Re-pair Delete+Insert lines by content similarity for display. The CRDT builder emits all Deletes before all Inserts within each Replace block (to preserve BRANCH_AFTER chain ordering). When the Myers diff creates multiple Replace blocks for one file, a deleted line and its matching insertion may land in different blocks. This post-pass collects ALL Removed and Added lines across the entire hun
(lines: Vec<HunkLine>)
| 709 | /// each paired Added line pulled forward to appear immediately after |
| 710 | /// its matching Removed line. |
| 711 | fn repair_diff_lines(lines: Vec<HunkLine>) -> Vec<HunkLine> { |
| 712 | use std::collections::{HashMap, HashSet}; |
| 713 | |
| 714 | // Helper: compute character bigrams for Jaccard similarity |
| 715 | fn bigrams(s: &str) -> HashSet<(u8, u8)> { |
| 716 | let bytes = s.trim().as_bytes(); |
| 717 | let mut set = HashSet::new(); |
| 718 | if bytes.len() >= 2 { |
| 719 | for w in bytes.windows(2) { |
| 720 | set.insert((w[0], w[1])); |
| 721 | } |
| 722 | } |
| 723 | set |
| 724 | } |
| 725 | |
| 726 | fn jaccard(a: &HashSet<(u8, u8)>, b: &HashSet<(u8, u8)>) -> f64 { |
| 727 | if a.is_empty() && b.is_empty() { |
| 728 | return 0.0; |
| 729 | } |
| 730 | let inter = a.intersection(b).count(); |
| 731 | let union = a.union(b).count(); |
| 732 | if union == 0 { |
| 733 | 0.0 |
| 734 | } else { |
| 735 | inter as f64 / union as f64 |
| 736 | } |
| 737 | } |
| 738 | |
| 739 | // 1. Collect all Removed and Added lines with their original indices |
| 740 | let mut rm_entries: Vec<(usize, &HunkLine)> = Vec::new(); |
| 741 | let mut add_entries: Vec<(usize, &HunkLine)> = Vec::new(); |
| 742 | |
| 743 | for (idx, line) in lines.iter().enumerate() { |
| 744 | if line.is_removed() { |
| 745 | rm_entries.push((idx, line)); |
| 746 | } else if line.is_added() { |
| 747 | add_entries.push((idx, line)); |
| 748 | } |
| 749 | } |
| 750 | |
| 751 | // Short-circuit: nothing to pair |
| 752 | if rm_entries.is_empty() || add_entries.is_empty() { |
| 753 | return lines; |
| 754 | } |
| 755 | |
| 756 | // 2. Compute bigrams |
| 757 | let rm_bigrams: Vec<HashSet<(u8, u8)>> = rm_entries |
| 758 | .iter() |
| 759 | .map(|(_, l)| bigrams(&l.content)) |
| 760 | .collect(); |
| 761 | let add_bigrams: Vec<HashSet<(u8, u8)>> = add_entries |
| 762 | .iter() |
| 763 | .map(|(_, l)| bigrams(&l.content)) |
| 764 | .collect(); |
| 765 | |
| 766 | // 3. Greedy best-match pairing across ALL removes and adds |
| 767 | let mut candidates: Vec<(usize, usize, f64)> = Vec::new(); |
| 768 | for (ri, rb) in rm_bigrams.iter().enumerate() { |