openlit 1.34.8__py3-none-any.whl → 1.34.11__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 +69 -0
- openlit/instrumentation/assemblyai/__init__.py +14 -18
- openlit/instrumentation/assemblyai/assemblyai.py +29 -120
- openlit/instrumentation/assemblyai/utils.py +142 -0
- openlit/instrumentation/elevenlabs/__init__.py +5 -27
- openlit/instrumentation/elevenlabs/async_elevenlabs.py +29 -119
- openlit/instrumentation/elevenlabs/elevenlabs.py +28 -118
- openlit/instrumentation/elevenlabs/utils.py +133 -0
- openlit/instrumentation/gpt4all/utils.py +3 -3
- openlit/instrumentation/groq/__init__.py +7 -9
- openlit/instrumentation/groq/async_groq.py +50 -374
- openlit/instrumentation/groq/groq.py +49 -373
- openlit/instrumentation/groq/utils.py +199 -0
- openlit/instrumentation/ollama/__init__.py +5 -6
- openlit/instrumentation/ollama/async_ollama.py +65 -61
- openlit/instrumentation/ollama/ollama.py +65 -61
- openlit/instrumentation/ollama/utils.py +184 -239
- openlit/instrumentation/premai/utils.py +3 -73
- openlit/instrumentation/reka/utils.py +3 -51
- {openlit-1.34.8.dist-info → openlit-1.34.11.dist-info}/METADATA +1 -1
- {openlit-1.34.8.dist-info → openlit-1.34.11.dist-info}/RECORD +23 -20
- {openlit-1.34.8.dist-info → openlit-1.34.11.dist-info}/LICENSE +0 -0
- {openlit-1.34.8.dist-info → openlit-1.34.11.dist-info}/WHEEL +0 -0
@@ -1,145 +1,55 @@
|
|
1
1
|
"""
|
2
|
-
Module for monitoring
|
2
|
+
Module for monitoring ElevenLabs API calls.
|
3
3
|
"""
|
4
4
|
|
5
|
-
import logging
|
6
5
|
import time
|
7
|
-
from opentelemetry.trace import SpanKind
|
8
|
-
from
|
9
|
-
from openlit.
|
10
|
-
get_audio_model_cost,
|
11
|
-
handle_exception,
|
12
|
-
create_metrics_attributes,
|
13
|
-
)
|
6
|
+
from opentelemetry.trace import SpanKind
|
7
|
+
from openlit.__helpers import handle_exception
|
8
|
+
from openlit.instrumentation.elevenlabs.utils import process_audio_response
|
14
9
|
from openlit.semcov import SemanticConvention
|
15
10
|
|
16
|
-
# Initialize logger for logging potential issues and operations
|
17
|
-
logger = logging.getLogger(__name__)
|
18
|
-
|
19
11
|
def async_generate(gen_ai_endpoint, version, environment, application_name,
|
20
|
-
|
12
|
+
tracer, pricing_info, capture_message_content, metrics, disable_metrics):
|
21
13
|
"""
|
22
|
-
Generates a telemetry wrapper for
|
23
|
-
|
24
|
-
Args:
|
25
|
-
version: Version of the monitoring package.
|
26
|
-
environment: Deployment environment (e.g., production, staging).
|
27
|
-
application_name: Name of the application using the ElevenLabs API.
|
28
|
-
tracer: OpenTelemetry tracer for creating spans.
|
29
|
-
pricing_info: Information used for calculating the cost of generating speech audio.
|
30
|
-
capture_message_content: Flag indicating whether to trace the input text and generated audio.
|
31
|
-
|
32
|
-
Returns:
|
33
|
-
A function that wraps the speech audio creation method to add telemetry.
|
14
|
+
Generates a telemetry wrapper for GenAI function call
|
34
15
|
"""
|
35
16
|
|
36
17
|
async def wrapper(wrapped, instance, args, kwargs):
|
37
18
|
"""
|
38
|
-
Wraps the
|
39
|
-
|
40
|
-
This collects metrics such as execution time, cost, and handles errors
|
41
|
-
gracefully, adding details to the trace for observability.
|
42
|
-
|
43
|
-
Args:
|
44
|
-
wrapped: The original 'generate' method to be wrapped.
|
45
|
-
instance: The instance of the class where the original method is defined.
|
46
|
-
args: Positional arguments for the 'generate' method.
|
47
|
-
kwargs: Keyword arguments for the 'generate' method.
