sij/khoj
1
0
Fork 0
mirror of https://github.com/khoj-ai/khoj.git synced 2024-11-27 09:25:06 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 5

Fix handling of command rate limit and improve rate limit messages

Browse source 
Command rate limit wouldn't be shown to user as server wouldn't be
able to handle HTTP exception in the middle of streaming.

Catch exception and render it as LLM response message instead for
visibility into command rate limiting to user on client

Log rate limmit messages for all rate limit events on server as info
messages

Convert exception messages into first person responses by Khoj to
prevent breaking the third wall and provide more details on wht
happened and possible ways to resolve them.
This commit is contained in:
Debanjum 2024-11-09 11:30:55 -08:00
parent 80ee35b9b1
commit d892ab3174
2 changed files with 50 additions and 26 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

9
src/khoj/routers/api_chat.py
View file

@ -738,8 +738,13 @@ async def chat(
conversation_commands.append(mode)
for cmd in conversation_commands:
try:
await conversation_command_rate_limiter.update_and_check_if_valid(request, cmd)
q = q.replace(f"/{cmd.value}", "").strip()
except HTTPException as e:
async for result in send_llm_response(str(e.detail)):
yield result
return
defiltered_query = defilter_query(q)
@ -775,10 +780,6 @@ async def chat(
# researched_results = await extract_relevant_info(q, researched_results, agent)
logger.info(f"Researched Results: {researched_results}")
for cmd in conversation_commands:
await conversation_command_rate_limiter.update_and_check_if_valid(request, cmd)
q = q.replace(f"/{cmd.value}", "").strip()
used_slash_summarize = conversation_commands == [ConversationCommand.Summarize]
file_filters = conversation.file_filters if conversation else []
# Skip trying to summarize if

