Skip to content

Make use of SPANDATA consistent in AI integrations.` #4376

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
May 9, 2025
Merged

Make use of SPANDATA consistent in AI integrations.` #4376

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions sentry_sdk/ai/monitoring.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -107,14 +107,14 @@ def record_token_usage(
if ai_pipeline_name:
span.set_attribute(SPANDATA.AI_PIPELINE_NAME, ai_pipeline_name)
if prompt_tokens is not None:
span.set_attribute("ai.prompt_tokens.used", prompt_tokens)
span.set_attribute(SPANDATA.AI_PROMPT_TOKENS_USED, prompt_tokens)
if completion_tokens is not None:
span.set_attribute("ai.completion_tokens.used", completion_tokens)
span.set_attribute(SPANDATA.AI_COMPLETION_TOKENS_USED, completion_tokens)
if (
total_tokens is None
and prompt_tokens is not None
and completion_tokens is not None
):
total_tokens = prompt_tokens + completion_tokens
if total_tokens is not None:
span.set_attribute("ai.total_tokens.used", total_tokens)
span.set_attribute(SPANDATA.AI_TOTAL_TOKENS_USED, total_tokens)
18 changes: 18 additions & 0 deletions sentry_sdk/consts.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -251,6 +251,24 @@ class SPANDATA:
Example: "qa-pipeline"
"""

AI_PROMPT_TOKENS_USED = "ai.prompt_tokens.used"
"""
The number of input prompt tokens used by the model.
Example: 10
"""

AI_COMPLETION_TOKENS_USED = "ai.completion_tokens.used"
"""
The number of output completion tokens used by the model.
Example: 10
"""

AI_TOTAL_TOKENS_USED = "ai.total_tokens.used"
"""
The total number of tokens (input + output) used by the request to the model.
Example: 20
"""

AI_TEXTS = "ai.texts"
"""
Raw text inputs provided to the model.
Expand Down
42 changes: 21 additions & 21 deletions tests/integrations/anthropic/test_anthropic.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -127,9 +127,9 @@ def test_nonstreaming_create_message(
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.AI_RESPONSES not in span["data"]

assert span["data"]["ai.prompt_tokens.used"] == 10
assert span["data"]["ai.completion_tokens.used"] == 20
assert span["data"]["ai.total_tokens.used"] == 30
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 10
assert span["data"][SPANDATA.AI_COMPLETION_TOKENS_USED] == 20
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 30
assert span["data"][SPANDATA.AI_STREAMING] is False


Expand Down Expand Up @@ -197,9 +197,9 @@ async def test_nonstreaming_create_message_async(
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.AI_RESPONSES not in span["data"]

assert span["data"]["ai.prompt_tokens.used"] == 10
assert span["data"]["ai.completion_tokens.used"] == 20
assert span["data"]["ai.total_tokens.used"] == 30
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 10
assert span["data"][SPANDATA.AI_COMPLETION_TOKENS_USED] == 20
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 30
assert span["data"][SPANDATA.AI_STREAMING] is False


Expand Down Expand Up @@ -299,9 +299,9 @@ def test_streaming_create_message(
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.AI_RESPONSES not in span["data"]

assert span["data"]["ai.prompt_tokens.used"] == 10
assert span["data"]["ai.completion_tokens.used"] == 30
assert span["data"]["ai.total_tokens.used"] == 40
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 10
assert span["data"][SPANDATA.AI_COMPLETION_TOKENS_USED] == 30
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 40
assert span["data"][SPANDATA.AI_STREAMING] is True


Expand Down Expand Up @@ -404,9 +404,9 @@ async def test_streaming_create_message_async(
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.AI_RESPONSES not in span["data"]

assert span["data"]["ai.prompt_tokens.used"] == 10
assert span["data"]["ai.completion_tokens.used"] == 30
assert span["data"]["ai.total_tokens.used"] == 40
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 10
assert span["data"][SPANDATA.AI_COMPLETION_TOKENS_USED] == 30
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 40
assert span["data"][SPANDATA.AI_STREAMING] is True


