agno 2.0.0rc1__tar.gz → 2.0.0rc2__tar.gz

{agno-2.0.0rc1 → agno-2.0.0rc2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agno
-Version: 2.0.0rc1
+Version: 2.0.0rc2
 Summary: Agno: a lightweight library for building Multi-Agent Systems
 Author-email: Ashpreet Bedi <ashpreet@agno.com>
 Project-URL: homepage, https://agno.com

{agno-2.0.0rc1 → agno-2.0.0rc2}/agno/agent/agent.py

@@ -129,10 +129,10 @@ class Agent:
     session_id: Optional[str] = None
     # Default session state (stored in the database to persist across runs)
     session_state: Optional[Dict[str, Any]] = None
-    # If True, the agent can update the session state
-    enable_agentic_state: bool = False
-    # If True, add the session state to the user prompt
+    # Set to True to add the session_state to the context
     add_session_state_to_context: bool = False
+    # Set to True to give the agent tools to update the session_state dynamically
+    enable_agentic_state: bool = False
     # If True, cache the current Agent session in memory for faster access
     cache_session: bool = False
 
@@ -321,8 +321,6 @@ class Agent:
     # --- If this Agent is part of a workflow ---
     # Optional workflow ID. Indicates this agent is part of a workflow.
     workflow_id: Optional[str] = None
-    # Set when this agent is part of a workflow.
-    workflow_session_id: Optional[str] = None
 
     # Metadata stored with this agent
     metadata: Optional[Dict[str, Any]] = None
@@ -345,7 +343,6 @@ class Agent:
         id: Optional[str] = None,
         introduction: Optional[str] = None,
         user_id: Optional[str] = None,
-        app_id: Optional[str] = None,
         session_id: Optional[str] = None,
         session_state: Optional[Dict[str, Any]] = None,
         add_session_state_to_context: bool = False,
@@ -429,7 +426,6 @@ class Agent:
         self.id = id
         self.introduction = introduction
         self.user_id = user_id
-        self.app_id = app_id
 
         self.session_id = session_id
         self.session_state = session_state
@@ -593,6 +589,15 @@ class Agent:
         if isinstance(input, Message):
             input = input.content  # type: ignore
 
+        # If input is a string, convert it to a dict
+        if isinstance(input, str):
+            import json
+
+            try:
+                input = json.loads(input)
+            except Exception as e:
+                raise ValueError(f"Failed to parse input. Is it a valid JSON string?: {e}")
+
         # Case 1: Message is already a BaseModel instance
         if isinstance(input, BaseModel):
             if isinstance(input, self.input_schema):
@@ -3231,6 +3236,12 @@ class Agent:
         if isinstance(model_response_event, tuple(get_args(RunOutputEvent))) or isinstance(
             model_response_event, tuple(get_args(TeamRunOutputEvent))
         ):
+            if model_response_event.event == RunEvent.custom_event:  # type: ignore
+                model_response_event.agent_id = self.id  # type: ignore
+                model_response_event.agent_name = self.name  # type: ignore
+                model_response_event.session_id = session.session_id  # type: ignore
+                model_response_event.run_id = run_response.run_id  # type: ignore
+
             # We just bubble the event up
             yield self._handle_event(model_response_event, run_response)  # type: ignore
         else:
@@ -4365,7 +4376,7 @@ class Agent:
 
             return agent_session
 
-        log_warning(f"AgentSession {session_id_to_load} not found in db")
+        log_debug(f"AgentSession {session_id_to_load} not found in db")
         return None
 
     def save_session(self, session: AgentSession) -> None:
@@ -6841,6 +6852,7 @@ class Agent:
 
         if self.output_schema is not None:
             markdown = False
+            markdown = False
 