|
48
|
-
|
49
|
-
Returns:
|
50
|
-
The response from the original 'generate' method.
|
19
|
+
Wraps the GenAI function call.
|
51
20
|
"""
|
52
21
|
|
53
22
|
server_address, server_port = "api.elevenlabs.io", 443
|
54
|
-
request_model = kwargs.get(
|
23
|
+
request_model = kwargs.get("model", kwargs.get("model_id", "eleven_multilingual_v2"))
|
55
24
|
|
56
|
-
span_name = f
|
25
|
+
span_name = f"{SemanticConvention.GEN_AI_OPERATION_TYPE_AUDIO} {request_model}"
|
57
26
|
|
58
|
-
with tracer.start_as_current_span(span_name, kind=
|
27
|
+
with tracer.start_as_current_span(span_name, kind=SpanKind.CLIENT) as span:
|
59
28
|
start_time = time.time()
|
60
|
-
response =
|
61
|
-
end_time = time.time()
|
29
|
+
response = wrapped(*args, **kwargs)
|
62
30
|
|
63
31
|
try:
|
64
|
-
|
65
|
-
|
66
|
-
|
67
|
-
|
68
|
-
|
69
|
-
|
70
|
-
|
71
|
-
|
72
|
-
|
73
|
-
|
74
|
-
|
75
|
-
|
76
|
-
|
77
|
-
|
78
|
-
|
79
|
-
|
80
|
-
|
81
|
-
request_model)
|
82
|
-
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_MODEL,
|
83
|
-
request_model)
|
84
|
-
span.set_attribute(SemanticConvention.GEN_AI_OUTPUT_TYPE,
|
85
|
-
'audio')
|
86
|
-
|
87
|
-
# Set Span attributes (Extras)
|
88
|
-
if gen_ai_endpoint == 'elevenlabs.generate':
|
89
|
-
if isinstance(kwargs.get('voice', 'Rachel'), str):
|
90
|
-
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_VOICE,
|
91
|
-
kwargs.get('voice', 'Rachel'))
|
92
|
-
else:
|
93
|
-
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_VOICE,
|
94
|
-
kwargs.get('voice_id', ''))
|
95
|
-
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_RESPONSE_FORMAT,
|
96
|
-
kwargs.get('output_format', 'mp3'))
|
97
|
-
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_SETTINGS,
|
98
|
-
str(kwargs.get('voice_settings', '')))
|
99
|
-
span.set_attribute(DEPLOYMENT_ENVIRONMENT,
|
100
|
-
environment)
|
101
|
-
span.set_attribute(SERVICE_NAME,
|
102
|
-
application_name)
|
103
|
-
span.set_attribute(SemanticConvention.GEN_AI_USAGE_COST,
|
104
|
-
cost)
|
105
|
-
span.set_attribute(SemanticConvention.GEN_AI_SDK_VERSION,
|
106
|
-
version)
|
107
|
-
if capture_message_content:
|
108
|
-
span.add_event(
|
109
|
-
name=SemanticConvention.GEN_AI_CONTENT_PROMPT_EVENT,
|
110
|
-
attributes={
|
111
|
-
SemanticConvention.GEN_AI_CONTENT_PROMPT: str(kwargs.get('text', '')),
|
112
|
-
},
|
113
|
-
)
|
114
|
-
|
115
|
-
span.set_status(Status(StatusCode.OK))
|
116
|
-
|
117
|
-
if disable_metrics is False:
|
118
|
-
attributes = create_metrics_attributes(
|
119
|
-
service_name=application_name,
|
120
|
-
deployment_environment=environment,
|
121
|
-
operation=SemanticConvention.GEN_AI_OPERATION_TYPE_AUDIO,
|
122
|
-
system=SemanticConvention.GEN_AI_SYSTEM_ELEVENLABS,
|
123
|
-
request_model=request_model,
|
124
|
-
server_address=server_address,
|
125
|
-
server_port=server_port,
|
126
|
-
response_model=request_model,
|
127
|
-
)
|
128
|
-
|
129
|
-
metrics['genai_client_operation_duration'].record(
|
130
|
-
end_time - start_time, attributes
|
131
|
-
)
|
132
|
-
metrics['genai_requests'].add(1, attributes)
|
133
|
-
metrics['genai_cost'].record(cost, attributes)
|
134
|
-
|
135
|
-
# Return original response
|
136
|
-
return response
|
32
|
+
response = process_audio_response(
|
33
|
+
response=response,
|
34
|
+
gen_ai_endpoint=gen_ai_endpoint,
|
35
|
+
pricing_info=pricing_info,
|
36
|
+
server_port=server_port,
|
37
|
+
server_address=server_address,
|
38
|
+
environment=environment,
|
39
|
+
application_name=application_name,
|
40
|
+
metrics=metrics,
|
41
|
+
start_time=start_time,
|
42
|
+
span=span,
|
43
|
+
args=args,
|
44
|
+
kwargs=kwargs,
|
45
|
+
capture_message_content=capture_message_content,
|
46
|
+
disable_metrics=disable_metrics,
|
47
|
+
version=version
|
48
|
+
)
|
137
49
|
|
138
50
|
except Exception as e:
|
139
51
|
handle_exception(span, e)
|
140
|
-
logger.error('Error in trace creation: %s', e)
|
141
52
|
|
142
|
-
|
143
|
-
return response
|
53
|
+
return response
|
144
54
|
|
145
55
|
return wrapper
|
@@ -1,145 +1,55 @@
|
|
1
1
|
"""
|
2
|
-
Module for monitoring
|
2
|
+
Module for monitoring ElevenLabs API calls.
