jj_lib/
conflicts.rs

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
// Copyright 2020 The Jujutsu Authors
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// https://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

#![allow(missing_docs)]

use std::io;
use std::io::Read;
use std::io::Write;
use std::iter::zip;

use bstr::BString;
use bstr::ByteSlice;
use futures::stream::BoxStream;
use futures::try_join;
use futures::Stream;
use futures::StreamExt;
use futures::TryStreamExt;
use itertools::Itertools;
use pollster::FutureExt;
use regex::bytes::Regex;
use regex::bytes::RegexBuilder;

use crate::backend::BackendError;
use crate::backend::BackendResult;
use crate::backend::CommitId;
use crate::backend::FileId;
use crate::backend::SymlinkId;
use crate::backend::TreeId;
use crate::backend::TreeValue;
use crate::copies::CopiesTreeDiffEntry;
use crate::copies::CopiesTreeDiffEntryPath;
use crate::diff::Diff;
use crate::diff::DiffHunk;
use crate::diff::DiffHunkKind;
use crate::files;
use crate::files::MergeResult;
use crate::merge::Merge;
use crate::merge::MergeBuilder;
use crate::merge::MergedTreeValue;
use crate::repo_path::RepoPath;
use crate::store::Store;

const CONFLICT_START_LINE: &str = "<<<<<<<";
const CONFLICT_END_LINE: &str = ">>>>>>>";
const CONFLICT_DIFF_LINE: &str = "%%%%%%%";
const CONFLICT_MINUS_LINE: &str = "-------";
const CONFLICT_PLUS_LINE: &str = "+++++++";
const CONFLICT_GIT_ANCESTOR_LINE: &str = "|||||||";
const CONFLICT_GIT_SEPARATOR_LINE: &str = "=======";
const CONFLICT_START_LINE_CHAR: u8 = CONFLICT_START_LINE.as_bytes()[0];
const CONFLICT_END_LINE_CHAR: u8 = CONFLICT_END_LINE.as_bytes()[0];
const CONFLICT_DIFF_LINE_CHAR: u8 = CONFLICT_DIFF_LINE.as_bytes()[0];
const CONFLICT_MINUS_LINE_CHAR: u8 = CONFLICT_MINUS_LINE.as_bytes()[0];
const CONFLICT_PLUS_LINE_CHAR: u8 = CONFLICT_PLUS_LINE.as_bytes()[0];
const CONFLICT_GIT_ANCESTOR_LINE_CHAR: u8 = CONFLICT_GIT_ANCESTOR_LINE.as_bytes()[0];
const CONFLICT_GIT_SEPARATOR_LINE_CHAR: u8 = CONFLICT_GIT_SEPARATOR_LINE.as_bytes()[0];

/// A conflict marker is one of the separators, optionally followed by a space
/// and some text.
// TODO: All the `{7}` could be replaced with `{7,}` to allow longer
// separators. This could be useful to make it possible to allow conflict
// markers inside the text of the conflicts.
static CONFLICT_MARKER_REGEX: once_cell::sync::Lazy<Regex> = once_cell::sync::Lazy::new(|| {
    RegexBuilder::new(r"^(<{7}|>{7}|%{7}|\-{7}|\+{7}|\|{7}|={7})( .*)?$")
        .multi_line(true)
        .build()
        .unwrap()
});

fn write_diff_hunks(hunks: &[DiffHunk], file: &mut dyn Write) -> io::Result<()> {
    for hunk in hunks {
        match hunk.kind {
            DiffHunkKind::Matching => {
                debug_assert!(hunk.contents.iter().all_equal());
                for line in hunk.contents[0].split_inclusive(|b| *b == b'\n') {
                    file.write_all(b" ")?;
                    file.write_all(line)?;
                }
            }
            DiffHunkKind::Different => {
                for line in hunk.contents[0].split_inclusive(|b| *b == b'\n') {
                    file.write_all(b"-")?;
                    file.write_all(line)?;
                }
                for line in hunk.contents[1].split_inclusive(|b| *b == b'\n') {
                    file.write_all(b"+")?;
                    file.write_all(line)?;
                }
            }
        }
    }
    Ok(())
}

