[owl] Fix docparse bad escape via plain string replacement (#910)

* plain string replacement

Author: haoshan98

services/api/src/owl/db/gen_executor.py

@@ -1383,7 +1383,10 @@ async def _load_files(self, message: ChatThreadEntry) -> ChatEntry:
                 continue
             for col_name, data in replacements.items():
                 _regex = r"(?<!\\)\${" + re.escape(col_name) + r"}"
-                c.text = re.sub(_regex, data, c.text)
+                # Use lambda to avoid re.sub interpreting backslashes in data as escape sequences
+                # 'lambda _, d=data: d' — '_' is the unused match object, d=data binds current
+                # loop value as default arg, and the lambda returns d as a literal string.
+                c.text = re.sub(_regex, lambda _, d=data: d, c.text)
             c.text = c.text.strip()
         # Re-assemble
         message = ChatEntry.user(content=contents)

services/api/tests/test_types.py

@@ -70,14 +70,7 @@ def test_id_string_bad(value: str):
 
 def test_string_normalisation():
     with pytest.raises(ValidationError, match="Text contains excessive combining marks"):
-        IdTest(
-            id=(
-                "H̵̛͕̞̦̰̜͍̰̥̟͆̏͂̌͑ͅä̷͔̟͓̬̯̟͍̭͉͈̮͙̣̯̬͚̞̭̍̀̾͠"
-                "m̴̡̧̛̝̯̹̗̹̤̲̺̟̥̈̏͊̔̑̍͆̌̀̚͝͝b̴̢̢̫̝̠̗̼̬̻̮̺̭͔̘͑̆̎̚"
-                "ư̵̧̡̥̙̭̿̈̀̒̐̊͒͑r̷̡̡̲̼̖͎̫̮̜͇̬͌͘g̷̹͍͎̬͕͓͕̐̃̈́̓̆̚͝"
-                "ẻ̵̡̼̬̥̹͇̭͔̯̉͛̈́̕r̸̮̖̻̮̣̗͚͖̝̂͌̾̓̀̿̔̀͋̈́͌̈́̋͜"
-            )
-        )
+        IdTest(id=("H̵̛͕̞̦̰̜͍̰̥̟͆̏͂̌͑ͅä̷͔̟͓̬̯̟͍̭͉͈̮͙̣̯̬͚̞̭̍̀̾͠m̴̡̧̛̝̯̹̗̹̤̲̺̟̥̈̏͊̔̑̍͆̌̀̚͝͝b̴̢̢̫̝̠̗̼̬̻̮̺̭͔̘͑̆̎̚ư̵̧̡̥̙̭̿̈̀̒̐̊͒͑r̷̡̡̲̼̖͎̫̮̜͇̬͌͘g̷̹͍͎̬͕͓͕̐̃̈́̓̆̚͝ẻ̵̡̼̬̥̹͇̭͔̯̉͛̈́̕r̸̮̖̻̮̣̗͚͖̝̂͌̾̓̀̿̔̀͋̈́͌̈́̋͜"))
 
     assert IdTest(id="مَرْحَبًا بِكُمْ").id == "مَرْحَبًا بِكُمْ"
     assert IdTest(id="สวัสดีครับ คามุย อิอิ").id == "สวัสดีครับ คามุย อิอิ"

services/api/tests/utils/test_lm.py

@@ -14,8 +14,7 @@ def test_inference_provider_should_prefer_vllm_cloud_over_owned_by() -> None:
     router = _make_router()
 
     assert (
-        router._inference_provider(CloudProvider.VLLM_CLOUD, "openai")
-        == CloudProvider.VLLM_CLOUD
+        router._inference_provider(CloudProvider.VLLM_CLOUD, "openai") == CloudProvider.VLLM_CLOUD
     )