llama-index-readers-confluence 0.7.0

llama-index readers confluence integration

Confluence Loader

pip install llama-index-readers-confluence

This loader loads pages from a given Confluence cloud instance. The user needs to specify the base URL for a Confluence instance to initialize the ConfluenceReader - base URL needs to end with /wiki.

The user can optionally specify OAuth 2.0 credentials to authenticate with the Confluence instance. If no credentials are specified, the loader will look for CONFLUENCE_API_TOKEN or CONFLUENCE_USERNAME/CONFLUENCE_PASSWORD environment variables to proceed with basic authentication.

[!NOTE] Keep in mind CONFLUENCE_PASSWORD is not your actual password, but an API Token obtained here: https://id.atlassian.com/manage-profile/security/api-tokens.

The following order is used for checking authentication credentials:

1. oauth2
2. api_token
3. cookies
4. user_name and password
5. Environment variable CONFLUENCE_API_TOKEN
6. Environment variable CONFLUENCE_USERNAME and CONFLUENCE_PASSWORD

For more on authenticating using OAuth 2.0, checkout:

• https://atlassian-python-api.readthedocs.io/index.html
• https://developer.atlassian.com/cloud/confluence/oauth-2-3lo-apps/

Confluence pages are obtained through one of 4 four mutually exclusive ways:

1. page_ids: Load all pages from a list of page ids
2. space_key: Load all pages from a space
3. label: Load all pages with a given label
4. cql: Load all pages that match a given CQL query (Confluence Query Language https://developer.atlassian.com/cloud/confluence/advanced-searching-using-cql/ ).

When page_ids is specified, include_children will cause the loader to also load all descendent pages. When space_key is specified, page_status further specifies the status of pages to load: None, 'current', 'archived', 'draft'.

limit (int): Deprecated, use max_num_results instead.

max_num_results (int): Maximum number of results to return. If None, return all results. Requests are made in batches to achieve the desired number of results.

start(int): Which offset we should jump to when getting pages, only works with space_key

cursor(str): An alternative to start for cql queries, the cursor is a pointer to the next "page" when searching atlassian products. The current one after a search can be found with get_next_cursor()

User can also specify a boolean include_attachments to include attachments, this is set to False by default, if set to True all attachments will be downloaded and ConfluenceReader will extract the text from the attachments and add it to the Document object. Currently supported attachment types are: PDF, PNG, JPEG/JPG, SVG, Word and Excel.

Advanced Configuration

The ConfluenceReader supports several advanced configuration options for customizing the reading behavior:

Custom Parsers: You can provide custom parsers for specific file types using the custom_parsers parameter. This allows you to override the default parsing behavior for attachments of different types.

Custom parsers must implement the LlamaIndex BaseReader interface. Here's an example for DOCX files using MarkItDown:

from typing import List, Union
import pathlib
from llama_index.core.readers.base import BaseReader
from llama_index.core.schema import Document
from markitdown import MarkItDown


class DocxParser(BaseReader):
    """DOCX parser using MarkItDown for text extraction."""

    def __init__(self):
        self.markitdown = MarkItDown()

    def load_data(
        self, file_path: Union[str, pathlib.Path], **kwargs
    ) -> List[Document]:
        """Load and parse a DOCX file."""
        result = self.markitdown.convert(source=file_path)

        return [
            Document(
                text=result.markdown, metadata={"file_path": str(file_path)}
            )
        ]


# Usage with ConfluenceReader - Multiple file type parsers
from parsers import DocxParser
from readers.confluence_reader import FileType as ConfluenceFileType

confluence_parsers = {
    # ConfluenceFileType.PRESENTATION: PPTXParser(),
    ConfluenceFileType.DOCUMENT: DocxParser(),
    # ConfluenceFileType.PDF: PDFParser(),
    # ConfluenceFileType.HTML: HTMLParser(),
    # ConfluenceFileType.CSV: CSVParser(),
    # ConfluenceFileType.SPREADSHEET: ExcelParser(),
    # ConfluenceFileType.MARKDOWN: MarkdownParser(),
    # ConfluenceFileType.TEXT: TextParser()
}

reader = ConfluenceReader(
    base_url="https://yoursite.atlassian.com/wiki",
    api_token="your_token",
    custom_parsers=confluence_parsers,
)

Processing Callbacks:

• process_attachment_callback: A callback function to control which attachments should be processed. The function receives the media type and file size as parameters and should return a tuple of (should_process: bool, reason: str).
• process_document_callback: A callback function to control which documents should be processed. The function receives the page ID as a parameter and should return a boolean indicating whether to process the document.

File Management:

• custom_folder: Specify a custom directory for storing temporary files during processing. Can only be used when custom_parsers are provided. Defaults to the current working directory if custom_parsers are used.

Error Handling:

• fail_on_error (default: True): Whether to raise exceptions when encountering processing errors or continue with warnings and skip problematic content.

