Skip to content

Latest commit

 

History

History
64 lines (53 loc) · 2.76 KB

README.md

File metadata and controls

64 lines (53 loc) · 2.76 KB
Raw

ProgramDataCollection

This repository contains code to create programs based on blockly library

Setup

Make sure you get the blockly submodule, modified to work with these programs. You can get it by cloning the repo recrusively.

git clone --recursive https://github.com/xavierpuigf/ProgramDataCollection.git

After cloning the repository. Download closure-library from google. This repo is based on version v20160619

git clone https://github.com/google/closure-library.git
git checkout tags/v20160619

Download bootstrap 2.3.2 and place it under libraries

cd libraries
wget https://getbootstrap.com/2.3.2/assets/bootstrap.zip
unzip bootstrap.zip

Contents

  • backend.js: starting point. Contains functions to create new actions, store them and submit tasks to AMT.
  • action.js: defines the class Action. Each action will contain a program and the block you want to store in it.
  • blocks.js: contains the definitions of the blocks. Each block starts with Blockly.Blocks['name_block'] you can create there your own blocks.
  • terminals.js: contains functions that fire when you modify the blocks, like setting the arguments etc. It also sets the toolbox to add blocks into the canvas.
  • knowledge_base/*: contains js files with information to help populate the blocks (number of actions, interactions etc).

Usage

Start annotation

Open interface.html in a browser as:

interface.html?filetasks={CONTENT}&mode={MODE}

Where {MODE} is a string containing the mode of annotation. It can be one of the following:

  • TASK_NAME: Creates a single task with name {CONTENT}.
  • TASK_FILE_NAME: Creates a task based on the contents of the file example_tasks/{CONTENT}
  • DATABASE: to debug. Checks from a database, based on which workers did what.
  • CREATED_FILE: to debug. Checks from a folder of saved programs

Submit annotations

When you click submit, the programs will be stored in a field answer as a string. You can check it as:

$('answer').val()

Note that the annotations are stored to work with AMT MATLAB annotation tool.

Parse annotations

Once you get the resulting string, you will need to parse it into a program. First create an .xml file from each program, either by checking the answer field or the save in ./php/xml_manager. Then parse the xml into a program running:

python parse_result.py {input_xml} {out_file}

As an example, you can run

python create_action_file.py test_programs/test_new_block.xml

and check the file out_prog.txt.

Note that the files test_programs/test{1,2,3}.xml follow an old format, which needs to be parsed differently.