openlit 1.31.1__py3-none-any.whl → 1.32.2__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/__init__.py +15 -0
- openlit/instrumentation/crewai/crewai.py +2 -0
- openlit/instrumentation/julep/__init__.py +80 -0
- openlit/instrumentation/julep/async_julep.py +111 -0
- openlit/instrumentation/julep/julep.py +112 -0
- openlit/instrumentation/mem0/__init__.py +79 -0
- openlit/instrumentation/mem0/mem0.py +115 -0
- openlit/instrumentation/multion/__init__.py +80 -0
- openlit/instrumentation/multion/async_multion.py +131 -0
- openlit/instrumentation/multion/multion.py +131 -0
- openlit/instrumentation/premai/__init__.py +51 -0
- openlit/instrumentation/premai/premai.py +507 -0
- openlit/instrumentation/reka/__init__.py +54 -0
- openlit/instrumentation/reka/async_reka.py +159 -0
- openlit/instrumentation/reka/reka.py +159 -0
- openlit/semcov/__init__.py +9 -0
- {openlit-1.31.1.dist-info → openlit-1.32.2.dist-info}/METADATA +11 -5
- {openlit-1.31.1.dist-info → openlit-1.32.2.dist-info}/RECORD +20 -7
- {openlit-1.31.1.dist-info → openlit-1.32.2.dist-info}/LICENSE +0 -0
- {openlit-1.31.1.dist-info → openlit-1.32.2.dist-info}/WHEEL +0 -0
@@ -0,0 +1,131 @@
|
|
1
|
+
# pylint: disable=duplicate-code, broad-exception-caught, too-many-statements, unused-argument, too-many-branches
|
2
|
+
"""
|
3
|
+
Module for monitoring multion calls.
|
4
|
+
"""
|
5
|
+
|
6
|
+
import logging
|
7
|
+
from opentelemetry.trace import SpanKind, Status, StatusCode
|
8
|
+
from opentelemetry.sdk.resources import TELEMETRY_SDK_NAME
|
9
|
+
from openlit.__helpers import (
|
10
|
+
handle_exception,
|
11
|
+
)
|
12
|
+
from openlit.semcov import SemanticConvetion
|
13
|
+
|
14
|
+
# Initialize logger for logging potential issues and operations
|
15
|
+
logger = logging.getLogger(__name__)
|
16
|
+
|
17
|
+
def async_multion_wrap(gen_ai_endpoint, version, environment, application_name,
|
18
|
+
tracer, pricing_info, trace_content, metrics, disable_metrics):
|
19
|
+
"""
|
20
|
+
Generates a telemetry wrapper for chat completions to collect metrics.
|
21
|
+
|
22
|
+
Args:
|
23
|
+
gen_ai_endpoint: Endpoint identifier for logging and tracing.
|
24
|
+
version: Version of the monitoring package.
|
25
|
+
environment: Deployment environment (e.g., production, staging).
|
26
|
+
application_name: Name of the application using the multion Agent.
|
27
|
+
tracer: OpenTelemetry tracer for creating spans.
|
28
|
+
pricing_info: Information used for calculating the cost of multion usage.
|
29
|
+
trace_content: Flag indicating whether to trace the actual content.
|
30
|
+
|
31
|
+
Returns:
|
32
|
+
A function that wraps the chat completions method to add telemetry.
|
33
|
+
"""
|
34
|
+
|
35
|
+
async def wrapper(wrapped, instance, args, kwargs):
|
36
|
+
"""
|
37
|
+
Wraps the 'chat.completions' API call to add telemetry.
|
38
|
+
|
39
|
+
This collects metrics such as execution time, cost, and token usage, and handles errors
|
40
|
+
gracefully, adding details to the trace for observability.
|
41
|
+
|
42
|
+
Args:
|
43
|
+
wrapped: The original 'chat.completions' method to be wrapped.
|
44
|
+
instance: The instance of the class where the original method is defined.
|
45
|
+
args: Positional arguments for the 'chat.completions' method.
|
46
|
+
kwargs: Keyword arguments for the 'chat.completions' method.
|
47
|
+
|
48
|
+
Returns:
|
49
|
+
The response from the original 'chat.completions' method.