|
3
3
|
"""
|
4
4
|
|
5
|
-
import logging
|
6
5
|
import time
|
7
|
-
from opentelemetry.trace import SpanKind
|
8
|
-
from
|
9
|
-
from openlit.
|
10
|
-
get_audio_model_cost,
|
11
|
-
handle_exception,
|
12
|
-
create_metrics_attributes,
|
13
|
-
)
|
6
|
+
from opentelemetry.trace import SpanKind
|
7
|
+
from openlit.__helpers import handle_exception
|
8
|
+
from openlit.instrumentation.elevenlabs.utils import process_audio_response
|
14
9
|
from openlit.semcov import SemanticConvention
|
15
10
|
|
16
|
-
# Initialize logger for logging potential issues and operations
|
17
|
-
logger = logging.getLogger(__name__)
|
18
|
-
|
19
11
|
def generate(gen_ai_endpoint, version, environment, application_name,
|
20
|
-
|
12
|
+
tracer, pricing_info, capture_message_content, metrics, disable_metrics):
|
21
13
|
"""
|
22
|
-
Generates a telemetry wrapper for
|
23
|
-
|
24
|
-
Args:
|
25
|
-
version: Version of the monitoring package.
|
26
|
-
environment: Deployment environment (e.g., production, staging).
|
27
|
-
application_name: Name of the application using the ElevenLabs API.
|
28
|
-
tracer: OpenTelemetry tracer for creating spans.
|
29
|
-
pricing_info: Information used for calculating the cost of generating speech audio.
|
30
|
-
capture_message_content: Flag indicating whether to trace the input text and generated audio.
|
31
|
-
|
32
|
-
Returns:
|
33
|
-
A function that wraps the speech audio creation method to add telemetry.
|
14
|
+
Generates a telemetry wrapper for GenAI function call
|
34
15
|
"""
|
35
16
|
|
36
17
|
def wrapper(wrapped, instance, args, kwargs):
|
37
18
|
"""
|
38
|
-
Wraps the
|
39
|
-
|
40
|
-
This collects metrics such as execution time, cost, and handles errors
|
41
|
-
gracefully, adding details to the trace for observability.
|
42
|
-
|
43
|
-
Args:
|
44
|
-
wrapped: The original 'generate' method to be wrapped.
|
45
|
-
instance: The instance of the class where the original method is defined.
|
46
|
-
args: Positional arguments for the 'generate' method.
|
47
|
-
kwargs: Keyword arguments for the 'generate' method.
|
48
|
-
|
49
|
-
Returns:
|
50
|
-
The response from the original 'generate' method.
|
19
|
+
Wraps the GenAI function call.
