Skip to main content

Blocks

General​

A Block represents a node within the Directed Acyclic Graph (DAG), each embodying a specific processing logic. Typically, a Block comprises several components:

  • Block Info:
    • Block Name: The name of the block.
    • Block ID: The unique identifier of the block.
    • Copy Block ID: A feature to copy the block's ID.
  • Conditional Port: Accepts True or False outputs from upstream condition type blocks, determining whether to execute this block based on the upstream result.
  • Inport: The information inputted into this block. Some blocks support multiple inports. Inports can handle text, list, dict data types.
    • Add an Inport: Adds an additional inport to the block. Manually added inports support connections of any data type.
  • Outport: The information outputted from this block. Outports typically support text, list, dict, boolean data types.
  • Parameters: Specific parameter details for the block, which vary from one block type to another.

Block Categories​

LinguFlow offers essential Block categories necessary for building LinguFlow applications, including:

  • Input & Output Category: For receiving inputs and sending outputs.
  • Data Process Category: For manipulating and processing data.
  • Condition Category: For making decisions based on certain conditions.
  • LLM Category: For integrating Large Language Model functionalities.
  • Invoke Category: For calling other blocks or applications within LinguFlow.
  • Tools Category: For utilizing third-party tools or services.

Input & Output Category​

Text_Input​

  • Description: Defines the input type for a LinguFlow App. It allows sending information of text type to the LinguFlow application during an API call.
  • Outport: text
  • Example:
- Outport: "Who are you?"

List_Input​

  • Description: Defines the input type for a LinguFlow App. It supports sending information of list type to the LinguFlow application during an API call.
  • Outport: list
  • Example:
- Outport: ["hi", "Hello, can I help you?", "Who are you?"]

Dict_Input​

  • Description: Defines the input type for a LinguFlow App. It facilitates sending information of dict type to the LinguFlow application during an API call.
  • Outport: dict
  • Example:
- Outport: {"key_1": "value_1", "key_2": "value_2"}

Text_Output​

  • Description: Specifies the output type for a LinguFlow App. After the LinguFlow application has finished running, it outputs a result of text type.
  • Inport: text
  • Example:
- Inport: "I'm LinguFlow."

Data Processing Category​

Dict_KeySelect_to_Text​

  • Description: Selects a specific key's value from a dict.
  • Inport: dict
  • Outport: text
  • Parameters:
    • Key: Specifies a particular key.
  • Example:
- Inport: {"key_1": "value_1", "key_2": "value_2"}
- Parameters: key="key_1"
- Outport: "value_1"

List_Join_to_Text​

  • Description: Merges a list into a single text string. The list typically contains multiple text elements.
  • Inport: Supports multiple inports, each corresponding to a list.
  • Outport: Merged text
  • Parameters:
    • Template: Describes how to render each list element.
    • Delimiter: Describes how to concatenate the rendered elements.
  • Example:
- Inport: ["text 1", "text 2", "text 3"]
- Parameters:
    - template="--{input}--"
    - delimiter='\n'
- Outport: "--text 1--\n--text 2--\n--text 3--\n"

List_Concat_to_List​

  • Description: Concatenates two lists in sequence into a single list.
  • Inport:
    • seq1: list
    • seq2: list
  • Outport: Concatenated list.
  • Example:
- Inport: ["a", "b"], ["1", "2"]
- Outport: ["a", "b", "1", "2"]

Text_Join_to_Dict​

  • Description: Combines multiple texts into a dict.
  • Inport: Supports multiple inports, each corresponding to a text.
  • Outport: Combined dict.
  • Example:
- Inport:
    - Inport name: "a", value: "b"
    - Inport name: "1", value: "2"
- Outport: {"a": "b", "1": "2"}

Text_Split_to_List​

  • Description: Parses text into a list based on specified rules.
  • Inport: text
  • Outport: list
  • Parameters:
    • Delim: The delimiter character in the text.
    • Prefix: Trims a prefix from each element.
    • Suffix: Trims a suffix from each element.
  • Example:
- Inport: "|abc-\n|def-\n|ghi-\n"
- Parameters:
    - delim='\n'
    - prefix='|'
    - suffix='-'
- Outport: ["abc", "def", "ghi"]

Condition Category​

Text_Condition​

  • Description: Evaluates a text input and outputs a boolean result based on the evaluation.
  • Inport: text
  • Outport: boolean
  • Parameters:
    • Comparator: Supports text evaluation logic, including "Text Equals", "Text Contains", "Text Has Prefix", "Text Has Suffix". Outputs "True" if the condition is met, "False" otherwise.
  • Example:
- Inport: "success"
- Parameters:
    - comparator: Text_Equal_With
        - value: "success"
- Outport: "True"

List_Condition​

  • Description: Evaluates a list input and outputs a boolean result based on the evaluation.
  • Inport: list
  • Outport: boolean
  • Parameters:
    • Comparator: Supports list evaluation logic, including "List Contains", "List Is Empty". Outputs "True" if the condition is met, "False" otherwise.
  • Example:
- Inport: ["abc", "def", "ghi"]
- Parameters:
    - comparator: List_Is_Empty
- Outport: "false"

LLM Category​

LLM​

  • Description: Interacts with an LLM.
  • Inport: Default inport is text. Additional inports can be created for input, which can be referenced in the template by inserting the inport name.
  • Outport: text
  • Parameters:
    • Model: Currently supports OpenAI's series of language models.
    • Prompt Template Type: Currently supports Zero_Shot_Prompt_Template and Few_Shot_Prompt_Template.
  • Example:
- Inport: "who are you?"
- Parameters:
    - model: OpenAI_Chat_LLM
        - openai_api_key: {key}
        - temperature: 0
        - max_tokens: 4096
        - model_name: "gpt-3.5-turbo"
    - prompt_template_type: Zero_Shot_Prompt_Template
        - prompt_template: "you are a useful assistant.\n question: {text}"
- Outport: "I'm LinguFlow."

Invoke Category​

Text_Invoke​

  • Description: Invokes another LinguFlow application, transferring text type content to it.
  • Inport: text
  • Outport: text
  • Parameters:
    • app_id: Enter the ID of the LinguFlow application you wish to invoke.
    • timeout: Invocation timeout in seconds.
  • Example:
- Inport: "who are you?"
- Parameters:
    - app_id: {id}
    - timeout: 300
- Outport: "I'm LinguFlow."

List_Invoke​

  • Description: Invokes another LinguFlow application, transferring list type content to it.
  • Inport: list
  • Outport: text
  • Parameters:
    • app_id: Enter the ID of the LinguFlow application you wish to invoke.
    • timeout: Invocation timeout in seconds.
  • Example:
- Inport: ["hi", "Hello, can I help you?", "who are you?"]
- Parameters:
    - app_id: {id}
    - timeout: 300
- Outport: "I'm LinguFlow."

Dict_Invoke​

  • Description: Invokes another LinguFlow application, transferring dict type content to it.
  • Inport: dict
  • Outport: text
  • Parameters:
    • app_id: Enter the ID of the LinguFlow application you wish to invoke.
    • timeout: Invocation timeout in seconds.
  • Example:
- Inport: {"key_1": "value_1", "key_2": "value_2"}
- Parameters:
    - app_id: {id}
    - timeout: 300
- Outport: "value_3"

Tools Category​

  • Description: Queries results using the Google search engine.
  • Inport: text
  • Outport: list
  • Parameters:
  • Example:
- Inport: "who are you?"
- Parameters:
    - search_engine_id: {id}
    - key: {key}
    - top_k: 2
- Outport: ["I'm LinguFlow.", "I'm LinguFlow."]