diff --git a/backend/danswer/background/update.py b/backend/danswer/background/update.py index ba310232a..6f95b4bf9 100755 --- a/backend/danswer/background/update.py +++ b/backend/danswer/background/update.py @@ -23,7 +23,6 @@ from danswer.db.connector_credential_pair import update_connector_credential_pai from danswer.db.credentials import backend_update_credential_json from danswer.db.engine import get_db_current_time from danswer.db.engine import get_sqlalchemy_engine -from danswer.db.feedback import create_document_metadata from danswer.db.index_attempt import create_index_attempt from danswer.db.index_attempt import get_index_attempt from danswer.db.index_attempt import get_inprogress_index_attempts diff --git a/backend/danswer/bots/slack/blocks.py b/backend/danswer/bots/slack/blocks.py new file mode 100644 index 000000000..5efe693de --- /dev/null +++ b/backend/danswer/bots/slack/blocks.py @@ -0,0 +1,155 @@ +from slack_sdk.models.blocks import ActionsBlock +from slack_sdk.models.blocks import Block +from slack_sdk.models.blocks import ButtonElement +from slack_sdk.models.blocks import SectionBlock + +from danswer.bots.slack.constants import DISLIKE_BLOCK_ACTION_ID +from danswer.bots.slack.constants import LIKE_BLOCK_ACTION_ID +from danswer.bots.slack.utils import build_block_id_from_query_event_id +from danswer.configs.app_configs import DANSWER_BOT_NUM_DOCS_TO_DISPLAY +from danswer.configs.constants import DocumentSource +from danswer.connectors.slack.utils import UserIdReplacer +from danswer.direct_qa.interfaces import DanswerQuote +from danswer.server.models import SearchDoc + + +def build_feedback_block(query_event_id: int) -> Block: + return ActionsBlock( + block_id=build_block_id_from_query_event_id(query_event_id), + elements=[ + ButtonElement( + action_id=LIKE_BLOCK_ACTION_ID, + text="👍", + style="primary", + ), + ButtonElement( + action_id=DISLIKE_BLOCK_ACTION_ID, + text="👎", + style="danger", + ), + ], + ) + + +_MAX_BLURB_LEN = 75 + + +def _build_custom_semantic_identifier( + semantic_identifier: str, blurb: str, source: str +) -> str: + """ + On slack, since we just show the semantic identifier rather than semantic + blurb, we need + to do some custom formatting to make sure the semantic identifier is unique and meaningful. + """ + if source == DocumentSource.SLACK.value: + truncated_blurb = ( + f"{blurb[:_MAX_BLURB_LEN]}..." if len(blurb) > _MAX_BLURB_LEN else blurb + ) + # NOTE: removing tags so that we don't accidentally tag users in Slack + + # so that it can be used as part of a link + truncated_blurb = UserIdReplacer.replace_tags_basic(truncated_blurb) + truncated_blurb = UserIdReplacer.replace_channels_basic(truncated_blurb) + truncated_blurb = UserIdReplacer.replace_special_mentions(truncated_blurb) + if truncated_blurb: + return f"#{semantic_identifier}: {truncated_blurb}" + else: + return f"#{semantic_identifier}" + + return semantic_identifier + + +def build_documents_block( + documents: list[SearchDoc], + already_displayed_doc_identifiers: list[str], + num_docs_to_display: int = DANSWER_BOT_NUM_DOCS_TO_DISPLAY, +) -> SectionBlock: + seen_docs_identifiers = set(already_displayed_doc_identifiers) + top_document_lines: list[str] = [] + for d in documents: + if d.document_id in seen_docs_identifiers: + continue + seen_docs_identifiers.add(d.document_id) + + custom_semantic_identifier = _build_custom_semantic_identifier( + semantic_identifier=d.semantic_identifier, + blurb=d.blurb, + source=d.source_type, + ) + + top_document_lines.append(f"- <{d.link}|{custom_semantic_identifier}>") + if len(top_document_lines) >= num_docs_to_display: + break + + return SectionBlock( + fields=[ + "*Other potentially relevant docs:*", + *top_document_lines, + ] + ) + + +def build_quotes_block( + quotes: list[DanswerQuote], +) -> tuple[list[Block], list[str]]: + quote_lines: list[str] = [] + doc_identifiers: list[str] = [] + for quote in quotes: + doc_id = quote.document_id + doc_link = quote.link + doc_name = quote.semantic_identifier + if doc_link and doc_name and doc_id and doc_id not in doc_identifiers: + doc_identifiers.append(doc_id) + custom_semantic_identifier = _build_custom_semantic_identifier( + semantic_identifier=doc_name, + blurb=quote.blurb, + source=quote.source_type, + ) + quote_lines.append(f"- <{doc_link}|{custom_semantic_identifier}>") + + if not quote_lines: + return [], [] + + return ( + [ + SectionBlock( + fields=[ + "*Sources:*", + *quote_lines, + ] + ) + ], + doc_identifiers, + ) + + +def build_qa_response_blocks( + query_event_id: int, + answer: str | None, + quotes: list[DanswerQuote] | None, + documents: list[SearchDoc], +) -> list[Block]: + doc_identifiers: list[str] = [] + quotes_blocks: list[Block] = [] + if not answer: + answer_block = SectionBlock( + text=f"Sorry, I was unable to find an answer, but I did find some potentially relevant docs 🤓" + ) + else: + answer_block = SectionBlock(text=answer) + if quotes: + quotes_blocks, doc_identifiers = build_quotes_block(quotes) + + # if no quotes OR `build_quotes_block()` did not give back any blocks + if not quotes_blocks: + quotes_blocks = [ + SectionBlock( + text="*Warning*: no sources were quoted for this answer, so it may be unreliable 😔" + ) + ] + + documents_block = build_documents_block(documents, doc_identifiers) + return ( + [answer_block] + + quotes_blocks + + [documents_block, build_feedback_block(query_event_id=query_event_id)] + ) diff --git a/backend/danswer/bots/slack/constants.py b/backend/danswer/bots/slack/constants.py new file mode 100644 index 000000000..5021ef1a1 --- /dev/null +++ b/backend/danswer/bots/slack/constants.py @@ -0,0 +1,2 @@ +LIKE_BLOCK_ACTION_ID = "feedback-like" +DISLIKE_BLOCK_ACTION_ID = "feedback-dislike" diff --git a/backend/danswer/bots/slack/handlers/handle_feedback.py b/backend/danswer/bots/slack/handlers/handle_feedback.py new file mode 100644 index 000000000..ff43c8163 --- /dev/null +++ b/backend/danswer/bots/slack/handlers/handle_feedback.py @@ -0,0 +1,16 @@ +from sqlalchemy.orm import Session + +from danswer.configs.constants import QAFeedbackType +from danswer.db.engine import get_sqlalchemy_engine +from danswer.db.feedback import update_query_event_feedback + + +def handle_qa_feedback(query_id: int, feedback_type: QAFeedbackType) -> None: + engine = get_sqlalchemy_engine() + with Session(engine) as db_session: + update_query_event_feedback( + feedback=feedback_type, + query_id=query_id, + user_id=None, # no "user" for Slack bot for now + db_session=db_session, + ) diff --git a/backend/danswer/bots/slack/handlers/handle_message.py b/backend/danswer/bots/slack/handlers/handle_message.py new file mode 100644 index 000000000..8f8d5561a --- /dev/null +++ b/backend/danswer/bots/slack/handlers/handle_message.py @@ -0,0 +1,115 @@ +import logging + +from retry import retry +from