async fn get_file_contents(
    store: &Store,
    path: &RepoPath,
    term: &Option<FileId>,
) -> BackendResult<BString> {
    match term {
        Some(id) => {
            let mut content = vec![];
            store
                .read_file_async(path, id)
                .await?
                .read_to_end(&mut content)
                .map_err(|err| BackendError::ReadFile {
                    path: path.to_owned(),
                    id: id.clone(),
                    source: err.into(),
                })?;
            Ok(BString::new(content))
        }
        // If the conflict had removed the file on one side, we pretend that the file
        // was empty there.
        None => Ok(BString::new(vec![])),
    }
}

pub async fn extract_as_single_hunk(
    merge: &Merge<Option<FileId>>,
    store: &Store,
    path: &RepoPath,
) -> BackendResult<Merge<BString>> {
    let builder: MergeBuilder<BString> = futures::stream::iter(merge.iter())
        .then(|term| get_file_contents(store, path, term))
        .try_collect()
        .await?;
    Ok(builder.build())
}

/// A type similar to `MergedTreeValue` but with associated data to include in
/// e.g. the working copy or in a diff.
pub enum MaterializedTreeValue {
    Absent,
    AccessDenied(Box<dyn std::error::Error + Send + Sync>),
    File {
        id: FileId,
        executable: bool,
        reader: Box<dyn Read>,
    },
    Symlink {
        id: SymlinkId,
        target: String,
    },
    FileConflict {
        id: Merge<Option<FileId>>,
        // TODO: or Vec<(FileId, Box<dyn Read>)> so that caller can stop reading
        // when null bytes found?
        contents: Merge<BString>,
        executable: bool,
    },
    OtherConflict {
        id: MergedTreeValue,
    },
    GitSubmodule(CommitId),
    Tree(TreeId),
}

impl MaterializedTreeValue {
    pub fn is_absent(&self) -> bool {
        matches!(self, MaterializedTreeValue::Absent)
    }

    pub fn is_present(&self) -> bool {
        !self.is_absent()
    }
}

/// Reads the data associated with a `MergedTreeValue` so it can be written to
/// e.g. the working copy or diff.
pub async fn materialize_tree_value(
    store: &Store,
    path: &RepoPath,
    value: MergedTreeValue,
) -> BackendResult<MaterializedTreeValue> {
    match materialize_tree_value_no_access_denied(store, path, value).await {
        Err(BackendError::ReadAccessDenied { source, .. }) => {
            Ok(MaterializedTreeValue::AccessDenied(source))
        }
        result => result,
    }
}

async fn materialize_tree_value_no_access_denied(
    store: &Store,
    path: &RepoPath,
    value: MergedTreeValue,
) -> BackendResult<MaterializedTreeValue> {
    match value.into_resolved() {
        Ok(None) => Ok(MaterializedTreeValue::Absent),
        Ok(Some(TreeValue::File { id, executable })) => {
            let reader = store.read_file_async(path, &id).await?;
            Ok(MaterializedTreeValue::File {
                id,
                executable,
                reader,
            })
        }
        Ok(Some(TreeValue::Symlink(id))) => {
            let target = store.read_symlink_async(path, &id).await?;
            Ok(MaterializedTreeValue::Symlink { id, target })
        }
        Ok(Some(TreeValue::GitSubmodule(id))) => Ok(MaterializedTreeValue::GitSubmodule(id)),
        Ok(Some(TreeValue::Tree(id))) => Ok(MaterializedTreeValue::Tree(id)),
        Ok(Some(TreeValue::Conflict(_))) => {
            panic!("cannot materialize legacy conflict object at path {path:?}");
        }
        Err(conflict) => {
            let Some(file_merge) = conflict.to_file_merge() else {
                return Ok(MaterializedTreeValue::OtherConflict { id: conflict });
            };
            let file_merge = file_merge.simplify();
            let contents = extract_as_single_hunk(&file_merge, store, path).await?;
            let executable = if let Some(merge) = conflict.to_executable_merge() {
                merge.resolve_trivial().copied().unwrap_or_default()
            } else {
                false
            };
            Ok(MaterializedTreeValue::FileConflict {
                id: file_merge,
                contents,
                executable,
            })
        }
    }
}

