bizon 0.1.2__py3-none-any.whl → 0.2.0__py3-none-any.whl

bizon/connectors/sources/notion/src/source.py

@@ -0,0 +1,1159 @@
+from concurrent.futures import ThreadPoolExecutor, as_completed
+from typing import Any, List, Optional, Tuple
+
+from loguru import logger
+from requests import Session
+from requests.adapters import HTTPAdapter
+from requests.auth import AuthBase
+from urllib3.util.retry import Retry
+
+from bizon.source.auth.builder import AuthBuilder
+from bizon.source.auth.config import AuthType
+from bizon.source.config import SourceConfig
+from bizon.source.models import SourceIteration, SourceRecord
+from bizon.source.source import AbstractSource
+
+from .config import NotionSourceConfig, NotionStreams
+
+NOTION_API_VERSION = "2025-09-03"
+BASE_URL = "https://api.notion.com/v1"
+
+
+class NotionSource(AbstractSource):
+    def __init__(self, config: NotionSourceConfig):
+        super().__init__(config)
+        self.config: NotionSourceConfig = config
+
+    def get_session(self) -> Session:
+        """Create a session with retry logic and required Notion headers."""
+        session = Session()
+        retries = Retry(
+            total=10,
+            backoff_factor=2,  # Exponential backoff: 2, 4, 8, 16, 32... seconds
+            status_forcelist=[429, 500, 502, 503, 504],
+            allowed_methods=["GET", "POST", "PATCH"],  # Retry on POST/PATCH too
+            respect_retry_after_header=True,  # Honor Notion's Retry-After header
+        )
+        session.mount("https://", HTTPAdapter(max_retries=retries))
+        session.headers.update(
+            {
+                "Notion-Version": NOTION_API_VERSION,
+                "Content-Type": "application/json",
+            }
+        )
+        return session
+
+    def get_authenticator(self) -> AuthBase:
+        if self.config.authentication.type.value in [AuthType.API_KEY, AuthType.BEARER]:
+            return AuthBuilder.token(params=self.config.authentication.params)
+        return None
+
+    @staticmethod
+    def streams() -> List[str]:
+        return [item.value for item in NotionStreams]
+
+    @staticmethod
+    def get_config_class() -> SourceConfig:
+        return NotionSourceConfig
+
+    def check_connection(self) -> Tuple[bool, Optional[Any]]:
+        """Test connection by listing users."""
+        try:
+            response = self.session.get(f"{BASE_URL}/users")
+            response.raise_for_status()
+            return True, None
+        except Exception as e:
+            return False, str(e)
+
+    def get_total_records_count(self) -> Optional[int]:
+        return None
+
+    # ==================== USERS STREAM ====================
+
+    def get_users(self, pagination: dict = None) -> SourceIteration:
+        """Fetch all users accessible to the integration."""
+        params = {"page_size": self.config.page_size}
+        if pagination and pagination.get("start_cursor"):
+            params["start_cursor"] = pagination["start_cursor"]
+
+        response = self.session.get(f"{BASE_URL}/users", params=params)
+        response.raise_for_status()
+        data = response.json()
+
+        records = [SourceRecord(id=user["id"], data=user) for user in data.get("results", [])]
+
+        next_pagination = {"start_cursor": data.get("next_cursor")} if data.get("has_more") else {}
+
+        return SourceIteration(records=records, next_pagination=next_pagination)
+
+    # ==================== DATABASES STREAM ====================
+
+    def get_database(self, database_id: str) -> dict:
+        """Fetch a single database by ID."""
+        response = self.session.get(f"{BASE_URL}/databases/{database_id}")
+        response.raise_for_status()
+        return response.json()
+
+    def get_databases(self, pagination: dict = None) -> SourceIteration:
+        """Fetch databases for the configured database_ids."""
+        if not self.config.database_ids:
+            logger.warning("No database_ids configured, returning empty results")
+            return SourceIteration(records=[], next_pagination={})
+
+        # Track progress through database_ids list
+        if pagination:
+            remaining_ids = pagination.get("remaining_ids", [])
+        else:
+            remaining_ids = list(self.config.database_ids)
+
+        if not remaining_ids:
+            return SourceIteration(records=[], next_pagination={})
+
+        # Process one database at a time
+        database_id = remaining_ids[0]
+        remaining_ids = remaining_ids[1:]
+
+        try:
+            database_data = self.get_database(database_id)
+            records = [SourceRecord(id=database_data["id"], data=database_data)]
+        except Exception as e:
+            logger.error(f"Failed to fetch database {database_id}: {e}")
+            records = []
+
+        next_pagination = {"remaining_ids": remaining_ids} if remaining_ids else {}
+
+        return SourceIteration(records=records, next_pagination=next_pagination)
+
+    # ==================== DATA SOURCES STREAM ====================
+
+    def get_data_sources(self, pagination: dict = None) -> SourceIteration:
+        """
+        Fetch data sources from databases.
+        In the 2025-09-03 API, databases contain a data_sources array.