slack_sdk import WebClient +from sqlalchemy.orm import Session + +from danswer.bots.slack.blocks import build_qa_response_blocks +from danswer.bots.slack.utils import respond_in_thread +from danswer.configs.app_configs import DANSWER_BOT_ANSWER_GENERATION_TIMEOUT +from danswer.configs.app_configs import DANSWER_BOT_DISABLE_DOCS_ONLY_ANSWER +from danswer.configs.app_configs import DANSWER_BOT_DISPLAY_ERROR_MSGS +from danswer.configs.app_configs import DANSWER_BOT_NUM_RETRIES +from danswer.configs.app_configs import DOCUMENT_INDEX_NAME +from danswer.db.engine import get_sqlalchemy_engine +from danswer.direct_qa.answer_question import answer_qa_query +from danswer.server.models import QAResponse +from danswer.server.models import QuestionRequest + + +def handle_message( + msg: str, + channel: str, + message_ts_to_respond_to: str, + client: WebClient, + logger: logging.Logger, + num_retries: int = DANSWER_BOT_NUM_RETRIES, + answer_generation_timeout: int = DANSWER_BOT_ANSWER_GENERATION_TIMEOUT, + should_respond_with_error_msgs: bool = DANSWER_BOT_DISPLAY_ERROR_MSGS, + disable_docs_only_answer: bool = DANSWER_BOT_DISABLE_DOCS_ONLY_ANSWER, +) -> None: + @retry( + tries=num_retries, + delay=0.25, + backoff=2, + logger=logger, + ) + def _get_answer(question: QuestionRequest) -> QAResponse: + engine = get_sqlalchemy_engine() + with Session(engine, expire_on_commit=False) as db_session: + answer = answer_qa_query( + question=question, + user=None, + db_session=db_session, + answer_generation_timeout=answer_generation_timeout, + ) + if not answer.error_msg: + return answer + else: + raise RuntimeError(answer.error_msg) + + try: + answer = _get_answer( + QuestionRequest( + query=msg, + collection=DOCUMENT_INDEX_NAME, + use_keyword=None, + filters=None, + offset=None, + ) + ) + except Exception as e: + logger.exception( + f"Unable to process message - did not successfully answer " + f"in {num_retries} attempts" + ) + # Optionally, respond in thread with the error message, Used primarily + # for debugging purposes + if should_respond_with_error_msgs: + respond_in_thread( + client=client, + channel=channel, + text=f"Encountered exception when trying to answer: \n\n```{e}```", + thread_ts=message_ts_to_respond_to, + ) + return + + if not answer.top_ranked_docs: + logger.error(f"Unable to answer question: '{msg}' - no documents found") + # Optionally, respond in thread with the error message, Used primarily + # for debugging purposes + if should_respond_with_error_msgs: + respond_in_thread( + client=client, + channel=channel, + text="Found no documents when trying to answer. Did you index any documents?", + thread_ts=message_ts_to_respond_to, + ) + return + + if not answer.answer and disable_docs_only_answer: + logger.info( + "Unable to find answer - not responding since the " + "`DANSWER_BOT_DISABLE_DOCS_ONLY_ANSWER` env variable is set" + ) + return + + # convert raw response into "nicely" formatted Slack message + blocks = build_qa_response_blocks( + query_event_id=answer.query_event_id, + answer=answer.answer, + documents=answer.top_ranked_docs, + quotes=answer.quotes, + ) + try: + respond_in_thread( + client=client, + channel=channel, + blocks=blocks, + thread_ts=message_ts_to_respond_to, + ) + except Exception: + logger.exception( + f"Unable to process message - could not respond in slack in {num_retries} attempts" + ) + return diff --git a/backend/danswer/bots/slack/listener.py b/backend/danswer/bots/slack/listener.py new file mode 100644 index 000000000..eda0f3819 --- /dev/null +++ b/backend/danswer/bots/slack/listener.py @@ -0,0 +1,184 @@ +import logging +import os +from collections.abc import MutableMapping +from typing import Any +from typing import cast + +from slack_sdk import WebClient +from slack_sdk.socket_mode import SocketModeClient +from slack_sdk.socket_mode.request import SocketModeRequest +from slack_sdk.socket_mode.response import SocketModeResponse + +from danswer.bots.slack.constants import DISLIKE_BLOCK_ACTION_ID +from danswer.bots.slack.constants import LIKE_BLOCK_ACTION_ID +from danswer.bots.slack.handlers.handle_feedback import handle_qa_feedback +from danswer.bots.slack.handlers.handle_message import handle_message +from danswer.bots.slack.utils import get_query_event_id_from_block_id +from danswer.configs.constants import QAFeedbackType +from danswer.utils.logger import setup_logger + +logger = setup_logger() + + +_CHANNEL_ID = "channel_id" + + +class _ChannelIdAdapter(logging.LoggerAdapter): + """This is used to add the channel ID to all log messages + emitted in this file""" + + def process( + self, msg: str, kwargs: MutableMapping[str, Any] + ) -> tuple[str, MutableMapping[str, Any]]: + channel_id = self.extra.get(_CHANNEL_ID) if self.extra else None + if channel_id: + return f"[Channel ID: {channel_id}] {msg}", kwargs + else: + return msg, kwargs + + +def _get_socket_client() -> SocketModeClient: + # For more info on how to set this up, checkout the docs: + # https://docs.danswer.dev/slack_bot_setup + app_token = os.environ.get("DANSWER_BOT_SLACK_APP_TOKEN") + if not app_token: + raise RuntimeError("DANSWER_BOT_SLACK_APP_TOKEN is not set") + bot_token = os.environ.get("DANSWER_BOT_SLACK_BOT_TOKEN") + if not bot_token: + raise RuntimeError("DANSWER_BOT_SLACK_BOT_TOKEN is not set") + return SocketModeClient( + # This app-level token will be used only for establishing a connection + app_token=app_token, + web_client=WebClient(token=bot_token), + ) + + +def _process_slack_event(client: SocketModeClient, req: SocketModeRequest) -> None: + logger.info(f"Received request of type: '{req.type}', with paylod: '{req.payload}'") + if req.type == "events_api": + # Acknowledge the request immediately + response = SocketModeResponse(envelope_id=req.envelope_id) + client.send_socket_mode_response(response) + + event = cast(dict[str, Any], req.payload.get("event", {})) + channel = cast(str | None, event.get("channel")) + channel_specific_logger = _ChannelIdAdapter( + logger, extra={_CHANNEL_ID: channel} + ) + + # Ensure that the message is a new message + of expected type + event_type = event.get("type") + if event_type != "message": + channel_specific_logger.info( + f"Ignoring non-message event of type '{event_type}' for channel '{channel}'" + ) + + # this should never happen, but we can't continue without a channel since + # we can't send a response without it + if not channel: + channel_specific_logger.error(f"Found message without channel - skipping") + return + + message_subtype = event.get("subtype") + # ignore things like channel_join, channel_leave, etc. + # NOTE: "file_share" is just a message with a file attachment, so we + # should not ignore it + if message_subtype not in [None, "file_share"]: + channel_specific_logger.info( + f"Ignoring message with subtype '{message_subtype}' since is is a