fix tests

adbar · Nov 6, 2024 · 38c85f1 · 38c85f1
1 parent fac72b5
commit 38c85f1
Show file tree

Hide file tree

Showing 3 changed files with 3 additions and 2 deletions.
diff --git a/simplemma/strategies/defaultrules/lv.py b/simplemma/strategies/defaultrules/lv.py
@@ -39,7 +39,7 @@
 
 def apply_lv(token: str) -> Optional[str]:
     "Apply pre-defined rules for Latvian."
-    if len(token) < 4:
+    if len(token) < 5:
         return None
 
     return apply_rules(token, DEFAULT_RULES)
diff --git a/tests/test_dictionary_pickler.py b/tests/test_dictionary_pickler.py
@@ -20,6 +20,7 @@ def test_logic() -> None:
     # log warning
     mydict = dictionary_pickler._read_dict(testfile, "zz", silent=False)
     assert len(mydict) == 3
+
     # different length
     mydict = dictionary_pickler._read_dict(testfile, "en", silent=True)
     assert len(mydict) == 5

diff --git a/training/dictionary_pickler.py b/training/dictionary_pickler.py
@@ -19,7 +19,7 @@
 
 LOGGER = logging.getLogger(__name__)
 
-INPUT_PUNCT = re.compile(r"[,:*/\+_]|.+-$|.+-\t")
+INPUT_PUNCT = re.compile(r"[,:*/\+_]|.+-$|.+-\t|^-.+")
 SAFE_LIMIT = {
     "cs",
     "da",