+        """
+        if not self.config.database_ids:
+            logger.warning("No database_ids configured, returning empty results")
+            return SourceIteration(records=[], next_pagination={})
+
+        if pagination:
+            remaining_ids = pagination.get("remaining_ids", [])
+        else:
+            remaining_ids = list(self.config.database_ids)
+
+        if not remaining_ids:
+            return SourceIteration(records=[], next_pagination={})
+
+        database_id = remaining_ids[0]
+        remaining_ids = remaining_ids[1:]
+
+        records = []
+        try:
+            database_data = self.get_database(database_id)
+            data_sources = database_data.get("data_sources", [])
+
+            for ds in data_sources:
+                # Enrich data source with parent database info
+                ds_record = {
+                    **ds,
+                    "parent_database_id": database_id,
+                    "parent_database_title": self._extract_title(database_data),
+                }
+                records.append(SourceRecord(id=ds["id"], data=ds_record))
+
+        except Exception as e:
+            logger.error(f"Failed to fetch data sources from database {database_id}: {e}")
+
+        next_pagination = {"remaining_ids": remaining_ids} if remaining_ids else {}
+
+        return SourceIteration(records=records, next_pagination=next_pagination)
+
+    # ==================== PAGES STREAM ====================
+
+    def query_data_source(self, data_source_id: str, start_cursor: str = None, filter: dict = None) -> dict:
+        """Query a data source for its pages."""
+        payload = {"page_size": self.config.page_size}
+        if start_cursor:
+            payload["start_cursor"] = start_cursor
+        if filter:
+            payload["filter"] = filter
+
+        response = self.session.post(f"{BASE_URL}/data_sources/{data_source_id}/query", json=payload)
+        response.raise_for_status()
+        return response.json()
+
+    def get_page(self, page_id: str) -> dict:
+        """Fetch a single page by ID."""
+        response = self.session.get(f"{BASE_URL}/pages/{page_id}")
+        response.raise_for_status()
+        return response.json()
+
+    def get_pages(self, pagination: dict = None) -> SourceIteration:
+        """
+        Fetch pages from data sources (querying databases) and/or specific page IDs.
+        """
+        records = []
+
+        if pagination:
+            # Continue previous pagination state
+            # remaining_data_sources is list of {"ds_id": str, "db_id": str}
+            remaining_data_sources = pagination.get("remaining_data_sources", [])
+            current_data_source = pagination.get("current_data_source")  # {"ds_id": str, "db_id": str}
+            data_source_cursor = pagination.get("data_source_cursor")
+            remaining_page_ids = pagination.get("remaining_page_ids", [])
+            data_sources_loaded = pagination.get("data_sources_loaded", False)
+        else:
+            remaining_data_sources = []
+            current_data_source = None
+            data_source_cursor = None
+            remaining_page_ids = list(self.config.page_ids)
+            data_sources_loaded = False
+
+        # First, load all data sources from databases if not done
+        if not data_sources_loaded and self.config.database_ids:
+            for db_id in self.config.database_ids:
+                try:
+                    db_data = self.get_database(db_id)
+                    for ds in db_data.get("data_sources", []):
+                        remaining_data_sources.append({"ds_id": ds["id"], "db_id": db_id})
+                except Exception as e:
+                    logger.error(f"Failed to get data sources from database {db_id}: {e}")
+            data_sources_loaded = True
+
+        # Process current data source if we have one with a cursor
+        if current_data_source and data_source_cursor:
+            try:
+                ds_filter = self.get_filter_for_database(current_data_source["db_id"])
+                result = self.query_data_source(current_data_source["ds_id"], data_source_cursor, filter=ds_filter)
+                for page in result.get("results", []):
+                    records.append(SourceRecord(id=page["id"], data=page))
+
+                if result.get("has_more"):
+                    return SourceIteration(
+                        records=records,
+                        next_pagination={
+                            "remaining_data_sources": remaining_data_sources,
+                            "current_data_source": current_data_source,
+                            "data_source_cursor": result.get("next_cursor"),
+                            "remaining_page_ids": remaining_page_ids,
+                            "data_sources_loaded": True,
+                        },
+                    )
+            except Exception as e:
+                logger.error(f"Failed to query data source {current_data_source['ds_id']}: {e}")
+
+        # Process next data source
+        if remaining_data_sources:
+            ds_info = remaining_data_sources[0]
+            remaining_data_sources = remaining_data_sources[1:]
+
+            try:
+                ds_filter = self.get_filter_for_database(ds_info["db_id"])
+                result = self.query_data_source(ds_info["ds_id"], filter=ds_filter)
+                for page in result.get("results", []):
+                    records.append(SourceRecord(id=page["id"], data=page))
+
+                if result.get("has_more"):
+                    return SourceIteration(
+                        records=records,
+                        next_pagination={
+                            "remaining_data_sources": remaining_data_sources,
+                            "current_data_source": ds_info,
+                            "data_source_cursor": result.get("next_cursor"),
+                            "remaining_page_ids": remaining_page_ids,
+                            "data_sources_loaded": True,
+                        },
+                    )
+
+                # If there are more data sources, continue
+                if remaining_data_sources:
+                    return SourceIteration(
+                        records=records,
+                        next_pagination={
+                            "remaining_data_sources": remaining_data_sources,
+                            "current_data_source": None,
+                            "data_source_cursor": None,
+                            "remaining_page_ids": remaining_page_ids,
+                            "data_sources_loaded": True,
+                        },
+                    )
+            except Exception as e:
+                logger.error(f"Failed to query data source {ds_info['ds_id']}: {e}")
+                # Continue with remaining data sources
+                if remaining_data_sources:
+                    return SourceIteration(
+                        records=records,
+                        next_pagination={
+                            "remaining_data_sources": remaining_data_sources,
+                            "current_data_source": None,
+                            "data_source_cursor": None,
+                            "remaining_page_ids": remaining_page_ids,
+                            "data_sources_loaded": True,
+                        },
+                    )
+
+        # Process individual page IDs
+        if remaining_page_ids:
+            page_id = remaining_page_ids[0]
+            remaining_page_ids = remaining_page_ids[1:]
+
+            try:
+                page_data = self.get_page(page_id)
+                records.append(SourceRecord(id=page_data["id"], data=page_data))
+            except Exception as e:
+                logger.error(f"Failed to fetch page {page_id}: {e}")
+
+            if remaining_page_ids:
+                return SourceIteration(
+                    records=records,
+                    next_pagination={
+                        "remaining_data_sources": [],
+                        "current_data_source": None,
+                        "data_source_cursor": None,
+                        "remaining_page_ids": remaining_page_ids,
+                        "data_sources_loaded": True,
+                    },
+                )
+
+        return SourceIteration(records=records, next_pagination={})
+
+    # ==================== BLOCKS STREAM ====================
+
+    def get_block_children(self, block_id: str, start_cursor: str = None) -> dict:
+        """Fetch children blocks of a block."""
