Transformation templates to automate aspects of computer programming
Implementations are described herein for building and/or applying a library of transformation templates to automate migration of source code. In various implementations, pre-migration and post-migration versions of source code that exist prior to and after migration of the source code may be analyzed. Based on the analysis, one or more transformations made to the pre-migration version of the source code to yield the post-migration version of the source code may be identified. A library of transformation templates that are applicable subsequently to automate migration of new source code may be built. In some implementations, for one or more of the transformations, a plurality of candidate transformation templates may be generated with different permutations of tokens being replaced with placeholders. One of the plurality of candidate transformation templates may be selected for inclusion in the library based on one or more criteria.
1. A method implemented using one or more processors, comprising:
analyzing pre-migration and post-migration versions of source code that exist prior to and after migration of the source code;
based on the analyzing, identifying one or more transformations made to the pre-migration version of the source code to yield the post-migration version of the source code; and
building a library of transformation templates that are applicable subsequently to automate migration of new source code, wherein the building includes, for one or more of the transformations:
generating a plurality of candidate transformation templates, wherein for each candidate transformation template, different permutations of tokens of the transformation are replaced with placeholders, and
selecting one of the plurality of candidate transformation templates for inclusion in the library of transformation templates, wherein the selecting is based on one or more criteria.
2. The method of claim 1 , wherein the library comprises a lattice of transformation templates.
3. The method of claim 1 , wherein the one or more criteria include successful application of the candidate transformation template to a pre-migration version training source code snippet to accurately generate a post-migration version of the training source code snippet.
4. The method of claim 1 , wherein the one or more criteria include preservation of a programming language keyword in the candidate transformation template.
5. The method of claim 1 , wherein the one or more criteria include a count of the transformations being implementable using the candidate transformation template.
6. The method of claim 1 , wherein the one or more transformations comprise a plurality of transformations, and the method further includes:
grouping the plurality of transformations into a plurality of clusters;
generating a plurality of candidate transformation templates for a given cluster of the plurality of clusters, and
selecting, based on one or more of the criteria, one of the plurality of candidate transformation templates generated for the given cluster for inclusion in the library of transformation templates.
7. The method of claim 6 , wherein the grouping includes encoding each transformation of the plurality of transformations into an embedding.
8. The method of claim 7 , wherein the encoding is based on a transformer network and a graph neural network (GNN).
9. The method of claim 1 , wherein one or more of the transformations includes contextual code surrounding the transformation.
10. The method of claim 1 , further comprising:
analyzing a pre-migration version of the new source code to match one or more transformation templates from the library to one or more snippets of the new source code; and
applying the matched one or more transformation templates to the one or more snippets of the new source code to generate a post-migration version of the new source code.
11. A system comprising one or more processors and memory storing instructions that, in response to execution of the instructions by the one or more processors, cause the one or more processors to:
analyze pre-migration and post-migration versions of source code that exist prior to and after migration of the source code;
based on the analysis, identify one or more transformations made to the pre-migration version of the source code to yield the post-migration version of the source code; and
build a library of transformation templates that are applicable subsequently to automate migration of new source code, wherein the instructions to build include, for one or more of the transformations, instructions to:
generate a plurality of candidate transformation templates, wherein for each candidate transformation template, different permutations of tokens of the transformation are replaced with placeholders, and
select one of the plurality of candidate transformation templates for inclusion in the library of transformation templates, wherein the selecting is based on one or more criteria.
12. The system of claim 11 , wherein the library comprises a lattice of transformation templates.
13. The system of claim 11 , wherein the one or more criteria include successful application of the candidate transformation template to a pre-migration version training source code snippet to accurately generate a post-migration version of the training source code snippet.
14. The system of claim 11 , wherein the one or more criteria include preservation of a programming language keyword in the candidate transformation template.
15. The system of claim 11 , wherein the one or more criteria include a count of the transformations being implementable using the candidate transformation template.
16. The system of claim 11 , wherein the one or more transformations comprise a plurality of transformations, and the instructions include instructions to:
group the plurality of transformations into a plurality of clusters;
generate a plurality of candidate transformation templates for a given cluster of the plurality of clusters, and
select, based on one or more of the criteria, one of the plurality of candidate transformation templates generated for the given cluster for inclusion in the library of transformation templates.
17. The system of claim 16 , further comprising instructions to encode each transformation of the plurality of transformations into an embedding.
18. The system of claim 17 , wherein the encoding is based on a transformer network and a graph neural network (GNN).
19. The system of claim 11 , wherein one or more of the transformations includes contextual code surrounding the transformation.
20. A non-transitory computer-readable medium comprising instructions that, in response to execution of the instructions by a processor, cause the processor to:
analyze pre-migration and post-migration versions of source code that exist prior to and after migration of the source code;
based on the analysis, identify one or more transformations made to the pre-migration version of the source code to yield the post-migration version of the source code; and
build a library of transformation templates that are applicable subsequently to automate migration of new source code, wherein the instructions to build include, for one or more of the transformations, instructions to:
generate a plurality of candidate transformation templates, wherein for each candidate transformation template, different permutations of tokens of the transformation are replaced with placeholders, and
select one of the plurality of candidate transformation templates for inclusion in the library of transformation templates, wherein the selecting is based on one or more criteria.