|
51
20
|
"""
|
52
21
|
|
53
22
|
server_address, server_port = "api.elevenlabs.io", 443
|
54
|
-
request_model = kwargs.get(
|
23
|
+
request_model = kwargs.get("model", kwargs.get("model_id", "eleven_multilingual_v2"))
|
55
24
|
|
56
|
-
span_name = f
|
25
|
+
span_name = f"{SemanticConvention.GEN_AI_OPERATION_TYPE_AUDIO} {request_model}"
|
57
26
|
|
58
|
-
with tracer.start_as_current_span(span_name, kind=
|
27
|
+
with tracer.start_as_current_span(span_name, kind=SpanKind.CLIENT) as span:
|
59
28
|
start_time = time.time()
|
60
29
|
response = wrapped(*args, **kwargs)
|
61
|
-
end_time = time.time()
|
62
30
|
|
63
31
|
try:
|
64
|
-
|
65
|
-
|
66
|
-
|
67
|
-
|
68
|
-
|
69
|
-
|
70
|
-
|
71
|
-
|
72
|
-
|
73
|
-
|
74
|
-
|
75
|
-
|
76
|
-
|
77
|
-
|
78
|
-
|
79
|
-
|
80
|
-
|
81
|
-
request_model)
|
82
|
-
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_MODEL,
|
83
|
-
request_model)
|
84
|
-
span.set_attribute(SemanticConvention.GEN_AI_OUTPUT_TYPE,
|
85
|
-
'audio')
|
86
|
-
|
87
|
-
# Set Span attributes (Extras)
|
88
|
-
if gen_ai_endpoint == 'elevenlabs.generate':
|
89
|
-
if isinstance(kwargs.get('voice', 'Rachel'), str):
|
90
|
-
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_VOICE,
|
91
|
-
kwargs.get('voice', 'Rachel'))
|
92
|
-
else:
|
93
|
-
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_VOICE,
|
94
|
-
kwargs.get('voice_id', ''))
|
95
|
-
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_RESPONSE_FORMAT,
|
96
|
-
kwargs.get('output_format', 'mp3'))
|
97
|
-
span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_SETTINGS,
|
98
|
-
str(kwargs.get('voice_settings', '')))
|
99
|
-
span.set_attribute(DEPLOYMENT_ENVIRONMENT,
|
100
|
-
environment)
|
101
|
-
span.set_attribute(SERVICE_NAME,
|
102
|
-
application_name)
|
103
|
-
span.set_attribute(SemanticConvention.GEN_AI_USAGE_COST,
|
104
|
-
cost)
|
105
|
-
span.set_attribute(SemanticConvention.GEN_AI_SDK_VERSION,
|
106
|
-
version)
|
107
|
-
if capture_message_content:
|
108
|
-
span.add_event(
|
109
|
-
name=SemanticConvention.GEN_AI_CONTENT_PROMPT_EVENT,
|
110
|
-
attributes={
|
111
|
-
SemanticConvention.GEN_AI_CONTENT_PROMPT: str(kwargs.get('text', '')),
|
112
|
-
},
|
113
|
-
)
|
114
|
-
|
115
|
-
span.set_status(Status(StatusCode.OK))
|
116
|
-
|
117
|
-
if disable_metrics is False:
|
118
|
-
attributes = create_metrics_attributes(
|
119
|
-
service_name=application_name,
|
120
|
-
deployment_environment=environment,
|
121
|
-
operation=SemanticConvention.GEN_AI_OPERATION_TYPE_AUDIO,
|
122
|
-
system=SemanticConvention.GEN_AI_SYSTEM_ELEVENLABS,
|
123
|
-
request_model=request_model,
|
124
|
-
server_address=server_address,
|
125
|
-
server_port=server_port,
|
126
|
-
response_model=request_model,
|
127
|
-
)
|
128
|
-
|
129
|
-
metrics['genai_client_operation_duration'].record(
|
130
|
-
end_time - start_time, attributes
|
131
|
-
)
|
132
|
-
metrics['genai_requests'].add(1, attributes)
|
133
|
-
metrics['genai_cost'].record(cost, attributes)
|
134
|
-
|
135
|
-
# Return original response
|
136
|
-
return response
|
32
|
+
response = process_audio_response(
|
33
|
+
response=response,
|
34
|
+
gen_ai_endpoint=gen_ai_endpoint,
|
35
|
+
pricing_info=pricing_info,
|
36
|
+
server_port=server_port,
|
37
|
+
server_address=server_address,
|
38
|
+
environment=environment,
|
39
|
+
application_name=application_name,
|
40
|
+
metrics=metrics,
|
41
|
+
start_time=start_time,
|
42
|
+
span=span,
|
43
|
+
args=args,
|
44
|
+
kwargs=kwargs,
|
45
|
+
capture_message_content=capture_message_content,
|
46
|
+
disable_metrics=disable_metrics,
|
47
|
+
version=version
|
48
|
+
)
|
137
49
|
|
138
50
|
except Exception as e:
|
139
51
|
handle_exception(span, e)
|
140
|
-
logger.error('Error in trace creation: %s', e)
|
141
52
|
|
142
|
-
|
143
|
-
return response
|
53
|
+
return response
|
144
54
|
|
145
55
|
return wrapper
|
@@ -0,0 +1,133 @@
|
|
1
|
+
"""
|
2
|
+
ElevenLabs OpenTelemetry instrumentation utility functions
|
3
|
+
"""
|
4
|
+
import time
|
5
|
+
|
6
|
+
from opentelemetry.sdk.resources import SERVICE_NAME, TELEMETRY_SDK_NAME, DEPLOYMENT_ENVIRONMENT
|
7
|
+
from opentelemetry.trace import Status, StatusCode
|
8
|
+
|
9
|
+
from openlit.__helpers import (
|
10
|
+
get_audio_model_cost,
|
11
|
+
create_metrics_attributes,
|
12
|
+
)
|
13
|
+
from openlit.semcov import SemanticConvention
|
14
|
+
|
15
|
+
def format_content(text):
|
16
|
+
"""
|
17
|
+
Process text input to extract content.