+        params = {"page_size": self.config.page_size}
+        if start_cursor:
+            params["start_cursor"] = start_cursor
+
+        response = self.session.get(f"{BASE_URL}/blocks/{block_id}/children", params=params)
+        response.raise_for_status()
+        return response.json()
+
+    def get_pages_from_database(self, database_id: str, apply_filter: bool = False) -> List[str]:
+        """Get all page IDs from a database by querying its data sources.
+
+        Args:
+            database_id: The database ID to fetch pages from
+            apply_filter: Whether to apply database_filters config (False for inline databases)
+        """
+        page_ids = []
+        db_filter = self.get_filter_for_database(database_id) if apply_filter else None
+        try:
+            db_data = self.get_database(database_id)
+            if not db_data:
+                return page_ids
+            for ds in db_data.get("data_sources") or []:
+                cursor = None
+                while True:
+                    result = self.query_data_source(ds["id"], cursor, filter=db_filter)
+                    if not result:
+                        break
+                    for page in result.get("results") or []:
+                        if page and page.get("id"):
+                            page_ids.append(page["id"])
+                    if result.get("has_more"):
+                        cursor = result.get("next_cursor")
+                    else:
+                        break
+        except Exception as e:
+            logger.error(f"Failed to get pages from database {database_id}: {e}")
+        return page_ids
+
+    def fetch_blocks_recursively(
+        self,
+        block_id: str,
+        parent_input_database_id: Optional[str] = None,
+        parent_input_page_id: Optional[str] = None,
+        source_page_id: Optional[str] = None,
+        current_depth: int = 0,
+        fetch_child_databases: bool = True,
+        global_order_counter: Optional[List[int]] = None,
+    ) -> List[dict]:
+        """
+        Fetch all blocks under a block_id recursively.
+        Also fetches content from child_database blocks.
+
+        Args:
+            block_id: The block/page ID to fetch children from
+            parent_input_database_id: The original input database ID from config
+            parent_input_page_id: The original input page ID from config
+            source_page_id: The immediate page this block belongs to
+            current_depth: Current recursion depth (0 = top level)
+            fetch_child_databases: Whether to recurse into child_database blocks (disable for all_* streams)
+            global_order_counter: Mutable counter [int] for tracking reading order across all blocks in a page
+
+        Returns:
+            Flat list of all blocks with lineage tracking fields
+        """
+        # Initialize counter on first call
+        if global_order_counter is None:
+            global_order_counter = [0]
+        # Check recursion depth limit
+        if current_depth >= self.config.max_recursion_depth:
+            logger.warning(
+                f"Max recursion depth {self.config.max_recursion_depth} reached for block {block_id}, stopping recursion"
+            )
+            return []
+
+        all_blocks = []
+        cursor = None
+        block_order = 0  # Track position within parent
+
+        while True:
+            result = self.get_block_children(block_id, cursor)
+            if not result:
+                break
+
+            for block in result.get("results") or []:
+                if not block:
+                    continue
+                # Add lineage tracking
+                block["parent_block_id"] = block_id
+                block["parent_input_database_id"] = parent_input_database_id
+                block["parent_input_page_id"] = parent_input_page_id
+                block["source_page_id"] = source_page_id
+                # Add depth and ordering
+                block["depth"] = current_depth
+                block["block_order"] = block_order
+                block["page_order"] = global_order_counter[0]
+                block_order += 1
+                global_order_counter[0] += 1
+
+                all_blocks.append(block)
+
+                # Handle child_database blocks - fetch their content in parallel
+                if (
+                    block.get("type") == "child_database"
+                    and self.config.fetch_blocks_recursively
+                    and fetch_child_databases
+                ):
+                    child_db_id = block.get("id")
+                    logger.info(f"Found inline database {child_db_id} at depth {current_depth}, fetching its content")
+
+                    try:
+                        # Get all pages from the inline database
+                        inline_page_ids = self.get_pages_from_database(child_db_id)
+
+                        # Fetch blocks from inline pages in parallel
+                        # Note: parallel execution means global_order_counter won't be sequential for inline DBs
+                        with ThreadPoolExecutor(max_workers=self.config.max_workers) as executor:
+                            futures = {
+                                executor.submit(
+                                    self.fetch_blocks_recursively,
+                                    block_id=inline_page_id,
+                                    parent_input_database_id=parent_input_database_id,
+                                    parent_input_page_id=parent_input_page_id,
+                                    source_page_id=inline_page_id,
+                                    current_depth=current_depth + 1,
+                                    fetch_child_databases=fetch_child_databases,
+                                    global_order_counter=global_order_counter,
+                                ): inline_page_id
+                                for inline_page_id in inline_page_ids
+                            }
+                            for future in as_completed(futures):
+                                try:
+                                    inline_blocks = future.result()
+                                    all_blocks.extend(inline_blocks)
+                                except Exception as e:
+                                    page_id = futures[future]
