openlit 1.33.18__py3-none-any.whl → 1.33.20__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- openlit/__helpers.py +11 -41
- openlit/__init__.py +3 -3
- openlit/evals/utils.py +7 -7
- openlit/guard/utils.py +7 -7
- openlit/instrumentation/ag2/ag2.py +24 -24
- openlit/instrumentation/ai21/ai21.py +3 -3
- openlit/instrumentation/ai21/async_ai21.py +3 -3
- openlit/instrumentation/ai21/utils.py +59 -59
- openlit/instrumentation/anthropic/anthropic.py +2 -2
- openlit/instrumentation/anthropic/async_anthropic.py +2 -2
- openlit/instrumentation/anthropic/utils.py +34 -34
- openlit/instrumentation/assemblyai/assemblyai.py +24 -24
- openlit/instrumentation/astra/astra.py +3 -3
- openlit/instrumentation/astra/async_astra.py +3 -3
- openlit/instrumentation/astra/utils.py +39 -39
- openlit/instrumentation/azure_ai_inference/async_azure_ai_inference.py +10 -10
- openlit/instrumentation/azure_ai_inference/azure_ai_inference.py +10 -10
- openlit/instrumentation/azure_ai_inference/utils.py +38 -38
- openlit/instrumentation/bedrock/__init__.py +2 -1
- openlit/instrumentation/bedrock/bedrock.py +32 -214
- openlit/instrumentation/bedrock/utils.py +252 -0
- openlit/instrumentation/chroma/chroma.py +57 -57
- openlit/instrumentation/cohere/async_cohere.py +88 -88
- openlit/instrumentation/cohere/cohere.py +88 -88
- openlit/instrumentation/controlflow/controlflow.py +15 -15
- openlit/instrumentation/crawl4ai/async_crawl4ai.py +14 -14
- openlit/instrumentation/crawl4ai/crawl4ai.py +14 -14
- openlit/instrumentation/crewai/crewai.py +22 -22
- openlit/instrumentation/dynamiq/dynamiq.py +19 -19
- openlit/instrumentation/elevenlabs/async_elevenlabs.py +24 -25
- openlit/instrumentation/elevenlabs/elevenlabs.py +23 -25
- openlit/instrumentation/embedchain/embedchain.py +15 -15
- openlit/instrumentation/firecrawl/firecrawl.py +10 -10
- openlit/instrumentation/google_ai_studio/async_google_ai_studio.py +33 -33
- openlit/instrumentation/google_ai_studio/google_ai_studio.py +33 -33
- openlit/instrumentation/gpt4all/gpt4all.py +78 -78
- openlit/instrumentation/gpu/__init__.py +8 -8
- openlit/instrumentation/groq/async_groq.py +74 -74
- openlit/instrumentation/groq/groq.py +74 -74
- openlit/instrumentation/haystack/haystack.py +6 -6
- openlit/instrumentation/julep/async_julep.py +14 -14
- openlit/instrumentation/julep/julep.py +14 -14
- openlit/instrumentation/langchain/async_langchain.py +39 -39
- openlit/instrumentation/langchain/langchain.py +39 -39
- openlit/instrumentation/letta/letta.py +26 -26
- openlit/instrumentation/litellm/async_litellm.py +94 -94
- openlit/instrumentation/litellm/litellm.py +94 -94
- openlit/instrumentation/llamaindex/llamaindex.py +7 -7
- openlit/instrumentation/mem0/mem0.py +13 -13
- openlit/instrumentation/milvus/milvus.py +47 -47
- openlit/instrumentation/mistral/async_mistral.py +88 -88
- openlit/instrumentation/mistral/mistral.py +88 -88
- openlit/instrumentation/multion/async_multion.py +21 -21
- openlit/instrumentation/multion/multion.py +21 -21
- openlit/instrumentation/ollama/async_ollama.py +3 -3
- openlit/instrumentation/ollama/ollama.py +3 -3
- openlit/instrumentation/ollama/utils.py +50 -50
- openlit/instrumentation/openai/async_openai.py +225 -225
- openlit/instrumentation/openai/openai.py +225 -225
- openlit/instrumentation/openai_agents/openai_agents.py +11 -11
- openlit/instrumentation/phidata/phidata.py +15 -15
- openlit/instrumentation/pinecone/pinecone.py +43 -43
- openlit/instrumentation/premai/premai.py +86 -86
- openlit/instrumentation/qdrant/async_qdrant.py +95 -95
- openlit/instrumentation/qdrant/qdrant.py +99 -99
- openlit/instrumentation/reka/async_reka.py +33 -33
- openlit/instrumentation/reka/reka.py +33 -33
- openlit/instrumentation/together/async_together.py +90 -90
- openlit/instrumentation/together/together.py +90 -90
- openlit/instrumentation/transformers/transformers.py +26 -26
- openlit/instrumentation/vertexai/async_vertexai.py +64 -64