         if stream is None:
             stream = self.stream or False
@@ -7172,10 +7184,12 @@ class Agent:
             image_artifacts = []
             for img in images:
                 try:
+                    artifact_id = img.id if hasattr(img, "id") and img.id else str(uuid4())
+
                     if img.url:
-                        image_artifacts.append(ImageArtifact(id=str(uuid4()), url=img.url))
+                        image_artifacts.append(ImageArtifact(id=artifact_id, url=img.url))
                     elif img.content:
-                        image_artifacts.append(ImageArtifact(id=str(uuid4()), content=img.content))
+                        image_artifacts.append(ImageArtifact(id=artifact_id, content=img.content))
                 except Exception as e:
                     log_warning(f"Error creating ImageArtifact: {e}")
                     continue
@@ -7185,10 +7199,12 @@ class Agent:
             video_artifacts = []
             for vid in videos:
                 try:
+                    artifact_id = vid.id if hasattr(vid, "id") and vid.id else str(uuid4())
+
                     if vid.url:
-                        video_artifacts.append(VideoArtifact(id=str(uuid4()), url=vid.url))
+                        video_artifacts.append(VideoArtifact(id=artifact_id, url=vid.url))
                     elif vid.content:
-                        video_artifacts.append(VideoArtifact(id=str(uuid4()), content=vid.content))
+                        video_artifacts.append(VideoArtifact(id=artifact_id, content=vid.content))
                 except Exception as e:
                     log_warning(f"Error creating VideoArtifact: {e}")
                     continue
@@ -7198,10 +7214,12 @@ class Agent:
             audio_artifacts = []
             for aud in audios:
                 try:
+                    artifact_id = aud.id if hasattr(aud, "id") and aud.id else str(uuid4())
+
                     if aud.url:
-                        audio_artifacts.append(AudioArtifact(id=str(uuid4()), url=aud.url))
+                        audio_artifacts.append(AudioArtifact(id=artifact_id, url=aud.url))
                     elif aud.content:
-                        audio_artifacts.append(AudioArtifact(id=str(uuid4()), content=aud.content))
+                        audio_artifacts.append(AudioArtifact(id=artifact_id, content=aud.content))
                 except Exception as e:
                     log_warning(f"Error creating AudioArtifact: {e}")
                     continue

{agno-2.0.0rc1 → agno-2.0.0rc2}/agno/db/mongo/mongo.py

@@ -672,7 +672,9 @@ class MongoDb(BaseDb):
             if result is None or not deserialize:
                 return result
 
-            return UserMemory.from_dict(result)
+            # Remove MongoDB's _id field before creating UserMemory object
+            result_filtered = {k: v for k, v in result.items() if k != "_id"}
+            return UserMemory.from_dict(result_filtered)
 
         except Exception as e:
             log_error(f"Exception reading from collection: {e}")
@@ -750,7 +752,8 @@ class MongoDb(BaseDb):
             if not deserialize:
                 return records, total_count
 
-            return [UserMemory.from_dict(record) for record in records]
+            # Remove MongoDB's _id field before creating UserMemory objects
+            return [UserMemory.from_dict({k: v for k, v in record.items() if k != "_id"}) for record in records]
 
         except Exception as e:
             log_error(f"Exception reading from collection: {e}")
@@ -861,7 +864,9 @@ class MongoDb(BaseDb):
             if not deserialize:
                 return update_doc
 
-            return UserMemory.from_dict(update_doc)
+            # Remove MongoDB's _id field before creating UserMemory object
+            update_doc_filtered = {k: v for k, v in update_doc.items() if k != "_id"}
+            return UserMemory.from_dict(update_doc_filtered)
 
         except Exception as e:
             log_error(f"Exception upserting user memory: {e}")