Logging:

• logger: Provide a custom logger instance for controlling log output during the reading process.

Event Monitoring (LlamaIndex Instrumentation): The ConfluenceReader integrates with LlamaIndex's instrumentation system to emit events during document and attachment processing. This allows you to monitor progress, handle errors, or integrate with external systems like databases or message queues.

Available event types:

• TotalPagesToProcessEvent: When the total number of pages to process is determined
• PageDataFetchStartedEvent: When processing of a page begins
• PageDataFetchCompletedEvent: When a page is successfully processed
• PageFailedEvent: When page processing fails
• PageSkippedEvent: When a page is skipped due to callback decision
• AttachmentProcessingStartedEvent: When attachment processing begins
• AttachmentProcessedEvent: When an attachment is successfully processed
• AttachmentSkippedEvent: When an attachment is skipped
• AttachmentFailedEvent: When attachment processing fails

To listen to events, create custom event handlers that inherit from BaseEventHandler and add them to the dispatcher:

from llama_index.core.instrumentation import get_dispatcher
from llama_index.core.instrumentation.event_handlers import BaseEventHandler


class MyEventHandler(BaseEventHandler):
    def handle(self, event):
        # Your event handling logic here
        print(f"Event: {event.class_name()}")


dispatcher = get_dispatcher("llama_index.readers.confluence.base")
dispatcher.add_event_handler(MyEventHandler())

Hint: space_key and page_id can both be found in the URL of a page in Confluence - https://yoursite.atlassian.com/wiki/spaces/<space_key>/pages/<page_id>

Usage

Here's an example usage of the ConfluenceReader.

# Example that reads the pages with the `page_ids`
from llama_index.readers.confluence import ConfluenceReader

token = {"access_token": "<access_token>", "token_type": "<token_type>"}
oauth2_dict = {"client_id": "<client_id>", "token": token}

base_url = "https://yoursite.atlassian.com/wiki"

page_ids = ["<page_id_1>", "<page_id_2>", "<page_id_3"]
space_key = "<space_key>"

reader = ConfluenceReader(
    base_url=base_url,
    oauth2=oauth2_dict,
    client_args={"backoff_and_retry": True},
)
documents = reader.load_data(
    space_key=space_key, include_attachments=True, page_status="current"
)
documents.extend(
    reader.load_data(
        page_ids=page_ids, include_children=True, include_attachments=True
    )
)

# Example that fetches the first 5, then the next 5 pages from a space
from llama_index.readers.confluence import ConfluenceReader

token = {"access_token": "<access_token>", "token_type": "<token_type>"}
oauth2_dict = {"client_id": "<client_id>", "token": token}

base_url = "https://yoursite.atlassian.com/wiki"

space_key = "<space_key>"

reader = ConfluenceReader(base_url=base_url, oauth2=oauth2_dict)
documents = reader.load_data(
    space_key=space_key,
    include_attachments=True,
    page_status="current",
    start=0,
    max_num_results=5,
)
documents.extend(
    reader.load_data(
        space_key=space_key,
        include_children=True,
        include_attachments=True,
        start=5,
        max_num_results=5,
    )
)

# Example that fetches the first 5 results from a cql query, the uses the cursor to pick up on the next element
from llama_index.readers.confluence import ConfluenceReader

token = {"access_token": "<access_token>", "token_type": "<token_type>"}
oauth2_dict = {"client_id": "<client_id>", "token": token}

base_url = "https://yoursite.atlassian.com/wiki"

cql = f'type="page" AND label="devops"'

reader = ConfluenceReader(base_url=base_url, oauth2=oauth2_dict)
documents = reader.load_data(cql=cql, max_num_results=5)
cursor = reader.get_next_cursor()
documents.extend(reader.load_data(cql=cql, cursor=cursor, max_num_results=5))

# Example with advanced configuration options
from llama_index.readers.confluence import ConfluenceReader
import logging


# Custom callback to filter attachments by size and type
def attachment_filter(
    media_type: str, file_size: int, title: str
) -> tuple[bool, str]:
    # Skip large files (>10MB)
    if file_size > 10 * 1024 * 1024:
        return False, f"File too large: {file_size} bytes"

    # Skip certain file types
    if media_type in ["application/x-zip-compressed", "application/zip"]:
        return False, f"Unsupported file type: {media_type}"

    return True, ""


# Custom callback to filter documents
def document_filter(page_id: str) -> bool:
    # Skip specific pages by ID
    excluded_pages = ["123456", "789012"]
    return page_id not in excluded_pages


# Setup custom logger
logger = logging.getLogger("confluence_reader")
logger.setLevel(logging.INFO)

reader = ConfluenceReader(
    base_url="https://yoursite.atlassian.com/wiki",
    api_token="your_api_token",
    process_attachment_callback=attachment_filter,
    process_document_callback=document_filter,
    custom_folder="/tmp/confluence_files",  # Custom temp directory
    logger=logger,
    fail_on_error=False,  # Continue processing even if some content fails
)

documents = reader.load_data(space_key="MYSPACE", include_attachments=True)