- openlit/instrumentation/vertexai/vertexai.py +64 -64
- openlit/instrumentation/vllm/vllm.py +24 -24
- openlit/otel/metrics.py +11 -11
- openlit/semcov/__init__.py +3 -3
- {openlit-1.33.18.dist-info → openlit-1.33.20.dist-info}/METADATA +8 -8
- openlit-1.33.20.dist-info/RECORD +131 -0
- {openlit-1.33.18.dist-info → openlit-1.33.20.dist-info}/WHEEL +1 -1
- openlit-1.33.18.dist-info/RECORD +0 -130
- {openlit-1.33.18.dist-info → openlit-1.33.20.dist-info}/LICENSE +0 -0
|
@@ -9,7 +9,7 @@ from opentelemetry.sdk.resources import SERVICE_NAME, TELEMETRY_SDK_NAME, DEPLOY
|
|
|
9
9
|
from openlit.__helpers import (
|
|
10
10
|
handle_exception,
|
|
11
11
|
)
|
|
12
|
-
from openlit.semcov import
|
|
12
|
+
from openlit.semcov import SemanticConvention
|
|
13
13
|
|
|
14
14
|
# Initialize logger for logging potential issues and operations
|
|
15
15
|
logger = logging.getLogger(__name__)
|
|
@@ -56,11 +56,11 @@ def dynamiq_wrap(gen_ai_endpoint, version, environment, application_name,
|
|
|
56
56
|
try:
|
|
57
57
|
# Set base span attribues
|
|
58
58
|
span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
|
|
59
|
-
span.set_attribute(
|
|
60
|
-
|
|
61
|
-
span.set_attribute(
|
|
62
|
-
|
|
63
|
-
span.set_attribute(
|
|
59
|
+
span.set_attribute(SemanticConvention.GEN_AI_SYSTEM,
|
|
60
|
+
SemanticConvention.GEN_AI_SYSTEM_DYNAMIQ)
|
|
61
|
+
span.set_attribute(SemanticConvention.GEN_AI_OPERATION,
|
|
62
|
+
SemanticConvention.GEN_AI_OPERATION_TYPE_AGENT)
|
|
63
|
+
span.set_attribute(SemanticConvention.GEN_AI_ENDPOINT,
|
|
64
64
|
gen_ai_endpoint)
|
|
65
65
|
span.set_attribute(SERVICE_NAME,
|
|
66
66
|
application_name)
|
|
@@ -68,32 +68,32 @@ def dynamiq_wrap(gen_ai_endpoint, version, environment, application_name,
|
|
|
68
68
|
environment)
|
|
69
69
|
|
|
70
70
|
if gen_ai_endpoint == "dynamiq.agent_run":
|
|
71
|
-
span.set_attribute(
|
|
71
|
+
span.set_attribute(SemanticConvention.GEN_AI_AGENT_ID,
|
|
72
72
|
getattr(instance, 'id', '') or '')
|
|
73
|
-
span.set_attribute(
|
|
73
|
+
span.set_attribute(SemanticConvention.GEN_AI_AGENT_ROLE,
|
|
74
74
|
getattr(instance, 'name', '') or '')
|
|
75
|
-
span.set_attribute(
|
|
75
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_MODEL,
|
|
76
76
|
getattr(getattr(instance, 'llm', None), 'model', '') or '')
|
|
77
|
-
span.set_attribute(
|
|
77
|
+
span.set_attribute(SemanticConvention.GEN_AI_AGENT_TYPE,
|
|
78
78
|
str(getattr(instance, 'type', '')) or '')
|
|
79
79
|
|
|
80
80
|
elif gen_ai_endpoint == "dynamiq.workflow_run":
|
|
81
|
-
span.set_attribute(
|
|
81
|
+
span.set_attribute(SemanticConvention.GEN_AI_AGENT_ID,
|
|
82
82
|
getattr(instance, 'id', '') or '')
|
|
83
|
-
span.set_attribute(
|
|
83
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_MODEL,
|
|
84
84
|
getattr(getattr(instance.flow, 'nodes', [None])[0], 'model', 'default_model'))
|
|
85
85
|
|
|
86
86
|
elif gen_ai_endpoint == "dynamiq.memory_add":
|
|
87
|
-
span.set_attribute(
|
|
88
|
-
|
|
89
|
-
span.set_attribute(
|
|
87
|
+
span.set_attribute(SemanticConvention.DB_OPERATION_NAME,
|
|
88
|
+
SemanticConvention.DB_OPERATION_ADD)
|
|
89
|
+
span.set_attribute(SemanticConvention.DB_METADATA, str(kwargs.get('metadata', '')))
|
|
90
90
|
|
|
91
91
|
elif gen_ai_endpoint == "dynamiq.memory_search":
|
|
92
92
|
query_value = kwargs.get('query', '') or (args[0] if args else '')
|
|
93
|
-
span.set_attribute(
|
|
94
|
-
|
|
95
|
-
span.set_attribute(
|
|
96
|
-
span.set_attribute(
|
|
93
|
+
span.set_attribute(SemanticConvention.DB_OPERATION_NAME,
|
|
94
|
+
SemanticConvention.DB_OPERATION_GET)
|
|
95
|
+
span.set_attribute(SemanticConvention.DB_FILTER, str(kwargs.get('filters', '')))
|
|
96
|
+
span.set_attribute(SemanticConvention.DB_STATEMENT, query_value)
|
|
97
97
|
|
|
98
98
|
span.set_status(Status(StatusCode.OK))
|
|
99
99
|
|
|
@@ -4,7 +4,6 @@ Module for monitoring Ollama API calls.