|
50
|
+
"""
|
51
|
+
|
52
|
+
# pylint: disable=line-too-long
|
53
|
+
with tracer.start_as_current_span(gen_ai_endpoint, kind= SpanKind.CLIENT) as span:
|
54
|
+
response = await wrapped(*args, **kwargs)
|
55
|
+
|
56
|
+
try:
|
57
|
+
# Set base span attribues
|
58
|
+
span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
|
59
|
+
span.set_attribute(SemanticConvetion.GEN_AI_SYSTEM,
|
60
|
+
SemanticConvetion.GEN_AI_SYSTEM_MULTION)
|
61
|
+
span.set_attribute(SemanticConvetion.GEN_AI_TYPE,
|
62
|
+
SemanticConvetion.GEN_AI_TYPE_AGENT)
|
63
|
+
span.set_attribute(SemanticConvetion.GEN_AI_ENDPOINT,
|
64
|
+
gen_ai_endpoint)
|
65
|
+
span.set_attribute(SemanticConvetion.GEN_AI_APPLICATION_NAME,
|
66
|
+
application_name)
|
67
|
+
|
68
|
+
if gen_ai_endpoint == "multion.browse":
|
69
|
+
span.set_attribute(SemanticConvetion.GEN_AI_AGENT_BROWSE_URL,
|
70
|
+
kwargs.get("url", ""))
|
71
|
+
span.set_attribute(SemanticConvetion.GEN_AI_AGENT_STEP_COUNT,
|
72
|
+
response.metadata.step_count)
|
73
|
+
span.set_attribute(SemanticConvetion.GEN_AI_AGENT_RESPONSE_TIME,
|
74
|
+
response.metadata.processing_time)
|
75
|
+
|
76
|
+
if trace_content:
|
77
|
+
span.add_event(
|
78
|
+
name=SemanticConvetion.GEN_AI_CONTENT_PROMPT_EVENT,
|
79
|
+
attributes={
|
80
|
+
SemanticConvetion.GEN_AI_CONTENT_PROMPT: kwargs.get("cmd", ""),
|
81
|
+
},
|
82
|
+
)
|
83
|
+
span.add_event(
|
84
|
+
name=SemanticConvetion.GEN_AI_CONTENT_COMPLETION_EVENT,
|
85
|
+
attributes={
|
86
|
+
SemanticConvetion.GEN_AI_CONTENT_COMPLETION: response.message,
|
87
|
+
},
|
88
|
+
)
|
89
|
+
elif gen_ai_endpoint == "multion.retrieve":
|
90
|
+
span.set_attribute(SemanticConvetion.GEN_AI_AGENT_BROWSE_URL,
|
91
|
+
kwargs.get("url", ""))
|
92
|
+
|
93
|
+
if trace_content:
|
94
|
+
span.add_event(
|
95
|
+
name=SemanticConvetion.GEN_AI_CONTENT_PROMPT_EVENT,
|
96
|
+
attributes={
|
97
|
+
SemanticConvetion.GEN_AI_CONTENT_PROMPT: kwargs.get("cmd", ""),
|
98
|
+
},
|
99
|
+
)
|
100
|
+
span.add_event(
|
101
|
+
name=SemanticConvetion.GEN_AI_CONTENT_COMPLETION_EVENT,
|
102
|
+
attributes={
|
103
|
+
SemanticConvetion.GEN_AI_CONTENT_COMPLETION: response.message,
|
104
|
+
},
|
105
|
+
)
|
106
|
+
|
107
|
+
elif gen_ai_endpoint == "multion.sessions.create":
|
108
|
+
span.set_attribute(SemanticConvetion.GEN_AI_AGENT_BROWSE_URL,
|
109
|
+
kwargs.get("url", ""))
|
110
|
+
|
111
|
+
if trace_content:
|
112
|
+
span.add_event(
|
113
|
+
name=SemanticConvetion.GEN_AI_CONTENT_COMPLETION_EVENT,
|
114
|
+
attributes={
|
115
|
+
SemanticConvetion.GEN_AI_CONTENT_COMPLETION: response.message,
|
116
|
+
},
|
117
|
+
)
|
118
|
+
|
119
|
+
span.set_status(Status(StatusCode.OK))
|
120
|
+
|
121
|
+
# Return original response
|
122
|
+
return response
|
123
|
+
|
124
|
+
except Exception as e:
|
125
|
+
handle_exception(span, e)
|
126
|
+
logger.error("Error in trace creation: %s", e)
|
127
|
+
|
128
|
+
# Return original response
|
129
|
+
return response
|
130
|
+
|
131
|
+
return wrapper
|
@@ -0,0 +1,131 @@
|
|
1
|
+
# pylint: disable=duplicate-code, broad-exception-caught, too-many-statements, unused-argument, too-many-branches
|
2
|
+
"""
|
3
|
+
Module for monitoring multion calls.