+                                    logger.error(f"Failed to fetch blocks from inline page {page_id}: {e}")
+
+                    except Exception as e:
+                        logger.error(f"Failed to fetch content from inline database {child_db_id}: {e}")
+
+                # Recursively fetch children if block has children
+                # Skip child_page and child_database - they are references, not containers with inline content
+                elif (
+                    block.get("has_children")
+                    and self.config.fetch_blocks_recursively
+                    and block.get("type") not in ("child_page", "child_database")
+                ):
+                    try:
+                        child_blocks = self.fetch_blocks_recursively(
+                            block_id=block["id"],
+                            parent_input_database_id=parent_input_database_id,
+                            parent_input_page_id=parent_input_page_id,
+                            source_page_id=source_page_id,
+                            current_depth=current_depth + 1,
+                            fetch_child_databases=fetch_child_databases,
+                            global_order_counter=global_order_counter,
+                        )
+                        all_blocks.extend(child_blocks)
+                    except Exception as e:
+                        # synced_block can return 404 if original block is inaccessible
+                        if block.get("type") == "synced_block":
+                            logger.warning(f"Skipping synced_block {block.get('id')} - children inaccessible: {e}")
+                        else:
+                            logger.error(f"Failed to fetch children of block {block.get('id')}: {e}")
+
+            if result.get("has_more"):
+                cursor = result.get("next_cursor")
+            else:
+                break
+
+        return all_blocks
+
+    def get_blocks(self, pagination: dict = None) -> SourceIteration:
+        """
+        Fetch blocks from databases and pages.
+        Blocks are fetched recursively if fetch_blocks_recursively is True.
+        Also fetches content from inline databases (child_database blocks).
+        """
+        if pagination:
+            # Each item is: {"block_id": str, "input_db_id": str|None, "input_page_id": str|None, "source_page_id": str|None}
+            items_to_process = pagination.get("items_to_process", [])
+            items_loaded = pagination.get("items_loaded", False)
+        else:
+            items_to_process = []
+            items_loaded = False
+
+        # First, collect all database IDs and page IDs to fetch blocks from
+        if not items_loaded:
+            # Add configured page_ids (these are direct input pages)
+            for page_id in self.config.page_ids:
+                items_to_process.append(
+                    {
+                        "block_id": page_id,
+                        "input_db_id": None,
+                        "input_page_id": page_id,
+                        "source_page_id": page_id,
+                    }
+                )
+
+            # Collect pages from databases
+            for db_id in self.config.database_ids:
+                try:
+                    db_filter = self.get_filter_for_database(db_id)
+                    db_data = self.get_database(db_id)
+                    for ds in db_data.get("data_sources", []):
+                        cursor = None
+                        while True:
+                            result = self.query_data_source(ds["id"], cursor, filter=db_filter)
+                            for page in result.get("results", []):
+                                items_to_process.append(
+                                    {
+                                        "block_id": page["id"],
+                                        "input_db_id": db_id,
+                                        "input_page_id": None,
+                                        "source_page_id": page["id"],
+                                    }
+                                )
+                            if result.get("has_more"):
+                                cursor = result.get("next_cursor")
+                            else:
+                                break
+                except Exception as e:
+                    logger.error(f"Failed to collect pages from database {db_id}: {e}")
+
+            items_loaded = True
+
+        if not items_to_process:
+            return SourceIteration(records=[], next_pagination={})
+
+        # Process a batch in parallel
+        batch_size = self.config.max_workers
+        batch = items_to_process[:batch_size]
+        items_to_process = items_to_process[batch_size:]
+
+        records = []
+
+        def fetch_item_blocks(item_info: dict) -> List[dict]:
+            """Fetch all blocks for a database or page."""
+            return self.fetch_blocks_recursively(
+                block_id=item_info["block_id"],
+                parent_input_database_id=item_info["input_db_id"],
+                parent_input_page_id=item_info["input_page_id"],
+                source_page_id=item_info["source_page_id"],
+            )
+
+        with ThreadPoolExecutor(max_workers=self.config.max_workers) as executor:
+            futures = {executor.submit(fetch_item_blocks, item_info): item_info for item_info in batch}
+            for future in as_completed(futures):
+                item_info = futures[future]
+                try:
+                    blocks = future.result()
+                    for block in blocks:
+                        records.append(SourceRecord(id=block["id"], data=block))
+                    logger.info(f"Fetched {len(blocks)} blocks from {item_info['block_id']}")
+                except Exception as e:
+                    logger.error(f"Failed to fetch blocks from {item_info['block_id']}: {e}")
+
+        next_pagination = {"items_to_process": items_to_process, "items_loaded": True} if items_to_process else {}
+
+        return SourceIteration(records=records, next_pagination=next_pagination)
+
+    # ==================== HELPERS ====================
+
+    def get_filter_for_database(self, database_id: str) -> Optional[dict]:
+        """Get the filter configured for a database, if any."""