Expand Down Expand Up @@ -536,9 +536,9 @@ def test_streaming_create_message_with_input_json_delta(
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.AI_RESPONSES not in span["data"]

assert span["data"]["ai.prompt_tokens.used"] == 366
assert span["data"]["ai.completion_tokens.used"] == 51
assert span["data"]["ai.total_tokens.used"] == 417
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 366
assert span["data"][SPANDATA.AI_COMPLETION_TOKENS_USED] == 51
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 417
assert span["data"][SPANDATA.AI_STREAMING] is True


Expand Down Expand Up @@ -675,9 +675,9 @@ async def test_streaming_create_message_with_input_json_delta_async(
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.AI_RESPONSES not in span["data"]

assert span["data"]["ai.prompt_tokens.used"] == 366
assert span["data"]["ai.completion_tokens.used"] == 51
assert span["data"]["ai.total_tokens.used"] == 417
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 366
assert span["data"][SPANDATA.AI_COMPLETION_TOKENS_USED] == 51
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 417
assert span["data"][SPANDATA.AI_STREAMING] is True


Expand Down Expand Up @@ -831,6 +831,6 @@ def test_add_ai_data_to_span_with_input_json_delta(sentry_init, capture_events):
[{"type": "text", "text": "{'test': 'data','more': 'json'}"}]
)
assert span["data"][SPANDATA.AI_STREAMING] is True
assert span["data"]["ai.prompt_tokens.used"] == 10
assert span["data"]["ai.completion_tokens.used"] == 20
assert span["data"]["ai.total_tokens.used"] == 30
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 10
assert span["data"][SPANDATA.AI_COMPLETION_TOKENS_USED] == 20
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 30
16 changes: 8 additions & 8 deletions tests/integrations/cohere/test_cohere.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,9 +65,9 @@ def test_nonstreaming_chat(
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.AI_RESPONSES not in span["data"]

assert span["data"]["ai.completion_tokens.used"] == 10
assert span["data"]["ai.prompt_tokens.used"] == 20
assert span["data"]["ai.total_tokens.used"] == 30
assert span["data"][SPANDATA.AI_COMPLETION_TOKENS_USED] == 10
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 20
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 30


# noinspection PyTypeChecker
Expand Down Expand Up @@ -137,9 +137,9 @@ def test_streaming_chat(sentry_init, capture_events, send_default_pii, include_p
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.AI_RESPONSES not in span["data"]

assert span["data"]["ai.completion_tokens.used"] == 10
assert span["data"]["ai.prompt_tokens.used"] == 20
assert span["data"]["ai.total_tokens.used"] == 30
assert span["data"][SPANDATA.AI_COMPLETION_TOKENS_USED] == 10
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 20
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 30


def test_bad_chat(sentry_init, capture_events):
Expand Down Expand Up @@ -201,8 +201,8 @@ def test_embed(sentry_init, capture_events, send_default_pii, include_prompts):
else:
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]

assert span["data"]["ai.prompt_tokens.used"] == 10
assert span["data"]["ai.total_tokens.used"] == 10
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 10
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 10


def test_span_origin_chat(sentry_init, capture_events):
Expand Down
4 changes: 2 additions & 2 deletions tests/integrations/huggingface_hub/test_huggingface_hub.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,7 +75,7 @@ def test_nonstreaming_chat_completion(
assert SPANDATA.AI_RESPONSES not in span["data"]

if details_arg:
assert span["data"]["ai.total_tokens.used"] == 10
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 10


