Colang 标准库 (CSL)
简介
Colang 标准库 (CSL) 提供了对底层事件和动作层的抽象，并提供了一个语义接口，用于设计 Bot 和用户之间的交互模式。目前，以下库文件位于 nemoguardrails/colang/v2_x/library/ 目录下 (Github 链接)
core.co: 基础核心 Flowtiming.co: 定时器相关 Flowavatars.co: 用于处理具有化身界面的多模态交互系统的 Flowllm.co: LLM 相关核心 Flowguardrails.co: Guardrail 相关 Flowutils.co: 一些有用的助手和实用工具 Flow
要使用这些库中定义的 Flow，您有两种选择
使用 import 语句导入标准库文件：例如
import llm将相应的 *.co 文件直接复制到您的 Colang 脚本目录中。
请注意，import <library> 语句将导入相应库的所有可用 Flow。
基础核心 Flow (core.co)
核心库，包含所有与用户和 Bot 话语事件和动作相关的 Flow。
用户事件 Flow
# Wait for a user to have said given text
flow user said $text -> $transcript
# Wait for a user to have said something
flow user said something -> $transcript
# Wait for a user to say given text while talking
flow user saying $text -> $transcript
# Wait for any ongoing user utterance
flow user saying something -> $transcript
# Wait for start of user utterance
flow user started saying something
# Wait for a user to have said something unexpected (no active match statement)
flow user said something unexpected -> $transcript
Bot 动作 Flow
# Trigger a specific bot utterance
flow bot say $text
# Trigger the bot to inform about something (semantic 'bot say' wrapper)
flow bot inform $text
# Trigger the bot to ask something (semantic 'bot say' wrapper)
flow bot ask $text
# Trigger the bot to express something (semantic 'bot say' wrapper)
flow bot express $text
# Trigger the bot to respond with given text (semantic 'bot say' wrapper)
flow bot respond $text
# Trigger the bot to clarify something (semantic 'bot say' wrapper)
flow bot clarify $text
# Trigger the bot to suggest something (semantic 'bot say' wrapper)
flow bot suggest $text
Bot 事件 Flow
# Wait for the bot starting with the given utterance
flow bot started saying $text
# Wait for the bot starting with any utterance
flow bot started saying something
# Wait for the bot to finish saying given utterance
flow bot said $text
# Wait for the bot to finish with any utterance
flow bot said something -> $text
# Wait for the bot to finish informing about something
flow bot informed something -> $text
# Wait for the bot to finish asking about something
flow bot asked something -> $text
# Wait for the bot to finish expressing something
flow bot expressed something -> $text
# Wait for the bot to finish responding something
flow bot responded something -> $text
# Wait for the bot to finish clarifying something
flow bot clarified something -> $text
# Wait for the bot to finish suggesting something
flow bot suggested something -> $text
状态跟踪 Flow
这些 Flow 在全局变量中跟踪 Bot 和用户状态。
# Track bot talking state in global variable $bot_talking_state
flow tracking bot talking state
# Track user utterance state in global variables: $user_talking_state, $last_user_transcript
flow tracking user talking state
开发助手 Flow
# A flow to notify about any runtime Colang errors
flow notification of colang errors
# A flow to notify about the start of an undefined flow
flow notification of undefined flow start
# A flow to notify about an unhandled user utterance
flow notification of unexpected user utterance
定时 Flow (timing.co)
# Little helper flow to wait indefinitely
flow wait indefinitely
# Wait the specified number of seconds before continuing
flow wait $time_s $timer_id="wait_timer_{uid()}"
# Start a repeating timer
flow repeating timer $timer_id $interval_s
# Wait for when user was silent for $time_s seconds
flow user was silent $time_s
# Wait for when user was silent for $time_s seconds while bot was silent
flow user didnt respond $time_s
# Wait for the bot to be silent (no utterance) for given time
flow bot was silent $time_s
# Trigger a specific bot gesture delayed
flow bot gesture with delay $gesture $delay
交互式化身模态 Flow (avatars.co)
用户事件 Flow
# Wait for a UI selection
flow user selected choice $choice_id -> $choice
# Wait for a UI selection to have happened (considering also choices that happened right before)
flow user has selected choice $choice_id
# Wait for user entering keystrokes in UI text field
flow user typing $text -> $inputs
# Wait for user to make a gesture
flow user gestured $gesture -> $final_gesture