special message type" + ) + return + + if event.get("bot_profile"): + channel_specific_logger.info("Ignoring message from bot") + return + + message_ts = event.get("ts") + thread_ts = event.get("thread_ts") + # pick the root of the thread (if a thread exists) + message_ts_to_respond_to = cast(str, thread_ts or message_ts) + if thread_ts and message_ts != thread_ts: + channel_specific_logger.info( + "Skipping message since it is not the root of a thread" + ) + return + + msg = cast(str | None, event.get("text")) + if not msg: + channel_specific_logger.error("Unable to process empty message") + return + + # TODO: message should be enqueued and processed elsewhere, + # but doing it here for now for simplicity + handle_message( + msg=msg, + channel=channel, + message_ts_to_respond_to=message_ts_to_respond_to, + client=client.web_client, + logger=cast(logging.Logger, channel_specific_logger), + ) + + channel_specific_logger.info( + f"Successfully processed message with ts: '{message_ts}'" + ) + + # handle button clicks + if req.type == "interactive" and req.payload.get("type") == "block_actions": + # Acknowledge the request immediately + response = SocketModeResponse(envelope_id=req.envelope_id) + client.send_socket_mode_response(response) + + actions = req.payload.get("actions") + if not actions: + logger.error("Unable to process block actions - no actions found") + return + + action = cast(dict[str, Any], actions[0]) + action_id = action.get("action_id") + if action_id == LIKE_BLOCK_ACTION_ID: + feedback_type = QAFeedbackType.LIKE + elif action_id == DISLIKE_BLOCK_ACTION_ID: + feedback_type = QAFeedbackType.DISLIKE + else: + logger.error( + f"Unable to process block action - unknown action_id: '{action_id}'" + ) + return + + block_id = cast(str, action.get("block_id")) + query_event_id = get_query_event_id_from_block_id(block_id) + handle_qa_feedback( + query_id=query_event_id, + feedback_type=feedback_type, + ) + + logger.info(f"Successfully handled QA feedback for event: {query_event_id}") + + +def process_slack_event(client: SocketModeClient, req: SocketModeRequest) -> None: + try: + _process_slack_event(client=client, req=req) + except Exception: + logger.exception("Failed to process slack event") + + +# Follow the guide (https://docs.danswer.dev/slack_bot_setup) to set up +# the slack bot in your workspace, and then add the bot to any channels you want to +# try and answer questions for. Running this file will setup Danswer to listen to all +# messages in those channels and attempt to answer them. As of now, it will only respond +# to messages sent directly in the channel - it will not respond to messages sent within a +# thread. +# +# NOTE: we are using Web Sockets so that you can run this from within a firewalled VPC +# without issue. +if __name__ == "__main__": + socket_client = _get_socket_client() + socket_client.socket_mode_request_listeners.append(process_slack_event) # type: ignore + # Establish a WebSocket connection to the Socket Mode servers + logger.info("Listening for messages from Slack...") + socket_client.connect() + + # Just not to stop this process + from threading import Event + + Event().wait() diff --git a/backend/danswer/bots/slack/utils.py b/backend/danswer/bots/slack/utils.py new file mode 100644 index 000000000..b85c058bd --- /dev/null +++ b/backend/danswer/bots/slack/utils.py @@ -0,0 +1,58 @@ +import logging +import random +import string +from typing import cast + +from retry import retry +from slack_sdk import WebClient +from slack_sdk.models.blocks import Block +from slack_sdk.models.metadata import Metadata + +from danswer.configs.app_configs import DANSWER_BOT_NUM_RETRIES +from danswer.connectors.slack.utils import make_slack_api_rate_limited +from danswer.utils.logger import setup_logger + + +logger = setup_logger() + + +@retry( + tries=DANSWER_BOT_NUM_RETRIES, + delay=0.25, + backoff=2, + logger=cast(logging.Logger, logger), +) +def respond_in_thread( + client: WebClient, + channel: str, + thread_ts: str, + text: str | None = None, + blocks: list[Block] | None = None, + metadata: Metadata | None = None, +) -> None: + if not text and not blocks: + raise ValueError("One of `text` or `blocks` must be provided") + + if text: + logger.info(f"Trying to send message: {text}") + if blocks: + logger.info(f"Trying to send blocks: {blocks}") + + slack_call = make_slack_api_rate_limited(client.chat_postMessage) + response = slack_call( + channel=channel, + text=text, + blocks=blocks, + thread_ts=thread_ts, + metadata=metadata, + ) + if not response.get("ok"): + raise RuntimeError(f"Unable to post message: {response}") + + +def build_block_id_from_query_event_id(query_event_id: int) -> str: + return f"{''.join(random.choice(string.ascii_letters) for _ in range(5))}:{query_event_id}" + + +def get_query_event_id_from_block_id(block_id: str) -> int: + return int(block_id.split(":")[-1]) diff --git a/backend/danswer/db/feedback.py b/backend/danswer/db/feedback.py index adc00546e..d52fe4642 100644 --- a/backend/danswer/db/feedback.py +++ b/backend/danswer/db/feedback.py @@ -27,15 +27,15 @@ def fetch_query_event_by_id(query_id: int, db_session: Session) -> QueryEvent: return query_event -def fetch_doc_m_by_id(doc_id: str, db_session: Session) -> DbDocument: +def fetch_docs_by_id(doc_id: str, db_session: Session) -> DbDocument: stmt = select(DbDocument).where(DbDocument.id == doc_id) result = db_session.execute(stmt) - doc_m = result.scalar_one_or_none() + doc = result.scalar_one_or_none() - if not doc_m: + if not doc: raise ValueError("Invalid Document provided for updating") - return doc_m + return doc def fetch_docs_ranked_by_boost( @@ -44,29 +44,19 @@ def fetch_docs_ranked_by_boost( order_func = asc if ascending else desc stmt = select(DbDocument).order_by(order_func(DbDocument.boost)).limit(limit) result = db_session.execute(stmt) - doc_m_list = result.scalars().all() + doc_list = result.scalars().all() - return list(doc_m_list) + return list(doc_list) -def create_document_metadata( - doc_id: str, - semantic_id: str, - link: str | None, - db_session: Session, -) -> None: - try: - fetch_doc_m_by_id(doc_id, db_session) - return - except ValueError: - # Document already exists, don't reset its data - pass +def update_document_boost(db_session: Session, document_id: str, boost: int) -> None: + stmt = select(DbDocument).where(DbDocument.id == document_id) + result = db_session.execute(stmt).scalar_one_or_none() + if result is None: + raise ValueError(f"No document found with ID: '{document_id}'") - DbDocument( - id=doc_id, - semantic_id=semantic_id, - link=link, - ) + result.boost = boost + db_session.commit() def create_query_event( @@ -121,7 +111,7 @@ def create_doc_retrieval_feedback( if user_id != query_event.user_id: raise ValueError("User trying to give feedback on a query run by another