plugin manager

Plugin discovery, loading, and lifecycle management system

---

source

PluginManager

 PluginManager (plugin_interface:Type[cjm_transcription_plugin_system.plug
                in_interface.PluginInterface]=<class 'cjm_transcription_pl
                ugin_system.plugin_interface.PluginInterface'>,
                entry_point_group:str='transcription.plugins')

Manages plugin discovery, loading, and lifecycle.

	Type	Default	Details
plugin_interface	Type	PluginInterface	The base class/interface plugins must implement
entry_point_group	str	transcription.plugins	The entry point group name for plugin discovery

---

source

reload_plugin

 reload_plugin (plugin_name:str, config:Optional[Dict[str,Any]]=None)

*Reload a plugin with optional new configuration.

This is useful when you want to completely restart a plugin, for example after updating its code during development.*

	Type	Default	Details
plugin_name	str		Name of the plugin to reload
config	Optional	None	Optional new configuration
Returns	bool		True if successful, False otherwise

---

source

get_all_plugin_schemas

 get_all_plugin_schemas ()

*Get configuration schemas for all loaded plugins.

Returns a dictionary where keys are plugin names and values are their configuration schemas.*

---

source

validate_plugin_config

 validate_plugin_config (plugin_name:str, config:Dict[str,Any])

*Validate a configuration dictionary for a plugin without applying it.

Returns: Tuple of (is_valid, error_message). If valid, error_message is None.*

	Type	Details
plugin_name	str	Name of the plugin
config	Dict	Configuration to validate
Returns	Tuple	(is_valid, error_message)

---

source

update_plugin_config

 update_plugin_config (plugin_name:str, config:Dict[str,Any],
                       merge:bool=True)

*Update a plugin’s configuration and reinitialize it.

Args: plugin_name: Name of the plugin to update config: New configuration dictionary merge: If True, merge with existing config. If False, replace entirely.

Returns: True if configuration was successfully updated, False otherwise.*

	Type	Default	Details
plugin_name	str		Name of the plugin
config	Dict		New configuration
merge	bool	True	Whether to merge with existing config or replace entirely
Returns	bool		True if successful, False otherwise

---

source

get_plugin_config

 get_plugin_config (plugin_name:str)

*Get the current configuration of a plugin.

Returns the actual configuration values being used by the plugin, including any defaults.*

	Type	Details
plugin_name	str	Name of the plugin
Returns	Optional	Current configuration or None if plugin not found

---

source

get_plugin_config_schema

 get_plugin_config_schema (plugin_name:str)

*Get the configuration schema for a plugin.

Returns the JSON Schema that describes all configuration options available for the specified plugin.*

	Type	Details
plugin_name	str	Name of the plugin
Returns	Optional	Configuration schema or None if plugin not found

---

source

get_streaming_plugins

 get_streaming_plugins ()

*Get a list of all loaded plugins that support streaming.

Returns: List of plugin names that have streaming capabilities.*

---

source

check_streaming_support

 check_streaming_support (plugin_name:str)

*Check if a plugin supports streaming transcription.

Returns: True if the plugin implements execute_stream and supports streaming, False otherwise.*

	Type	Details
plugin_name	str	Name of the plugin to check
Returns	bool	True if plugin supports streaming

---

source

execute_plugin_stream

 execute_plugin_stream (plugin_name:str, audio:Any, **kwargs)

*Execute a plugin with streaming support if available.

This method will use the plugin’s execute_stream method if the plugin supports streaming, otherwise it falls back to the regular execute method.

Args: plugin_name: Name of the plugin to execute audio: Audio data or path to audio file **kwargs: Additional plugin-specific parameters

Yields: str: Partial transcription text chunks as they become available

Returns: Final transcription result from the plugin

Example: >>> # Stream transcription in real-time >>> for chunk in manager.execute_plugin_stream(“gemini”, audio_file): … print(chunk, end=““, flush=True)*

	Type	Details
plugin_name	str	Name of the plugin to execute
audio	Any	Audio data or path
kwargs	VAR_KEYWORD
Returns	Generator	Generator yielding text chunks, returns final result

logging.basicConfig(level=logging.INFO)
from nbdev.config import get_config
project_config = get_config()

test_files_dir = project_config.config_path/"test_files"
file_path = test_files_dir/"example_plugin.py"

# Create plugin manager
plugin_manager = PluginManager()

# Discover installed plugins
print("Discovering plugins...")
discovered = plugin_manager.discover_plugins()
print(f"Found {len(discovered)} plugins")

# Load discovered plugins
for plugin_meta in discovered:
    plugin_manager.load_plugin(plugin_meta, config={"debug": True})

print(plugin_manager.load_plugin_from_module(file_path))
example_plugin_name = plugin_manager.list_plugins()[0].name
print(f"Plugin Name: {example_plugin_name}")

