FeatLens builds a feature index that links natural-language feature descriptions to function-level code entities and dynamically constructs a task-specific seed graph from the feature index and applies semantic-structural graph reasoning with personalized PageRank to select a compact reasoning graph.
Abstract
Recent code generation research has moved from isolated function completion toward repository-level generation in existing codebases. To implement a target function correctly, an LLM must identify reusable repository dependencies such as existing functions, APIs, and cross-file definitions. Existing retrieval methods provide such context through code similarity search, persistent whole-repository graphs, or LLM-driven graph exploration, but often incur high graph construction, reasoning, and token costs. Feature-oriented methods offer a natural view of software functionality, yet they mainly support requirement decomposition, planning, or feature editing rather than code dependency retrieval. This paper presents \textbf{FeatLens}, a feature-guided dynamic code graph construction and retrieval approach for repository-level code generation. FeatLens builds a feature index that links natural-language feature descriptions to function-level code entities. Given a generation task, it dynamically constructs a task-specific seed graph from the feature index and applies semantic-structural graph reasoning with personalized PageRank to select a compact reasoning graph. This design replaces persistent whole-repository graph maintenance and LLM exploration with deterministic and lightweight dependency retrieval. Experiments on DevEval and EvoCodeBench show that FeatLens achieves the best DR@15 among sparse, dense, and graph-based baselines (0.501 and 0.460). On DevEval generation, it obtains the highest DIR@1, reaching 52.91\% with DeepSeek-V3.2 and 53.58\% with GPT-5-mini, while maintaining competitive Pass@1 and producing shorter code. Compared with the strongest graph-based baseline, FeatLens reduces graph nodes by 61.0\%, edges by 86.2\%, and total token overhead by 45.9\%, with no LLM tokens used during retrieval.
CrossCoder is proposed, a cross-repository code generation framework that explicitly incorporates external libraries into the retrieval context through a unified knowledge graph over repository and library entities that consistently improves both functional correctness and robustness to dependency-version changes.
Minh Le-Anh, Nam Le Hai, Quyen Tran et al.· 0 citations
Large language models (LLMs) have significantly advanced automatic code generation. Although they have demonstrated remarkable ability in writing standalone code, they still struggle with real-world repository-level coding tasks due to the lack of repository-specific knowledge. Existing studies typically employ retriev...
Tanghaoran Zhang, Xin-Jun Mao, Yu-Xin Zhao et al.· ACM Transactions on Software...· 0 citations
JDQuery is proposed, a query-driven defect localization framework for Java source code based on a code knowledge graph that parses Java source code into abstract syntax trees, extracts software entities and their semantic relationships according to a formalized domain ontology, and constructs a unified code knowledge g...
Despite recent progress in function-level docstring generation, code summarization systems have mostly focused on creating concise descriptions for individual functions using automatically generated docstrings, leaving the difficult task of generating structured repository-level summaries unaddressed. We propose a nove...
Eric Binu Koshy, Sunitha E. V.· International Conference Inn...· 0 citations
RepoNav is introduced, a lightweight post-retrieval interface that reorganizes retrieved snippets into a file-centered navigation scaffold that improves function-level localization and narrows the file-to-function gap.
Hong-Zheng Chai, Jia-Kun Li, Hong Yu et al.· 0 citations
StructFix is proposed, a structure-aware APR framework that grounds masked patch generation in Code Property Graphs (CPGs), and explicitly coupling structural dependencies with masked generation improves repair effectiveness and enables transfer across datasets.
Mengtian Cui, Yang-Fan Liu, Zhibo Lu et al.· International Conference on...· 0 citations