@pytest.mark.parametrize(
Expand Down Expand Up @@ -134,7 +134,7 @@ def test_streaming_chat_completion(
assert SPANDATA.AI_RESPONSES not in span["data"]

if details_arg:
assert span["data"]["ai.total_tokens.used"] == 10
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 10


def test_bad_chat_completion(sentry_init, capture_events):
Expand Down
8 changes: 4 additions & 4 deletions tests/integrations/langchain/test_langchain.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -181,13 +181,13 @@ def test_langchain_agent(
assert len(list(x for x in tx["spans"] if x["op"] == "ai.run.langchain")) > 0

if use_unknown_llm_type:
assert "ai.prompt_tokens.used" in chat_spans[0]["data"]
assert "ai.total_tokens.used" in chat_spans[0]["data"]
assert SPANDATA.AI_PROMPT_TOKENS_USED in chat_spans[0]["data"]
assert SPANDATA.AI_TOTAL_TOKENS_USED in chat_spans[0]["data"]
else:
# important: to avoid double counting, we do *not* measure
# tokens used if we have an explicit integration (e.g. OpenAI)
assert "ai.prompt_tokens.used" not in chat_spans[0]["data"]
assert "ai.total_tokens.used" not in chat_spans[0]["data"]
assert SPANDATA.AI_PROMPT_TOKENS_USED not in chat_spans[0]["data"]
assert SPANDATA.AI_TOTAL_TOKENS_USED not in chat_spans[0]["data"]

if send_default_pii and include_prompts:
assert (
Expand Down
32 changes: 16 additions & 16 deletions tests/integrations/openai/test_openai.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -90,9 +90,9 @@ def test_nonstreaming_chat_completion(
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.AI_RESPONSES not in span["data"]

assert span["data"]["ai.completion_tokens.used"] == 10
assert span["data"]["ai.prompt_tokens.used"] == 20
assert span["data"]["ai.total_tokens.used"] == 30
assert span["data"][SPANDATA.AI_COMPLETION_TOKENS_USED] == 10
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 20
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 30


@pytest.mark.asyncio
Expand Down Expand Up @@ -132,9 +132,9 @@ async def test_nonstreaming_chat_completion_async(
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]
assert SPANDATA.AI_RESPONSES not in span["data"]

assert span["data"]["ai.completion_tokens.used"] == 10
assert span["data"]["ai.prompt_tokens.used"] == 20
assert span["data"]["ai.total_tokens.used"] == 30
assert span["data"][SPANDATA.AI_COMPLETION_TOKENS_USED] == 10
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 20
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 30


def tiktoken_encoding_if_installed():
Expand Down Expand Up @@ -228,9 +228,9 @@ def test_streaming_chat_completion(
try:
import tiktoken # type: ignore # noqa # pylint: disable=unused-import

assert span["data"]["ai.completion_tokens.used"] == 2
assert span["data"]["ai.prompt_tokens.used"] == 1
assert span["data"]["ai.total_tokens.used"] == 3
assert span["data"][SPANDATA.AI_COMPLETION_TOKENS_USED] == 2
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 1
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 3
except ImportError:
pass # if tiktoken is not installed, we can't guarantee token usage will be calculated properly

Expand Down Expand Up @@ -324,9 +324,9 @@ async def test_streaming_chat_completion_async(
try:
import tiktoken # type: ignore # noqa # pylint: disable=unused-import

assert span["data"]["ai.completion_tokens.used"] == 2
assert span["data"]["ai.prompt_tokens.used"] == 1
assert span["data"]["ai.total_tokens.used"] == 3
assert span["data"][SPANDATA.AI_COMPLETION_TOKENS_USED] == 2
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 1
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 3
except ImportError:
pass # if tiktoken is not installed, we can't guarantee token usage will be calculated properly

Expand Down Expand Up @@ -410,8 +410,8 @@ def test_embeddings_create(
else:
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]

assert span["data"]["ai.prompt_tokens.used"] == 20
assert span["data"]["ai.total_tokens.used"] == 30
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 20
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 30


@pytest.mark.asyncio
Expand Down Expand Up @@ -458,8 +458,8 @@ async def test_embeddings_create_async(
else:
assert SPANDATA.AI_INPUT_MESSAGES not in span["data"]

assert span["data"]["ai.prompt_tokens.used"] == 20
assert span["data"]["ai.total_tokens.used"] == 30
assert span["data"][SPANDATA.AI_PROMPT_TOKENS_USED] == 20
assert span["data"][SPANDATA.AI_TOTAL_TOKENS_USED] == 30


@pytest.mark.forked
Expand Down
Loading