Syntactic Completions with Material Obligations

2025-08-22

AI Summary

Existing code editors often fail under syntactic errors, and conventional recovery techniques—such as panic-mode parsing or multi-option repair—are either overly aggressive or trigger combinatorial explosion in code completion. This paper introduces *tylr*, a parser-and-edit-generation framework enabling precise autocompletion for erroneous code. Its core innovation lies in formalizing “syntactic obligations”—e.g., missing operands, operators, or keywords—as a unified abstraction; identifying obligations via grammar traversal, resolving ambiguities with grammar zippers, and generating minimal-obligation-driven completions via tile-based operator-precedence parsing. We further design a visual obligation-aware interface that unifies textual and structural editing. The approach is formally specified and empirically validated through a user study, demonstrating significant improvements in edit usability and practicality. *tylr* establishes a new paradigm for human–computer collaborative syntactic repair.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Planning, Routing, and Scheduling: Replanning and Plan Repair

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Code editors provide essential services that help developers understand, navigate, and modify programs. However, these services often fail in the presence of syntax errors. Existing syntax error recovery techniques, like panic mode and multi-option repairs, are either too coarse, e.g. in deleting large swathes of code, or lead to a proliferation of possible completions. This paper introduces $ exttt{tylr}$, a parser and editor generator that completes arbitrarily malformed code by inserting obligations, which generalize holes to cover missing operands, operators, mixfix keywords, and sort transitions. $ exttt{tylr}$ is backed by a novel theory of tile-based parsing, which extends operator-precedence parsing in two ways. First, traditional token precedence comparisons are replaced by a notion of grammar walks, which form the basis for generating obligations. Second, a distinct "molding" system based on grammar zippers expand grammar expressivity by allowing the system to disambiguate between possible parses and completions based on an obligation minimization criterion. In addition to serving as a novel approach to error correction, $ exttt{tylr}$'s design enables the development of an editor that visually materializes obligations to the human user, serving as a novel hybrid between a text editor and a structure editor. We introduce $ exttt{tylr}$ by example, then formalize its key ideas. Finally, we conduct a human subjects study to evaluate the extent to which an editor like $ exttt{tylr}$ that materializes syntactic obligations might be usable and useful, finding both points of positivity and interesting new avenues for future work.

Problem

• Addressing syntax error recovery limitations in code editors
• Completing malformed code by inserting generalized obligations
• Enhancing parsing with grammar walks and obligation minimization

Innovation

• Tile-based parsing with grammar walks
• Molding system using grammar zippers
• Materializing obligations as hybrid editor