A CLI tool in Java that runs inference using published pre-trained LambdaNet model (ref: https://github.com/MrVPlusOne/LambdaNet) \w user-defined types. The tool receives a path to the pre-trained model, a path to the folder containing parsingFromFile.js (parsingFromFile.ts, pre-compiled with tsc) and all its dependencies (parsing.ts, etc.), and an input Typescript project folder as named CLI arguments and print the results to stdout. The output includes a specific line of code, a place in that line where the type suggestion should be added, Top5 most probable type suggestions.
- Download the zip file containing all the required JARs from here: https://drive.google.com/file/d/1GIpybi1s2aryDYtAKSG6ogRTHUY7jD7F/view?usp=sharing
- Unzip the zip file at any location (let's say 'lib' folder in the current directory)
- Download Driver.java from this repository and compile it using using the following command:
javac -cp lib\* Driver.java - Run the code using the following command:
java -cp lib\*;. Driver dir1 dir2 dir3
Here,
dir1 = Path to the model, i.e. ~\models\newParsing-GAT1-fc2-newSim-decay-6 in the folder extracted from models.zip which can be downloaded from: https://drive.google.com/file/d/1NvEVQ4-5tC3Nc-Mzpu3vYeyEcaM_zEgV/view?usp=sharing
dir2 = Path to the scripts, i.e. a folder containing these: (https://github.com/MrVPlusOne/LambdaNet/tree/master/scripts/ts) files, don't forget to pre-compile parsingFromFile.ts to parsingFromFile.js using tsc command, if not already compiled.
dir3 = the TypeScript project directory, where all the TypeScript files on which inference is to be run, are kept. - You'll get output as shown in the following image
The above screenshot was taken with regard to following folder structure:
