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>)
| 497 | /// each paired Added line pulled forward to appear immediately after |
| 498 | /// its matching Removed line. |
| 499 | fn repair_diff_lines(lines: Vec<HunkLine>) -> Vec<HunkLine> { |
| 500 | use std::collections::{HashMap, HashSet}; |
| 501 | |
| 502 | // Helper: compute character bigrams for Jaccard similarity |
| 503 | fn bigrams(s: &str) -> HashSet<(u8, u8)> { |
| 504 | let bytes = s.trim().as_bytes(); |
| 505 | let mut set = HashSet::new(); |
| 506 | if bytes.len() >= 2 { |
| 507 | for w in bytes.windows(2) { |
| 508 | set.insert((w[0], w[1])); |
| 509 | } |
| 510 | } |
| 511 | set |
| 512 | } |
| 513 | |
| 514 | fn jaccard(a: &HashSet<(u8, u8)>, b: &HashSet<(u8, u8)>) -> f64 { |
| 515 | if a.is_empty() && b.is_empty() { |
| 516 | return 0.0; |
| 517 | } |
| 518 | let inter = a.intersection(b).count(); |
| 519 | let union = a.union(b).count(); |
| 520 | if union == 0 { |
| 521 | 0.0 |
| 522 | } else { |
| 523 | inter as f64 / union as f64 |
| 524 | } |
| 525 | } |
| 526 | |
| 527 | // 1. Collect all Removed and Added lines with their original indices |
| 528 | let mut rm_entries: Vec<(usize, &HunkLine)> = Vec::new(); |
| 529 | let mut add_entries: Vec<(usize, &HunkLine)> = Vec::new(); |
| 530 | |
| 531 | for (idx, line) in lines.iter().enumerate() { |
| 532 | if line.is_removed() { |
| 533 | rm_entries.push((idx, line)); |
| 534 | } else if line.is_added() { |
| 535 | add_entries.push((idx, line)); |
| 536 | } |
| 537 | } |
| 538 | |
| 539 | // Short-circuit: nothing to pair |
| 540 | if rm_entries.is_empty() || add_entries.is_empty() { |
| 541 | return lines; |
| 542 | } |
| 543 | |
| 544 | // 2. Compute bigrams |
| 545 | let rm_bigrams: Vec<HashSet<(u8, u8)>> = rm_entries |
| 546 | .iter() |
| 547 | .map(|(_, l)| bigrams(&l.content)) |
| 548 | .collect(); |
| 549 | let add_bigrams: Vec<HashSet<(u8, u8)>> = add_entries |
| 550 | .iter() |
| 551 | .map(|(_, l)| bigrams(&l.content)) |
| 552 | .collect(); |
| 553 | |
| 554 | // 3. Greedy best-match pairing across ALL removes and adds |
| 555 | let mut candidates: Vec<(usize, usize, f64)> = Vec::new(); |
| 556 | for (ri, rb) in rm_bigrams.iter().enumerate() { |