|
|
|
4
4
|
|
|
5
5
|
import logging
|
|
6
6
|
import time
|
|
7
|
-
from urllib.parse import urlparse
|
|
8
7
|
from opentelemetry.trace import SpanKind, Status, StatusCode
|
|
9
8
|
from opentelemetry.sdk.resources import SERVICE_NAME, TELEMETRY_SDK_NAME, DEPLOYMENT_ENVIRONMENT
|
|
10
9
|
from openlit.__helpers import (
|
|
@@ -12,7 +11,7 @@ from openlit.__helpers import (
|
|
|
12
11
|
handle_exception,
|
|
13
12
|
create_metrics_attributes,
|
|
14
13
|
)
|
|
15
|
-
from openlit.semcov import
|
|
14
|
+
from openlit.semcov import SemanticConvention
|
|
16
15
|
|
|
17
16
|
# Initialize logger for logging potential issues and operations
|
|
18
17
|
logger = logging.getLogger(__name__)
|
|
@@ -50,11 +49,11 @@ def async_generate(gen_ai_endpoint, version, environment, application_name,
|
|
|
50
49
|
Returns:
|
|
51
50
|
The response from the original 'generate' method.
|
|
52
51
|
"""
|
|
53
|
-
|
|
54
|
-
server_address, server_port =
|
|
52
|
+
|
|
53
|
+
server_address, server_port = "api.elevenlabs.io", 443
|
|
55
54
|
request_model = kwargs.get('model', kwargs.get('model_id', 'eleven_multilingual_v2'))
|
|
56
55
|
|
|
57
|
-
span_name = f'{
|
|
56
|
+
span_name = f'{SemanticConvention.GEN_AI_OPERATION_TYPE_AUDIO} {request_model}'
|
|
58
57
|
|
|
59
58
|
with tracer.start_as_current_span(span_name, kind= SpanKind.CLIENT) as span:
|
|
60
59
|
start_time = time.time()
|
|
@@ -68,48 +67,48 @@ def async_generate(gen_ai_endpoint, version, environment, application_name,
|
|
|
68
67
|
|
|
69
68
|
# Set Span attributes
|
|
70
69
|
span.set_attribute(TELEMETRY_SDK_NAME, 'openlit')
|
|
71
|
-
span.set_attribute(
|
|
72
|
-
|
|
73
|
-
span.set_attribute(
|
|
74
|
-
|
|
75
|
-
span.set_attribute(
|
|
70
|
+
span.set_attribute(SemanticConvention.GEN_AI_OPERATION,
|
|
71
|
+
SemanticConvention.GEN_AI_OPERATION_TYPE_AUDIO)
|
|
72
|
+
span.set_attribute(SemanticConvention.GEN_AI_SYSTEM,
|
|
73
|
+
SemanticConvention.GEN_AI_SYSTEM_ASSEMBLYAI)
|
|
74
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_MODEL,
|
|
76
75
|
request_model)
|
|
77
|
-
span.set_attribute(
|
|
76
|
+
span.set_attribute(SemanticConvention.SERVER_ADDRESS,
|
|
78
77
|
server_address)
|
|
79
|
-
span.set_attribute(
|
|
78
|
+
span.set_attribute(SemanticConvention.SERVER_PORT,
|
|
80
79
|
server_port)
|
|
81
|
-
span.set_attribute(
|
|
80
|
+
span.set_attribute(SemanticConvention.GEN_AI_RESPONSE_MODEL,
|
|
82
81
|
request_model)
|
|
83
|
-
span.set_attribute(
|
|
82
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_MODEL,
|
|
84
83
|
request_model)
|
|
85
|
-
span.set_attribute(
|
|
84
|
+
span.set_attribute(SemanticConvention.GEN_AI_OUTPUT_TYPE,
|
|
86
85
|
'audio')
|
|
87
86
|
|
|
88
87
|
# Set Span attributes (Extras)
|
|
89
88
|
if gen_ai_endpoint == 'elevenlabs.generate':
|
|
90
89
|
if isinstance(kwargs.get('voice', 'Rachel'), str):