user.") - doc_m = fetch_doc_m_by_id(document_id, db_session) + doc_m = fetch_docs_by_id(document_id, db_session) retrieval_feedback = DocumentRetrievalFeedback( qa_event_id=qa_event_id, diff --git a/backend/danswer/listeners/slack_listener.py b/backend/danswer/listeners/slack_listener.py deleted file mode 100644 index 0e5ab0a67..000000000 --- a/backend/danswer/listeners/slack_listener.py +++ /dev/null @@ -1,332 +0,0 @@ -import logging -import os -from collections.abc import MutableMapping -from typing import Any -from typing import cast - -from retry import retry -from slack_sdk import WebClient -from slack_sdk.socket_mode import SocketModeClient -from slack_sdk.socket_mode.request import SocketModeRequest -from slack_sdk.socket_mode.response import SocketModeResponse -from sqlalchemy.orm import Session - -from danswer.configs.app_configs import DANSWER_BOT_ANSWER_GENERATION_TIMEOUT -from danswer.configs.app_configs import DANSWER_BOT_DISABLE_DOCS_ONLY_ANSWER -from danswer.configs.app_configs import DANSWER_BOT_DISPLAY_ERROR_MSGS -from danswer.configs.app_configs import DANSWER_BOT_NUM_DOCS_TO_DISPLAY -from danswer.configs.app_configs import DANSWER_BOT_NUM_RETRIES -from danswer.configs.app_configs import DOCUMENT_INDEX_NAME -from danswer.configs.constants import DocumentSource -from danswer.connectors.slack.utils import make_slack_api_rate_limited -from danswer.connectors.slack.utils import UserIdReplacer -from danswer.db.engine import get_sqlalchemy_engine -from danswer.direct_qa.answer_question import answer_qa_query -from danswer.direct_qa.interfaces import DanswerQuote -from danswer.server.models import QAResponse -from danswer.server.models import QuestionRequest -from danswer.server.models import SearchDoc -from danswer.utils.logger import setup_logger - -logger = setup_logger() - - -_CHANNEL_ID = "channel_id" - - -class _ChannelIdAdapter(logging.LoggerAdapter): - """This is used to add the channel ID to all log messages - emitted in this file""" - - def process( - self, msg: str, kwargs: MutableMapping[str, Any] - ) -> tuple[str, MutableMapping[str, Any]]: - channel_id = self.extra.get(_CHANNEL_ID) if self.extra else None - if channel_id: - return f"[Channel ID: {channel_id}] {msg}", kwargs - else: - return msg, kwargs - - -def _get_socket_client() -> SocketModeClient: - # For more info on how to set this up, checkout the docs: - # https://docs.danswer.dev/slack_bot_setup - app_token = os.environ.get("DANSWER_BOT_SLACK_APP_TOKEN") - if not app_token: - raise RuntimeError("DANSWER_BOT_SLACK_APP_TOKEN is not set") - bot_token = os.environ.get("DANSWER_BOT_SLACK_BOT_TOKEN") - if not bot_token: - raise RuntimeError("DANSWER_BOT_SLACK_BOT_TOKEN is not set") - return SocketModeClient( - # This app-level token will be used only for establishing a connection - app_token=app_token, - web_client=WebClient(token=bot_token), - ) - - -_MAX_BLURB_LEN = 25 - - -def _build_custom_semantic_identifier( - semantic_identifier: str, blurb: str, source: str -) -> str: - """ - On slack, since we just show the semantic identifier rather than semantic + blurb, we need - to do some custom formatting to make sure the semantic identifier is unique and meaningful. - """ - if source == DocumentSource.SLACK.value: - truncated_blurb = ( - f"{blurb[:_MAX_BLURB_LEN]}..." if len(blurb) > _MAX_BLURB_LEN else blurb - ) - # NOTE: removing tags so that we don't accidentally tag users in Slack + - # so that it can be used as part of a link - truncated_blurb = UserIdReplacer.replace_tags_basic(truncated_blurb) - truncated_blurb = UserIdReplacer.replace_channels_basic(truncated_blurb) - truncated_blurb = UserIdReplacer.replace_special_mentions(truncated_blurb) - if truncated_blurb: - return f"#{semantic_identifier}: {truncated_blurb}" - else: - return f"#{semantic_identifier}" - - return semantic_identifier - - -def _process_quotes(quotes: list[DanswerQuote] | None) -> tuple[str | None, list[str]]: - if not quotes: - return None, [] - - quote_lines: list[str] = [] - doc_identifiers: list[str] = [] - for quote in quotes: - doc_id = quote.document_id - doc_link = quote.link - doc_name = quote.semantic_identifier - if doc_link and doc_name and doc_id and doc_id not in doc_identifiers: - doc_identifiers.append(doc_id) - custom_semantic_identifier = _build_custom_semantic_identifier( - semantic_identifier=doc_name, - blurb=quote.blurb, - source=quote.source_type, - ) - quote_lines.append(f"- <{doc_link}|{custom_semantic_identifier}>") - - if not quote_lines: - return None, [] - - return "\n".join(quote_lines), doc_identifiers - - -def _process_documents( - documents: list[SearchDoc] | None, - already_displayed_doc_identifiers: list[str], - num_docs_to_display: int = DANSWER_BOT_NUM_DOCS_TO_DISPLAY, -) -> str | None: - if not documents: - return None - - seen_docs_identifiers = set(already_displayed_doc_identifiers) - top_document_lines: list[str] = [] - for d in documents: - if d.document_id in seen_docs_identifiers: - continue - seen_docs_identifiers.add(d.document_id) - - custom_semantic_identifier = _build_custom_semantic_identifier( - semantic_identifier=d.semantic_identifier, - blurb=d.blurb, - source=d.source_type, - ) - - top_document_lines.append(f"- <{d.link}|{custom_semantic_identifier}>") - if len(top_document_lines) >= num_docs_to_display: - break - - return "\n".join(top_document_lines) - - -@retry( - tries=DANSWER_BOT_NUM_RETRIES, - delay=0.25, - backoff=2, - logger=cast(logging.Logger, logger), -) -def _respond_in_thread( - client: SocketModeClient, - channel: str, - text: str, - thread_ts: str, -) -> None: - logger.info(f"Trying to send message: {text}") - slack_call = make_slack_api_rate_limited(client.web_client.chat_postMessage) - response = slack_call( - channel=channel, - text=text, - thread_ts=thread_ts, - ) - if not response.get("ok"): - raise RuntimeError(f"Unable to post message: {response}") - - -def process_slack_event(client: SocketModeClient, req: SocketModeRequest) -> None: - if req.type == "events_api": - # Acknowledge the request anyway - response = SocketModeResponse(envelope_id=req.envelope_id) - client.send_socket_mode_response(response) - - event = cast(dict[str, Any], req.payload.get("event", {})) - channel = cast(str | None, event.get("channel")) - channel_specific_logger = _ChannelIdAdapter( - logger, extra={_CHANNEL_ID: channel} - ) - - # Ensure that the message is a new message + of expected type - event_type = event.get("type") - if event_type != "message": - channel_specific_logger.info( - f"Ignoring non-message event of type '{event_type}' for channel '{channel}'" - ) - - # this should never happen, but we can't continue without a channel since - # we can't send a response without it - if not channel: - channel_specific_logger.error(f"Found message without channel - skipping") - return - - message_subtype = event.get("subtype") - # ignore things like channel_join, channel_leave, etc. - # NOTE: "file_share" is just a message with a file attachment, so we - # should not ignore it - if message_subtype not in [None, "file_share"]: - channel_specific_logger.info( - f"Ignoring message with subtype '{message_subtype}' since is is a special message type" - ) - return - - if event.get("bot_profile"): - channel_specific_logger.info("Ignoring message from bot") - return - - message_ts = event.get("ts") - thread_ts = event.get("thread_ts") - # pick the root of the thread (if a thread exists) - message_ts_to_respond_to = cast(str, thread_ts or message_ts) - if thread_ts and message_ts != thread_ts: - channel_specific_logger.info( - "Skipping message since it is not the root of a thread" - ) - return - - msg = cast(str | None, event.get("text")) - if not msg: - channel_specific_logger.error("Unable to process empty message") - return - - # TODO: message should be enqueued and processed elsewhere, - # but doing it here for now for simplicity - - @retry( - tries=DANSWER_BOT_NUM_RETRIES, - delay=0.25, - backoff=2, - logger=cast(logging.Logger, logger), - ) - def _get_answer(question: QuestionRequest) -> QAResponse: - engine = get_sqlalchemy_engine() - with Session(engine, expire_on_commit=False) as db_session: - answer = answer_qa_query( - question=question, - user=None, - db_session=db_session, - answer_generation_timeout=DANSWER_BOT_ANSWER_GENERATION_TIMEOUT, - ) - if not answer.error_msg: - return answer - else: - raise RuntimeError(answer.error_msg) - - try: - answer = _get_answer( - QuestionRequest( - query=msg, - collection=DOCUMENT_INDEX_NAME, - use_keyword=None, - filters=None, - offset=None, - ) - ) - except Exception as e: - channel_specific_logger.exception( - f"Unable to process message - did not successfully answer " - f"in {DANSWER_BOT_NUM_RETRIES} attempts" - ) - # Optionally, respond in thread with the error message, Used primarily - # for debugging purposes - if DANSWER_BOT_DISPLAY_ERROR_MSGS: - _respond_in_thread( - client=client, - channel=channel, - text=f"Encountered exception when trying to answer: \n\n```{e}```", - thread_ts=message_ts_to_respond_to, - ) - return - - # convert raw response into "nicely" formatted Slack message - quote_str, doc_identifiers = _process_quotes(answer.quotes) - top_documents_str = _process_documents(answer.top_ranked_docs, doc_identifiers) - - if not answer.answer: - if DANSWER_BOT_DISABLE_DOCS_ONLY_ANSWER: - logger.info( - "Unable to find answer - not responding since the " - "`DANSWER_BOT_DISABLE_DOCS_ONLY_ANSWER` env variable is set" - ) - return - - text = f"Sorry, I was unable to find an answer, but I did find some potentially relevant docs 🤓\n\n{top_documents_str}" - else: - top_documents_str_with_header = ( - f"*Other potentially relevant docs:*\n{top_documents_str}" - ) - if quote_str: - text = f"{answer.answer}\n\n*Sources:*\n{quote_str}\n\n{top_documents_str_with_header}" - else: - text = f"{answer.answer}\n\n*Warning*: no sources were quoted for this answer, so it may be unreliable 😔\n\n{top_documents_str_with_header}" - - try: - _respond_in_thread( - client=client, - channel=channel, - text=text, - thread_ts=message_ts_to_respond_to, - ) - except Exception: - channel_specific_logger.exception( - f"Unable to process message - could not respond in slack in {DANSWER_BOT_NUM_RETRIES} attempts" - ) - return - - channel_specific_logger.info( - f"Successfully processed message with ts: '{message_ts}'" - ) - - -# Follow the guide (https://docs.danswer.dev/slack_bot_setup) to set up -# the slack bot in your workspace, and then add the bot to any channels you want to -# try and answer questions for. Running this file will setup Danswer to listen to all -# messages in those channels and attempt to answer them. As of now, it will only respond -# to messages sent directly in the channel - it will not respond to messages sent within a -# thread. -# -# NOTE: we are using Web Sockets so that you can run this from within a firewalled VPC -# without issue. -if __name__ == "__main__": - socket_client = _get_socket_client() - socket_client.socket_mode_request_listeners.append(process_slack_event) # type: ignore - # Establish a WebSocket connection to the Socket Mode servers - logger.info("Listening for messages from Slack...") - socket_client.connect() - - # Just not to stop this process - from threading import Event - - Event().wait() diff --git a/backend/danswer/search/semantic_search.py b/backend/danswer/search/semantic_search.py index f7ea4a462..c6122b9b4 100644 --- a/backend/danswer/search/semantic_search.py +++ b/backend/danswer/search/semantic_search.py @@ -85,6 +85,7 @@ def retrieve_ranked_documents( f"Semantic search returned no results with filters: {filters_log_msg}" ) return None, None + logger.info(top_chunks) ranked_chunks = semantic_reranking(query, top_chunks[:num_rerank]) top_docs = [ diff --git a/backend/danswer/server/manage.py b/backend/danswer/server/manage.py index a4d897083..56dfb817d 100644 --- a/backend/danswer/server/manage.py +++ b/backend/danswer/server/manage.py @@ -50,6 +50,7 @@ from danswer.db.deletion_attempt import create_deletion_attempt from danswer.db.deletion_attempt import get_deletion_attempts from danswer.db.engine import get_session from danswer.db.feedback import fetch_docs_ranked_by_boost +from danswer.db.feedback import update_document_boost from danswer.db.index_attempt import create_index_attempt from danswer.db.index_attempt import get_latest_index_attempts from danswer.db.models import DeletionAttempt @@ -63,6 +64,7 @@ from danswer.server.models import ApiKey from danswer.server.models import AuthStatus from danswer.server.models import AuthUrl from danswer.server.models import BoostDoc +from danswer.server.models import BoostUpdateRequest from danswer.server.models import ConnectorBase from danswer.server.models import ConnectorCredentialPairIdentifier from danswer.server.models import ConnectorIndexingStatus @@ -104,6 +106,7 @@ def get_most_boosted_docs( BoostDoc( document_id=doc.id, semantic_id=doc.semantic_id, + # source=doc.source, link=doc.link or "", boost=doc.boost, hidden=doc.hidden, @@ -112,6 +115,22 @@ def get_most_boosted_docs( ] +@router.post("/admin/doc-boosts") +def document_boost_update( + boost_update: BoostUpdateRequest, + _: User | None = Depends(current_admin_user), + db_session: Session = Depends(get_session), +) -> None: + try: + update_document_boost( + db_session=db_session, + document_id=boost_update.document_id, + boost=boost_update.boost, + ) + except ValueError as e: + raise HTTPException(status_code=400, detail=str(e)) + + @router.get("/admin/connector/google-drive/app-credential") def check_google_app_credentials_exist( _: User = Depends(current_admin_user), diff --git a/backend/danswer/server/models.py