fix: strip spoken terms before validating and sending them

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Author: hive-bot

README.md

@@ -641,10 +641,11 @@ The `translate_document()`, `translate_document_from_filepath()` and
 A spoken-terms collection holds terms, such as product names or technical
 words, that improve speech transcription. Each collection has one term list
 per language; a term list is a list of strings. Before sending, the library
-checks each term like a glossary term (`validate_glossary_term()`): an empty
-term or one with a control character, such as a newline, raises `ValueError`.
-The API validates the rest (at most 300 characters per list, no duplicates, no
-leading or trailing whitespace). Methods that take a collection accept either its ID or a
+strips leading and trailing whitespace from each term and sends the stripped
+term, then checks it like a glossary term (`validate_glossary_term()`): an
+empty term or one with a control character, such as a newline, raises
+`ValueError`. The API validates the rest (at most 300 characters per list, no
+duplicates). Methods that take a collection accept either its ID or a
 `SpokenTerms` object.
 
 ```python

deepl/deepl_client.py

@@ -736,8 +736,9 @@ def create_spoken_terms(
             terms of one language.
         :return: SpokenTerms containing information about the created
             collection.
-        :raises ValueError: If a term is empty or contains a control
-            character, as validate_glossary_term() checks.
+        :raises ValueError: If a term, after stripping leading and trailing
+            whitespace, is empty or contains a control character, as
+            validate_glossary_term() checks.
         """
         request_data = {
             "name": name,
@@ -797,8 +798,9 @@ def update_spoken_terms(
             are merged into the term list of the same language.
         :return: SpokenTerms containing information about the updated
             collection.
-        :raises ValueError: If a term is empty or contains a control
-            character, as validate_glossary_term() checks.
+        :raises ValueError: If a term, after stripping leading and trailing
+            whitespace, is empty or contains a control character, as
+            validate_glossary_term() checks.
         """
         spoken_terms_id = self._spoken_terms_id(spoken_terms)
         request_data: Dict[str, Any] = {}
@@ -846,8 +848,9 @@ def replace_spoken_terms_list(
         :param spoken_terms: SpokenTerms or ID of the collection.
         :param entries: SpokenTermsEntries with the language and its terms.
         :return: SpokenTermsList containing information about the term list.
-        :raises ValueError: If a term is empty or contains a control
-            character, as validate_glossary_term() checks.
+        :raises ValueError: If a term, after stripping leading and trailing
+            whitespace, is empty or contains a control character, as
+            validate_glossary_term() checks.
         """
         spoken_terms_id = self._spoken_terms_id(spoken_terms)
         status, content, json = self._api_call(
@@ -892,9 +895,10 @@ def delete_spoken_terms_list(
     def _spoken_terms_json(entries: SpokenTermsEntries) -> dict:
         # Terms are sent newline-separated: a term with a newline would
         # silently become several terms.
-        for term in entries.entries:
+        terms = [term.strip() for term in entries.entries]
+        for term in terms:
             util.validate_glossary_term(term)
-        return entries.to_json()
+        return SpokenTermsEntries(entries.lang, terms).to_json()
 
     @staticmethod
     def _spoken_terms_id(spoken_terms: Union[str, SpokenTerms]) -> str:

tests/offline/test_spoken_terms.py

@@ -71,3 +71,22 @@ def test_url_encodes_the_id_in_the_path(mocked):
     mocked.delete(url("v3/spoken-terms/a%2Fb%3Fc"), status=204)
 
     make_deepl_client().delete_spoken_terms("a/b?c")
+
+
+def test_strips_terms_before_sending(mocked):
+    mocked.put(
+        url("v3/spoken-terms/st-1/term-lists"),
+        json={"lang": "en", "entry_count": 2},
+        match=json_request({"lang": "en", "entries": "DeepL\nwebhook"}),
+    )
+
+    make_deepl_client().replace_spoken_terms_list(
+        "st-1", deepl.SpokenTermsEntries("en", [" DeepL ", "webhook\t"])
+    )
+
+
+def test_rejects_a_whitespace_only_term(mocked):
+    with pytest.raises(ValueError, match="not a valid string"):
+        make_deepl_client().create_spoken_terms(
+            "terms", [deepl.SpokenTermsEntries("en", ["a", "  "])]
+        )