Sending a JSON object as API input

[dylanwe]

Is it possible to send a JSON object as API flow input_value like so:

curl -X POST \
    "http://localhost/api/v1/run/my-flow" \
    -H 'Content-Type: application/json'\
    -d '{"input_value": {
        "name": "Dante"
        "job": "Pilot",
        "message": "What could I do to make my passengers more comfortable?"
    },
    "output_type": "chat",
    "input_type": "chat",
    "tweaks": {}
}'

This could help change the flow based on JSON values.

[dosubot[bot]]

To parse the message and job fields within a Langflow flow, you can modify the run_flow_from_json function to include the extraction and processing logic. Here is an example of how you can achieve this:

import json
from pathlib import Path
from typing import List, Optional, Union

from dotenv import load_dotenv
from loguru import logger

from langflow.graph import Graph
from langflow.graph.schema import RunOutputs
from langflow.processing.process import process_tweaks, run_graph
from langflow.utils.logger import configure
from langflow.utils.util import update_settings

def run_flow_from_json(
    flow: Union[Path, str, dict],
    input_value: str,
    tweaks: Optional[dict] = None,
    input_type: str = "chat",
    output_type: str = "chat",
    output_component: Optional[str] = None,
    log_level: Optional[str] = None,
    log_file: Optional[str] = None,
    env_file: Optional[str] = None,
    cache: Optional[str] = None,
    disable_logs: Optional[bool] = True,
    fallback_to_env_vars: bool = False,
) -> List[RunOutputs]:
    """
    Run a flow from a JSON file or dictionary.

    Args:
        flow (Union[Path, str, dict]): The path to the JSON file or the JSON dictionary representing the flow.
        input_value (str): The input value to be processed by the flow.
        tweaks (Optional[dict], optional): Optional tweaks to be applied to the flow. Defaults to None.
        input_type (str, optional): The type of the input value. Defaults to "chat".
        output_type (str, optional): The type of the output value. Defaults to "chat".
        output_component (Optional[str], optional): The specific component to output. Defaults to None.
        log_level (Optional[str], optional): The log level to use. Defaults to None.
        log_file (Optional[str], optional): The log file to write logs to. Defaults to None.
        env_file (Optional[str], optional): The environment file to load. Defaults to None.
        cache (Optional[str], optional): The cache directory to use. Defaults to None.
        disable_logs (Optional[bool], optional): Whether to disable logs. Defaults to True.
        fallback_to_env_vars (bool, optional): Whether Global Variables should fallback to environment variables if not found. Defaults to False.

    Returns:
        List[RunOutputs]: A list of RunOutputs objects representing the results of running the flow.
    """
    # Set all streaming to false
    try:
        import nest_asyncio  # type: ignore
        nest_asyncio.apply()
    except Exception as e:
        logger.warning(f"Could not apply nest_asyncio: {e}")
    if tweaks is None:
        tweaks = {}
    tweaks["stream"] = False
    graph = load_flow_from_json(
        flow=flow,
        tweaks=tweaks,
        log_level=log_level,
        log_file=log_file,
        env_file=env_file,
        cache=cache,
        disable_logs=disable_logs,
    )
    
    # Extract and process the `message` and `job` fields from the input_value
    input_json = json.loads(input_value)
    message = input_json.get("message")
    job = input_json.get("job")
    
    # Process the extracted fields as needed
    # For example, you can log them or pass them to the flow
    logger.info(f"Message: {message}")
    logger.info(f"Job: {job}")
    
    result = run_graph(
        graph=graph,
        input_value=input_value,
        input_type=input_type,
        output_type=output_type,
        output_component=output_component,
        fallback_to_env_vars=fallback_to_env_vars,
    )
    return result

In this example, the run_flow_from_json function is modified to include the extraction of the message and job fields from the input_value JSON string. The extracted fields are then logged, but you can process them further as needed ^[1].

To continue talking to Dosu, mention @dosu.

[dylanwe]

@dosu How would I parse these values so I can use them in my flow?

[dosubot[bot]]

To parse the message and job values from the JSON input and use them effectively in your flow, you can modify the run_flow_from_json function to extract these values and pass them to the flow. Here is an updated version of the function:

import json
from pathlib import Path
from typing import List, Optional, Union

from dotenv import load_dotenv
from loguru import logger

from langflow.graph import Graph
from langflow.graph.schema import RunOutputs
from langflow.processing.process import process_tweaks, run_graph
from langflow.utils.logger import configure
from langflow.utils.util import update_settings

def run_flow_from_json(
    flow: Union[Path, str, dict],
    input_value: str,
    tweaks: Optional[dict] = None,
    input_type: str = "chat",
    output_type: str = "chat",
    output_component: Optional[str] = None,
    log_level: Optional[str] = None,
    log_file: Optional[str] = None,
    env_file: Optional[str] = None,
    cache: Optional[str] = None,
    disable_logs: Optional[bool] = True,
    fallback_to_env_vars: bool = False,
) -> List[RunOutputs]:
    """
    Run a flow from a JSON file or dictionary.

    Args:
        flow (Union[Path, str, dict]): The path to the JSON file or the JSON dictionary representing the flow.
        input_value (str): The input value to be processed by the flow.
        tweaks (Optional[dict], optional): Optional tweaks to be applied to the flow. Defaults to None.
        input_type (str, optional): The type of the input value. Defaults to "chat".
        output_type (str, optional): The type of the output value. Defaults to "chat".
        output_component (Optional[str], optional): The specific component to output. Defaults to None.
        log_level (Optional[str], optional): The log level to use. Defaults to None.
        log_file (Optional[str], optional): The log file to write logs to. Defaults to None.
        env_file (Optional[str], optional): The environment file to load. Defaults to None.
        cache (Optional[str], optional): The cache directory to use. Defaults to None.
        disable_logs (Optional[bool], optional): Whether to disable logs. Defaults to True.
        fallback_to_env_vars (bool, optional): Whether Global Variables should fallback to environment variables if not found. Defaults to False.

    Returns:
        List[RunOutputs]: A list of RunOutputs objects representing the results of running the flow.
    """
    # Set all streaming to false
    try:
        import nest_asyncio  # type: ignore
        nest_asyncio.apply()
    except Exception as e:
        logger.warning(f"Could not apply nest_asyncio: {e}")
    if tweaks is None:
        tweaks = {}
    tweaks["stream"] = False

    # Parse message and job from the input_value JSON string
    input_json = json.loads(input_value)
    message = input_json.get("message")
    job = input_json.get("job")
    logger.info(f"Parsed message: {message}, job: {job}")

    graph = load_flow_from_json(
        flow=flow,
        tweaks=tweaks,
        log_level=log_level,
        log_file=log_file,
        env_file=env_file,
        cache=cache,
        disable_logs=disable_logs,
    )
    result = run_graph(
        graph=graph,
        input_value=input_value,
        input_type=input_type,
        output_type=output_type,
        output_component=output_component,
        fallback_to_env_vars=fallback_to_env_vars,
    )
    return result

This modification adds the extraction of the message and job values from the input_value JSON string and logs them for verification. You can further process these values as needed within your flow ^[1][2].

To continue talking to Dosu, mention @dosu.