# List loaded plugins
print("\nLoaded plugins:")
for meta in plugin_manager.list_plugins():
    print(f"  - {meta.name} v{meta.version} (enabled: {meta.enabled})")

# Execute a plugin
if plugin_manager.get_plugin(example_plugin_name):
    result = plugin_manager.execute_plugin(example_plugin_name, "test", key="value")
    print(f"\nPlugin result: {result}")

# Disable/enable plugins
plugin_manager.disable_plugin(example_plugin_name)
plugin_manager.enable_plugin(example_plugin_name)

INFO:example_plugin.WhisperPlugin:Initializing Whisper with config: {'model': 'base', 'device': 'auto', 'compute_type': 'default', 'language': None, 'task': 'transcribe', 'temperature': 0.0, 'beam_size': 5, 'best_of': 5, 'patience': 1.0, 'length_penalty': None, 'suppress_tokens': '-1', 'initial_prompt': None, 'condition_on_previous_text': True, 'no_speech_threshold': 0.6, 'compression_ratio_threshold': 2.4, 'logprob_threshold': -1.0, 'word_timestamps': False, 'prepend_punctuations': '"\'“¿([{-', 'append_punctuations': '"\'.。,，!！?？:：”)]}、', 'vad_filter': False, 'vad_parameters': {}}
INFO:__main__.PluginManager:Loaded plugin from module: whisper
INFO:example_plugin.WhisperPlugin:Transcribing with Whisper model: WhisperModel-base
INFO:example_plugin.WhisperPlugin:Execution config: {'model': 'base', 'device': 'auto', 'compute_type': 'default', 'language': None, 'task': 'transcribe', 'temperature': 0.0, 'beam_size': 5, 'best_of': 5, 'patience': 1.0, 'length_penalty': None, 'suppress_tokens': '-1', 'initial_prompt': None, 'condition_on_previous_text': True, 'no_speech_threshold': 0.6, 'compression_ratio_threshold': 2.4, 'logprob_threshold': -1.0, 'word_timestamps': False, 'prepend_punctuations': '"\'“¿([{-', 'append_punctuations': '"\'.。,，!！?？:：”)]}、', 'vad_filter': False, 'vad_parameters': {}, 'key': 'value'}

Discovering plugins...
Found 0 plugins
True
Plugin Name: whisper

Loaded plugins:
  - whisper v1.0.0 (enabled: True)

Plugin result: TranscriptionResult(text='Mock transcription using base model', confidence=0.95, segments=[{'start': 0.0, 'end': 2.5, 'text': 'Mock transcription', 'confidence': 0.96}, {'start': 2.5, 'end': 5.0, 'text': 'using base model', 'confidence': 0.94}], metadata={'model': 'base', 'language': None, 'device': 'auto', 'task': 'transcribe'})

True

Configuration Management Example

This section demonstrates how to use the configuration schema methods to manage plugin configurations.

# Demonstration of plugin configuration management
import json
from pathlib import Path

# Setup
logging.basicConfig(level=logging.INFO)
from nbdev.config import get_config
project_config = get_config()
test_files_dir = project_config.config_path/"test_files"
file_path = test_files_dir/"example_plugin.py"

# Create plugin manager and load the Whisper example plugin
plugin_manager = PluginManager()
plugin_manager.load_plugin_from_module(file_path)

print("="*60)
print("PLUGIN CONFIGURATION MANAGEMENT DEMONSTRATION")
print("="*60)

# 1. Get plugin configuration schema
schema = plugin_manager.get_plugin_config_schema("whisper")
print("\n1. Configuration Schema (first 5 properties):")
print("-" * 40)
if schema:
    props = list(schema.get("properties", {}).items())[:5]
    for prop_name, prop_schema in props:
        print(f"  {prop_name}:")
        print(f"    Type: {prop_schema.get('type')}")
        print(f"    Default: {prop_schema.get('default')}")
        if 'enum' in prop_schema:
            print(f"    Options: {prop_schema['enum'][:3]}...")

