Qianxiang Wang

The impressive performance of Large Language Models (LLMs) on code-related tasks has shown the potential of fully automated software development. In light of this, we introduce a new software engineering task, namely Natural Language to Code Repository (NL2Repo). This task aims to generate an entire code repository from its natural language requirements. To address this task, we propose a simple yet effective framework CodeS , which decomposes NL2Repo into multiple sub-tasks by a multi-layer sketch. Specifically, CodeS includes three modules: RepoSketcher, FileSketcher, and SketchFiller. RepoSketcher first generates a repository’s directory structure for given requirements; FileSketcher then generates a file sketch for each file in the generated structure; SketchFiller finally fills in the details for each function in the generated file sketch. To rigorously assess CodeS on the NL2Repo task, we carry out evaluations through both automated benchmarking and manual feedback analysis. For benchmark-based evaluation, we craft a repository-oriented benchmark, SketchEval, and design an evaluation metric, SketchBLEU. For feedback-based evaluation, we develop a VSCode plugin for CodeS and engage 30 participants in conducting empirical studies. Extensive experiments prove the effectiveness and practicality of CodeS on the NL2Repo task.

Code intelligence is an emerging domain in software engineering, aiming to improve the effectiveness and efficiency of various code-related tasks. Recent research suggests that incorporating contextual information beyond the basic original task inputs (i.e., source code) can substantially enhance model performance. Such contextual signals may be obtained directly or indirectly from sources such as API documentation or intermediate representations like abstract syntax trees can significantly improve the effectiveness of code intelligence. Despite growing academic interest, there is a lack of systematic analysis of context in code intelligence. To address this gap, we conduct an extensive literature review of 146 relevant studies published between September 2007 and August 2024. Our investigation yields four main contributions. (1) A quantitative analysis of the research landscape, including publication trends, venues, and the explored domains; (2) A novel taxonomy of context types used in code intelligence; (3) A task-oriented analysis investigating context integration strategies across diverse code intelligence tasks; (4) A critical evaluation of evaluation methodologies for context-aware methods. Based on these findings, we identify fundamental challenges in context utilization in current code intelligence systems and propose a research roadmap that outlines key opportunities for future research.

The e X tensible M arkup L anguage (XML) is a file format widely used for data transmission in modern software development. In recent years, embedding SQL statements in XML files (i.e., XML-SQL) has become a popular way for developing applications with database access capability. Typically, XML-SQL code snippets demonstrate similar functionalities and structures, leading to repetitive programming work. Therefore, leveraging pre-trained code models for automated code generation presents a promising way to alleviate duplicated efforts and enhance the efficiency of developing XML-SQL code. However, XML-SQL code has strong domain-specific characteristics that general pre-trained code models typically struggle to fully harness, thereby leading to limited overall performance of general pre-trained code models. In this article, we aim to address the challenge of handling this domain-specific knowledge. First, we propose a code updating task and construct the corresponding TwinXSQL dataset to better evaluate the model’s code generation performance in the XML-SQL domain. Then, we leverage the common characteristics of XML-SQL and other programming languages (i.e., all programming languages impose grammar constraints on behavior) to design a bipartite-grammar–aware training framework (named BGA) for unsupervised pre-training, thereby improving the transfer of general-purpose code models to the XML-SQL domain. Specifically, we divide the XML-SQL code into two types of grammatical components: structure components and value components. During pre-training, we undertake three tasks, each designed to learn the internal information of these grammatical components and the relationships between them, enabling the pre-training process to better incorporate previously unlearned domain-specific knowledge of XML-SQL code. Our experimental results show that our trained model XSQLT5-base (220M) improves accuracy by 13.8% compared to the similarly sized CodeT5-base (220M). Additionally, our experiments reveal that ChatGPT, due to its inability to fully learn the XML-SQL domain knowledge, achieves a much lower generation accuracy even with few-shot samples compared to our XSQLT5-base (220M) model.

To address security vulnerabilities arising from third-party libraries, security researchers maintain databases monitoring and curating vulnerability reports. Application developers can identify libraries affected by vulnerability reports (in short, affected libraries) by directly querying the databases with their used libraries. However, the querying results of affected libraries are not reliable due to the incompleteness of vulnerability reports. Thus, current approaches model the task of identifying affected libraries as a named-entity-recognition (NER) task or an extreme multi-label learning (XML) task. These approaches suffer from highly inaccurate results in identifying affected libraries with complex and similar names, e.g., Java libraries. To address these limitations, in this article, we propose VulLibMiner, the first to identify affected libraries from textual descriptions of both vulnerabilities and libraries, together with VulLib, a Java vulnerability dataset with their affected libraries. VulLibMiner consists of a TF-IDF matcher to efficiently screen out a small set of candidate libraries and a BERT-FNN model to effectively identify affected libraries from these candidates. We evaluate VulLibMiner using four state-of-the-art/practice approaches of identifying affected libraries on both their dataset named VeraJava and our VulLib dataset. Our evaluation results show that VulLibMiner can effectively identify affected libraries with an average F1 score of 0.669 while the state-of-the-art/practice approaches achieve only 0.547.