+        return self.config.database_filters.get(database_id)
+
+    def _extract_title(self, database_data: dict) -> str:
+        """Extract plain text title from database object."""
+        title_parts = database_data.get("title", [])
+        return "".join(part.get("plain_text", "") for part in title_parts)
+
+    # ==================== MARKDOWN CONVERSION ====================
+
+    def _extract_rich_text(self, rich_text_array: List[dict]) -> str:
+        """Convert Notion rich text array to markdown string with formatting."""
+        if not rich_text_array:
+            return ""
+
+        result = []
+        for item in rich_text_array:
+            text = item.get("plain_text", "")
+            annotations = item.get("annotations", {})
+            href = item.get("href")
+
+            # Apply formatting in order: code, bold, italic, strikethrough
+            if annotations.get("code"):
+                text = f"`{text}`"
+            if annotations.get("bold"):
+                text = f"**{text}**"
+            if annotations.get("italic"):
+                text = f"*{text}*"
+            if annotations.get("strikethrough"):
+                text = f"~~{text}~~"
+            if href:
+                text = f"[{text}]({href})"
+
+            result.append(text)
+
+        return "".join(result)
+
+    def _block_to_markdown(self, block: dict) -> str:
+        """Convert a single Notion block to markdown string."""
+        block_type = block.get("type", "")
+        content = block.get(block_type) or {}
+
+        # Text blocks
+        if block_type == "paragraph":
+            return self._extract_rich_text(content.get("rich_text", []))
+
+        elif block_type == "heading_1":
+            return f"# {self._extract_rich_text(content.get('rich_text', []))}"
+
+        elif block_type == "heading_2":
+            return f"## {self._extract_rich_text(content.get('rich_text', []))}"
+
+        elif block_type == "heading_3":
+            return f"### {self._extract_rich_text(content.get('rich_text', []))}"
+
+        elif block_type == "bulleted_list_item":
+            return f"- {self._extract_rich_text(content.get('rich_text', []))}"
+
+        elif block_type == "numbered_list_item":
+            return f"1. {self._extract_rich_text(content.get('rich_text', []))}"
+
+        elif block_type == "to_do":
+            checkbox = "[x]" if content.get("checked") else "[ ]"
+            return f"- {checkbox} {self._extract_rich_text(content.get('rich_text', []))}"
+
+        elif block_type == "quote":
+            return f"> {self._extract_rich_text(content.get('rich_text', []))}"
+
+        elif block_type == "callout":
+            icon = content.get("icon") or {}
+            emoji = icon.get("emoji", "💡")
+            text = self._extract_rich_text(content.get("rich_text", []))
+            return f"> {emoji} {text}"
+
+        elif block_type == "code":
+            language = content.get("language", "")
+            code_text = self._extract_rich_text(content.get("rich_text", []))
+            return f"```{language}\n{code_text}\n```"
+
+        elif block_type == "equation":
+            return f"$$ {content.get('expression', '')} $$"
+
+        elif block_type == "divider":
+            return "---"
+
+        elif block_type == "toggle":
+            return f"<details><summary>{self._extract_rich_text(content.get('rich_text', []))}</summary></details>"
+
+        # Media blocks
+        elif block_type == "image":
+            url = content.get("external", {}).get("url") or content.get("file", {}).get("url", "")
+            caption = self._extract_rich_text(content.get("caption", []))
+            return f"![{caption}]({url})"
+
+        elif block_type == "video":
+            url = content.get("external", {}).get("url") or content.get("file", {}).get("url", "")
+            return f"[Video]({url})"
+
+        elif block_type == "file":
+            url = content.get("external", {}).get("url") or content.get("file", {}).get("url", "")
+            caption = self._extract_rich_text(content.get("caption", [])) or "File"
+            return f"[{caption}]({url})"
+
+        elif block_type == "pdf":
+            url = content.get("external", {}).get("url") or content.get("file", {}).get("url", "")
+            return f"[PDF]({url})"
+
+        elif block_type == "bookmark":
+            url = content.get("url", "")
+            caption = self._extract_rich_text(content.get("caption", [])) or url
+            return f"[{caption}]({url})"
+
+        elif block_type == "embed":
+            return f"[Embed]({content.get('url', '')})"
+
+        elif block_type == "link_preview":
+            return f"[Link Preview]({content.get('url', '')})"
+
+        # Table blocks
+        elif block_type == "table":
+            return "[Table - see child blocks for rows]"
+
+        elif block_type == "table_row":
+            cells = content.get("cells", [])
+            row = " | ".join(self._extract_rich_text(cell) for cell in cells)
+            return f"| {row} |"
+
+        # Database/page references
+        elif block_type == "child_page":
+            return f"[Page: {content.get('title', 'Untitled')}]"
+
+        elif block_type == "child_database":
+            return f"[Database: {content.get('title', 'Untitled')}]"
+
+        elif block_type == "link_to_page":
+            page_id = content.get("page_id") or content.get("database_id", "")
+            return f"[Link to page: {page_id}]"
+
+        elif block_type == "table_of_contents":
+            return "[Table of Contents]"
+
+        elif block_type == "breadcrumb":
+            return "[Breadcrumb]"
+
+        elif block_type == "synced_block":
+            return "[Synced Block]"
+
+        elif block_type == "template":
+            return "[Template]"
+
+        elif block_type == "column_list":
+            return ""  # Column list is just a container
+
+        elif block_type == "column":
+            return ""  # Column is just a container
+
+        else:
+            return f"[Unsupported block type: {block_type}]"
+
+    def get_blocks_markdown(self, pagination: dict = None) -> SourceIteration:
+        """
+        Fetch blocks and convert them to markdown.