|
|
91
|
-
span.set_attribute(
|
|
90
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_VOICE,
|
|
92
91
|
kwargs.get('voice', 'Rachel'))
|
|
93
92
|
else:
|
|
94
|
-
span.set_attribute(
|
|
93
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_VOICE,
|
|
95
94
|
kwargs.get('voice_id', ''))
|
|
96
|
-
span.set_attribute(
|
|
95
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_RESPONSE_FORMAT,
|
|
97
96
|
kwargs.get('output_format', 'mp3'))
|
|
98
|
-
span.set_attribute(
|
|
97
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_SETTINGS,
|
|
99
98
|
str(kwargs.get('voice_settings', '')))
|
|
100
99
|
span.set_attribute(DEPLOYMENT_ENVIRONMENT,
|
|
101
100
|
environment)
|
|
102
101
|
span.set_attribute(SERVICE_NAME,
|
|
103
102
|
application_name)
|
|
104
|
-
span.set_attribute(
|
|
103
|
+
span.set_attribute(SemanticConvention.GEN_AI_USAGE_COST,
|
|
105
104
|
cost)
|
|
106
|
-
span.set_attribute(
|
|
105
|
+
span.set_attribute(SemanticConvention.GEN_AI_SDK_VERSION,
|
|
107
106
|
version)
|
|
108
107
|
if capture_message_content:
|
|
109
108
|
span.add_event(
|
|
110
|
-
name=
|
|
109
|
+
name=SemanticConvention.GEN_AI_CONTENT_PROMPT_EVENT,
|
|
111
110
|
attributes={
|
|
112
|
-
|
|
111
|
+
SemanticConvention.GEN_AI_CONTENT_PROMPT: str(kwargs.get('text', '')),
|
|
113
112
|
},
|
|
114
113
|
)
|
|
115
114
|
|
|
@@ -119,8 +118,8 @@ def async_generate(gen_ai_endpoint, version, environment, application_name,
|
|
|
119
118
|
attributes = create_metrics_attributes(
|
|
120
119
|
service_name=application_name,
|
|
121
120
|
deployment_environment=environment,
|
|
122
|
-
operation=
|
|
123
|
-
system=
|
|
121
|
+
operation=SemanticConvention.GEN_AI_OPERATION_TYPE_AUDIO,
|
|
122
|
+
system=SemanticConvention.GEN_AI_SYSTEM_ELEVENLABS,
|
|
124
123
|
request_model=request_model,
|
|
125
124
|
server_address=server_address,
|
|
126
125
|
server_port=server_port,
|
|
@@ -4,7 +4,6 @@ Module for monitoring Ollama API calls.
|
|
|
4
4
|
|
|
5
5
|
import logging
|
|
6
6
|
import time
|
|
7
|
-
from urllib.parse import urlparse
|
|
8
7
|
from opentelemetry.trace import SpanKind, Status, StatusCode
|
|
9
8
|
from opentelemetry.sdk.resources import SERVICE_NAME, TELEMETRY_SDK_NAME, DEPLOYMENT_ENVIRONMENT
|
|
10
9
|
from openlit.__helpers import (
|
|
@@ -12,7 +11,7 @@ from openlit.__helpers import (
|
|
|
12
11
|
handle_exception,
|
|
13
12
|
create_metrics_attributes,
|
|
14
13
|
)
|
|
15
|
-
from openlit.semcov import
|
|
14
|
+
from openlit.semcov import SemanticConvention
|
|
16
15
|
|
|
17
16
|
# Initialize logger for logging potential issues and operations
|
|
18
17
|
logger = logging.getLogger(__name__)
|
|
@@ -51,11 +50,10 @@ def generate(gen_ai_endpoint, version, environment, application_name,
|
|
|
51
50
|
The response from the original 'generate' method.
