Code translation tools, namely transpilers, are developed for automatic source-to-source translation. Latest learning-based transpilers have shown impressive enhancement against rule-based counterparts in both translation accuracy and readability, owing to their task-specific pre-training on extensive monolingual corpora. Nevertheless, their current performance still remains unsatisfactory for practical deployment, and the associated training resources are also prohibitively expensive. Large Language Models (LLMs), pre-trained on huge amounts of human-written code/text, have shown remarkable performance in many code intelligence tasks due to their powerful generality, even without task-specific re-training/fine-tuning. Thus, LLMs can potentially circumvent the above limitations, but they have not been exhaustively explored yet. This paper investigates diverse LLMs and learning-based transpilers for automated code translation tasks, finding that: although certain LLMs have outperformed current transpilers, they still have some accuracy issues, where most of the failures are induced by a lack of comprehension of source programs (38.51%), missing clear instructions on I/O types in translation (14.94%), and ignoring discrepancies between source and target programs (41.38%). Enlightened by the above findings, we further propose UniTrans , a Unified code Translation framework, applicable to various LLMs, for unleashing their power in this field. Specifically, UniTrans first crafts a series of test cases for target programs with the assistance of source programs. Next, it harnesses the above auto-generated test cases to augment the code translation and then evaluate their correctness via execution. Afterward, UniTrans further (iteratively) repairs incorrectly translated programs prompted by test case execution results. Extensive experiments are conducted on six settings of translation datasets between Python, Java, and C++. Three recent LLMs of diverse sizes, including GPT-3.5 and LLaMA-13B/7B, are tested with UniTrans , and all achieve substantial improvements in terms of computational accuracy and exact match accuracy among almost all translation settings, showing the universal effectiveness of UniTrans in practice.

Software comments sometimes are not promptly updated in sync when the associated code is changed. The inconsistency between code and comments may mislead the developers and result in future bugs. Thus, studies concerning code-comment synchronization have become highly important, which aims to automatically synchronize comments with code changes. Existing code-comment synchronization approaches mainly contain two types, i.e., (1) deep learning-based (e.g., CUP), and (2) heuristic-based (e.g., HebCUP). The former constructs a neural machine translation-structured semantic model, which has a more generalized capability on synchronizing comments with software evolution and growth. However, the latter designs a series of rules for performing token-level replacements on old comments, which can generate the completely correct comments for the samples fully covered by their fine-designed heuristic rules. In this article, we propose a composite approach named CBS (i.e., Classifying Before Synchronizing ) to further improve the code-comment synchronization performance, which combines the advantages of CUP and HebCUP with the assistance of inferred categories of Code-Comment Inconsistent (CCI) samples. Specifically, we firstly define two categories (i.e., heuristic-prone and non-heuristic-prone) for CCI samples and propose five features to assist category prediction. The samples whose comments can be correctly synchronized by HebCUP are heuristic-prone, while others are non-heuristic-prone. Then, CBS employs our proposed Multi-Subsets Ensemble Learning (MSEL) classification algorithm to alleviate the class imbalance problem and construct the category prediction model. Next, CBS uses the trained MSEL to predict the category of the new sample. If the predicted category is heuristic-prone, CBS employs HebCUP to conduct the code-comment synchronization for the sample, otherwise, CBS allocates CUP to handle it. Our extensive experiments demonstrate that CBS statistically significantly outperforms CUP and HebCUP, and obtains an average improvement of 23.47%, 22.84%, 3.04%, 3.04%, 1.64%, and 19.39% in terms of Accuracy, Recall@5, Average Edit Distance (AED) , Relative Edit Distance (RED) , BLEU-4, and Effective Synchronized Sample (ESS) ratio, respectively, which highlights that category prediction for CCI samples can boost the code-comment synchronization performance.

In recent years, Large Language Models (LLMs) have dramatically advanced the performance of automated code translation, making their computational accuracy score reach up to over 80% on many previous benchmarks. However, most code samples in these benchmarks are short, standalone, statement/method-level, and algorithmic, which is not aligned with practical coding tasks. Therefore, it is still unknown the actual capability of LLMs in translating code samples written for daily development. To achieve this, we construct a class-level code translation benchmark, ClassEval-T, and make the first attempt to extensively assess recent LLMs' performance on class-level code translation. ClassEval-T is extended from ClassEval, a well-known class-level Python code generation benchmark consisting of multiple practical coding topics, such as database operation and game design, and diverse contextual dependencies (e.g., fields, methods, and libraries). It cost us 360 person-hours to accomplish the manual migration to Java and C++ with complete code samples and associated test suites. Subsequently, we design three translation strategies (i.e., holistic, min-dependency, and standalone) for class-level code translations and evaluate eight recent LLMs of commercial, general, and code kinds in diverse families and sizes on ClassEval-T. Experimental results demonstrate a remarkable performance drop compared with the most widely studied method-level code translation benchmark, and obvious discrepancies among LLMs appear, showing the effectiveness of ClassEval-T in measuring recent LLMs. Afterwards, we further discuss the usage scenarios for diverse translation strategies and LLMs' ability to dependency awareness when translating class samples. Finally, 1,243 failure cases made by the best-performing LLM under test are thoroughly analyzed and categorized in this paper for practical guidance and future enlightenment.

The rise of Large Language Models (LLMs) has significantly advanced various applications on software engineering tasks, particularly in code generation. Despite the promising performance, LLMs are prone to generate hallucinations, which means LLMs might produce outputs that deviate from users’ intent, exhibit internal inconsistencies, or misaligned with the real world knowledge, making the deployment of LLMs potentially risky in a wide range of applications. Existing work mainly focuses on investigating the hallucination in the domain of Natural Language Generation (NLG), leaving a gap in comprehensively understanding the types, causes, and impacts of hallucinations in the context of code generation. To bridge the gap, we conducted a thematic analysis of the LLM-generated code to summarize and categorize the hallucinations, as well as their causes and impacts. Our study established a comprehensive taxonomy of code hallucinations, encompassing 3 primary categories and 12 specific categories. Furthermore, we systematically analyzed the distribution of hallucinations, exploring variations among different LLMs and benchmarks. Moreover, we perform an indepth analysis on the causes and impacts of various hallucinations, aiming to provide valuable insights into hallucination mitigation. Finally, to enhance the correctness and reliability of LLM-generated code in a lightweight manner, we explore a training-free hallucination mitigation approach by prompt enhancing techniques. We believe our findings will shed light on future research about hallucination evaluation and mitigation, ultimately paving the way for building more effective and reliable code LLMs in the future. The replication package is available at https://github.com/Lorien1128/code hallucination.