+        Returns one record per block with its markdown content.
+        """
+        if pagination:
+            items_to_process = pagination.get("items_to_process", [])
+            items_loaded = pagination.get("items_loaded", False)
+        else:
+            items_to_process = []
+            items_loaded = False
+
+        # Collect all database IDs and page IDs to fetch blocks from
+        if not items_loaded:
+            for page_id in self.config.page_ids:
+                items_to_process.append(
+                    {
+                        "block_id": page_id,
+                        "input_db_id": None,
+                        "input_page_id": page_id,
+                        "source_page_id": page_id,
+                    }
+                )
+
+            for db_id in self.config.database_ids:
+                # Collect pages from database's data_sources
+                try:
+                    db_filter = self.get_filter_for_database(db_id)
+                    db_data = self.get_database(db_id)
+                    for ds in db_data.get("data_sources", []):
+                        cursor = None
+                        while True:
+                            result = self.query_data_source(ds["id"], cursor, filter=db_filter)
+                            for page in result.get("results", []):
+                                items_to_process.append(
+                                    {
+                                        "block_id": page["id"],
+                                        "input_db_id": db_id,
+                                        "input_page_id": None,
+                                        "source_page_id": page["id"],
+                                    }
+                                )
+                            if result.get("has_more"):
+                                cursor = result.get("next_cursor")
+                            else:
+                                break
+                except Exception as e:
+                    logger.error(f"Failed to collect pages from database {db_id}: {e}")
+
+            items_loaded = True
+
+        if not items_to_process:
+            return SourceIteration(records=[], next_pagination={})
+
+        # Process a batch in parallel
+        batch_size = self.config.max_workers
+        batch = items_to_process[:batch_size]
+        items_to_process = items_to_process[batch_size:]
+
+        records = []
+
+        def fetch_and_convert_item(item_info: dict) -> List[dict]:
+            """Fetch blocks for a database or page and convert each to markdown."""
+            blocks = self.fetch_blocks_recursively(
+                block_id=item_info["block_id"],
+                parent_input_database_id=item_info["input_db_id"],
+                parent_input_page_id=item_info["input_page_id"],
+                source_page_id=item_info["source_page_id"],
+                fetch_child_databases=False,
+            )
+
+            # Convert each block to markdown record
+            block_records = []
+            for block in blocks or []:
+                if not block:
+                    continue
+                md = self._block_to_markdown(block)
+                block_records.append(
+                    {
+                        "block_id": block.get("id"),
+                        "block_type": block.get("type"),
+                        "markdown": md,
+                        "source_page_id": block.get("source_page_id"),
+                        "parent_block_id": block.get("parent_block_id"),
+                        "parent_input_database_id": block.get("parent_input_database_id"),
+                        "parent_input_page_id": block.get("parent_input_page_id"),
+                        "depth": block.get("depth"),
+                        "block_order": block.get("block_order"),
+                        "page_order": block.get("page_order"),
+                        "block_raw": block,
+                    }
+                )
+            return block_records
+
+        with ThreadPoolExecutor(max_workers=self.config.max_workers) as executor:
+            futures = {executor.submit(fetch_and_convert_item, item_info): item_info for item_info in batch}
+            for future in as_completed(futures):
+                item_info = futures[future]
+                try:
+                    block_records = future.result()
+                    for block_record in block_records:
+                        records.append(SourceRecord(id=block_record.get("block_id"), data=block_record))
+                    logger.info(f"Converted {len(block_records)} blocks to markdown from {item_info['block_id']}")
+                except Exception as e:
+                    import traceback
+
+                    logger.error(
+                        f"Failed to fetch/convert blocks from {item_info['block_id']}: {e}\n{traceback.format_exc()}"
+                    )
+
+        next_pagination = {"items_to_process": items_to_process, "items_loaded": True} if items_to_process else {}
+
+        return SourceIteration(records=records, next_pagination=next_pagination)
+
+    # ==================== SEARCH API (ALL_* STREAMS) ====================
+
+    def search(self, start_cursor: str = None) -> dict:
+        """
+        Search all pages and databases accessible to the integration.
+
+        Args:
+            start_cursor: Pagination cursor
+
+        Returns:
+            Raw search results (filter client-side by object type)
+        """
+        payload = {"page_size": self.config.page_size}
+        if start_cursor:
+            payload["start_cursor"] = start_cursor
+
+        response = self.session.post(f"{BASE_URL}/search", json=payload)
+        response.raise_for_status()
+        return response.json()
+
+    def search_by_type(self, object_type: str, start_cursor: str = None) -> dict:
+        """
+        Search and filter by object type client-side.