|
|
52
51
|
"""
|
|
53
52
|
|
|
54
|
-
|
|
55
|
-
server_address, server_port = url.hostname, url.port or 443
|
|
53
|
+
server_address, server_port = "api.elevenlabs.io", 443
|
|
56
54
|
request_model = kwargs.get('model', kwargs.get('model_id', 'eleven_multilingual_v2'))
|
|
57
55
|
|
|
58
|
-
span_name = f'{
|
|
56
|
+
span_name = f'{SemanticConvention.GEN_AI_OPERATION_TYPE_AUDIO} {request_model}'
|
|
59
57
|
|
|
60
58
|
with tracer.start_as_current_span(span_name, kind= SpanKind.CLIENT) as span:
|
|
61
59
|
start_time = time.time()
|
|
@@ -69,48 +67,48 @@ def generate(gen_ai_endpoint, version, environment, application_name,
|
|
|
69
67
|
|
|
70
68
|
# Set Span attributes
|
|
71
69
|
span.set_attribute(TELEMETRY_SDK_NAME, 'openlit')
|
|
72
|
-
span.set_attribute(
|
|
73
|
-
|
|
74
|
-
span.set_attribute(
|
|
75
|
-
|
|
76
|
-
span.set_attribute(
|
|
70
|
+
span.set_attribute(SemanticConvention.GEN_AI_OPERATION,
|
|
71
|
+
SemanticConvention.GEN_AI_OPERATION_TYPE_AUDIO)
|
|
72
|
+
span.set_attribute(SemanticConvention.GEN_AI_SYSTEM,
|
|
73
|
+
SemanticConvention.GEN_AI_SYSTEM_ASSEMBLYAI)
|
|
74
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_MODEL,
|
|
77
75
|
request_model)
|
|
78
|
-
span.set_attribute(
|
|
76
|
+
span.set_attribute(SemanticConvention.SERVER_ADDRESS,
|
|
79
77
|
server_address)
|
|
80
|
-
span.set_attribute(
|
|
78
|
+
span.set_attribute(SemanticConvention.SERVER_PORT,
|
|
81
79
|
server_port)
|
|
82
|
-
span.set_attribute(
|
|
80
|
+
span.set_attribute(SemanticConvention.GEN_AI_RESPONSE_MODEL,
|
|
83
81
|
request_model)
|
|
84
|
-
span.set_attribute(
|
|
82
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_MODEL,
|
|
85
83
|
request_model)
|
|
86
|
-
span.set_attribute(
|
|
84
|
+
span.set_attribute(SemanticConvention.GEN_AI_OUTPUT_TYPE,
|
|
87
85
|
'audio')
|
|
88
86
|
|
|
89
87
|
# Set Span attributes (Extras)
|
|
90
88
|
if gen_ai_endpoint == 'elevenlabs.generate':
|
|
91
89
|
if isinstance(kwargs.get('voice', 'Rachel'), str):
|
|
92
|
-
span.set_attribute(
|
|
90
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_VOICE,
|
|
93
91
|
kwargs.get('voice', 'Rachel'))
|
|
94
92
|
else:
|
|
95
|
-
span.set_attribute(
|
|
93
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_VOICE,
|
|
96
94
|
kwargs.get('voice_id', ''))
|
|
97
|
-
span.set_attribute(
|
|
95
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_RESPONSE_FORMAT,
|
|
98
96
|
kwargs.get('output_format', 'mp3'))
|
|
99
|
-
span.set_attribute(
|
|
97
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_SETTINGS,
|
|
100
98
|
str(kwargs.get('voice_settings', '')))
|
|
101
99
|
span.set_attribute(DEPLOYMENT_ENVIRONMENT,
|
|
102
100
|
environment)
|
|
103
101
|
span.set_attribute(SERVICE_NAME,
|
|
104
102
|
application_name)
|
|
105
|
-
span.set_attribute(
|
|
103
|
+
span.set_attribute(SemanticConvention.GEN_AI_USAGE_COST,
|
|
106
104
|
cost)
|
|
107
|
-
span.set_attribute(
|
|
105
|
+
span.set_attribute(SemanticConvention.GEN_AI_SDK_VERSION,
|
|
108
106
|
version)
|
|
109
107
|
if capture_message_content:
|
|
110
108
|
span.add_event(
|
|
111
|
-
name=
|
|
109
|
+
name=SemanticConvention.GEN_AI_CONTENT_PROMPT_EVENT,
|
|
112
110
|
attributes={
|
|
113
|
-
|
|
111
|
+
SemanticConvention.GEN_AI_CONTENT_PROMPT: str(kwargs.get('text', '')),
|
|
114
112
|
},
|
|
115
113
|
)
|
|
116
114
|
|
|
@@ -120,8 +118,8 @@ def generate(gen_ai_endpoint, version, environment, application_name,
|
|
|
120
118
|
attributes = create_metrics_attributes(
|
|
121
119
|
service_name=application_name,
|
|
122
120
|
deployment_environment=environment,
|
|
123
|
-
operation=
|
|
124
|
-
system=
|
|
121
|
+
operation=SemanticConvention.GEN_AI_OPERATION_TYPE_AUDIO,
