qoa_client
qoa4ml.qoa_client
¶
Classes¶
QoaClient
¶
Functions¶
__init__(report_cls=MLReport, config_dict=None, config_path=None, registration_url=None)
¶
Initialize the QoA Client with configuration settings and a report class.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
report_cls
¶ |
type[T]
|
The class type for reports, default is MLReport. |
MLReport
|
config_dict
¶ |
dict
|
A dictionary to load the client's configuration from. |
None
|
config_path
¶ |
str
|
Path to a JSON configuration file. |
None
|
registration_url
¶ |
str
|
URL for registering the client and receiving configuration data. |
None
|
Notes
- If both
config_dictandconfig_pathare provided, theconfig_dictwill take precedence. - If neither
config_dictnorconfig_pathis provided, the client may attempt to fetch configurations from theregistration_url. - The method will raise an exception if the necessary configuration details are not found.
__str__()
¶
Returns a string representation of the client's configuration and connectors.
Returns:
| Type | Description |
|---|---|
str
|
JSON representation of the client configuration and a string representation of the connector list. |
Notes
This method is particularly useful for debugging and logging purposes.
asyn_report(body_mess, connectors=None)
¶
Asynchronously send a report through the connectors.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
body_mess
¶ |
str
|
The message body to be sent. |
required |
connectors
¶ |
list
|
A list of connectors to send the report through. If None, the default connector is used. |
None
|
Notes
Uses threading to send reports asynchronously.
get_client_config()
¶
Get the current client configuration.
Returns:
| Type | Description |
|---|---|
ClientConfig
|
The client's current configuration settings. |
import_previous_report(reports)
¶
Import and process previous reports.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
reports
¶ |
Union[dict, list[dict]]
|
A single report or a list of reports to be processed. |
required |
init_connector(configuration)
¶
Initialize a connector based on the configuration provided.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
configuration
¶ |
ConnectorConfig
|
Configuration settings for initializing the connector. |
required |
Returns:
| Type | Description |
|---|---|
BaseConnector
|
An instance of the connector (e.g., AMQP, Debug). |
Raises:
| Type | Description |
|---|---|
RuntimeError
|
If the connector configuration type is not supported. |
init_probes(probe_config_list, client_info)
¶
Initialize monitoring probes based on the provided probe configuration list.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
probe_config_list
¶ |
list[ProbeConfig]
|
A list of configuration settings for each probe. |
required |
client_info
¶ |
ClientInfo
|
Information about the client to be passed to each probe. |
required |
Returns:
| Type | Description |
|---|---|
list[Probe]
|
A list of initialized probe instances. |
Raises:
| Type | Description |
|---|---|
ValueError
|
If an unsupported probe configuration type is provided. |
observe_inference(inference_value)
¶
Observe and record inference data.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
inference_value
¶ |
Any
|
The value of the inference to be observed. |
required |
Notes
This method is used to record predictions or inference results for later analysis.
observe_inference_metric(metric_name, value)
¶
Observe and report a specific inference metric.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
metric_name
¶ |
MetricNameEnum
|
The name of the inference metric being observed. |
required |
value
¶ |
Any
|
The value of the observed metric. |
required |
Notes
This method can be used to log performance metrics, evaluation scores, etc. during inference.
observe_metric(metric_name, value, category=0, description='')
¶
Observe and report a metric.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
metric_name
¶ |
MetricNameEnum
|
The name of the metric being observed. |
required |
value
¶ |
Any
|
The value of the observed metric. |
required |
category
¶ |
int
|
The category of the metric (0: service, 1: data, 2: security), default is 0. |
0
|
description
¶ |
str
|
An optional description of the observed metric, default is "". |
''
|
Raises:
| Type | Description |
|---|---|
RuntimeError
|
If the category type is not supported. |
registration(url)
¶
Registers the client with the monitoring service and retrieves connector configurations.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
url
¶ |
str
|
The registration URL to fetch the connector configuration from. |
required |
Returns:
| Type | Description |
|---|---|
Response
|
The response from the registration service, containing connector configurations. |
Notes
This method sends a POST request to the given URL with the client's configuration in JSON format.
report(report=None, connectors=None, submit=False, reset=True, corr_id=None)
¶
Generate a report and optionally submit it through the default connector.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
report
¶ |
dict
|
The report data to be submitted. If None, a report will be generated. |
None
|
connectors
¶ |
list
|
A list of connectors through which to send the report, default is None. |
None
|
submit
¶ |
bool
|
Whether to submit the report, default is False. |
False
|
reset
¶ |
bool
|
Whether to reset the report state after submission, default is True. |
True
|
corr_id
¶ |
str
|
The correlation ID for the report, default is None. |
None
|
Returns:
| Type | Description |
|---|---|
str
|
The JSON-encoded report. |
Notes
The method will create a report based on the current state if none is provided.
If submit is True, the report will be sent through the default or specified connectors.
set_config(key, value)
¶
Update a specific configuration setting by key.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
key
¶ |
str
|
The configuration attribute name to be updated. |
required |
value
¶ |
Any
|
The value to set for the specified key. |
required |
Raises:
| Type | Description |
|---|---|
Exception
|
Logs an error if setting the configuration value fails. |
start_all_probes()
¶
Start all probes for monitoring, running them in the background.
Raises:
| Type | Description |
|---|---|
RuntimeError
|
If no probes have been initialized. |
Notes
If the probe takes a long time to report and the main process exits, no report may be sent.
stop_all_probes()
¶
Stop all running probes.
Raises:
| Type | Description |
|---|---|
RuntimeError
|
If no probes have been initialized. |
Notes
This method stops the background monitoring activities of all active probes.
timer()
¶
Start or stop a timer and record the response time.
Returns:
| Type | Description |
|---|---|
dict
|
A dictionary containing the start time and response time. |
Notes
- When called for the first time, it starts the timer.
- When called again, it stops the timer and records the response time as a metric.