Precise Condition Synthesis for Program Repair (1608.07754v5)

Abstract: Due to the difficulty of repairing defect, many research efforts have been devoted into automatic defect repair. Given a buggy program that fails some test cases, a typical automatic repair technique tries to modify the program to make all tests pass. However, since the test suites in real world projects are usually insufficient, aiming at passing the test suites often leads to incorrect patches. In this paper we aim to produce precise patches, that is, any patch we produce has a relatively high probability to be correct. More concretely, we focus on condition synthesis, which was shown to be able to repair more than half of the defects in existing approaches. Our key insight is threefold. First, it is important to know what variables in a local context should be used in an "if" condition, and we propose a sorting method based on the dependency relations between variables. Second, we observe that the API document can be used to guide the repair process, and propose document analysis technique to further filter the variables. Third, it is important to know what predicates should be performed on the set of variables, and we propose to mine a set of frequently used predicates in similar contexts from existing projects. We develop a novel program repair system, ACS, that could generate precise conditions at faulty locations. Furthermore, given the generated conditions are very precise, we can perform a repair operation that is previously deemed to be too overfitting: directly returning the test oracle to repair the defect. Using our approach, we successfully repaired 17 defects on four projects of Defects4J, which is the largest number of fully automatically repaired defects reported on the dataset so far. More importantly, the precision of our approach in the evaluation is 73.9%, which is significantly higher than previous approaches, which are usually less than 40%.

• The paper introduces ACS, a novel system that synthesizes precise conditions to repair program defects, achieving a 73.9% success rate.
• It employs variable dependency analysis and API documentation to filter relevant candidate variables for enhanced patch correctness.
• The method mines predicates from existing projects to rank corrections early, reducing human verification efforts in defect repair.

Precise Condition Synthesis for Program Repair: An Analytical Synopsis

The work, "Precise Condition Synthesis for Program Repair," addresses an intricate problem within the automatic program repair domain: the synthesis of accurate conditions to generate program patches. This research aims to overcome challenges associated with the production of plausible but incorrect patches that arise from weak test suites. This is achieved by proposing and evaluating a method that emphasizes precise condition synthesis, thereby significantly enhancing patch correctness.

Core Contributions

The paper proposes a novel system called ACS (Accurate Condition Synthesis), which emphasizes three key insights to synthesize precise conditional expressions, critical for generating reliable program repairs:

1. Variable Dependency Analysis: ACS introduces a variable sorting method predicated on dependency analysis. Recognizing which variables are relevant to a local context is achieved using dependency relations, allowing the system to focus on the appropriate variables for condition synthesis.
2. API Document Utilization: The authors present an innovative method by which API documentation assists in filtering candidate variables. This document analysis contributes to the precision of condition synthesis by aligning the semantic information contained in documentation with program requirements.
3. Predicate Mining from Existing Projects: The system mines predicates frequently utilized in comparable contexts from existing projects. This heuristic leverages historical source code data to suggest relevant predicates for condition synthesis, emphasizing a data-driven approach to enhance accuracy.

Methodology and Results

The ACS system employs a two-step template approach for defect repair:

• Returning the Test Oracle: This approach is considered for boundary cases where the precision in condition synthesis allows direct use of test oracle values.
• Condition Modification: Existing conditions are refined or extended using synthesized conditions that meet program specification demands.

Empirical evaluations on four projects from the Defects4J benchmark revealed ACS's adeptness at repairing 17 defects automatically—a significant accomplishment when compared to existing systems, demonstrating a precision rate of 73.9% as opposed to sub-40% seen in other methodologies. The research emphasizes that most corrections were identified in the early stages of ranking, highlighting the efficacy of the proposed variable and predicate ranking strategies.

Implications and Future Work

The implications of precise condition synthesis are substantial for both theoretical advancements and practical applications in AI-assisted programming. Enhanced precision in defect repair translates directly into reduced human verification effort, potentially automating aspects of software maintenance. Furthermore, this work lays foundational blocks for future exploration into refined ranking techniques across different repair paradigms. The authors suggest further investigation into broader predicate spaces and integration with models that consider semantic differences to optimize patch generation further.

The paper provides a salient contribution to the field of automatic program repair by setting a new benchmark in precision. It effectively challenges the existing paradigms by focusing on controlling the granularity of condition synthesis through systematic variable and predicate selection strategies. With its data-driven insights and thorough methodological approach, this research elevates the discussion on high-precision automated program repair techniques. Future explorations could strategically build upon this framework to solve more complex programming errors across diverse languages and environments.