# 2. Get current plugin configuration
current_config = plugin_manager.get_plugin_config("whisper")
print("\n2. Current Configuration (subset):")
print("-" * 40)
if current_config:
    config_subset = {k: current_config[k] for k in ["model", "device", "language", "temperature"] if k in current_config}
    print(json.dumps(config_subset, indent=2))

# 3. Validate configurations
print("\n3. Configuration Validation:")
print("-" * 40)
test_configs = [
    ({"model": "small"}, "Valid partial config"),
    ({"model": "invalid"}, "Invalid model value"),
    ({"temperature": 2.0}, "Temperature out of range"),
]

for config, description in test_configs:
    is_valid, error = plugin_manager.validate_plugin_config("whisper", config)
    print(f"{description}:")
    print(f"  Config: {config}")
    print(f"  Valid: {is_valid}")
    if error:
        print(f"  Error: {error[:100]}...")  # Truncate long errors
    print()

# 4. Update plugin configuration
print("\n4. Updating Plugin Configuration:")
print("-" * 40)
new_config = {
    "model": "large-v3",
    "device": "cuda",
    "temperature": 0.2,
    "word_timestamps": True
}
success = plugin_manager.update_plugin_config("whisper", new_config, merge=True)
print(f"Update successful: {success}")

if success:
    updated_config = plugin_manager.get_plugin_config("whisper")
    config_subset = {k: updated_config[k] for k in ["model", "device", "temperature", "word_timestamps"] if k in updated_config}
    print("Updated configuration:")
    print(json.dumps(config_subset, indent=2))

# 5. Get all plugin schemas (useful for UI generation)
print("\n5. All Plugin Schemas:")
print("-" * 40)
all_schemas = plugin_manager.get_all_plugin_schemas()
for plugin_name, schema in all_schemas.items():
    print(f"Plugin: {plugin_name}")
    print(f"  Title: {schema.get('title', 'N/A')}")
    print(f"  Required fields: {schema.get('required', [])}")
    print(f"  Number of properties: {len(schema.get('properties', {}))}")

# 6. Execute plugin with runtime config override
print("\n6. Execute with Runtime Config Override:")
print("-" * 40)
result = plugin_manager.execute_plugin("whisper", "test_audio.wav", temperature=0.5, language="es")
print(f"Result: {result.text}")

INFO:example_plugin.WhisperPlugin:Initializing Whisper with config: {'model': 'base', 'device': 'auto', 'compute_type': 'default', 'language': None, 'task': 'transcribe', 'temperature': 0.0, 'beam_size': 5, 'best_of': 5, 'patience': 1.0, 'length_penalty': None, 'suppress_tokens': '-1', 'initial_prompt': None, 'condition_on_previous_text': True, 'no_speech_threshold': 0.6, 'compression_ratio_threshold': 2.4, 'logprob_threshold': -1.0, 'word_timestamps': False, 'prepend_punctuations': '"\'“¿([{-', 'append_punctuations': '"\'.。,，!！?？:：”)]}、', 'vad_filter': False, 'vad_parameters': {}}
INFO:__main__.PluginManager:Loaded plugin from module: whisper
INFO:example_plugin.WhisperPlugin:Cleaning up Whisper model
INFO:example_plugin.WhisperPlugin:Initializing Whisper with config: {'model': 'large-v3', 'device': 'cuda', 'compute_type': 'default', 'language': None, 'task': 'transcribe', 'temperature': 0.2, 'beam_size': 5, 'best_of': 5, 'patience': 1.0, 'length_penalty': None, 'suppress_tokens': '-1', 'initial_prompt': None, 'condition_on_previous_text': True, 'no_speech_threshold': 0.6, 'compression_ratio_threshold': 2.4, 'logprob_threshold': -1.0, 'word_timestamps': True, 'prepend_punctuations': '"\'“¿([{-', 'append_punctuations': '"\'.。,，!！?？:：”)]}、', 'vad_filter': False, 'vad_parameters': {'threshold': 0.5, 'min_speech_duration_ms': 250, 'max_speech_duration_s': 3600}}
INFO:__main__.PluginManager:Updated configuration for plugin: whisper
INFO:example_plugin.WhisperPlugin:Transcribing with Whisper model: WhisperModel-large-v3
INFO:example_plugin.WhisperPlugin:Execution config: {'model': 'large-v3', 'device': 'cuda', 'compute_type': 'default', 'language': 'es', 'task': 'transcribe', 'temperature': 0.5, 'beam_size': 5, 'best_of': 5, 'patience': 1.0, 'length_penalty': None, 'suppress_tokens': '-1', 'initial_prompt': None, 'condition_on_previous_text': True, 'no_speech_threshold': 0.6, 'compression_ratio_threshold': 2.4, 'logprob_threshold': -1.0, 'word_timestamps': True, 'prepend_punctuations': '"\'“¿([{-', 'append_punctuations': '"\'.。,，!！?？:：”)]}、', 'vad_filter': False, 'vad_parameters': {'threshold': 0.5, 'min_speech_duration_ms': 250, 'max_speech_duration_s': 3600}}

============================================================
PLUGIN CONFIGURATION MANAGEMENT DEMONSTRATION
============================================================

1. Configuration Schema (first 5 properties):
----------------------------------------

2. Current Configuration (subset):
----------------------------------------
{
  "model": "base",
  "device": "auto",
  "language": null,
  "temperature": 0.0
}

3. Configuration Validation:
----------------------------------------
Valid partial config:
  Config: {'model': 'small'}
  Valid: True

Invalid model value:
  Config: {'model': 'invalid'}
  Valid: False
  Error: 'invalid' is not one of ['tiny', 'tiny.en', 'base', 'base.en', 'small', 'small.en', 'medium', 'mediu...