+
+        Args:
+            object_type: "page" or "database"
+            start_cursor: Pagination cursor
+
+        Returns:
+            Filtered results matching object_type
+        """
+        result = self.search(start_cursor=start_cursor)
+
+        # Filter results by object type
+        filtered_results = [item for item in result.get("results", []) if item.get("object") == object_type]
+
+        return {
+            "results": filtered_results,
+            "has_more": result.get("has_more", False),
+            "next_cursor": result.get("next_cursor"),
+        }
+
+    def get_all_databases(self, pagination: dict = None) -> SourceIteration:
+        """
+        Fetch all databases accessible to the integration.
+        In 2025-09-03 API, we get data_sources from search and fetch their parent databases.
+        """
+        if pagination:
+            db_ids_to_fetch = pagination.get("db_ids_to_fetch", [])
+            dbs_loaded = pagination.get("dbs_loaded", False)
+        else:
+            db_ids_to_fetch = []
+            dbs_loaded = False
+
+        # Collect unique database IDs from data_sources
+        if not dbs_loaded:
+            seen_db_ids = set()
+            search_cursor = None
+
+            while True:
+                result = self.search_by_type(object_type="data_source", start_cursor=search_cursor)
+                for ds in result.get("results", []):
+                    # Data sources have a parent.database_id
+                    parent = ds.get("parent", {})
+                    if parent.get("type") == "database_id":
+                        db_id = parent.get("database_id")
+                        if db_id and db_id not in seen_db_ids:
+                            seen_db_ids.add(db_id)
+                            db_ids_to_fetch.append(db_id)
+
+                if result.get("has_more"):
+                    search_cursor = result.get("next_cursor")
+                else:
+                    break
+
+            dbs_loaded = True
+            logger.info(f"Found {len(db_ids_to_fetch)} unique databases from data_sources")
+
+        if not db_ids_to_fetch:
+            return SourceIteration(records=[], next_pagination={})
+
+        # Fetch one database at a time
+        db_id = db_ids_to_fetch[0]
+        db_ids_to_fetch = db_ids_to_fetch[1:]
+
+        records = []
+        try:
+            db_data = self.get_database(db_id)
+            records.append(SourceRecord(id=db_data["id"], data=db_data))
+        except Exception as e:
+            logger.error(f"Failed to fetch database {db_id}: {e}")
+
+        next_pagination = {"db_ids_to_fetch": db_ids_to_fetch, "dbs_loaded": True} if db_ids_to_fetch else {}
+
+        return SourceIteration(records=records, next_pagination=next_pagination)
+
+    def get_all_data_sources(self, pagination: dict = None) -> SourceIteration:
+        """Fetch all data_sources accessible to the integration."""
+        cursor = pagination.get("start_cursor") if pagination else None
+
+        result = self.search_by_type(object_type="data_source", start_cursor=cursor)
+
+        records = [SourceRecord(id=ds["id"], data=ds) for ds in result.get("results", [])]
+
+        next_pagination = {"start_cursor": result.get("next_cursor")} if result.get("has_more") else {}
+
+        return SourceIteration(records=records, next_pagination=next_pagination)
+
+    def get_all_pages(self, pagination: dict = None) -> SourceIteration:
+        """Fetch all pages accessible to the integration."""
+        cursor = pagination.get("start_cursor") if pagination else None
+
+        result = self.search_by_type(object_type="page", start_cursor=cursor)
+
+        records = [SourceRecord(id=page["id"], data=page) for page in result.get("results", [])]
+
+        next_pagination = {"start_cursor": result.get("next_cursor")} if result.get("has_more") else {}
+
+        return SourceIteration(records=records, next_pagination=next_pagination)
+
+    def get_all_blocks_markdown(self, pagination: dict = None) -> SourceIteration:
+        """
+        Fetch all databases and pages accessible to the integration and convert their blocks to markdown.
+        Includes databases and pages from search API AND pages from all databases via data_sources.
+        """
+        if pagination:
+            items_to_process = pagination.get("items_to_process", [])
+            items_loaded = pagination.get("items_loaded", False)
+        else:
+            items_to_process = []
+            items_loaded = False
+
+        # Collect databases and pages from search API
+        if not items_loaded:
+            seen_ids = set()
+
+            # 1. Get pages from search API
+            search_cursor = None
+            while True:
+                result = self.search_by_type(object_type="page", start_cursor=search_cursor)
+                for page in result.get("results", []):
+                    if page["id"] not in seen_ids:
+                        seen_ids.add(page["id"])
+                        items_to_process.append(
+                            {
+                                "block_id": page["id"],
+                                "input_db_id": None,
+                                "input_page_id": None,
+                                "source_page_id": page["id"],
+                            }
+                        )
+
+                if result.get("has_more"):
+                    search_cursor = result.get("next_cursor")
+                else:
+                    break
+
+            logger.info(f"Found {len(items_to_process)} pages from search API")
+
+            # 2. Get all data_sources and their parent databases + query for pages
+            ds_search_cursor = None
+            while True:
+                result = self.search_by_type(object_type="data_source", start_cursor=ds_search_cursor)
+                for ds in result.get("results", []):
+                    ds_id = ds["id"]
+                    # Get parent database_id from data_source
+                    parent = ds.get("parent", {})
+                    parent_db_id = parent.get("database_id") if parent.get("type") == "database_id" else None
+
+                    # Add the parent database to fetch its blocks (headers, descriptions, etc.)