# Wait for user to be detected as present (e.g. camera ROI)
flow user became present -> $user_id
# Wait for when the user talked while bot is speaking
flow user interrupted bot talking $sentence_length=15
Bot 动作 Flow
# Trigger a specific bot gesture
flow bot gesture $gesture
# Trigger a specific bot posture
flow bot posture $posture
# Show a 2D UI with some options to select from
flow scene show choice $prompt $options
# Show a 2D UI with detailed information
flow scene show textual information $title $text $header_image
# Show a 2D UI with a short information
flow scene show short information $info
# Show a 2D UI with some input fields to be filled in
flow scene show form $prompt $inputs
Bot 事件 Flow
# Wait for the bot to start with the given gesture
flow bot started gesture $gesture
# Wait for the bot to start with any gesture
flow bot started a gesture -> $gesture
# Wait for the bot to start with the given posture
flow bot started posture $posture
# Wait for the bot to start with any posture
flow bot started a posture -> $posture
# Wait for the bot to start with any action
flow bot started an action -> $action
状态跟踪 Flow
这些 Flow 在全局变量中跟踪 Bot 和用户状态。
# Track most recent visual choice selection state in global variable $choice_selection_state
flow tracking visual choice selection state
助手 & 实用工具 Flow
这些是一些有用的助手和实用工具 Flow
# Stops all the current bot actions
flow finish all bot actions
# Stops all the current scene actions
flow finish all scene actions
# Handling the bot talking interruption reaction
flow handling bot talking interruption $mode="inform"
姿势管理 Flow
# Activates all the posture management
flow managing bot postures
# Start and stop listening posture
flow managing listening posture
# Start and stop talking posture
flow managing talking posture
# Start and stop thinking posture
flow managing thinking posture
# Start and stop idle posture
flow managing idle posture
LLM Flow (llm.co)
LLM 启用的 Bot 动作
# Trigger a bot utterance similar to given text
flow bot say something like $text
LLM 实用工具
# Start response polling for all LLM related calls to receive the LLM responses an act on that
flow polling llm request response $interval=1.0
交互继续
用于在未处理的用户动作/意图或未定义的 Flow 时继续当前交互的 Flow。
# Activate all LLM based interaction continuations
flow llm continuation
# Generate a user intent event (finish flow event) for unhandled user utterance
flow generating user intent for unhandled user utterance
# Wait for the end of any flow with the name starting with 'user ' (considered a user intent)
flow unhandled user intent -> $intent
# Generate and start new flow to continue the interaction for an unhandled user intent
flow continuation on unhandled user intent
# Generate and start a new flow to continue the interaction for the start of an undefined flow
flow continuation on undefined flow
# Generate a flow that continues the current interaction
flow llm generate interaction continuation flow -> $flow_name
# Generate and continue with a suitable interaction
flow llm continue interaction
交互历史记录
用于记录交互历史记录以创建 LLM Prompt 所需上下文的 Flow。
# Activate all automated user and bot intent flows logging based on flow naming
flow automating bot user intent logging
# Marking user intent flows using only naming convention
flow marking user intent flows
# Generate user intent logging for marked flows that finish by themselves
flow logging marked user intent flows
# Marking bot intent flows using only naming convention
flow marking bot intent flows
# Generate user intent logging for marked flows that finish by themselves
flow logging marked bot intent flows
状态跟踪 Flow
这些 Flow 在全局变量中跟踪 Bot 和用户状态。
# Track most recent unhandled user intent state in global variable $user_intent_state
flow tracking unhandled user intent state
Guardrail Flow (guardrails.co)
用于 Guardrail 用户输入和 LLM 响应的 Flow。
# Check user utterances before they get further processed
flow run input rails $input_text
# Check llm responses before they get further processed
flow run output rails $output_text
实用工具 Flow (utils.co)
一些有用的通用助手和实用工具 Flow。
# Start a flow with the provided name and wait for it to finish
flow await_flow_by_name $flow_name