b/backend/danswer/server/models.py index 1052b7e23..aee909844 100644 --- a/backend/danswer/server/models.py +++ b/backend/danswer/server/models.py @@ -115,6 +115,11 @@ class BoostDoc(BaseModel): hidden: bool +class BoostUpdateRequest(BaseModel): + document_id: str + boost: int + + class SearchDoc(BaseModel): document_id: str semantic_identifier: str diff --git a/backend/supervisord.conf b/backend/supervisord.conf index b42240619..32e4debe7 100644 --- a/backend/supervisord.conf +++ b/backend/supervisord.conf @@ -29,7 +29,7 @@ autorestart=true # If not setup, this will just fail 5 times and then stop. # More details on setup here: https://docs.danswer.dev/slack_bot_setup [program:slack_bot_listener] -command=python danswer/listeners/slack_listener.py +command=python danswer/bots/slack/listener.py stdout_logfile=/var/log/slack_bot_listener.log stdout_logfile_maxbytes=52428800 redirect_stderr=true diff --git a/web/src/app/admin/connectors/google-drive/GoogleDriveConnectorsTable.tsx b/web/src/app/admin/connectors/google-drive/GoogleDriveConnectorsTable.tsx index da88e0bff..803d9fc70 100644 --- a/web/src/app/admin/connectors/google-drive/GoogleDriveConnectorsTable.tsx +++ b/web/src/app/admin/connectors/google-drive/GoogleDriveConnectorsTable.tsx @@ -2,7 +2,7 @@ import { Button } from "@/components/Button"; import { BasicTable } from "@/components/admin/connectors/BasicTable"; import { PopupSpec } from "@/components/admin/connectors/Popup"; import { StatusRow } from "@/components/admin/connectors/table/ConnectorsTable"; -import { PencilIcon } from "@/components/icons/icons"; +import { EditIcon } from "@/components/icons/icons"; import { deleteConnector } from "@/lib/connector"; import { GoogleDriveConfig, @@ -44,7 +44,7 @@ const EditableColumn = ({ connectorIndexingStatus }: EditableColumnProps) => { className="cursor-pointer" >
- +
diff --git a/web/src/app/admin/documents/feedback/page.tsx b/web/src/app/admin/documents/feedback/page.tsx new file mode 100644 index 000000000..d2505225b --- /dev/null +++ b/web/src/app/admin/documents/feedback/page.tsx @@ -0,0 +1,239 @@ +"use client"; + +import { LoadingAnimation } from "@/components/Loading"; +import { PageSelector } from "@/components/PageSelector"; +import { BasicTable } from "@/components/admin/connectors/BasicTable"; +import { PopupSpec, usePopup } from "@/components/admin/connectors/Popup"; +import { + CheckmarkIcon, + EditIcon, + ThumbsUpIcon, +} from "@/components/icons/icons"; +import { useMostReactedToDocuments } from "@/lib/hooks"; +import { DocumentBoostStatus, User } from "@/lib/types"; +import { useState } from "react"; + +const numPages = 8; +const numToDisplay = 10; + +const updateBoost = async (documentId: string, boost: number) => { + const response = await fetch("/api/manage/admin/doc-boosts", { + method: "POST", + headers: { + "Content-Type": "application/json", + }, + body: JSON.stringify({ + document_id: documentId, + boost, + }), + }); + if (response.ok) { + return null; + } + const responseJson = await response.json(); + return responseJson.message || responseJson.detail || "Unknown error"; +}; + +const ScoreSection = ({ + documentId, + initialScore, + setPopup, + refresh, +}: { + documentId: string; + initialScore: number; + setPopup: (popupSpec: PopupSpec | null) => void; + refresh: () => void; +}) => { + const [isOpen, setIsOpen] = useState(false); + const [score, setScore] = useState(initialScore); + + const onSubmit = async () => { + const errorMsg = await updateBoost(documentId, score); + if (errorMsg) { + setPopup({ + message: errorMsg, + type: "error", + }); + } else { + setPopup({ + message: "Updated score!", + type: "success", + }); + refresh(); + setIsOpen(false); + } + }; + + if (isOpen) { + return ( +
+ { + if (!isNaN(Number(e.target.value))) { + setScore(Number(e.target.value)); + } + }} + onKeyDown={(e) => { + if (e.key === "Enter") { + onSubmit(); + } + }} + className="border bg-slate-700 text-gray-200 border-gray-300 rounded py-1 px-3 w-16" + /> +
+ +
+
+ ); + } + + return ( +
+
+
+
{initialScore}
+
+
setIsOpen(true)}> + +
+
+
+ ); +}; + +interface DocumentFeedbackTableProps { + documents: DocumentBoostStatus[]; + refresh: () => void; +} + +const DocumentFeedbackTable = ({ + documents, + refresh, +}: DocumentFeedbackTableProps) => { + const [page, setPage] = useState(1); + const { popup, setPopup } = usePopup(); + + return ( +
+ {popup} + { + return { + name: ( + + {document.semantic_id} + + ), + score: ( +
+
+ +
+
+ ), + }; + })} + /> +
+
+ setPage(newPage)} + /> +
+
+
+ ); +}; + +const Main = () => { + const { + data: mostLikedDocuments, + isLoading: isMostLikedDocumentsLoading, + error: mostLikedDocumentsError, + refreshDocs: refreshMostLikedDocuments, + } = useMostReactedToDocuments(false, numToDisplay * numPages); + + const { + data: mostDislikedDocuments, + isLoading: isMostLikedDocumentLoading, + error: mostDislikedDocumentsError, + refreshDocs: refreshMostDislikedDocuments, + } = useMostReactedToDocuments(true, numToDisplay * numPages); + + const refresh = () => { + refreshMostLikedDocuments(); + refreshMostDislikedDocuments(); + }; + + if (isMostLikedDocumentsLoading || isMostLikedDocumentLoading) { + return ; + } + + if ( + mostLikedDocumentsError || + mostDislikedDocumentsError || + !mostLikedDocuments || + !mostDislikedDocuments + ) { + return ( +
+ Error loading documents -{" "} + {mostDislikedDocumentsError || mostLikedDocumentsError} +
+ ); + } + + return ( +
+

Most Liked Documents

+ + +

Most Disliked Documents

+ +
+ ); +}; + +const Page = () => { + return ( +
+
+ +

Document Feedback

+
+ +
+
+ ); +}; + +export default Page; diff --git a/web/src/app/admin/layout.tsx b/web/src/app/admin/layout.tsx index 761b3dd04..7c9edba2d 100644 --- a/web/src/app/admin/layout.tsx +++ b/web/src/app/admin/layout.tsx @@ -18,6 +18,7 @@ import { ProductboardIcon, LinearIcon, UsersIcon, + ThumbsUpIcon, } from "@/components/icons/icons"; import { DISABLE_AUTH } from "@/lib/constants"; import { getCurrentUserSS } from "@/lib/userSS"; @@ -219,6 +220,20 @@ export default async function AdminLayout({ }, ], }, + { + name: "Document Management", + items: [ + { + name: ( +
+ +
Feedback
+
+ ), + link: "/admin/documents/feedback", + }, + ], + }, ]} />
