11个,1个-11个
4143
prefix = number + (rmspace + units).ques + to

0 commit comments

Comments
 (0)