Event System Integration

For monitoring the processing with custom event handlers, you can import the event types directly:

from llama_index.readers.confluence.event import (
    TotalPagesToProcessEvent,
    PageDataFetchStartedEvent,
    PageDataFetchCompletedEvent,
    PageFailedEvent,
    PageSkippedEvent,
    AttachmentProcessingStartedEvent,
    AttachmentProcessedEvent,
    AttachmentSkippedEvent,
    AttachmentFailedEvent,
    FileType,
)

# Example using LlamaIndex's instrumentation system for event monitoring
from llama_index.readers.confluence import ConfluenceReader
from llama_index.core.instrumentation import get_dispatcher
from llama_index.core.instrumentation.event_handlers import BaseEventHandler
from llama_index.readers.confluence.event import (
    TotalPagesToProcessEvent,
    PageDataFetchStartedEvent,
    PageDataFetchCompletedEvent,
    PageFailedEvent,
    PageSkippedEvent,
    AttachmentProcessingStartedEvent,
    AttachmentProcessedEvent,
    AttachmentSkippedEvent,
    AttachmentFailedEvent,
)


# Custom event handler class
class ConfluenceEventHandler(BaseEventHandler):
    """Custom event handler for Confluence reader events."""

    @classmethod
    def class_name(cls) -> str:
        return "ConfluenceEventHandler"

    def handle(self, event) -> None:
        """Handle different types of events."""
        if isinstance(event, TotalPagesToProcessEvent):
            print(f"Total pages to process: {event.total_pages}")
        elif isinstance(event, PageDataFetchStartedEvent):
            print(f"Started processing page: {event.page_id}")
        elif isinstance(event, PageDataFetchCompletedEvent):
            print(f"Completed processing page: {event.page_id}")
            print(
                f"Document title: {event.document.metadata.get('title', 'Unknown')}"
            )
        elif isinstance(event, PageSkippedEvent):
            print(f"Skipped page: {event.page_id}")
        elif isinstance(event, PageFailedEvent):
            print(f"Failed to process page {event.page_id}: {event.error}")
        elif isinstance(event, AttachmentProcessingStartedEvent):
            print(f"Started processing attachment: {event.attachment_name}")
        elif isinstance(event, AttachmentProcessedEvent):
            print(f"Processed attachment: {event.attachment_name}")
            print(f"Attachment type: {event.attachment_type}")
            print(f"Attachment size: {event.attachment_size}")
        elif isinstance(event, AttachmentSkippedEvent):
            print(
                f"Skipped attachment {event.attachment_name}: {event.reason}"
            )
        elif isinstance(event, AttachmentFailedEvent):
            print(
                f"Failed to process attachment {event.attachment_name}: {event.error}"
            )


# Alternative: Create separate handlers for different event types
class PageEventHandler(BaseEventHandler):
    """Handler specifically for page-related events."""

    @classmethod
    def class_name(cls) -> str:
        return "PageEventHandler"

    def handle(self, event) -> None:
        if isinstance(
            event,
            (
                PageDataFetchStartedEvent,
                PageDataFetchCompletedEvent,
                PageFailedEvent,
                PageSkippedEvent,
            ),
        ):
            print(f"Page event: {event.class_name()} for page {event.page_id}")


class AttachmentEventHandler(BaseEventHandler):
    """Handler specifically for attachment-related events."""

    @classmethod
    def class_name(cls) -> str:
        return "AttachmentEventHandler"

    def handle(self, event) -> None:
        if isinstance(
            event,
            (
                AttachmentProcessingStartedEvent,
                AttachmentProcessedEvent,
                AttachmentSkippedEvent,
                AttachmentFailedEvent,
            ),
        ):
            print(
                f"Attachment event: {event.class_name()} for {event.attachment_name}"
            )


# Create reader and set up event monitoring
reader = ConfluenceReader(
    base_url="https://yoursite.atlassian.com/wiki", api_token="your_api_token"
)

# Get the dispatcher and add event handlers
dispatcher = get_dispatcher("llama_index.readers.confluence.base")

# Add handlers to the dispatcher
confluence_handler = ConfluenceEventHandler()
page_handler = PageEventHandler()
attachment_handler = AttachmentEventHandler()

dispatcher.add_event_handler(confluence_handler)
dispatcher.add_event_handler(page_handler)
dispatcher.add_event_handler(attachment_handler)

# Load data - events will be emitted during processing
documents = reader.load_data(space_key="MYSPACE", include_attachments=True)

print(f"Processing completed. Total documents: {len(documents)}")

This loader is designed to be used as a way to load data into LlamaIndex.