|
|
122
|
+
system=SemanticConvention.GEN_AI_SYSTEM_ELEVENLABS,
|
|
125
123
|
request_model=request_model,
|
|
126
124
|
server_address=server_address,
|
|
127
125
|
server_port=server_port,
|
|
@@ -7,7 +7,7 @@ import logging
|
|
|
7
7
|
from opentelemetry.trace import SpanKind, Status, StatusCode
|
|
8
8
|
from opentelemetry.sdk.resources import SERVICE_NAME, TELEMETRY_SDK_NAME, DEPLOYMENT_ENVIRONMENT
|
|
9
9
|
from openlit.__helpers import handle_exception
|
|
10
|
-
from openlit.semcov import
|
|
10
|
+
from openlit.semcov import SemanticConvention
|
|
11
11
|
|
|
12
12
|
# Initialize logger for logging potential issues and operations
|
|
13
13
|
logger = logging.getLogger(__name__)
|
|
@@ -59,21 +59,21 @@ def evaluate(gen_ai_endpoint, version, environment, application_name,
|
|
|
59
59
|
|
|
60
60
|
try:
|
|
61
61
|
span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
|
|
62
|
-
span.set_attribute(
|
|
63
|
-
|
|
64
|
-
span.set_attribute(
|
|
62
|
+
span.set_attribute(SemanticConvention.GEN_AI_SYSTEM,
|
|
63
|
+
SemanticConvention.GEN_AI_SYSTEM_EMBEDCHAIN)
|
|
64
|
+
span.set_attribute(SemanticConvention.GEN_AI_ENDPOINT,
|
|
65
65
|
gen_ai_endpoint)
|
|
66
66
|
span.set_attribute(DEPLOYMENT_ENVIRONMENT,
|
|
67
67
|
environment)
|
|
68
|
-
span.set_attribute(
|
|
69
|
-
|
|
68
|
+
span.set_attribute(SemanticConvention.GEN_AI_OPERATION,
|
|
69
|
+
SemanticConvention.GEN_AI_OPERATION_TYPE_FRAMEWORK)
|
|
70
70
|
span.set_attribute(SERVICE_NAME,
|
|
71
71
|
application_name)
|
|
72
|
-
span.set_attribute(
|
|
72
|
+
span.set_attribute(SemanticConvention.GEN_AI_EVAL_CONTEXT_RELEVANCY,
|
|
73
73
|
response["context_relevancy"])
|
|
74
|
-
span.set_attribute(
|
|
74
|
+
span.set_attribute(SemanticConvention.GEN_AI_EVAL_ANSWER_RELEVANCY,
|
|
75
75
|
response["answer_relevancy"])
|
|
76
|
-
span.set_attribute(
|
|
76
|
+
span.set_attribute(SemanticConvention.GEN_AI_EVAL_GROUNDEDNESS,
|
|
77
77
|
response["groundedness"])
|
|
78
78
|
|
|
79
79
|
span.set_status(Status(StatusCode.OK))
|
|
@@ -137,17 +137,17 @@ def get_data_sources(gen_ai_endpoint, version, environment, application_name,
|
|
|
137
137
|
|
|
138
138
|
try:
|
|
139
139
|
span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
|
|
140
|
-
span.set_attribute(
|
|
141
|
-
|
|
142
|
-
span.set_attribute(
|
|
140
|
+
span.set_attribute(SemanticConvention.GEN_AI_SYSTEM,
|
|
141
|
+
SemanticConvention.GEN_AI_SYSTEM_EMBEDCHAIN)
|
|
142
|
+
span.set_attribute(SemanticConvention.GEN_AI_ENDPOINT,
|
|
143
143
|
gen_ai_endpoint)
|
|
144
144
|
span.set_attribute(DEPLOYMENT_ENVIRONMENT,
|
|
145
145
|
environment)
|
|
146
|
-
span.set_attribute(
|
|
147
|
-
|
|
146
|
+
span.set_attribute(SemanticConvention.GEN_AI_OPERATION,
|
|
147
|
+
SemanticConvention.GEN_AI_OPERATION_TYPE_FRAMEWORK)
|
|
148
148
|
span.set_attribute(SERVICE_NAME,
|
|
149
149
|
application_name)
|
|
150
|
-
span.set_attribute(
|
|
150
|
+
span.set_attribute(SemanticConvention.GEN_AI_DATA_SOURCES,
|
|
151
151
|
len(response))
|
|
152
152
|
|
|
153
153
|
span.set_status(Status(StatusCode.OK))
|
|
@@ -9,7 +9,7 @@ from opentelemetry.sdk.resources import SERVICE_NAME, TELEMETRY_SDK_NAME, DEPLOY
|
|
|
9
9
|
from openlit.__helpers import (
|
|
10
10
|
handle_exception,
|
|
11
11
|
)
|
|
12
|
-
from openlit.semcov import
|
|
12
|
+
from openlit.semcov import SemanticConvention
|
|
13
13
|
|
|
14
14
|
# Initialize logger for logging potential issues and operations
|
|
15
15
|
logger = logging.getLogger(__name__)
|
|
@@ -56,24 +56,24 @@ def wrap_crawl(gen_ai_endpoint, version, environment, application_name,
|
|
|
56
56
|
try:
|
|
57
57
|
# Set base span attribues
|
|
58
58