|
18
|
+
"""
|
19
|
+
return str(text) if text else ""
|
20
|
+
|
21
|
+
def common_span_attributes(scope, gen_ai_operation, gen_ai_system, server_address, server_port,
|
22
|
+
request_model, response_model, environment, application_name, is_stream, tbt, ttft, version):
|
23
|
+
"""
|
24
|
+
Set common span attributes for both chat and RAG operations.
|
25
|
+
"""
|
26
|
+
|
27
|
+
scope._span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
|
28
|
+
scope._span.set_attribute(SemanticConvention.GEN_AI_OPERATION, gen_ai_operation)
|
29
|
+
scope._span.set_attribute(SemanticConvention.GEN_AI_SYSTEM, gen_ai_system)
|
30
|
+
scope._span.set_attribute(SemanticConvention.SERVER_ADDRESS, server_address)
|
31
|
+
scope._span.set_attribute(SemanticConvention.SERVER_PORT, server_port)
|
32
|
+
scope._span.set_attribute(SemanticConvention.GEN_AI_REQUEST_MODEL, request_model)
|
33
|
+
scope._span.set_attribute(SemanticConvention.GEN_AI_RESPONSE_MODEL, scope._response_model)
|
34
|
+
scope._span.set_attribute(DEPLOYMENT_ENVIRONMENT, environment)
|
35
|
+
scope._span.set_attribute(SERVICE_NAME, application_name)
|
36
|
+
scope._span.set_attribute(SemanticConvention.GEN_AI_REQUEST_IS_STREAM, is_stream)
|
37
|
+
scope._span.set_attribute(SemanticConvention.GEN_AI_SERVER_TBT, scope._tbt)
|
38
|
+
scope._span.set_attribute(SemanticConvention.GEN_AI_SERVER_TTFT, scope._ttft)
|
39
|
+
scope._span.set_attribute(SemanticConvention.GEN_AI_SDK_VERSION, version)
|
40
|
+
|
41
|
+
def record_audio_metrics(metrics, gen_ai_operation, gen_ai_system, server_address, server_port,
|
42
|
+
request_model, response_model, environment, application_name, start_time, end_time, cost):
|
43
|
+
"""
|
44
|
+
Record audio generation metrics for the operation.
