all-nodes 1.7.1

PySide2-based visual programming for Python 3

Features • How to • Execution logic

all_nodes is a visual programming node editor designed to be used in the most flexible way possible. It also aims to prevent re-duplication and re-writing of code inside of a project, as well as making debugging easier with its one-node-at-a-time approach to execution.

💽 Install and run

Install from PyPI with: pip install all-nodes

And then run with all-nodes

💡 Features

✅ Drag and drop creation of nodes	✅ Class search
✅ Visual feedback of execution	✅ Execution in separate thread
✅ Preview of images and other elements	✅ Attribute editor
✅ Deactivation of nodes	✅ Annotations

📖 How to

Adding more classes/nodes

The places where all_nodes will search for node classes, scenes, etc is defined by the environment variable ALL_NODES_LIB_PATH

Folder structure example

Example of organization of a folder, that contains two libraries

Folder structure	Classes representation	Scenes

This folder has:

• A *node_lib named folder with node libraries, in this case 2 libraries: lib_1 and lib_2. Inside each library we can have:

  • As many py modules as needed
  • A styles.yml file to define default styles for these modules (default colors, icons...)
  • A icons folder for the classes contained in the modules

• A *scene_lib named folder with scenes saved in yaml format

Defining a node

Inside each module, nodes/classes can be defined as easy as this:

class GetEnvVariable(GeneralLogicNode):
    
    NICE_NAME = "Get env variable"
    HELP = "Get the value of an environment variable, with possibility of a fallback" \
           "value if the variable is not defined"

    INPUTS_DICT = {
        "env_variable_name": {"type": str},
        "fallback_value": {"type": str, "optional": True},
    }

    OUTPUTS_DICT = {"env_variable_value": {"type": str}}

    def run(self):
        import os

        env_variable_name = self.get_attribute_value("env_variable_name")
        fallback_value = self.get_attribute_value("fallback_value")
        env_var_value = os.getenv(env_variable_name)
        if env_var_value is None:
            if fallback_value:
                env_var_value = fallback_value
            else:
                self.fail(
                    "No environment variable '{}' found".format(env_variable_name)
                )
                return

        self.set_output("env_variable_value", env_var_value)

Key elements that a node MUST have:

• A very unique name to itself, that cleary indicates what it does
• A run method

Key elements that a node CAN have:

• A NICE_NAME attribute, for this node to be displayed more readable
• A HELP attribute to explain how the node works
• The INPUTS_DICT dictionary, if the node needs inputs
• The OUTPUTS_DICT dictionary, if the node needs outputs
• The INTERNALS_DICT dictionary, if the node needs inputs/previews through GUI

Other considerations:

• The import statements are kept inside the run method, so no ImportError is met when editing nodes outside the software they are meant for.

Defining a context

A context is a "node of nodes", or a node that contains a scene inside itself. To define it, just two ingredients are needed:

• As usual, a class in a .py file, with the IS_CONTEXT attribute set
• In the same folder as the .py file, a .ctx file (just a yaml file but with that extension instead) with the inner scene of the node

Once "inside" the context, its input attributes can be "grabbed" and its output attributes set by making use of the special ctx nodes:

Defining a scene

Defining a scene is very easy, it is a simple and readable yaml file.

To look for scenes, all_nodes will:

• Examine each path defined in ALL_NODES_LIB_PATH
• Then in that path, will look for a folder that contains "scene_lib" in its name
• From that folder, will recursively scan for and add all .yml files found

Example of scene:

# SCENE environ_to_yaml
# ---------------------
# Description: Simple scene to write the environ out to a yaml file

# Nodes section: overall list of nodes to be created
nodes:
- EnvironToYmlCtx_1:
    class_name: EnvironToYmlCtx
    x_pos: -904
    y_pos: -284
- StartFile_1:
    class_name: StartFile
    x_pos: -611
    y_pos: -304

# Connections section: connections to be done between nodes
connections:
- EnvironToYmlCtx_1.yaml_filepath -> StartFile_1.file_path


# Scene created at: 2023-01-24 23:13:10.371378
# Created by: jaime.rvq

The scene just needs two keys:

• nodes: with a list of nodes to be created
• connections: to establish the connections between the created nodes

Note how the -> symbol is used to make the connections section more easily readable.

Batch execution

Any .yml scene file can be executed in batch mode without the need to open the graphic editor.

This can be done by launching the main file and making use of the -f flag.

Example: main.py -f D:/scene_examples/my_scene.yml

Also, when a scene belongs to a repo/folder that has been defined in the ALL_NODES_LIB_PATH variable, it can be found just by its "alias". This alias is just the file name, with no need for the complete path or the ,yml extension.

Example: main.py -f environ_to_yaml

Finally, we can use the -s argument to set all the attributes we want to change in the scene prior to running. This flag accepts as many arguments as we need. Here is an example setting one node's attribute prior to running:

Example: main.py -f environ_to_yaml -s OPEN_FILE.internal_bool true

▶️ Execution logic

In a scene, the execution starts from nodes that are recognized as "starting nodes". Those are nodes that:

• Have no dependency/inputs/connetions from others.
• Don't have any non-optional attributes not set

In these example nodes, none of them have any incoming connections. Also, the only input attribute they all have is the START attribute, but since it is an optional attribute, the node can start executing even if it is not set.

Another example:

The node has a non-optional input. It is not connected, but a value has been defined to it, so the node can start executing.

Examples of execution order after the starting nodes

When a start node is executed, and it is successful, the nodes connected to its outputs will try to execute as well. Have a look at this example:

What will happen is:

• IntInput_1 is executed
• It is successful, so its output value "out_int" is propagated to the connected nodes
• IntInput_1 now launches the execution of the connected nodes
• The connected nodes are executed in random order. This time it starts with IntToStr_1
• IntToStr_1 runs successfully
• IntToStr_2 is executed, runs successfully

However, changing the scene a little bit, makes things more complex:

• IntInput_1 is executed
• It is successful, so its output value "out_int" is propagated to the connected nodes. Also, this time the "COMPLETED" attribute is propagated to the EmptyNode_1 connected node.
• IntInput_1 now launches the execution of the connected nodes
• The connected nodes are executed in random order. It has 3 connected nodes. This time it starts with IntToStr_2
• IntToStr_2 runs successfully
• IntToStr_1 is attempted to be executed, but has not recieved a value in its START attribute. It will not run now.
• EmptyNode_1 runs successfully. It propagates its COMPLETED value to IntToStr_1.
• EmptyNode_1 now launches the execution of its only connected node, IntToStr_1.
• IntToStr_1 now has all its connected inputs set, it can be executed. It runs successfully

Create and run node networks on the fly

For simple setups and tests, node networks can be created and run at the same time in a python script.

🐍 Example:

from all_nodes.logic.logic_scene import LogicScene
 
# Create a scene
logic_scene = LogicScene()

# Create a node and set one of its attributes
n_1 = logic_scene.add_node_by_name("GetEnvVariable")
n_1["env_variable_name"].set_value("USER")

# Create another node
n_2 = logic_scene.add_node_by_name("PrintToConsole")

# Connect the nodes
n_1["env_variable_value"].connect_to_other(n_2["in_object_0"])

# Run!
logic_scene.run_all_nodes()

📈 Analytics

Automatically generated once a month (for personal use, they are not gathered from any other user)