{agno-2.0.0rc1 → agno-2.0.0rc2}/agno/eval/accuracy.py

@@ -97,11 +97,18 @@ class AccuracyResult:
             title_justify="center",
         )
         summary_table.add_row("Number of Runs", f"{len(self.results)}")
-        summary_table.add_row("Average Score", f"{self.avg_score:.2f}")
-        summary_table.add_row("Mean Score", f"{self.mean_score:.2f}")
-        summary_table.add_row("Minimum Score", f"{self.min_score:.2f}")
-        summary_table.add_row("Maximum Score", f"{self.max_score:.2f}")
-        summary_table.add_row("Standard Deviation", f"{self.std_dev_score:.2f}")
+
+        if self.avg_score is not None:
+            summary_table.add_row("Average Score", f"{self.avg_score:.2f}")
+        if self.mean_score is not None:
+            summary_table.add_row("Mean Score", f"{self.mean_score:.2f}")
+        if self.min_score is not None:
+            summary_table.add_row("Minimum Score", f"{self.min_score:.2f}")
+        if self.max_score is not None:
+            summary_table.add_row("Maximum Score", f"{self.max_score:.2f}")
+        if self.std_dev_score is not None:
+            summary_table.add_row("Standard Deviation", f"{self.std_dev_score:.2f}")
+
         console.print(summary_table)
 
     def print_results(self, console: Optional["Console"] = None):

{agno-2.0.0rc1 → agno-2.0.0rc2}/agno/knowledge/chunking/strategy.py

@@ -35,13 +35,13 @@ class ChunkingStrategy(ABC):
 class ChunkingStrategyType(str, Enum):
     """Enumeration of available chunking strategies."""
 
-    AGENTIC_CHUNKING = "AgenticChunking"
-    DOCUMENT_CHUNKING = "DocumentChunking"
-    RECURSIVE_CHUNKING = "RecursiveChunking"
-    SEMANTIC_CHUNKING = "SemanticChunking"
-    FIXED_SIZE_CHUNKING = "FixedSizeChunking"
-    ROW_CHUNKING = "RowChunking"
-    MARKDOWN_CHUNKING = "MarkdownChunking"
+    AGENTIC_CHUNKER = "AgenticChunker"
+    DOCUMENT_CHUNKER = "DocumentChunker"
+    RECURSIVE_CHUNKER = "RecursiveChunker"
+    SEMANTIC_CHUNKER = "SemanticChunker"
+    FIXED_SIZE_CHUNKER = "FixedSizeChunker"
+    ROW_CHUNKER = "RowChunker"
+    MARKDOWN_CHUNKER = "MarkdownChunker"
 
     @classmethod
     def from_string(cls, strategy_name: str) -> "ChunkingStrategyType":
@@ -63,13 +63,13 @@ class ChunkingStrategyFactory:
     def create_strategy(cls, strategy_type: ChunkingStrategyType, **kwargs) -> ChunkingStrategy:
         """Create an instance of the chunking strategy with the given parameters."""
         strategy_map = {
-            ChunkingStrategyType.AGENTIC_CHUNKING: cls._create_agentic_chunking,
-            ChunkingStrategyType.DOCUMENT_CHUNKING: cls._create_document_chunking,
-            ChunkingStrategyType.RECURSIVE_CHUNKING: cls._create_recursive_chunking,
-            ChunkingStrategyType.SEMANTIC_CHUNKING: cls._create_semantic_chunking,
-            ChunkingStrategyType.FIXED_SIZE_CHUNKING: cls._create_fixed_chunking,
-            ChunkingStrategyType.ROW_CHUNKING: cls._create_row_chunking,
-            ChunkingStrategyType.MARKDOWN_CHUNKING: cls._create_markdown_chunking,
+            ChunkingStrategyType.AGENTIC_CHUNKER: cls._create_agentic_chunking,
+            ChunkingStrategyType.DOCUMENT_CHUNKER: cls._create_document_chunking,
+            ChunkingStrategyType.RECURSIVE_CHUNKER: cls._create_recursive_chunking,
+            ChunkingStrategyType.SEMANTIC_CHUNKER: cls._create_semantic_chunking,
+            ChunkingStrategyType.FIXED_SIZE_CHUNKER: cls._create_fixed_chunking,
+            ChunkingStrategyType.ROW_CHUNKER: cls._create_row_chunking,
+            ChunkingStrategyType.MARKDOWN_CHUNKER: cls._create_markdown_chunking,
         }
         return strategy_map[strategy_type](**kwargs)
 