|
45
|
+
"""
|
46
|
+
|
47
|
+
attributes = create_metrics_attributes(
|
48
|
+
operation=gen_ai_operation,
|
49
|
+
system=gen_ai_system,
|
50
|
+
server_address=server_address,
|
51
|
+
server_port=server_port,
|
52
|
+
request_model=request_model,
|
53
|
+
response_model=response_model,
|
54
|
+
service_name=application_name,
|
55
|
+
deployment_environment=environment,
|
56
|
+
)
|
57
|
+
metrics["genai_client_operation_duration"].record(end_time - start_time, attributes)
|
58
|
+
metrics["genai_requests"].add(1, attributes)
|
59
|
+
metrics["genai_cost"].record(cost, attributes)
|
60
|
+
|
61
|
+
def common_audio_logic(scope, gen_ai_endpoint, pricing_info, environment, application_name,
|
62
|
+
metrics, capture_message_content, disable_metrics, version):
|
63
|
+
"""
|
64
|
+
Process audio generation request and generate Telemetry
|
65
|
+
"""
|
66
|
+
|
67
|
+
text = format_content(scope._kwargs.get("text", ""))
|
68
|
+
request_model = scope._kwargs.get("model", scope._kwargs.get("model_id", "eleven_multilingual_v2"))
|
69
|
+
is_stream = False # ElevenLabs audio generation is not streaming
|
70
|
+
|
71
|
+
cost = get_audio_model_cost(request_model, pricing_info, text)
|
72
|
+
|
73
|
+
# Common Span Attributes
|
74
|
+
common_span_attributes(scope,
|
75
|
+
SemanticConvention.GEN_AI_OPERATION_TYPE_AUDIO, SemanticConvention.GEN_AI_SYSTEM_ELEVENLABS,
|
76
|
+
scope._server_address, scope._server_port, request_model, request_model,
|
77
|
+
environment, application_name, is_stream, scope._tbt, scope._ttft, version)
|
78
|
+
|
79
|
+
# Span Attributes for Cost and Tokens
|
80
|
+
scope._span.set_attribute(SemanticConvention.GEN_AI_USAGE_COST, cost)
|
81
|
+
|
82
|
+
# Span Attributes for Response parameters
|
83
|
+
scope._span.set_attribute(SemanticConvention.GEN_AI_OUTPUT_TYPE, scope._kwargs.get("output_format", "mp3_44100_128"))
|
84
|
+
|
85
|
+
# Audio-specific span attributes
|
86
|
+
scope._span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_VOICE, scope._kwargs.get("voice_id", ""))
|
87
|
+
scope._span.set_attribute(SemanticConvention.GEN_AI_REQUEST_AUDIO_SETTINGS, str(scope._kwargs.get("voice_settings", "")))
|
88
|
+
|
89
|
+
# Span Attributes for Content
|
90
|
+
if capture_message_content:
|
91
|
+
scope._span.set_attribute(SemanticConvention.GEN_AI_CONTENT_PROMPT, text)
|
92
|
+
|
93
|
+
# To be removed once the change to span_attributes (from span events) is complete
|
94
|
+
scope._span.add_event(
|
95
|
+
name=SemanticConvention.GEN_AI_CONTENT_PROMPT_EVENT,
|
96
|
+
attributes={
|
97
|
+
SemanticConvention.GEN_AI_CONTENT_PROMPT: text,
|
98
|
+
},
|
99
|
+
)
|
100
|
+
|
101
|
+
scope._span.set_status(Status(StatusCode.OK))
|
102
|
+
|
103
|
+
# Metrics
|
104
|
+
if not disable_metrics:
|
105
|
+
record_audio_metrics(metrics, SemanticConvention.GEN_AI_OPERATION_TYPE_AUDIO, SemanticConvention.GEN_AI_SYSTEM_ELEVENLABS,
|
106
|
+
scope._server_address, scope._server_port, request_model, request_model, environment,
|
107
|
+
application_name, scope._start_time, scope._end_time, cost)
|
108
|
+
|
109
|
+
def process_audio_response(response, gen_ai_endpoint, pricing_info, server_port, server_address,
|
110
|
+
environment, application_name, metrics, start_time, span, args, kwargs, capture_message_content=False,
|
111
|
+
disable_metrics=False, version="1.0.0"):
|
112
|
+
"""
|
113
|
+
Process audio generation request and generate Telemetry
|
114
|
+
"""
|
115
|
+
|
116
|
+
scope = type("GenericScope", (), {})()
|
117
|
+
|
118
|
+
scope._start_time = start_time
|
119
|
+
scope._end_time = time.time()
|
120
|
+
scope._span = span
|
121
|
+
scope._server_address, scope._server_port = server_address, server_port
|
122
|
+
scope._kwargs = kwargs
|
123
|
+
scope._args = args
|
124
|
+
|
125
|
+