diff --git a/web/src/components/PageSelector.tsx b/web/src/components/PageSelector.tsx new file mode 100644 index 000000000..78d296eff --- /dev/null +++ b/web/src/components/PageSelector.tsx @@ -0,0 +1,149 @@ +import React from "react"; + +const PAGINATION_OPTIONS_ON_EACH_SIDE = 2; + +const getPaginationOptions = ( + currentPage: number, + pageCount: number +): number[] => { + const paginationOptions = [currentPage]; + // if (currentPage !== 1) { + // paginationOptions.push(currentPage) + // } + + let offset = 1; + + // Add one because currentPage is included + const maxPaginationOptions = PAGINATION_OPTIONS_ON_EACH_SIDE * 2 + 1; + while (paginationOptions.length < maxPaginationOptions) { + let added = false; + if (currentPage + offset <= pageCount) { + paginationOptions.push(currentPage + offset); + added = true; + } + if (currentPage - offset >= 1) { + paginationOptions.unshift(currentPage - offset); + added = true; + } + if (!added) { + break; + } + offset++; + } + + return paginationOptions; +}; + +const scrollUp = () => { + setTimeout(() => window.scrollTo({ top: 0 }), 50); +}; + +type PageLinkProps = { + linkText: string | number; + pageChangeHandler?: () => void; + active?: boolean; + unclickable?: boolean; +}; + +const PageLink = ({ + linkText, + pageChangeHandler, + active, + unclickable, +}: PageLinkProps) => ( +
{ + if (pageChangeHandler) { + pageChangeHandler(); + } + }} + > + {linkText} +
+); + +interface PageSelectorProps { + currentPage: number; + totalPages: number; + onPageChange: (newPage: number) => void; + shouldScroll?: boolean; +} + +export const PageSelector = ({ + currentPage, + totalPages, + onPageChange, + shouldScroll = false, +}: PageSelectorProps) => { + const paginationOptions = getPaginationOptions(currentPage, totalPages); + const modifiedScrollUp = () => { + if (shouldScroll) { + scrollUp(); + } + }; + + return ( +
+ { + onPageChange(Math.max(currentPage - 1, 1)); + modifiedScrollUp(); + }} + /> + {!paginationOptions.includes(1) && ( + <> + { + onPageChange(1); + modifiedScrollUp(); + }} + /> + + + )} + {(!paginationOptions.includes(1) + ? paginationOptions.slice(2) + : paginationOptions + ).map((page) => { + return ( + { + onPageChange(page); + modifiedScrollUp(); + }} + /> + ); + })} + { + onPageChange(Math.min(currentPage + 1, totalPages)); + modifiedScrollUp(); + }} + /> +
+ ); +}; diff --git a/web/src/components/admin/connectors/BasicTable.tsx b/web/src/components/admin/connectors/BasicTable.tsx index 3cf00b044..e3e4600bb 100644 --- a/web/src/components/admin/connectors/BasicTable.tsx +++ b/web/src/components/admin/connectors/BasicTable.tsx @@ -3,7 +3,7 @@ import React, { FC } from "react"; type Column = { header: string; key: string; - width?: number; + width?: number | string; }; type TableData = { @@ -25,10 +25,10 @@ export const BasicTable: FC = ({ columns, data }) => { {column.header} @@ -44,8 +44,8 @@ export const BasicTable: FC = ({ columns, data }) => { {row[column.key]} diff --git a/web/src/components/icons/icons.tsx b/web/src/components/icons/icons.tsx index 60ff4eeb9..00f093fb3 100644 --- a/web/src/components/icons/icons.tsx +++ b/web/src/components/icons/icons.tsx @@ -14,7 +14,10 @@ import { X, Question, Users, + ThumbsUp, + ThumbsDown, } from "@phosphor-icons/react"; +import { FiCheck, FiEdit, FiThumbsDown, FiThumbsUp } from "react-icons/fi"; import { SiBookstack } from "react-icons/si"; import { FaFile, FaGlobe } from "react-icons/fa"; import Image from "next/image"; @@ -122,11 +125,11 @@ export const BrainIcon = ({ return ; }; -export const PencilIcon = ({ +export const EditIcon = ({ size = 16, className = defaultTailwindCSS, }: IconProps) => { - return ; + return ; }; export const XIcon = ({ @@ -136,6 +139,27 @@ export const XIcon = ({ return ; }; +export const ThumbsUpIcon = ({ + size = 16, + className = defaultTailwindCSS, +}: IconProps) => { + return ; +}; + +export const ThumbsDownIcon = ({ + size = 16, + className = defaultTailwindCSS, +}: IconProps) => { + return ; +}; + +export const CheckmarkIcon = ({ + size = 16, + className = defaultTailwindCSS, +}: IconProps) => { + return ; +}; + // // COMPANY LOGOS // diff --git a/web/src/components/search/DocumentDisplay.tsx b/web/src/components/search/DocumentDisplay.tsx new file mode 100644 index 000000000..c4a01606c --- /dev/null +++ b/web/src/components/search/DocumentDisplay.tsx @@ -0,0 +1,57 @@ +import { DanswerDocument } from "@/lib/search/interfaces"; +import { DocumentFeedbackBlock } from "./DocumentFeedbackBlock"; +import { getSourceIcon } from "../source"; +import { useState } from "react"; +import { usePopup } from "../admin/connectors/Popup"; + +interface DocumentDisplayProps { + document: DanswerDocument; + queryEventId: number | null; +} + +export const DocumentDisplay = ({ + document, + queryEventId, +}: DocumentDisplayProps) => { + const { popup, setPopup } = usePopup(); + const [isHovered, setIsHovered] = useState(false); + + return ( +
{ + setIsHovered(true); + }} + onMouseLeave={() => setIsHovered(false)} + > + {popup} + +

{document.blurb}

+
+ ); +}; diff --git a/web/src/components/search/DocumentFeedbackBlock.tsx b/web/src/components/search/DocumentFeedbackBlock.tsx new file mode 100644 index 000000000..df076f7cc --- /dev/null +++ b/web/src/components/search/DocumentFeedbackBlock.tsx @@ -0,0 +1,123 @@ +import { PopupSpec } from "../admin/connectors/Popup"; +import { ThumbsDownIcon, ThumbsUpIcon } from "../icons/icons"; + +type DocumentFeedbackType = "endorse" | "reject" | "hide" | "unhide"; + +const giveDocumentFeedback = async ( + documentId: string, + queryId: number, + searchFeedback: DocumentFeedbackType +): Promise => { + const response = await fetch("/api/doc-retrieval-feedback", { + method: "POST", + headers: { + "Content-Type": "application/json", + }, + body: JSON.stringify({ + query_id: queryId, + search_feedback: searchFeedback, + click: false, + document_rank: 0, + document_id: documentId, + }), + }); + return response.ok + ? null + : response.statusText || (await response.json()).message; +}; + +interface DocumentFeedbackIconProps { + documentId: string; + queryId: number; + setPopup: (popupSpec: PopupSpec | null) => void; + feedbackType: DocumentFeedbackType; +} + +const DocumentFeedback = ({ + documentId, + queryId, + setPopup, + feedbackType, +}: DocumentFeedbackIconProps) => { + let icon = null; + const size = 16; + if (feedbackType === "endorse") { + icon = ( + + ); + } + if (feedbackType === "reject") { + icon = ( + + ); + } + if (!icon) { + // TODO: support other types of feedback + return null; + } + + return ( +
{ + console.log("HI"); + const errorMsg = await giveDocumentFeedback( + documentId, + queryId, + feedbackType + ); + console.log(errorMsg); + if (!errorMsg) { + setPopup({ + message: "Thanks for your feedback!", + type: "success", + }); + } else { + setPopup({ + message: `Error giving feedback - ${errorMsg}`, + type: "error", + }); + } + }} + className="cursor-pointer" + > + {icon} +
+ ); +}; + +interface DocumentFeedbackBlockProps { + documentId: string; + queryId: number; + setPopup: (popupSpec: PopupSpec | null) => void; +} + +export const DocumentFeedbackBlock = ({ + documentId, + queryId, + setPopup, +}: DocumentFeedbackBlockProps) => { + return ( +
+ +
+ +
+