|
span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
|
|
59
|
-
span.set_attribute(
|
|
60
|
-
|
|
61
|
-
span.set_attribute(
|
|
62
|
-
|
|
63
|
-
span.set_attribute(
|
|
59
|
+
span.set_attribute(SemanticConvention.GEN_AI_SYSTEM,
|
|
60
|
+
SemanticConvention.GEN_AI_SYSTEM_FIRECRAWL)
|
|
61
|
+
span.set_attribute(SemanticConvention.GEN_AI_OPERATION,
|
|
62
|
+
SemanticConvention.GEN_AI_OPERATION_TYPE_AGENT)
|
|
63
|
+
span.set_attribute(SemanticConvention.GEN_AI_ENDPOINT,
|
|
64
64
|
gen_ai_endpoint)
|
|
65
65
|
span.set_attribute(SERVICE_NAME,
|
|
66
66
|
application_name)
|
|
67
67
|
span.set_attribute(DEPLOYMENT_ENVIRONMENT,
|
|
68
68
|
environment)
|
|
69
|
-
span.set_attribute(
|
|
70
|
-
|
|
71
|
-
span.set_attribute(
|
|
69
|
+
span.set_attribute(SemanticConvention.GEN_AI_AGENT_TYPE,
|
|
70
|
+
SemanticConvention.GEN_AI_AGENT_TYPE_BROWSER)
|
|
71
|
+
span.set_attribute(SemanticConvention.GEN_AI_AGENT_PARAMS,
|
|
72
72
|
str(kwargs.get("params")))
|
|
73
73
|
|
|
74
74
|
url = kwargs.get("url") if "url" in kwargs else str(args[0]) if args else None
|
|
75
75
|
if url is not None:
|
|
76
|
-
span.set_attribute(
|
|
76
|
+
span.set_attribute(SemanticConvention.GEN_AI_AGENT_BROWSE_URL, url)
|
|
77
77
|
|
|
78
78
|
span.set_status(Status(StatusCode.OK))
|
|
79
79
|
|
|
@@ -13,7 +13,7 @@ from openlit.__helpers import (
|
|
|
13
13
|
create_metrics_attributes,
|
|
14
14
|
set_server_address_and_port
|
|
15
15
|
)
|
|
16
|
-
from openlit.semcov import
|
|
16
|
+
from openlit.semcov import SemanticConvention
|
|
17
17
|
|
|
18
18
|
# Initialize logger for logging potential issues and operations
|
|
19
19
|
logger = logging.getLogger(__name__)
|
|
@@ -56,7 +56,7 @@ def async_generate(version, environment, application_name,
|
|
|
56
56
|
server_address, server_port = set_server_address_and_port(instance, "generativelanguage.googleapis.com", 443)
|
|
57
57
|
request_model = kwargs.get("model", "gemini-2.0-flash")
|
|
58
58
|
|
|
59
|
-
span_name = f"{
|
|
59
|
+
span_name = f"{SemanticConvention.GEN_AI_OPERATION_TYPE_CHAT} {request_model}"
|
|
60
60
|
|
|
61
61
|
with tracer.start_as_current_span(span_name, kind=SpanKind.CLIENT) as span:
|
|
62
62
|
start_time = time.time()
|
|
@@ -110,26 +110,26 @@ def async_generate(version, environment, application_name,
|
|
|
110
110
|
|
|
111
111
|
# Set base span attribues (OTel Semconv)
|
|
112
112
|
span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
|
|
113
|
-
span.set_attribute(
|
|
114
|
-
|
|
115
|
-
span.set_attribute(
|
|
116
|
-
|
|
117
|
-
span.set_attribute(
|
|
113
|
+
span.set_attribute(SemanticConvention.GEN_AI_OPERATION,
|
|
114
|
+
SemanticConvention.GEN_AI_OPERATION_TYPE_CHAT)
|
|
115
|
+
span.set_attribute(SemanticConvention.GEN_AI_SYSTEM,
|
|
116
|
+
SemanticConvention.GEN_AI_SYSTEM_GEMINI)
|
|
117
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_MODEL,
|
|
118
118
|
request_model)
|
|
119
|
-
span.set_attribute(
|
|
119
|
+
span.set_attribute(SemanticConvention.SERVER_PORT,
|
|
120
120
|
server_port)
|
|
121
121
|
|
|
122
122
|
inference_config = kwargs.get('config', {})
|
|
123
123
|
|
|
124
124
|
# List of attributes and their config keys
|
|
125
125
|
attributes = [
|
|
126
|
-
(
|
|
127
|
-
(
|
|
128
|
-
(
|
|
129
|
-
(
|
|
130
|
-
(
|
|
131
|
-
(
|
|
132
|
-
(
|
|
126
|
+
(SemanticConvention.GEN_AI_REQUEST_FREQUENCY_PENALTY, 'frequency_penalty'),
|
|
127
|
+
(SemanticConvention.GEN_AI_REQUEST_MAX_TOKENS, 'max_tokens'),
|
|
128
|
+
(SemanticConvention.GEN_AI_REQUEST_PRESENCE_PENALTY, 'presence_penalty'),
|
|
129
|
+
(SemanticConvention.GEN_AI_REQUEST_STOP_SEQUENCES, 'stop_sequences'),
|
|
130
|
+
(SemanticConvention.GEN_AI_REQUEST_TEMPERATURE, 'temperature'),
|
|
131
|
+
(SemanticConvention.GEN_AI_REQUEST_TOP_P, 'top_p'),
|
|
132
|
+
(SemanticConvention.GEN_AI_REQUEST_TOP_K, 'top_k'),
|
|
133
133
|
]
|
|
134
134
|
|
|
135
135
|
# Set each attribute if the corresponding value exists and is not None
|
|
@@ -139,15 +139,15 @@ def async_generate(version, environment, application_name,
|
|
|
139
139
|
if value is not None:
|
|
140
140
|
span.set_attribute(attribute, value)
|
|
141
141
|
|
|
142
|
-
span.set_attribute(
|
|
142
|
+
span.set_attribute(SemanticConvention.GEN_AI_RESPONSE_MODEL,
|
|
143
143
|
response_dict.get('model_version'))
|
|
144
|
-
span.set_attribute(
|
|
144
|
+
span.set_attribute(SemanticConvention.GEN_AI_USAGE_INPUT_TOKENS,
|
|
145
145
|
input_tokens)
|
|
146
|
-
span.set_attribute(
|
|
146
|
+
span.set_attribute(SemanticConvention.GEN_AI_USAGE_OUTPUT_TOKENS,
|
|
147
147
|
output_tokens)
|
|
148
|
-
span.set_attribute(
|
|
148
|
+
span.set_attribute(SemanticConvention.SERVER_ADDRESS,
|
|
149
149
|
server_address)
|
|
150
|
-
span.set_attribute(
|
|
150
|
+
span.set_attribute(SemanticConvention.GEN_AI_RESPONSE_FINISH_REASON,
|
|
151
151
|
[str(response_dict.get('candidates')[0].get('finish_reason'))])
|
|
152
152
|
|
|
153
153
|
# Set base span attribues (Extras)
|
|
@@ -155,35 +155,35 @@ def async_generate(version, environment, application_name,
|
|
|
155
155
|
environment)
|
|
156
156
|
span.set_attribute(SERVICE_NAME,
|
|
157
157
|
application_name)
|
|
158
|
-
span.set_attribute(
|
|
158
|
+
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_IS_STREAM,
|
|
159
159
|
False)
|
|
160
|
-
span.set_attribute(
|
|
160
|
+
span.set_attribute(SemanticConvention.GEN_AI_USAGE_TOTAL_TOKENS,
|
|
161
161
|
input_tokens + output_tokens)
|
|
162
|
-
span.set_attribute(
|
|
162
|
+
span.set_attribute(SemanticConvention.GEN_AI_USAGE_COST,
|
|
163
163
|
cost)
|
|
164
|
-
span.set_attribute(
|
|
164
|
+
span.set_attribute(SemanticConvention.GEN_AI_SERVER_TTFT,
|
|
165
165
|
end_time - start_time)
|
|
166
|
-
span.set_attribute(
|
|
166
|
+
span.set_attribute(SemanticConvention.GEN_AI_SDK_VERSION,
|
|
167
167
|
version)
|
|
168
168
|
if capture_message_content:
|
|
169
169
|
span.add_event(
|
|
170
|
-
name=
|
|
170
|
+
name=SemanticConvention.GEN_AI_CONTENT_PROMPT_EVENT,
|
|
171
171
|
attributes={
|
|
172
|
-
|
|
172
|
+
SemanticConvention.GEN_AI_CONTENT_PROMPT: prompt,
|
|
173
173
|
},
|
|
174
174
|
)
|
|
175
175
|
span.add_event(
|
|
176
|
-
name=
|
|
176
|
+
name=SemanticConvention.GEN_AI_CONTENT_COMPLETION_EVENT,
|
|
177
177
|
attributes={
|
|
178
|
-
|
|
178
|
+
SemanticConvention.GEN_AI_CONTENT_COMPLETION: response.text,
|
|
179
179
|
},
|
|
180
180
|
)
|
|
181
181
|
|
|
182
182
|
if isinstance(response_dict.get('text'), str):
|
|
183
|
-
span.set_attribute(
|
|
183
|
+
span.set_attribute(SemanticConvention.GEN_AI_OUTPUT_TYPE,
|
|
184
184
|
"text")
|
|
185
185
|
elif response_dict.get('text') is not None:
|
|
186
|
-
span.set_attribute(
|
|
186
|
+
span.set_attribute(SemanticConvention.GEN_AI_OUTPUT_TYPE,
|
|
187
187
|
"json")
|
|
188
188
|
|
|
189
189
|
span.set_status(Status(StatusCode.OK))
|
|
@@ -192,8 +192,8 @@ def async_generate(version, environment, application_name,
|
|
|
192
192
|
attributes = create_metrics_attributes(
|
|
193
193
|
service_name=application_name,
|
|
194
194
|
deployment_environment=environment,
|
|
195
|
-
operation=
|
|
196
|
-
system=
|
|
195
|
+
operation=SemanticConvention.GEN_AI_OPERATION_TYPE_CHAT,
|
|
196
|
+
system=SemanticConvention.GEN_AI_SYSTEM_GEMINI,
|
|
197
197
|
request_model=request_model,
|
|
198
198
|
server_address=server_address,
|
|
199
199
|
server_port=server_port,
|