ImageMagick

This is a project used to demonstrate the sysnthesis functionality provided by APE.

The use case aims to describe the domain knowledge that has to be provided, comprising Domain ontology and Tool annotations). In addition it will demonstrate the usefulness of the synthesis approach for solving a workflow discovery problem with ImageMagick, an open-source software suite for displaying, creating, converting and modifying images.

Note

In order to be able to execute the generated workflows on the machine, Imagemagick needs to be installed (https://imagemagick.org/index.php). However, this step is not required for the workflow synthesis, considering that the synthesis of conadidate workflows and their execution are 2 separate steps.

In order to use the APE library from the command line, simply run the APE-<version>-executable.jar file using command:

java -jar APE-<version>-executable.jar [path_to_ape_configuration_file]

As an example, if you would download the APE-2.3.0-executable.jar to the root of APE_UseCases repository on your local machine, you could run this demo by executing the following command:

cd ~/git/APE_UseCases
java -jar APE-2.3.0-executable.jar ImageMagick/Example1/config.json

The results of the synthesis would be stored under the directory specified in the configuration file (solutions_dir_path parameter). The results of the synthesis would be:

solutions_dir_path/solutions.txt - First X candidate solutions in textual format, where X is the number of solutions specified in the config file (``solutions`` parameter)
solutions_dir_path/Figures/        - Workflow figures corresponding to the first Y solutions, where Y is the number of solutions specified in the config file (``number_of_generated_graphs`` parameter, 0 if not specified))
solutions_dir_path/Executables/  - Executable shell scripts corresponding to the first Z solution, where Z is the number of solutions specified in the config file (``number_of_execution_scripts`` parameter, 0 if not specified))
solutions_dir_path/CWL/ - CWL files corresponding to the first Q solution, where Q is the number of solutions specified in the config file (``number_of_cwl_files`` parameter, 0 if not specified)

Domain Model

The current use case ontology (see below) contains the following structure:

• thing (root class in the OWL file)

  • Tool (name provided as modulesTaxonomyRoot in config file)

  • Data (usage of this class is optional)

    • Type (name provided under dataDimensionsTaxonomyRoots in config file)

    • Format (name provided under dataDimensionsTaxonomyRoots in config file)

where the Tool represents the root of the operations taxonomy, and the Type and Format, represent roots of taxonomies that classify types and formats of data, respectively. The class Data can be ommited, considering that it is ignored by APE library, however it was part of the doman classification and we decided to keep it.

APE loads the domain ontology from a file in Web Ontology Language (OWL) format. Note that the the annotated tools (provided in the tool annotations are included in the image below as blue leafs, although they are not part in the OWL file.

Example 1

In this example a postcard is generated by providing a APE with an Image and some Content as input. By setting the right constraint, we make sure the Image will get a border.

Configuration

Input / output

{
    "inputs": [
        { "Type": ["Image"], "Format": ["PNG"] },
        { "Type": ["Content"] }
    ],
    "outputs": [
        { "Type": ["Image"], "Format": ["LosslessImageFormat"]  }
    ]
}

Constraints

At least one implemented tool of the abstract class Borders must be used in the solution workflow.

{
    "constraints": [
        {
            "constraintid": "use_m",
            "parameters": [
                { "Tool" : ["Borders"]}
            ]
        }
    ]
}

Results

Generated Workflow

Generated Script

workflowSolution_0.sh

#!/bin/bash
if [ $# -ne 2 ]
        then
                echo "2 argument(s) expected."
                exit
fi
node579867570=$1
node1548883689=$2

# [generate_color]
read -p "Enter Color [default:Cyan]: " node173380607
node173380607=${node173380607:-Cyan}

# [generate_font]
read -p "Enter FontFamily name [default:Helvetica]: " node1695080704
node1695080704=${node1695080704:-Helvetica}

# [add_title]
node579870453='node579870453.png'
node579867570_height=$(($(identify -format '%h' $node579867570)/8))
convert $node579867570 -fill $node173380607 -stroke black -font $node1695080704 -pointsize $node579867570_height -gravity north -annotate 0 "$node1548883689" $node579870453

# [add_large_border]
node579871414='node579871414.png'
node579870453_height=$(($(identify -format '%h' $node579870453)/10))
convert $node579870453 -bordercolor $node173380607 -border $node579870453_height $node579871414

echo "1. output is: $node579871414"

Executing the script

> bash workflowSolution_0.sh input.png "Greetings from the\nNetherlands!"
> Enter Color [Cyan]: Orange
> Enter FontFamily name [Helvetica]:
> output is: node579871414.png

INPUT (input.png):

add_title (node579870453.png)

add_large_border OUTPUT (node579871414.png):

Example 2

In this example we will try to replace the Color red by the Color blue.

Configuration

Input / output

{
        "inputs": [
                { "Type": ["Image"], "Format": ["JPG"] },
                { "Type": ["Color"], "Format": ["String"] },
                { "Type": ["Color"], "Format": ["String"] }
        ],
        "outputs": [
                { "Type": ["Image"] }
        ]
}

Constraints

Using our foreknowledge about images we assume there must be some kind of filtering involved, so we add the constraint to use a type Filter in the workflow solution.

After generating multiple workflows we notice that our provided Colors are all being used to add borders to our image, so we add a constraint to make sure that Border tools are not allowed in the workflow solution.

{
        "constraints": [
        {
                        "constraintid": "use_t",
                        "parameters": [
                                ["Filter"]
                        ]
                },
                {
                        "constraintid": "nuse_m",
                        "parameters": [
                                ["Borders"]
                        ]
                },
                {
                        "constraintid": "next_m",
                        "parameters": [
                                ["cut"],
                                ["overlap"]
                        ]
                }
        ]
}

Results

Generated Workflows

SolutionNo_1_length_4.pngSolutionNo_0_length_4.png

Generated scripts

workflowSolution_1.shworkflowSolution_0.sh

#!/bin/bash
if [ $# -ne 3 ]
    then
        echo "3 argument(s) expected."
        exit
fi
node539597562=$1
node173381567=$2
node173381566=$3

node19521642='node19521642.png'
convert $node539597562 -fuzz 35% -fill black +opaque $node173381566 +fuzz -fill white +opaque black $node19521642

node579869492='node579869492.png'
convert $node539597562 +level-colors $node173381567, $node579869492

node579870453='node579870453.png'
convert $node579869492 $node19521642 -compose multiply -composite -transparent black $node579870453

node579871414='node579871414.png'
convert $node539597562 $node579870453 -background none -layers flatten $node579871414

echo "1. output is: $node579871414"

Executing the script

> bash workflowSolution_1.sh input.jpg "Blue" "Red"
> output is: node579871414.png

INPUT (input.jpg):

color_filter (node19521642.png):

level_colors (node579869492.png):

cut (node579870453.png):

overlap (node579871414.png):