+ ); +}; diff --git a/web/src/components/search/QAFeedback.tsx b/web/src/components/search/QAFeedback.tsx new file mode 100644 index 000000000..e1aabea76 --- /dev/null +++ b/web/src/components/search/QAFeedback.tsx @@ -0,0 +1,92 @@ +import { useState } from "react"; +import { PopupSpec, usePopup } from "../admin/connectors/Popup"; +import { ThumbsDownIcon, ThumbsUpIcon } from "../icons/icons"; + +type Feedback = "like" | "dislike"; + +const giveFeedback = async ( + queryId: number, + feedback: Feedback +): Promise => { + const response = await fetch("/api/query-feedback", { + method: "POST", + headers: { + "Content-Type": "application/json", + }, + body: JSON.stringify({ + query_id: queryId, + feedback, + }), + }); + return response.ok; +}; + +interface QAFeedbackIconProps { + queryId: number; + setPopup: (popupSpec: PopupSpec | null) => void; + feedbackType: Feedback; +} + +const QAFeedback = ({ + queryId, + setPopup, + feedbackType, +}: QAFeedbackIconProps) => { + const [isHovered, setIsHovered] = useState(false); + + const size = isHovered ? 22 : 20; + const paddingY = isHovered ? "" : "py-0.5 "; + + return ( +
{ + const isSuccessful = await giveFeedback(queryId, feedbackType); + if (isSuccessful) { + setPopup({ + message: "Thanks for your feedback!", + type: "success", + }); + } + }} + onMouseEnter={() => { + setIsHovered(true); + }} + onMouseLeave={() => setIsHovered(false)} + className={"cursor-pointer " + paddingY} + > + {feedbackType === "like" ? ( + + ) : ( + + )} +
+ ); +}; + +interface QAFeedbackBlockProps { + queryId: number; +} + +export const QAFeedbackBlock = ({ queryId }: QAFeedbackBlockProps) => { + const { popup, setPopup } = usePopup(); + + return ( +
+ {popup} + +
+ +
+
+ ); +}; diff --git a/web/src/components/search/SearchResultsDisplay.tsx b/web/src/components/search/SearchResultsDisplay.tsx index 854c1571b..afff8905f 100644 --- a/web/src/components/search/SearchResultsDisplay.tsx +++ b/web/src/components/search/SearchResultsDisplay.tsx @@ -9,6 +9,8 @@ import { FlowType, SearchDefaultOverrides, } from "@/lib/search/interfaces"; +import { QAFeedbackBlock } from "./QAFeedback"; +import { DocumentDisplay } from "./DocumentDisplay"; const removeDuplicateDocs = (documents: DanswerDocument[]) => { const seen = new Set(); @@ -37,7 +39,7 @@ export const SearchResultsDisplay: React.FC = ({ return null; } - const { answer, quotes, documents, error } = searchResponse; + const { answer, quotes, documents, error, queryEventId } = searchResponse; if (isFetching && !answer && !documents) { return ( @@ -132,6 +134,10 @@ export const SearchResultsDisplay: React.FC = ({
)} + +
+ +
)} @@ -145,27 +151,12 @@ export const SearchResultsDisplay: React.FC = ({
Results
- {removeDuplicateDocs(documents).map((doc) => ( -
- - {getSourceIcon(doc.source_type, 20)} -

- {doc.semantic_identifier || doc.document_id} -

- -

{doc.blurb}

-
+ {removeDuplicateDocs(documents).map((document) => ( + ))} )} diff --git a/web/src/components/search/SearchSection.tsx b/web/src/components/search/SearchSection.tsx index 32e0d4b34..f84f7e135 100644 --- a/web/src/components/search/SearchSection.tsx +++ b/web/src/components/search/SearchSection.tsx @@ -65,6 +65,7 @@ export const SearchSection: React.FC = ({ suggestedSearchType: null, suggestedFlowType: null, error: null, + queryEventId: null, }; const updateCurrentAnswer = (answer: string) => setSearchResponse((prevState) => ({ @@ -96,6 +97,11 @@ export const SearchSection: React.FC = ({ ...(prevState || initialSearchResponse), error, })); + const updateQueryEventId = (queryEventId: number) => + setSearchResponse((prevState) => ({ + ...(prevState || initialSearchResponse), + queryEventId, + })); let lastSearchCancellationToken = useRef(null); const onSearch = async ({ @@ -141,6 +147,10 @@ export const SearchSection: React.FC = ({ cancellationToken: lastSearchCancellationToken.current, fn: updateError, }), + updateQueryEventId: cancellable({ + cancellationToken: lastSearchCancellationToken.current, + fn: updateQueryEventId, + }), selectedSearchType: searchType ?? selectedSearchType, offset: offset ?? defaultOverrides.offset, }); diff --git a/web/src/lib/hooks.ts b/web/src/lib/hooks.ts index 0dd0237d6..35ecd7326 100644 --- a/web/src/lib/hooks.ts +++ b/web/src/lib/hooks.ts @@ -1,5 +1,5 @@ -import { Credential } from "@/lib/types"; -import useSWR, { useSWRConfig } from "swr"; +import { Credential, DocumentBoostStatus } from "@/lib/types"; +import useSWR, { mutate, useSWRConfig } from "swr"; import { fetcher } from "./fetcher"; const CREDENTIAL_URL = "/api/manage/admin/credential"; @@ -13,3 +13,22 @@ export const usePublicCredentials = () => { refreshCredentials: () => mutate(CREDENTIAL_URL), }; }; + +const MOST_REACTED_DOCS_URL = "/api/manage/doc-boosts"; + +const buildReactedDocsUrl = (ascending: boolean, limit: number) => { + return `/api/manage/admin/doc-boosts?ascending=${ascending}&limit=${limit}`; +}; + +export const useMostReactedToDocuments = ( + ascending: boolean, + limit: number +) => { + const url = buildReactedDocsUrl(ascending, limit); + const swrResponse = useSWR(url, fetcher); + + return { + ...swrResponse, + refreshDocs: () => mutate(url), + }; +}; diff --git a/web/src/lib/search/interfaces.ts b/web/src/lib/search/interfaces.ts index 02bc8b004..379226c6a 100644 --- a/web/src/lib/search/interfaces.ts +++ b/web/src/lib/search/interfaces.ts @@ -36,6 +36,7 @@ export interface SearchResponse { quotes: Quote[] | null; documents: DanswerDocument[] | null; error: string | null; + queryEventId: number | null; } export interface Source { @@ -57,6 +58,7 @@ export interface SearchRequestArgs { updateSuggestedSearchType: (searchType: SearchType) => void; updateSuggestedFlowType: (flowType: FlowType) => void; updateError: (error: string) => void; + updateQueryEventId: (queryEventID: number) => void; selectedSearchType: SearchType | null; offset: number | null; } diff --git a/web/src/lib/search/streamingQa.ts b/web/src/lib/search/streamingQa.ts index 50c0fe0d1..8225c4dd6 100644 --- a/web/src/lib/search/streamingQa.ts +++ b/web/src/lib/search/streamingQa.ts @@ -60,6 +60,7 @@ export const searchRequestStreamed = async ({ updateSuggestedSearchType, updateSuggestedFlowType, updateError, + updateQueryEventId, selectedSearchType, offset, }: SearchRequestArgs) => { @@ -174,6 +175,12 @@ export const searchRequestStreamed = async ({ return; } + // check for query ID section + if (chunk.query_event_id) { + updateQueryEventId(chunk.query_event_id); + return; + } + // should never reach this console.log("Unknown chunk:", chunk); }); diff --git a/web/src/lib/types.ts b/web/src/lib/types.ts index 86dd7c792..b357e8ff6 100644 --- a/web/src/lib/types.ts +++ b/web/src/lib/types.ts @@ -29,6 +29,14 @@ export type ValidStatuses = | "in_progress" | "not_started"; +export interface DocumentBoostStatus { + document_id: string; + semantic_id: string; + link: string; + boost: number; + hidden: boolean; +} + // CONNECTORS export interface ConnectorBase { name: string;