Code generation models based on the pre-training and fine-tuning paradigm have been increasingly attempted by both academia and industry, resulting in well-known industrial models such as Codex, CodeGen, and PanGu-Coder. After being pre-trained on a large-scale corpus of code, a model is further fine-tuned with datasets specifically for the target downstream task, e.g., generating code from natural language description. The target code being generated can be classified into two types: a standalone function, i.e., a function that invokes or accesses only built-in functions and standard libraries, and a non-standalone function, i.e., a function that invokes or accesses user-defined functions or third-party libraries. To effectively generate code especially non-standalone functions (largely ignored by existing work), in this article, we present Wenwang, an approach to improving the capability of a pre-trained model on generating code beyond standalone functions. Wenwang consists of two components: a fine-tuning dataset named WenwangData and a fine-tuned model named WenwangCoder. Compared with existing fine-tuning datasets, WenwangData additionally covers non-standalone functions. Besides the docstring and code snippet for a function, WenwangData also includes its contextual information collected via program analysis. Based on PanGu-Coder, we produce WenwangCoder by fine-tuning PanGu-Coder on WenwangData with our context-aware fine-tuning technique so that the contextual information can be fully leveraged during code generation. On CoderEval and HumanEval, WenwangCoder outperforms three state-of-the-art models with similar parameter sizes (at the scale of around 300M), namely CodeGen, PanGu-Coder, and PanGu-FT. Although WenwangCoder does not outperform ChatGPT on HumanEval, WenwangCoder with smaller model parameter sizes can achieve similar effects to ChatGPT on CoderEval. Our experimental results also shed light on a number of promising optimization directions based on existing pre-trained models.

Unit testing is an essential activity in software development for verifying the correctness of software components. However, manually writing unit tests is challenging and time-consuming. The emergence of Large Language Models (LLMs) offers a new direction for automating unit test generation. Existing research primarily focuses on closed-source LLMs (e.g., ChatGPT and CodeX) with fixed prompting strategies, leaving the capabilities of advanced open-source LLMs with various prompting settings unexplored. Particularly, open-source LLMs offer advantages in data privacy protection and have demonstrated superior performance in some tasks. Moreover, effective prompting is crucial for maximizing LLMs' capabilities. In this paper, we conduct the first empirical study to fill this gap, based on 17 Java projects, five widely-used open-source LLMs with different structures and parameter sizes, and comprehensive evaluation metrics. Our findings highlight the significant influence of various prompt factors, show the performance of open-source LLMs compared to the commercial GPT-4 and the traditional Evosuite, and identify limitations in LLM-based unit test generation. We then derive a series of implications from our study to guide future research and practical use of LLM-based unit test generation.

Code generation models based on the pre-training and fine-tuning paradigm have been increasingly attempted by both academia and industry, resulting in well-known industrial models such as Codex, CodeGen, and PanGu-Coder. To evaluate the effectiveness of these models, multiple existing benchmarks (e.g., HumanEval and AiXBench) are proposed, including only cases of generating a standalone function, i.e., a function that may invoke or access only built-in functions and standard libraries. However, non-standalone functions, which typically are not included in the existing benchmarks, constitute more than 70% of the functions in popular open-source projects, and evaluating models' effectiveness on standalone functions cannot reflect these models' effectiveness on pragmatic code generation scenarios (i.e., code generation for real settings of open source or proprietary code). To help bridge the preceding gap, in this paper, we propose a benchmark named CoderEval, consisting of 230 Python and 230 Java code generation tasks carefully curated from popular real-world open-source projects and a self-contained execution platform to automatically assess the functional correctness of generated code. CoderEval supports code generation tasks from six levels of context dependency, where context refers to code elements such as types, APIs, variables, and consts defined outside the function under generation but within the dependent third-party libraries, current class, file, or project. CoderEval can be used to evaluate the effectiveness of models in generating code beyond only standalone functions. By evaluating three state-of-the-art code generation models (CodeGen, PanGu-Coder, and ChatGPT) on CoderEval and HumanEval, we find that the effectiveness of these models in generating standalone functions is substantially higher than that in generating non-standalone functions. Our analysis highlights the current progress and pinpoints future directions to further improve a model's effectiveness by leveraging contextual information for pragmatic code generation.

In recent years, applying deep learning to detect semantic code clones has received substantial attention from the research community. Accordingly, various evaluation benchmark datasets, with the most popular one as BigCloneBench, are constructed and selected as benchmarks to assess and compare different deep learning models for detecting semantic clones. However, there is no study to investigate whether an evaluation benchmark dataset such as BigCloneBench is properly used to evaluate models for detecting semantic code clones. In this article, we present an experimental study to show that BigCloneBench typically includes semantic clone pairs that use the same identifier names, which however are not used in non-semantic-clone pairs. Subsequently, we propose an undesirable-by-design Linear-Model that considers only which identifiers appear in a code fragment; this model can achieve high effectiveness for detecting semantic clones when evaluated on BigCloneBench, even comparable to state-of-the-art deep learning models recently proposed for detecting semantic clones. To alleviate these issues, we abstract a subset of the identifier names (including type, variable, and method names) in BigCloneBench to result in AbsBigCloneBench and use AbsBigCloneBench to better assess the effectiveness of deep learning models on the task of detecting semantic clones.