Temperature out of range:
  Config: {'temperature': 2.0}
  Valid: False
  Error: 'model' is a required property

Failed validating 'required' in schema:
    {'$schema': 'http://json...


4. Updating Plugin Configuration:
----------------------------------------
Update successful: True
Updated configuration:
{
  "model": "large-v3",
  "device": "cuda",
  "temperature": 0.2,
  "word_timestamps": true
}

5. All Plugin Schemas:
----------------------------------------
Plugin: whisper
  Title: Whisper Configuration
  Required fields: ['model']
  Number of properties: 21

6. Execute with Runtime Config Override:
----------------------------------------
Result: Mock transcription using large-v3 model

Simple Configuration Workflow Example

# Simple example: Loading a plugin with configuration
from pathlib import Path

# Setup
from nbdev.config import get_config
project_config = get_config()
test_files_dir = project_config.config_path/"test_files"

# Create plugin manager
manager = PluginManager()

# Load plugin with initial configuration
initial_config = {
    "model": "tiny",  # Start with fast model
    "device": "cpu",
    "language": "en"
}

manager.load_plugin_from_module(
    test_files_dir/"example_plugin.py", 
    config=initial_config
)

# Check what configuration is being used
config = manager.get_plugin_config("whisper")
print(f"Model: {config['model']}")
print(f"Device: {config['device']}")

# Later, update to use a better model
manager.update_plugin_config("whisper", {"model": "base"})

# Execute transcription
result = manager.execute_plugin("whisper", "audio.wav")
print(f"Transcription: {result.text}")

INFO:example_plugin.WhisperPlugin:Initializing Whisper with config: {'model': 'tiny', 'device': 'cpu', 'compute_type': 'default', 'language': 'en', 'task': 'transcribe', 'temperature': 0.0, 'beam_size': 5, 'best_of': 5, 'patience': 1.0, 'length_penalty': None, 'suppress_tokens': '-1', 'initial_prompt': None, 'condition_on_previous_text': True, 'no_speech_threshold': 0.6, 'compression_ratio_threshold': 2.4, 'logprob_threshold': -1.0, 'word_timestamps': False, 'prepend_punctuations': '"\'“¿([{-', 'append_punctuations': '"\'.。,，!！?？:：”)]}、', 'vad_filter': False, 'vad_parameters': {}}
INFO:__main__.PluginManager:Loaded plugin from module: whisper
INFO:example_plugin.WhisperPlugin:Cleaning up Whisper model
INFO:example_plugin.WhisperPlugin:Initializing Whisper with config: {'model': 'base', 'device': 'cpu', 'compute_type': 'default', 'language': 'en', 'task': 'transcribe', 'temperature': 0.0, 'beam_size': 5, 'best_of': 5, 'patience': 1.0, 'length_penalty': None, 'suppress_tokens': '-1', 'initial_prompt': None, 'condition_on_previous_text': True, 'no_speech_threshold': 0.6, 'compression_ratio_threshold': 2.4, 'logprob_threshold': -1.0, 'word_timestamps': False, 'prepend_punctuations': '"\'“¿([{-', 'append_punctuations': '"\'.。,，!！?？:：”)]}、', 'vad_filter': False, 'vad_parameters': {'threshold': 0.5, 'min_speech_duration_ms': 250, 'max_speech_duration_s': 3600}}
INFO:__main__.PluginManager:Updated configuration for plugin: whisper
INFO:example_plugin.WhisperPlugin:Transcribing with Whisper model: WhisperModel-base
INFO:example_plugin.WhisperPlugin:Execution config: {'model': 'base', 'device': 'cpu', 'compute_type': 'default', 'language': 'en', 'task': 'transcribe', 'temperature': 0.0, 'beam_size': 5, 'best_of': 5, 'patience': 1.0, 'length_penalty': None, 'suppress_tokens': '-1', 'initial_prompt': None, 'condition_on_previous_text': True, 'no_speech_threshold': 0.6, 'compression_ratio_threshold': 2.4, 'logprob_threshold': -1.0, 'word_timestamps': False, 'prepend_punctuations': '"\'“¿([{-', 'append_punctuations': '"\'.。,，!！?？:：”)]}、', 'vad_filter': False, 'vad_parameters': {'threshold': 0.5, 'min_speech_duration_ms': 250, 'max_speech_duration_s': 3600}}

Model: tiny
Device: cpu
Transcription: Mock transcription using base model