{agno-2.0.0rc1 → agno-2.0.0rc2}/agno/knowledge/knowledge.py

@@ -5,16 +5,21 @@ import time
 from dataclasses import dataclass
 from enum import Enum
 from functools import cached_property
+from io import BytesIO
+from os.path import basename
 from pathlib import Path
 from typing import Any, Dict, List, Optional, Set, Tuple, Union, cast, overload
 from uuid import uuid4
 
+from httpx import AsyncClient
+
 from agno.db.base import BaseDb
 from agno.db.schemas.knowledge import KnowledgeRow
 from agno.knowledge.content import Content, ContentAuth, ContentStatus, FileData
 from agno.knowledge.document import Document
 from agno.knowledge.reader import Reader, ReaderFactory
 from agno.knowledge.remote_content.remote_content import GCSContent, RemoteContent, S3Content
+from agno.utils.http import async_fetch_with_retry
 from agno.utils.log import log_debug, log_error, log_info, log_warning
 from agno.vectordb import VectorDb
 
@@ -421,20 +426,31 @@ class Knowledge:
         upsert: bool,
         skip_if_exists: bool,
     ):
+        """Load the content in the contextual URL
+
+        1. Set content hash
+        2. Validate the URL
+        3. Read the content
+        4. Prepare and insert the content in the vector database
+        """
         log_info(f"Adding content from URL {content.url}")
         content.file_type = "url"
 
+        if not content.url:
+            raise ValueError("No url provided")
+
         if self.vector_db.__class__.__name__ == "LightRag":
             await self._process_lightrag_content(content, KnowledgeContentOrigin.URL)
             return
 
+        # 1. Set content hash
         content.content_hash = self._build_content_hash(content)
         if self.vector_db and self.vector_db.content_hash_exists(content.content_hash) and skip_if_exists:
             log_info(f"Content {content.content_hash} already exists, skipping")
             return
         self._add_to_contents_db(content)
 
-        # Validate URL
+        # 2. Validate URL
         try:
             from urllib.parse import urlparse
 
@@ -450,61 +466,47 @@ class Knowledge:
             self._update_content(content)
             log_warning(f"Invalid URL: {content.url} - {str(e)}")
 
-        # Determine file type from URL
-        url_path = Path(parsed_url.path)  # type: ignore
-        file_extension = url_path.suffix.lower()
-        read_documents = []
-        try:
-            if content.url.endswith("llms-full.txt") or content.url.endswith("llms.txt"):  # type: ignore
-                log_info("Detected llms, using url reader")
-                reader = content.reader or self.url_reader
-                if reader is not None:
-                    # TODO: We will refactor this to eventually pass authorization to all readers
-                    import inspect
+        # 3. Fetch and load content
+        async with AsyncClient() as client:
+            response = await async_fetch_with_retry(content.url, client=client)
+        bytes_content = BytesIO(response.content)
 
-                    read_signature = inspect.signature(reader.read)
-                    if "password" in read_signature.parameters and content.auth and content.auth.password:
-                        read_documents = reader.read(content.url, name=content.name, password=content.auth.password)
-                    else:
-                        read_documents = reader.read(content.url, name=content.name)
-
-            elif file_extension and file_extension is not None:
-                log_info(f"Detected file type: {file_extension} from URL: {content.url}")
-                if content.reader:
-                    reader = content.reader
-                else:
-                    reader = self._select_url_file_reader(file_extension)
-                if reader is not None:
-                    log_info(f"Selected reader: {reader.__class__.__name__}")
-                    # TODO: We will refactor this to eventually pass authorization to all readers
-                    import inspect
-
-                    read_signature = inspect.signature(reader.read)
-                    if "password" in read_signature.parameters and content.auth and content.auth.password:
-                        read_documents = reader.read(content.url, name=content.name, password=content.auth.password)
-                    else:
-                        read_documents = reader.read(content.url, name=content.name)
-                else:
-                    log_info(f"No reader found for file extension: {file_extension}")
+        # 4. Select reader
+        # If a reader was provided by the user, use it
+        reader = content.reader
+        name = content.name
+        # Else select based on file extension
+        if reader is None:
+            url_path = Path(parsed_url.path)
+            file_extension = url_path.suffix.lower()
+            if file_extension == ".csv":
+                name = basename(parsed_url.path) or "data.csv"
+                reader = self.csv_reader
+            elif file_extension == ".pdf":
+                reader = self.pdf_reader
+            elif file_extension == ".docx":
+                reader = self.docx_reader
+            elif file_extension == ".json":
+                reader = self.json_reader
+            elif file_extension == ".markdown":
+                reader = self.markdown_reader
             else:
-                log_info(f"No file extension found for URL: {content.url}, determining website type")
-                if content.reader:
-                    reader = content.reader
-                else:
-                    reader = self._select_url_reader(content.url)  # type: ignore
-                if reader is not None:
-                    log_info(f"Selected reader: {reader.__class__.__name__}")
-                    # TODO: We will refactor this to eventually pass authorization to all readers
-                    import inspect
+                reader = self.text_reader
 
