SketchRegex

Data and code for the paper "Sketch-Driven Regular Expression Generation from Natural Language and Examples ".

@InProceedings{sketchregex,
 title = {Sketch-Driven Regular Expression Generation from Natural Language and Examples},
 author = {Xi Ye, Qiaochu Chen, Xinyu Wang, Isil Dillig, and Greg Durrett},
 booktitle = {Transactions of the Association for Computational Linguistics (TACL)},
 year = {2020},
}

Prerequisites

• pytorch > 1.0.0
• Z3. Make sure you have Z3 installed with the Java binding.
• JAVA 1.8.0

Code

Our sketch-driven framework can be instatiated with either

• a neural parser [DeepSketch]

or

• a grammar-based parser [GrammarSketch]. (Code comming soon. For now, please refer to Regal, "Multi-Modal Synthesis of Regular Expressions", which uses the same grammar based-parser with slightly different grammar rules.)

Please refer to the README of each module for details.