|
4
|
+
"""
|
5
|
+
|
6
|
+
import logging
|
7
|
+
from opentelemetry.trace import SpanKind, Status, StatusCode
|
8
|
+
from opentelemetry.sdk.resources import TELEMETRY_SDK_NAME
|
9
|
+
from openlit.__helpers import (
|
10
|
+
handle_exception,
|
11
|
+
)
|
12
|
+
from openlit.semcov import SemanticConvetion
|
13
|
+
|
14
|
+
# Initialize logger for logging potential issues and operations
|
15
|
+
logger = logging.getLogger(__name__)
|
16
|
+
|
17
|
+
def multion_wrap(gen_ai_endpoint, version, environment, application_name,
|
18
|
+
tracer, pricing_info, trace_content, metrics, disable_metrics):
|
19
|
+
"""
|
20
|
+
Generates a telemetry wrapper for chat completions to collect metrics.
|
21
|
+
|
22
|
+
Args:
|
23
|
+
gen_ai_endpoint: Endpoint identifier for logging and tracing.
|
24
|
+
version: Version of the monitoring package.
|
25
|
+
environment: Deployment environment (e.g., production, staging).
|
26
|
+
application_name: Name of the application using the multion Agent.
|
27
|
+
tracer: OpenTelemetry tracer for creating spans.
|
28
|
+
pricing_info: Information used for calculating the cost of multion usage.
|
29
|
+
trace_content: Flag indicating whether to trace the actual content.
|
30
|
+
|
31
|
+
Returns:
|
32
|
+
A function that wraps the chat completions method to add telemetry.
|
33
|
+
"""
|
34
|
+
|
35
|
+
def wrapper(wrapped, instance, args, kwargs):
|
36
|
+
"""
|
37
|
+
Wraps the 'chat.completions' API call to add telemetry.
|
38
|
+
|
39
|
+
This collects metrics such as execution time, cost, and token usage, and handles errors
|
40
|
+
gracefully, adding details to the trace for observability.
|
41
|
+
|
42
|
+
Args:
|
43
|
+
wrapped: The original 'chat.completions' method to be wrapped.
|
44
|
+
instance: The instance of the class where the original method is defined.
|
45
|
+
args: Positional arguments for the 'chat.completions' method.
|
46
|
+
kwargs: Keyword arguments for the 'chat.completions' method.
|
47
|
+
|
48
|
+
Returns:
|
49
|
+
The response from the original 'chat.completions' method.
|
50
|
+
"""
|
51
|
+
|
52
|
+
# pylint: disable=line-too-long
|
53
|
+
with tracer.start_as_current_span(gen_ai_endpoint, kind= SpanKind.CLIENT) as span:
|
54
|
+
response = wrapped(*args, **kwargs)
|
55
|
+
|
56
|
+
try:
|
57
|
+
# Set base span attribues
|
58
|
+
span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
|
59
|
+
span.set_attribute(SemanticConvetion.GEN_AI_SYSTEM,
|
60
|
+
SemanticConvetion.GEN_AI_SYSTEM_MULTION)
|
61
|
+
span.set_attribute(SemanticConvetion.GEN_AI_TYPE,
|
62
|
+
SemanticConvetion.GEN_AI_TYPE_AGENT)
|
63
|
+
span.set_attribute(SemanticConvetion.GEN_AI_ENDPOINT,
|
64
|
+
gen_ai_endpoint)
|
65
|
+
span.set_attribute(SemanticConvetion.GEN_AI_APPLICATION_NAME,
|
66
|
+
application_name)
|
67
|
+
|
68
|
+
if gen_ai_endpoint == "multion.browse":
|
69
|
+
span.set_attribute(SemanticConvetion.GEN_AI_AGENT_BROWSE_URL,
|
70
|
+
kwargs.get("url", ""))
|
71
|
+
span.set_attribute(SemanticConvetion.GEN_AI_AGENT_STEP_COUNT,
|
72
|
+
response.metadata.step_count)
|
73
|
+
span.set_attribute(SemanticConvetion.GEN_AI_AGENT_RESPONSE_TIME,
|
74
|
+
response.metadata.processing_time)
|
75
|
+
|
76
|
+
if trace_content:
|
77
|
+
span.add_event(
|
78
|
+
name=SemanticConvetion.GEN_AI_CONTENT_PROMPT_EVENT,
|
79
|
+
attributes={
|
80
|
+
SemanticConvetion.GEN_AI_CONTENT_PROMPT: kwargs.get("cmd", ""),
|
81
|
+
},
|
82
|
+
)
|
83
|
+
span.add_event(
|
84
|
+
name=SemanticConvetion.GEN_AI_CONTENT_COMPLETION_EVENT,
|
85
|
+
attributes={
|
86
|