# Initialize streaming and timing values for ElevenLabs audio generation
|
126
|
+
scope._response_model = kwargs.get("model", kwargs.get("model_id", "eleven_multilingual_v2"))
|
127
|
+
scope._tbt = 0.0
|
128
|
+
scope._ttft = scope._end_time - scope._start_time
|
129
|
+
|
130
|
+
common_audio_logic(scope, gen_ai_endpoint, pricing_info, environment, application_name,
|
131
|
+
metrics, capture_message_content, disable_metrics, version)
|
132
|
+
|
133
|
+
return response
|
@@ -109,7 +109,7 @@ def record_embedding_metrics(metrics, gen_ai_operation, gen_ai_system, server_ad
|
|
109
109
|
metrics["genai_prompt_tokens"].add(input_tokens, attributes)
|
110
110
|
metrics["genai_cost"].record(cost, attributes)
|
111
111
|
|
112
|
-
def
|
112
|
+
def common_t2s_logic(scope, pricing_info, environment, application_name, metrics,
|
113
113
|
capture_message_content, disable_metrics, version, is_stream):
|
114
114
|
"""
|
115
115
|
Process generate request and generate Telemetry
|
@@ -228,7 +228,7 @@ def process_streaming_generate_response(scope, pricing_info, environment, applic
|
|
228
228
|
"""
|
229
229
|
Process generate request and generate Telemetry
|
230
230
|
"""
|
231
|
-
|
231
|
+
common_t2s_logic(scope, pricing_info, environment, application_name, metrics,
|
232
232
|
capture_message_content, disable_metrics, version, is_stream=True)
|
233
233
|
|
234
234
|
def process_generate_response(response, request_model, pricing_info, server_port, server_address,
|
@@ -252,7 +252,7 @@ def process_generate_response(response, request_model, pricing_info, server_port
|
|
252
252
|
scope._args = args
|
253
253
|
scope._tools = None
|
254
254
|
|
255
|
-
|
255
|
+
common_t2s_logic(scope, pricing_info, environment, application_name, metrics,
|
256
256
|
capture_message_content, disable_metrics, version, is_stream=False)
|
257
257
|
|
258
258
|
return response
|
@@ -1,4 +1,3 @@
|
|
1
|
-
# pylint: disable=useless-return, bad-staticmethod-argument, disable=duplicate-code
|
2
1
|
"""Initializer of Auto Instrumentation of Groq Functions"""
|
3
2
|
|
4
3
|
from typing import Collection
|
@@ -13,15 +12,15 @@ _instruments = ("groq >= 0.5.0",)
|
|
13
12
|
|
14
13
|
class GroqInstrumentor(BaseInstrumentor):
|
15
14
|
"""
|
16
|
-
An instrumentor for Groq
|
15
|
+
An instrumentor for Groq client library.
|
17
16
|
"""
|
18
17
|
|
19
18
|
def instrumentation_dependencies(self) -> Collection[str]:
|
20
19
|
return _instruments
|
21
20
|
|
22
21
|
def _instrument(self, **kwargs):
|
23
|
-
application_name = kwargs.get("application_name", "
|
24
|
-
environment = kwargs.get("environment", "
|
22
|
+
application_name = kwargs.get("application_name", "default")
|
23
|
+
environment = kwargs.get("environment", "default")
|
25
24
|
tracer = kwargs.get("tracer")
|
26
25
|
metrics = kwargs.get("metrics_dict")
|
27
26
|
pricing_info = kwargs.get("pricing_info", {})
|
@@ -29,22 +28,21 @@ class GroqInstrumentor(BaseInstrumentor):
|
|
29
28
|
disable_metrics = kwargs.get("disable_metrics")
|
30
29
|
version = importlib.metadata.version("groq")
|
31
30
|
|
32
|
-
#
|
31
|
+
# Chat completions
|
33
32
|
wrap_function_wrapper(
|
34
33
|
"groq.resources.chat.completions",
|
35
34
|
"Completions.create",
|
36
35
|
chat(version, environment, application_name,
|
37
|
-
|
36
|
+
tracer, pricing_info, capture_message_content, metrics, disable_metrics),
|
38
37
|
)
|
39
38
|
|
40
|
-
#
|
39
|
+
# Chat completions
|
41
40
|
wrap_function_wrapper(
|
42
41
|
"groq.resources.chat.completions",
|
43
42
|
"AsyncCompletions.create",
|
44
43
|
async_chat(version, environment, application_name,
|
45
|
-
|
44
|
+
tracer, pricing_info, capture_message_content, metrics, disable_metrics),
|
46
45
|
)
|
47
46
|
|
48
47
|
def _uninstrument(self, **kwargs):
|
49
|
-
# Proper uninstrumentation logic to revert patched methods
|
50
48
|
pass
|