+                    if parent_db_id and parent_db_id not in seen_ids:
+                        seen_ids.add(parent_db_id)
+                        items_to_process.append(
+                            {
+                                "block_id": parent_db_id,
+                                "input_db_id": parent_db_id,
+                                "input_page_id": None,
+                                "source_page_id": None,
+                            }
+                        )
+
+                    try:
+                        # Query data_source for pages (no filter for all_* streams)
+                        ds_cursor = None
+                        while True:
+                            ds_result = self.query_data_source(ds_id, ds_cursor)
+                            for page in ds_result.get("results", []):
+                                if page["id"] not in seen_ids:
+                                    seen_ids.add(page["id"])
+                                    items_to_process.append(
+                                        {
+                                            "block_id": page["id"],
+                                            "input_db_id": parent_db_id,
+                                            "input_page_id": None,
+                                            "source_page_id": page["id"],
+                                        }
+                                    )
+                            if ds_result.get("has_more"):
+                                ds_cursor = ds_result.get("next_cursor")
+                            else:
+                                break
+                    except Exception as e:
+                        logger.error(f"Failed to get pages from data_source {ds_id}: {e}")
+
+                if result.get("has_more"):
+                    ds_search_cursor = result.get("next_cursor")
+                else:
+                    break
+
+            items_loaded = True
+            logger.info(
+                f"Total {len(items_to_process)} unique items (databases + pages) to process for all_blocks_markdown"
+            )
+
+        if not items_to_process:
+            return SourceIteration(records=[], next_pagination={})
+
+        # Process a batch in parallel
+        batch_size = self.config.max_workers
+        batch = items_to_process[:batch_size]
+        items_to_process = items_to_process[batch_size:]
+
+        records = []
+
+        def fetch_and_convert_item(item_info: dict) -> List[dict]:
+            """Fetch blocks for a database or page and convert each to markdown."""
+            # fetch_child_databases=False because all_blocks_markdown already collects
+            # all pages from all data_sources, so we don't need to recurse into child_database blocks
+            blocks = self.fetch_blocks_recursively(
+                block_id=item_info["block_id"],
+                parent_input_database_id=item_info["input_db_id"],
+                parent_input_page_id=item_info["input_page_id"],
+                source_page_id=item_info["source_page_id"],
+                fetch_child_databases=False,
+            )
+
+            # Convert each block to markdown record
+            block_records = []
+            for block in blocks or []:
+                if not block:
+                    continue
+                md = self._block_to_markdown(block)
+                block_records.append(
+                    {
+                        "block_id": block.get("id"),
+                        "block_type": block.get("type"),
+                        "markdown": md,
+                        "source_page_id": block.get("source_page_id"),
+                        "parent_block_id": block.get("parent_block_id"),
+                        "parent_input_database_id": block.get("parent_input_database_id"),
+                        "parent_input_page_id": block.get("parent_input_page_id"),
+                        "depth": block.get("depth"),
+                        "block_order": block.get("block_order"),
+                        "page_order": block.get("page_order"),
+                        "block_raw": block,
+                    }
+                )
+            return block_records
+
+        with ThreadPoolExecutor(max_workers=self.config.max_workers) as executor:
+            futures = {executor.submit(fetch_and_convert_item, item_info): item_info for item_info in batch}
+            for future in as_completed(futures):
+                item_info = futures[future]
+                try:
+                    block_records = future.result()
+                    for block_record in block_records:
+                        records.append(SourceRecord(id=block_record.get("block_id"), data=block_record))
+                    logger.info(f"Converted {len(block_records)} blocks to markdown from {item_info['block_id']}")
+                except Exception as e:
+                    logger.error(f"Failed to fetch/convert blocks from {item_info['block_id']}: {e}")
+
+        next_pagination = {"items_to_process": items_to_process, "items_loaded": True} if items_to_process else {}
+
+        return SourceIteration(records=records, next_pagination=next_pagination)
+
+    # ==================== MAIN DISPATCH ====================
+
+    def get(self, pagination: dict = None) -> SourceIteration:
+        if self.config.stream == NotionStreams.USERS:
+            return self.get_users(pagination)
+        elif self.config.stream == NotionStreams.DATABASES:
+            return self.get_databases(pagination)
+        elif self.config.stream == NotionStreams.DATA_SOURCES:
+            return self.get_data_sources(pagination)
+        elif self.config.stream == NotionStreams.PAGES:
+            return self.get_pages(pagination)
+        elif self.config.stream == NotionStreams.BLOCKS:
+            return self.get_blocks(pagination)
+        elif self.config.stream == NotionStreams.BLOCKS_MARKDOWN:
+            return self.get_blocks_markdown(pagination)
+        elif self.config.stream == NotionStreams.ALL_PAGES:
+            return self.get_all_pages(pagination)
+        elif self.config.stream == NotionStreams.ALL_DATABASES:
+            return self.get_all_databases(pagination)
+        elif self.config.stream == NotionStreams.ALL_DATA_SOURCES:
+            return self.get_all_data_sources(pagination)
+        elif self.config.stream == NotionStreams.ALL_BLOCKS_MARKDOWN:
+            return self.get_all_blocks_markdown(pagination)
+
+        raise NotImplementedError(f"Stream {self.config.stream} not implemented for Notion")