-                    read_signature = inspect.signature(reader.read)
-                    if "password" in read_signature.parameters and content.auth and content.auth.password:
-                        read_documents = reader.read(content.url, name=content.name, password=content.auth.password)
-                    else:
-                        read_documents = reader.read(content.url, name=content.name)
+        # 5. Read content
+        try:
+            read_documents = []
+            if reader is not None:
+                # TODO: We will refactor this to eventually pass authorization to all readers
+                import inspect
+
+                read_signature = inspect.signature(reader.read)
+                if reader.__class__.__name__ == "YouTubeReader":
+                    read_documents = reader.read(content.url, name=name)
+                elif "password" in read_signature.parameters and content.auth and content.auth.password:
+                    read_documents = reader.read(bytes_content, name=name, password=content.auth.password)
                 else:
-                    log_info(f"No reader found for URL: {content.url}")
-
+                    read_documents = reader.read(bytes_content, name=name)
         except Exception as e:
             log_error(f"Error reading URL: {content.url} - {str(e)}")
             content.status = ContentStatus.FAILED
@@ -512,13 +514,17 @@ class Knowledge:
             self._update_content(content)
             return
 
+        # 6. Chunk documents if needed
+        if reader and not reader.chunk:
+            read_documents = await reader.chunk_documents_async(read_documents)
+
+        # 7. Prepare and insert the content in the vector database
         file_size = 0
         if read_documents:
             for read_document in read_documents:
                 if read_document.size:
                     file_size += read_document.size
                 read_document.content_id = content.id
-
         await self._handle_vector_db_insert(content, read_documents, upsert)
 
     async def _load_from_content(
@@ -699,21 +705,23 @@ class Knowledge:
             log_warning(f"Unsupported remote content type: {type(remote_content)}")
 
     async def _load_from_s3(self, content: Content, upsert: bool, skip_if_exists: bool):
-        from agno.aws.resource.s3.object import S3Object  # type: ignore
-
-        if content.reader is None:
-            reader = self.s3_reader
-        else:
-            reader = content.reader
-
-        if reader is None:
-            log_warning("No reader provided for content")
-            return
+        """Load the contextual S3 content.
+
+        1. Identify objects to read
+        2. Setup Content object
+        3. Hash content and add it to the contents database
+        4. Select reader
+        5. Fetch and load the content
+        6. Read the content
+        7. Prepare and insert the content in the vector database
+        8. Remove temporary file if needed
+        """
+        from agno.cloud.aws.s3.object import S3Object
 
         remote_content: S3Content = cast(S3Content, content.remote_content)
 
+        # 1. Identify objects to read
         objects_to_read: List[S3Object] = []
-
         if remote_content.bucket is not None:
             if remote_content.key is not None:
                 _object = S3Object(bucket_name=remote_content.bucket.name, name=remote_content.key)
@@ -725,10 +733,11 @@ class Knowledge:
             else:
                 objects_to_read.extend(remote_content.bucket.get_objects())
 
-        for object in objects_to_read:
+        for s3_object in objects_to_read:
+            # 2. Setup Content object
             id = str(uuid4())
             content_name = content.name or ""
-            content_name += "_" + (object.name or "")
+            content_name += "_" + (s3_object.name or "")
             content_entry = Content(
                 id=id,
                 name=content_name,
@@ -738,63 +747,123 @@ class Knowledge:
                 file_type="s3",
             )
 
+            # 3. Hash content and add it to the contents database
             content_hash = self._build_content_hash(content_entry)
             if self.vector_db and self.vector_db.content_hash_exists(content_hash) and skip_if_exists:
                 log_info(f"Content {content_hash} already exists, skipping")
                 continue
-
             self._add_to_contents_db(content_entry)
 
-            read_documents = reader.read(content_entry.name, object)
+            # 4. Select reader
+            reader = content.reader
+            if reader is None:
+                if s3_object.uri.endswith(".pdf"):
+                    reader = self.pdf_reader
+                elif s3_object.uri.endswith(".csv"):
+                    reader = self.csv_reader
+                elif s3_object.uri.endswith(".docx"):
+                    reader = self.docx_reader
+                elif s3_object.uri.endswith(".json"):
+                    reader = self.json_reader
+                elif s3_object.uri.endswith(".markdown"):
+                    reader = self.markdown_reader
+                else:
+                    reader = self.text_reader
+            reader = cast(Reader, reader)
+
+            # 5. Fetch and load the content
+            temporary_file = None
+            obj_name = content_name or s3_object.name.split("/")[-1]
+            readable_content: Optional[Union[BytesIO, Path]] = None
+            if s3_object.uri.endswith(".pdf"):
+                readable_content = BytesIO(s3_object.get_resource().get()["Body"].read())
+            else:
+                temporary_file = Path("storage").joinpath(obj_name)
+                readable_content = temporary_file
+                s3_object.download(readable_content)  # type: ignore
+
+            # 6. Read the content
+            read_documents = reader.read(readable_content, name=obj_name)
 
+            # 7. Prepare and insert the content in the vector database
             for read_document in read_documents:
                 read_document.content_id = content.id
-
             await self._handle_vector_db_insert(content_entry, read_documents, upsert)
 
-    async def _load_from_gcs(self, content: Content, upsert: bool, skip_if_exists: bool):
-        if content.reader is None:
-            reader = self.gcs_reader
-        else:
-            reader = content.reader
-
-        if reader is None:
-            log_warning("No reader provided for content")
-            return
+            # 8. Remove temporary file if needed
+            if temporary_file:
+                temporary_file.unlink()
 
+    async def _load_from_gcs(self, content: Content, upsert: bool, skip_if_exists: bool):
+        """Load the contextual GCS content.
+
+        1. Identify objects to read
+        2. Setup Content object
+        3. Hash content and add it to the contents database
+        4. Select reader
+        5. Fetch and load the content
+        6. Read the content
+        7. Prepare and insert the content in the vector database
+        """
         remote_content: GCSContent = cast(GCSContent, content.remote_content)
-        objects_to_read = []
 
+        # 1. Identify objects to read
+        objects_to_read = []
         if remote_content.blob_name is not None:
-            objects_to_read.append(remote_content.bucket.blob(remote_content.blob_name))
+            objects_to_read.append(remote_content.bucket.blob(remote_content.blob_name))  # type: ignore
         elif remote_content.prefix is not None:
-            objects_to_read.extend(remote_content.bucket.list_blobs(prefix=remote_content.prefix))
+            objects_to_read.extend(remote_content.bucket.list_blobs(prefix=remote_content.prefix))  # type: ignore
         else:
-            objects_to_read.extend(remote_content.bucket.list_blobs())
+            objects_to_read.extend(remote_content.bucket.list_blobs())  # type: ignore
 
