Detect file paths preceded by Unicode/CJK punctuation

[SagarSDagdu]

Description

Clickable file path detection in the terminal grid relied on a hardcoded ASCII separator set, so a path immediately preceded by CJK / full-width punctuation was not detected as clickable. This affects CJK prose and AI CLI output like:

路径：/Users/me/project/plan.md   ← was NOT clickable (fullwidth ：touching /)
路径: /Users/me/project/plan.md   ← clickable (ASCII : + space)
/Users/me/project/plan.md         ← clickable (bare)

This change replaces every FILE_LINK_SEPARATORS.contains(&c) site with a new is_file_link_separator(c) helper that also accepts non-ASCII Unicode whitespace and general categories Po | Ps | Pe | Pi | Pf. Connectors (Pc), dashes (Pd), and CJK letters (Lo) are deliberately excluded so paths containing _, -, or CJK characters in their names (e.g. /path/音楽/テスト.txt) continue to detect as a single token.

Two latent issues that became visible once multi-byte separators are accepted are also fixed:

• possible_file_paths_in_word indexed past a separator with + 1, which is invalid for multi-byte UTF-8 punctuation (： is 3 bytes). The substring enumeration now tracks run_starts / run_ends separately and advances by c.len_utf8().
• The separator fragment emitted by line_to_fragments hardcoded total_cell_width = 1. Full-width punctuation visually occupies two cells, so this is now UnicodeWidthChar::width(cell.c).

Linked Issue

Fixes #10245.

• The linked issue is labeled ready-to-spec or ready-to-implement.

Screenshots / Videos

n/a — the existing unit tests in app/src/util/link_detection_test.rs cover the failing repros.

Testing

• New test_possible_file_paths_in_word_cjk_punctuation covers the issue's full table — fullwidth colon, fullwidth parentheses, CJK corner brackets, ideographic full stop, fullwidth comma — plus a negative case asserting that CJK letters in path names don't fragment the path.
• The pre-existing test_possible_file_paths_in_word_multibyte still passes (CJK letters remain non-separators).
• All 80 terminal::model::grid::grid_handler::tests and all 7 util::link_detection::tests pass under cargo test --features local_fs --test-threads=1.
• Manually verified end-to-end: built target/debug/warp-oss, ran printf '路径：/tmp/warp-repro.md\n路径: /tmp/warp-repro.md\n/tmp/warp-repro.md\n', and confirmed Cmd+Click works on all three lines (only the latter two worked before).

Agent Mode

• Warp Agent Mode - This PR was created via Warp's AI Agent Mode

[oz-for-oss]

@SagarSDagdu

I'm starting a first review of this pull request.

You can view the conversation on Warp.

I reviewed this pull request and requested human review from: @warpdotdev/oss-maintainers.

Comment /oz-review on this pull request to retrigger a review (up to 3 times on the same pull request).

Powered by Oz

[oz-for-oss]

Overview

This PR updates file-path boundary detection to recognize non-ASCII whitespace and selected Unicode punctuation, fixes candidate substring indexing for multi-byte separators, and uses Unicode display width for separator fragments.

Concerns

• No blocking correctness, security, or performance concerns found in the annotated diff.

Verdict

Found: 0 critical, 0 important, 0 suggestions

Approve

Comment /oz-review on this pull request to retrigger a review (up to 3 times on the same pull request).

Powered by Oz

[vorporeal]

hey @SagarSDagdu - #10004 (merged earlier today) also expanded the set of file link separators and added general support for non-ASCII ones.

mind rebasing on top of those changes? i think it'll mostly simplify your PR down to is_file_link_separator() (which we'll want to also handle the unicode line drawing characters covered by that PR).

[SagarSDagdu]

Rebased on master to pick up #10004's multi-byte indexing refactor that already covered the byte-range fix I'd duplicated, so this PR now reduces to:

• New is_file_link_separator(c) helper that wraps the existing FILE_LINK_SEPARATORS set (ASCII + box-drawing) and additionally accepts non-ASCII Unicode whitespace plus general categories Po | Ps | Pe | Pi | Pf. Pc/Pd/Lo are deliberately excluded so _, -, and CJK letters don't fragment paths.
• All 5 FILE_LINK_SEPARATORS.contains(&c) sites (4 in grid_handler, 1 in link_detection) now go through the helper, so the box-drawing chars from fix: detect tree output filenames as file links #10004 and the CJK punctuation from this PR share the same gate.
• line_to_fragments now uses UnicodeWidthChar::width for the separator fragment (full-width punctuation occupies two cells).
• New test_possible_file_paths_in_word_cjk_punctuation covers the issue's table plus a negative case for CJK letters in path names. The tree-output tests added in fix: detect tree output filenames as file links #10004 still pass.

Please review @vorporeal.

[vorporeal]

looks great; thank you!

[SagarSDagdu]

@vorporeal I think you'll have to kickoff the workflows manually on this PR to merge it, as I had to resolve some conflicts