Refactor _AzureSearchSettings initialization to ensure proper inheritance from BaseSettings and DatasourcePayloadConstructor; update regex pattern in PdfTextSplitter for improved URL matching

Author: Kingshuk-Microsoft

scripts/data_utils.py

@@ -161,7 +161,7 @@ def find_urls(string):
                 r"(?i)\b("
                 r"(?:https?://|www\d{0,3}[.]|[a-z0-9.\-]+[.][a-z]{2,4}/)"
                 r"(?:[^()\s<>]+|\(([^()\s<>]+|(\([^()\s<>]+\)))*\))+"
-                r"(?:\(([^()\s<>]+|(\([^()\s<>]+\)))*\)|[^()\s`!()\[\]{};:'\".,<>?«»“”‘’])"
+                r"(?:\(([^()\s<>]+|(\([^()\s<>]+\)))*\)|[^()\s`!\[\]{};:'\".,<>?«»“”‘’])"
                 r")"
             )
             urls = re.findall(regex, string)

src/backend/settings.py

@@ -275,11 +275,6 @@ class _AzureSearchSettings(BaseSettings, DatasourcePayloadConstructor):
     endpoint_suffix: str = Field(default="search.windows.net", exclude=True)
     connection_name: Optional[str] = None
     index: str = Field(serialization_alias="index_name")
-
-    def __init__(self, settings: "_AppSettings", **data):
-        # Ensure both BaseSettings and DatasourcePayloadConstructor are initialized
-        super().__init__(settings=settings, **data)
-
     key: Optional[str] = Field(default=None, exclude=True)
     use_semantic_search: bool = Field(default=False, exclude=True)
     semantic_search_config: str = Field(
@@ -308,6 +303,10 @@ def __init__(self, settings: "_AppSettings", **data):
     fields_mapping: Optional[dict] = None
     filter: Optional[str] = Field(default=None, exclude=True)
 
+    def __init__(self, settings: "_AppSettings", **data):
+        # Ensure both BaseSettings and DatasourcePayloadConstructor are initialized
+        super().__init__(settings=settings, **data)
+
     @field_validator("content_columns", "vector_columns", mode="before")
     @classmethod
     def split_columns(cls, comma_separated_string: str) -> List[str]: