AI-assisted engineering/Reviewing AI code

Fix the panic scan for a Rust config loader

Review the AI's codeEasy12 min

Your task

An AI explained a Rust config loader as safe: "It returns a Result, so any error goes to the caller. It cannot crash the process." Your team checks a claim like that by scanning the function body for every call that panics, which stops the thread instead of returning an error.

An assistant wrote find_panic_sites(body_lines). It is in the editor. It has a bug. Find it and fix it.

Input and output:

body_lines: the lines of a Rust function body, in order.
Return a list of findings. Each finding is "<line>:<name>", where <line> is the line number counting from 1.

Rules the tests check:

On each line, ignore everything from the first // onward. It is a comment.
These exact texts are panic sites, each with the name it reports:
- .unwrap( reports unwrap
- .expect( reports expect
- panic!( reports panic
- unreachable!( reports unreachable
- todo!( reports todo
Report every occurrence. A line with two panic sites gives two findings.
Findings go in line order, and left to right within a line.
A match inside a string literal still counts. Nothing else is a panic site, so .unwrap_or( is not one.

Examples

find_panic_sites(body_lines: list[str]) -> list[str]

Example 1

Input

find_panic_sites(["fn load_config(path: &str) -> Result<Config, ConfigError> {", "    let text = std::fs::read_to_string(path).unwrap();", "    let cfg: Config = serde_json::from_str(&text).expect(\"valid config\");", "    Ok(cfg)", "}"])

Output["2:unwrap", "3:expect"]

Example 2

Input

find_panic_sites(["fn load_config(path: &str) -> Result<Config, ConfigError> {", "    let text = std::fs::read_to_string(path)?;", "    let cfg: Config = serde_json::from_str(&text)?;", "    Ok(cfg)", "}"])

Output[]