+
SemanticConvetion.GEN_AI_CONTENT_COMPLETION: response.message,
|
87
|
+
},
|
88
|
+
)
|
89
|
+
elif gen_ai_endpoint == "multion.retrieve":
|
90
|
+
span.set_attribute(SemanticConvetion.GEN_AI_AGENT_BROWSE_URL,
|
91
|
+
kwargs.get("url", ""))
|
92
|
+
|
93
|
+
if trace_content:
|
94
|
+
span.add_event(
|
95
|
+
name=SemanticConvetion.GEN_AI_CONTENT_PROMPT_EVENT,
|
96
|
+
attributes={
|
97
|
+
SemanticConvetion.GEN_AI_CONTENT_PROMPT: kwargs.get("cmd", ""),
|
98
|
+
},
|
99
|
+
)
|
100
|
+
span.add_event(
|
101
|
+
name=SemanticConvetion.GEN_AI_CONTENT_COMPLETION_EVENT,
|
102
|
+
attributes={
|
103
|
+
SemanticConvetion.GEN_AI_CONTENT_COMPLETION: response.message,
|
104
|
+
},
|
105
|
+
)
|
106
|
+
|
107
|
+
elif gen_ai_endpoint == "multion.sessions.create":
|
108
|
+
span.set_attribute(SemanticConvetion.GEN_AI_AGENT_BROWSE_URL,
|
109
|
+
kwargs.get("url", ""))
|
110
|
+
|
111
|
+
if trace_content:
|
112
|
+
span.add_event(
|
113
|
+
name=SemanticConvetion.GEN_AI_CONTENT_COMPLETION_EVENT,
|
114
|
+
attributes={
|
115
|
+
SemanticConvetion.GEN_AI_CONTENT_COMPLETION: response.message,
|
116
|
+
},
|
117
|
+
)
|
118
|
+
|
119
|
+
span.set_status(Status(StatusCode.OK))
|
120
|
+
|
121
|
+
# Return original response
|
122
|
+
return response
|
123
|
+
|
124
|
+
except Exception as e:
|
125
|
+
handle_exception(span, e)
|
126
|
+
logger.error("Error in trace creation: %s", e)
|
127
|
+
|
128
|
+
# Return original response
|
129
|
+
return response
|
130
|
+
|
131
|
+
return wrapper
|
@@ -0,0 +1,51 @@
|
|
1
|
+
# pylint: disable=useless-return, bad-staticmethod-argument, disable=duplicate-code
|
2
|
+
"""Initializer of Auto Instrumentation of Prem AI Functions"""
|
3
|
+
|
4
|
+
from typing import Collection
|
5
|
+
import importlib.metadata
|
6
|
+
from opentelemetry.instrumentation.instrumentor import BaseInstrumentor
|
7
|
+
from wrapt import wrap_function_wrapper
|
8
|
+
|
9
|
+
from openlit.instrumentation.premai.premai import (
|
10
|
+
chat, embedding
|
11
|
+
)
|
12
|
+
|
13
|
+
_instruments = ("premai >= 0.3.79",)
|
14
|
+
|
15
|
+
class PremAIInstrumentor(BaseInstrumentor):
|
16
|
+
"""
|
17
|
+
An instrumentor for Prem AI's client library.
|
18
|
+
"""
|
19
|
+
|
20
|
+
def instrumentation_dependencies(self) -> Collection[str]:
|
21
|
+
return _instruments
|
22
|
+
|
23
|
+
def _instrument(self, **kwargs):
|
24
|
+
application_name = kwargs.get("application_name", "default_application")
|
25
|
+
environment = kwargs.get("environment", "default_environment")
|
26
|
+
tracer = kwargs.get("tracer")
|
27
|
+
metrics = kwargs.get("metrics_dict")
|
28
|
+
pricing_info = kwargs.get("pricing_info", {})
|
29
|
+
trace_content = kwargs.get("trace_content", False)
|
30
|
+
disable_metrics = kwargs.get("disable_metrics")
|
31
|
+
version = importlib.metadata.version("premai")
|
32
|
+
|
33
|
+
# sync chat
|
34
|
+
wrap_function_wrapper(
|
35
|
+
"premai.api",
|
36
|
+
"ChatCompletionsModule.create",
|
37
|
+
chat("premai.chat.completions", version, environment, application_name,
|
38
|
+
tracer, pricing_info, trace_content, metrics, disable_metrics),
|
39
|
+
)
|
40
|
+
|
41
|
+
# sync embedding
|
42
|
+
wrap_function_wrapper(
|
43
|
+
"premai.api",
|
44
|
+
"EmbeddingsModule.create",
|
45
|
+
embedding("premai.embeddings", version, environment, application_name,
|
46
|
+
tracer, pricing_info, trace_content, metrics, disable_metrics),
|
47
|
+
)
|
48
|
+
|
49
|
+
def _uninstrument(self, **kwargs):
|
50
|
+
# Proper uninstrumentation logic to revert patched methods
|
51
|
+
pass
|