Story 7.6: Starship-compatible format field (#56)

[stephenleo]

Summary

• Add format field to CshipConfig supporting Starship-style format strings with $line_break for multi-row output and [content](style) ANSI styled spans
• Fix parse_line to scan for both [ and $ simultaneously, preventing styled spans after literal text from being silently consumed as plain text
• Handle $fill as empty with one-shot warning, $line_break splitting at the render() level, and format priority over lines with tracing::warn!

Closes #56

Test plan

• cargo clippy -- -D warnings clean
• cargo fmt --check clean
• cargo test — 324 total (259 unit + 65 integration), 0 failures
• 12 new tests: styled span parsing (content, empty, spaces, unclosed bracket, mixed, after-literal, after-native), $line_break split, format priority, lines backwards compat, $fill empty, integration format+line_break

🤖 Generated with Claude Code

[claude]

Code review

Bug: $token name parser doesn't stop at [, causing styled spans immediately adjacent to tokens to be silently dropped

File: src/renderer.rs (L59–L64)

When parse_line processes a $token, it terminates the token name only at whitespace:

let after_dollar = &remaining[1..];
let name_end = after_dollar
    .find(char::is_whitespace)
    .unwrap_or(after_dollar.len());
let name = &after_dollar[..name_end];

Because [ is not a stop character here, a format string like $cship.model[sep](fg:red) (no space before the styled span) will set name = "cship.model[sep](fg:red)" — the entire styled span is absorbed into the token name. The StyledSpan is never emitted; instead a Native lookup for a nonexistent module name is pushed, producing empty output.

This inconsistency is introduced by this PR: before it, [ had no special meaning, so consuming it into a token name was harmless. Now that [ opens a styled span, the $ name parser must also stop at it.

The existing tests all place a space between a token and a span (e.g. $cship.model [sep](fg:red)), so the no-space case is untested and broken.

Suggested fix:

let name_end = after_dollar
    .find(|c: char| c.is_whitespace() || c == '[')
    .unwrap_or(after_dollar.len());

---

🤖 Generated with Claude Code