From 8ae5394bbaa383d630912e79d0822aa7c8ac7dca Mon Sep 17 00:00:00 2001
From: Yi Ding <yi.s.ding@gmail.com>
Date: Thu, 11 Jun 2026 22:46:01 -0700
Subject: [PATCH 1/2] fix(spacy): preserve positional language argument

---
 sentencesplit/spacy_component.py | 10 +++++++++-
 tests/test_spacy_component.py    | 14 ++++++++++++++
 2 files changed, 23 insertions(+), 1 deletion(-)

diff --git a/sentencesplit/spacy_component.py b/sentencesplit/spacy_component.py
index 24f759b..6934ad5 100644
--- a/sentencesplit/spacy_component.py
+++ b/sentencesplit/spacy_component.py
@@ -3,11 +3,19 @@
 
 import re
 
+from sentencesplit.languages import LANGUAGE_CODES
+
+_DEFAULT_COMPONENT_NAME = "sentencesplit"
+
 
 class SentenceSplitFactory:
     """sentencesplit as a spacy component through entrypoints"""
 
-    def __init__(self, nlp, name: str = "sentencesplit", language: str = "en") -> None:
+    def __init__(self, nlp, name: str = _DEFAULT_COMPONENT_NAME, language: str = "en") -> None:
+        if language == "en" and name != _DEFAULT_COMPONENT_NAME and name in LANGUAGE_CODES:
+            language = name
+            name = _DEFAULT_COMPONENT_NAME
+
         self.nlp = nlp
         self.name = name
         # Deferred import avoids circular dependency with sentencesplit.__init__
diff --git a/tests/test_spacy_component.py b/tests/test_spacy_component.py
index a7711e3..822da4b 100644
--- a/tests/test_spacy_component.py
+++ b/tests/test_spacy_component.py
@@ -22,6 +22,20 @@ def test_create_sentencesplit_is_importable_without_spacy_runtime():
     assert isinstance(factory, SentenceSplitFactory)
 
 
+def test_spacy_component_preserves_positional_language_argument():
+    factory = SentenceSplitFactory(None, "fr")
+
+    assert factory.name == "sentencesplit"
+    assert factory.seg.language == "fr"
+
+
+def test_create_sentencesplit_preserves_spacy_name_and_language():
+    factory = create_sentencesplit(None, "custom_sentences", "fr")
+
+    assert factory.name == "custom_sentences"
+    assert factory.seg.language == "fr"
+
+
 def test_spacy_component_reads_doc_text():
     doc = FakeDoc("Hello. World.", [0, 1, 7])
     factory = SentenceSplitFactory(None)

From 4414c460a2e0486a559cfb140fceb86431fed360 Mon Sep 17 00:00:00 2001
From: Yi Ding <yi.s.ding@gmail.com>
Date: Thu, 11 Jun 2026 23:20:34 -0700
Subject: [PATCH 2/2] fix(spacy): preserve explicit language config

---
 sentencesplit/spacy_component.py | 16 +++++++++++-----
 tests/test_spacy_component.py    | 14 ++++++++++++++
 2 files changed, 25 insertions(+), 5 deletions(-)

diff --git a/sentencesplit/spacy_component.py b/sentencesplit/spacy_component.py
index 228fc82..7369f83 100644
--- a/sentencesplit/spacy_component.py
+++ b/sentencesplit/spacy_component.py
@@ -6,22 +6,28 @@
 from sentencesplit.languages import LANGUAGE_CODES
 
 _DEFAULT_COMPONENT_NAME = "sentencesplit"
+_DEFAULT_LANGUAGE = object()
 
 
 class SentenceSplitFactory:
     """sentencesplit as a spacy component through entrypoints"""
 
-    def __init__(self, nlp, name: str = _DEFAULT_COMPONENT_NAME, language: str = "en") -> None:
-        if language == "en" and name != _DEFAULT_COMPONENT_NAME and name in LANGUAGE_CODES:
-            language = name
-            name = _DEFAULT_COMPONENT_NAME
+    def __init__(self, nlp, name: str = _DEFAULT_COMPONENT_NAME, language: str | object = _DEFAULT_LANGUAGE) -> None:
+        if language is _DEFAULT_LANGUAGE:
+            if name != _DEFAULT_COMPONENT_NAME and name in LANGUAGE_CODES:
+                language_code = name
+                name = _DEFAULT_COMPONENT_NAME
+            else:
+                language_code = "en"
+        else:
+            language_code = language
 
         self.nlp = nlp
         self.name = name
         # Deferred import avoids circular dependency with sentencesplit.__init__
         from sentencesplit import Segmenter
 
-        self.seg = Segmenter(language=language, clean=False)
+        self.seg = Segmenter(language=language_code, clean=False)
 
     def __call__(self, doc):
         sents_char_spans = self.seg.segment_spans(doc.text)
diff --git a/tests/test_spacy_component.py b/tests/test_spacy_component.py
index 822da4b..fc82dc7 100644
--- a/tests/test_spacy_component.py
+++ b/tests/test_spacy_component.py
@@ -36,6 +36,20 @@ def test_create_sentencesplit_preserves_spacy_name_and_language():
     assert factory.seg.language == "fr"
 
 
+def test_spacy_component_preserves_explicit_english_with_language_code_name():
+    factory = SentenceSplitFactory(None, "fr", "en")
+
+    assert factory.name == "fr"
+    assert factory.seg.language == "en"
+
+
+def test_create_sentencesplit_preserves_explicit_english_with_language_code_name():
+    factory = create_sentencesplit(None, "fr", "en")
+
+    assert factory.name == "fr"
+    assert factory.seg.language == "en"
+
+
 def test_spacy_component_reads_doc_text():
     doc = FakeDoc("Hello. World.", [0, 1, 7])
     factory = SentenceSplitFactory(None)