fix(cli): preserve case-distinct paths-file entries

[mstykow]

Summary

• Preserve the original case of each normalized --paths-file entry when deduplicating the direct collection frontier.
• Keep case-insensitive selection matching independently deduplicated, avoiding redundant filters without dropping case-distinct files.
• Cover case-distinct collection and exact normalized repeat deduplication in a focused regression test.

Issues

• Closes --paths-file silently drops files whose paths differ only in capitalization #1430

Scope and exclusions

• Included: --paths-file resolution, frontier construction, and regression coverage through collection and selection filtering.
• Explicit exclusions: --include/--exclude glob semantics and ordinary directory discovery are unchanged.

How to verify

• On a case-sensitive filesystem, run the issue reproduction with Example.js and example.js; confirm both file entries appear with their respective license detections. Add ./Example.js to the paths file and confirm the normalized repeat does not add a third entry.

Expected-output fixture changes

• Files changed: None.
• Why the new expected output is correct: The behavior is covered by focused Rust assertions rather than serialized golden output.

🤖 Generated with Claude Code

[greptile-apps]

Greptile Summary

The PR corrects --paths-file deduplication by separating collection-frontier identity from case-insensitive selection identity.

• Canonical filesystem paths deduplicate aliases that resolve to the same file.
• Case-distinct files remain separate collection-frontier entries on case-sensitive filesystems.
• Selection and missing-entry deduplication remain independently scoped.
• A focused regression test covers case-sensitive and case-insensitive filesystem behavior.

Confidence Score: 5/5

The PR appears safe to merge; the updated deduplication preserves case-distinct files while avoiding duplicate collection of filesystem aliases.

No actionable new issue remains. The previous duplicate-file finding was manually resolved without explanation, and the current canonical-path deduplication also addresses its underlying case-alias behavior.

Important Files Changed

Filename	Overview
src/scan_result_shaping/selection.rs	Uses canonical filesystem identity for frontier deduplication while preserving independent selection and missing-entry sets.
src/scan_result_shaping/selection_test.rs	Verifies case-distinct collection, normalized-repeat deduplication, alias handling, and downstream selection filtering across filesystem semantics.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart LR
    A[Normalized paths-file entry] --> B{Entry exists?}
    B -->|Yes| C[Canonicalize filesystem path]
    C --> D[Deduplicate collection frontier by file identity]
    B -->|Yes| E[Build case-insensitive selection key]
    E --> F[Deduplicate selection filters]
    B -->|No| G[Deduplicate missing normalized entries]

Loading

_{Reviews (2): Last reviewed commit: "fix(cli): preserve case-distinct paths-f..." | Re-trigger Greptile}