-        for object in objects_to_read:
+        for gcs_object in objects_to_read:
+            # 2. Setup Content object
             id = str(uuid4())
+            name = (content.name or "content") + "_" + gcs_object.name
             content_entry = Content(
                 id=id,
-                name=(content.name or "content") + "_" + object.name,
+                name=name,
                 description=content.description,
                 status=ContentStatus.PROCESSING,
                 metadata=content.metadata,
                 file_type="gcs",
             )
 
+            # 3. Hash content and add it to the contents database
             content_hash = self._build_content_hash(content_entry)
             if self.vector_db and self.vector_db.content_hash_exists(content_hash) and skip_if_exists:
                 log_info(f"Content {content_hash} already exists, skipping")
                 continue
 
+            # 4. Add it to the contents database
             self._add_to_contents_db(content_entry)
 
-            read_documents = reader.read(content_entry.name, object)
+            # 5. Select reader
+            reader = content.reader
+            if reader is None:
+                if gcs_object.name.endswith(".pdf"):
+                    reader = self.pdf_reader
+                elif gcs_object.name.endswith(".csv"):
+                    reader = self.csv_reader
+                elif gcs_object.name.endswith(".docx"):
+                    reader = self.docx_reader
+                elif gcs_object.name.endswith(".json"):
+                    reader = self.json_reader
+                elif gcs_object.name.endswith(".markdown"):
+                    reader = self.markdown_reader
+                else:
+                    reader = self.text_reader
+            reader = cast(Reader, reader)
+
+            # 5. Fetch and load the content
+            readable_content = BytesIO(gcs_object.download_as_bytes())
+
+            # 6. Read the content
+            read_documents = reader.read(readable_content, name=name)
 
+            # 7. Prepare and insert the content in the vector database
             for read_document in read_documents:
                 read_document.content_id = content.id
-
             await self._handle_vector_db_insert(content_entry, read_documents, upsert)
 
     async def _handle_vector_db_insert(self, content, read_documents, upsert):
@@ -1006,7 +1075,7 @@ class Knowledge:
         elif content_type == KnowledgeContentOrigin.URL:
             log_info(f"Uploading file to LightRAG from URL: {content.url}")
             try:
-                reader = self.url_reader
+                reader = content.reader or self.website_reader
                 if reader is None:
                     log_error("No URL reader available")
                     content.status = ContentStatus.FAILED
@@ -1354,14 +1423,6 @@ class Knowledge:
         log_info(f"Selecting reader for extension: {extension}")
         return ReaderFactory.get_reader_for_extension(extension)
 
-    def _select_url_reader(self, url: str) -> Reader:
-        """Select the appropriate reader for a URL."""
-        return ReaderFactory.get_reader_for_url(url)
-
-    def _select_url_file_reader(self, extension: str) -> Reader:
-        """Select the appropriate reader for a URL file extension."""
-        return ReaderFactory.get_reader_for_url_file(extension)
-
     def get_filters(self) -> List[str]:
         return [
             "filter_tag_1",
@@ -1484,32 +1545,7 @@ class Knowledge:
         """Firecrawl reader - lazy loaded via factory."""
         return self._get_reader("firecrawl")
 
-    @property
-    def url_reader(self) -> Optional[Reader]:
-        """URL reader - lazy loaded via factory."""
-        return self._get_reader("url")
-
-    @property
-    def pdf_url_reader(self) -> Optional[Reader]:
-        """PDF URL reader - lazy loaded via factory."""
-        return self._get_reader("pdf_url")
-
     @property
     def youtube_reader(self) -> Optional[Reader]:
         """YouTube reader - lazy loaded via factory."""
         return self._get_reader("youtube")
-
-    @property
-    def csv_url_reader(self) -> Optional[Reader]:
-        """CSV URL reader - lazy loaded via factory."""
-        return self._get_reader("csv_url")
-
-    @property
-    def s3_reader(self) -> Optional[Reader]:
-        """S3 reader - lazy loaded via factory."""
-        return self._get_reader("s3")
-
-    @property
-    def gcs_reader(self) -> Optional[Reader]:
-        """GCS reader - lazy loaded via factory."""
-        return self._get_reader("gcs")