43
src/khoj/routers/helpers.py
View file

@ -1306,25 +1306,28 @@ class ApiUserRateLimiter:
# Check if the user has exceeded the rate limit
if subscribed and count_requests >= self.subscribed_requests:
logger.info(
f"Rate limit: {count_requests} requests in {self.window} seconds for user: {user}. Limit is {self.subscribed_requests} requests."
f"Rate limit: {count_requests}/{self.subscribed_requests} requests not allowed in {self.window} seconds for subscribed user: {user}."
)
raise HTTPException(
status_code=429,
detail="I'm glad you're enjoying interacting with me! You've unfortunately exceeded your usage limit for today. But let's chat more tomorrow?",
)
raise HTTPException(status_code=429, detail="Slow down! Too Many Requests")
if not subscribed and count_requests >= self.requests:
if self.requests >= self.subscribed_requests:
logger.info(
f"Rate limit: {count_requests} requests in {self.window} seconds for user: {user}. Limit is {self.subscribed_requests} requests."
f"Rate limit: {count_requests}/{self.subscribed_requests} requests not allowed in {self.window} seconds for user: {user}."
)
raise HTTPException(
status_code=429,
detail="Slow down! Too Many Requests",
detail="I'm glad you're enjoying interacting with me! You've unfortunately exceeded your usage limit for today. But let's chat more tomorrow?",
)
logger.info(
f"Rate limit: {count_requests} requests in {self.window} seconds for user: {user}. Limit is {self.subscribed_requests} requests."
f"Rate limit: {count_requests}/{self.requests} requests not allowed in {self.window} seconds for user: {user}."
)
raise HTTPException(
status_code=429,
detail="I'm glad you're enjoying interacting with me! But you've exceeded your usage limit for today. Come back tomorrow or subscribe to increase your usage limit via [your settings](https://app.khoj.dev/settings).",
detail="I'm glad you're enjoying interacting with me! You've unfortunately exceeded your usage limit for today. You can subscribe to increase your usage limit via [your settings](https://app.khoj.dev/settings) or we can continue our conversation tomorrow?",
)
# Add the current request to the cache
@ -1350,6 +1353,7 @@ class ApiImageRateLimiter:
# Check number of images
if len(body.images) > self.max_images:
logger.info(f"Rate limit: {len(body.images)}/{self.max_images} images not allowed per message.")
raise HTTPException(
status_code=429,
detail=f"Those are way too many images for me! I can handle up to {self.max_images} images per message.",
@ -1370,6 +1374,7 @@ class ApiImageRateLimiter:
total_size_mb += len(image_bytes) / (1024 * 1024) # Convert bytes to MB
if total_size_mb > self.max_combined_size_mb:
logger.info(f"Data limit: {total_size_mb}MB/{self.max_combined_size_mb}MB size not allowed per message.")
raise HTTPException(
status_code=429,
detail=f"Those images are way too large for me! I can handle up to {self.max_combined_size_mb}MB of images per message.",
@ -1405,13 +1410,19 @@ class ConversationCommandRateLimiter:
if subscribed and count_requests >= self.subscribed_rate_limit:
logger.info(
f"Rate limit: {count_requests} requests in 24 hours for user: {user}. Limit is {self.subscribed_rate_limit} requests."
f"Rate limit: {count_requests}/{self.subscribed_rate_limit} requests not allowed in 24 hours for subscribed user: {user}."
)
raise HTTPException(status_code=429, detail="Slow down! Too Many Requests")
if not subscribed and count_requests >= self.trial_rate_limit:
raise HTTPException(
status_code=429,
detail=f"We're glad you're enjoying Khoj! You've exceeded your `/{conversation_command.value}` command usage limit for today. Subscribe to increase your usage limit via [your settings](https://app.khoj.dev/settings).",
detail=f"I'm glad you're enjoying interacting with me! You've unfortunately exceeded your `/{conversation_command.value}` command usage limit for today. Maybe we can talk about something else for today?",
)
if not subscribed and count_requests >= self.trial_rate_limit:
logger.info(
f"Rate limit: {count_requests}/{self.trial_rate_limit} requests not allowed in 24 hours for user: {user}."
)
raise HTTPException(
status_code=429,
detail=f"I'm glad you're enjoying interacting with me! You've unfortunately exceeded your `/{conversation_command.value}` command usage limit for today. You can subscribe to increase your usage limit via [your settings](https://app.khoj.dev/settings) or we can talk about something else for today?",
)
await UserRequests.objects.acreate(user=user, slug=command_slug)
return
@ -1457,16 +1468,28 @@ class ApiIndexedDataLimiter:
logger.info(f"Deleted {num_deleted_entries} entries for user: {user}.")
if subscribed and incoming_data_size_mb >= self.subscribed_num_entries_size:
logger.info(
f"Data limit: {incoming_data_size_mb}MB incoming will exceed {self.subscribed_num_entries_size}MB allowed for subscribed user: {user}."
)
raise HTTPException(status_code=429, detail="Too much data indexed.")
if not subscribed and incoming_data_size_mb >= self.num_entries_size:
logger.info(
f"Data limit: {incoming_data_size_mb}MB incoming will exceed {self.num_entries_size}MB allowed for user: {user}."
)
raise HTTPException(
status_code=429, detail="Too much data indexed. Subscribe to increase your data index limit."
)
user_size_data = EntryAdapters.get_size_of_indexed_data_in_mb(user)
if subscribed and user_size_data + incoming_data_size_mb >= self.subscribed_total_entries_size:
logger.info(
f"Data limit: {incoming_data_size_mb}MB incoming + {user_size_data}MB existing will exceed {self.subscribed_total_entries_size}MB allowed for subscribed user: {user}."
)
raise HTTPException(status_code=429, detail="Too much data indexed.")
if not subscribed and user_size_data + incoming_data_size_mb >= self.total_entries_size_limit:
logger.info(
f"Data limit: {incoming_data_size_mb}MB incoming + {user_size_data}MB existing will exceed {self.subscribed_total_entries_size}MB allowed for non subscribed user: {user}."
)
raise HTTPException(
status_code=429, detail="Too much data indexed. Subscribe to increase your data index limit."
)