fix: don't convert - to 减 in non-math context (#347)

* fix: don't convert - to 减 in non-math context (#294)

Split math operators into ambiguous (-, +, =) and unambiguous (×, ÷,
≥, ≤, etc.). Ambiguous operators now require number context on both
sides. Unambiguous operators can still match independently.

Before: wide-awake => wide减awake
After: wide-awake => wide-awake

* ci: only test macOS/Windows on schedule and manual dispatch

* ci: run macOS/Windows only on manual dispatch, Linux on PR

Author: pengzhendong

.github/workflows/unittest.yml

@@ -15,7 +15,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        os: [ubuntu-latest, macos-latest, windows-latest]
+        os: ${{ github.event_name == 'pull_request' && fromJSON('["ubuntu-latest"]') || fromJSON('["ubuntu-latest", "macos-latest", "windows-latest"]') }}
         python-version: ['3.9']
     steps:
       - uses: actions/checkout@v4

tn/chinese/data/math/operator_unambiguous.tsv

@@ -0,0 +1,6 @@
+×	乘
+÷	除
+≥	大于等于
+≤	小于等于
+>=	大于等于
+<=	小于等于

tn/chinese/rules/math.py

@@ -30,10 +30,11 @@ def __init__(self, cardinal=None):
 
     def build_tagger(self):
         operator = string_file(get_abs_path("chinese/data/math/operator.tsv"))
+        unambiguous_operator = string_file(get_abs_path("chinese/data/math/operator_unambiguous.tsv"))
         symbols = cross("~", "到") | cross(":", "比") | cross("<", "小于") | cross(">", "大于")
 
         number = self.cardinal.number
         tagger = number + (delete(" ").ques + (operator | symbols) + delete(" ").ques + number).star
-        tagger |= operator
+        tagger |= unambiguous_operator
         tagger = insert('value: "') + tagger + insert('"')
         self.tagger = self.add_tokens(tagger)