/// Describes what style should be used when materializing conflicts.
#[derive(Clone, Copy, PartialEq, Eq, Debug, Default, serde::Deserialize)]
#[serde(rename_all = "kebab-case")]
pub enum ConflictMarkerStyle {
    /// Style which shows a snapshot and a series of diffs to apply.
    #[default]
    Diff,
    /// Style which shows a snapshot for each base and side.
    Snapshot,
    /// Style which replicates Git's "diff3" style to support external tools.
    Git,
}

pub fn materialize_merge_result<T: AsRef<[u8]>>(
    single_hunk: &Merge<T>,
    conflict_marker_style: ConflictMarkerStyle,
    output: &mut dyn Write,
) -> io::Result<()> {
    let merge_result = files::merge(single_hunk);
    match &merge_result {
        MergeResult::Resolved(content) => output.write_all(content),
        MergeResult::Conflict(hunks) => {
            materialize_conflict_hunks(hunks, conflict_marker_style, output)
        }
    }
}

pub fn materialize_merge_result_to_bytes<T: AsRef<[u8]>>(
    single_hunk: &Merge<T>,
    conflict_marker_style: ConflictMarkerStyle,
) -> BString {
    let merge_result = files::merge(single_hunk);
    match merge_result {
        MergeResult::Resolved(content) => content,
        MergeResult::Conflict(hunks) => {
            let mut output = Vec::new();
            materialize_conflict_hunks(&hunks, conflict_marker_style, &mut output)
                .expect("writing to an in-memory buffer should never fail");
            output.into()
        }
    }
}

fn materialize_conflict_hunks(
    hunks: &[Merge<BString>],
    conflict_marker_style: ConflictMarkerStyle,
    output: &mut dyn Write,
) -> io::Result<()> {
    let num_conflicts = hunks
        .iter()
        .filter(|hunk| hunk.as_resolved().is_none())
        .count();
    let mut conflict_index = 0;
    for hunk in hunks {
        if let Some(content) = hunk.as_resolved() {
            output.write_all(content)?;
        } else {
            conflict_index += 1;
            let conflict_info = format!("Conflict {conflict_index} of {num_conflicts}");

            match (conflict_marker_style, hunk.as_slice()) {
                // 2-sided conflicts can use Git-style conflict markers
                (ConflictMarkerStyle::Git, [left, base, right]) => {
                    materialize_git_style_conflict(left, base, right, &conflict_info, output)?;
                }
                _ => {
                    materialize_jj_style_conflict(
                        hunk,
                        &conflict_info,
                        conflict_marker_style,
                        output,
                    )?;
                }
            }
        }
    }
    Ok(())
}

fn materialize_git_style_conflict(
    left: &[u8],
    base: &[u8],
    right: &[u8],
    conflict_info: &str,
    output: &mut dyn Write,
) -> io::Result<()> {
    writeln!(output, "{CONFLICT_START_LINE} Side #1 ({conflict_info})")?;
    output.write_all(left)?;
    writeln!(output, "{CONFLICT_GIT_ANCESTOR_LINE} Base")?;
    output.write_all(base)?;
    // VS Code doesn't seem to support any trailing text on the separator line
    writeln!(output, "{CONFLICT_GIT_SEPARATOR_LINE}")?;
    output.write_all(right)?;
    writeln!(output, "{CONFLICT_END_LINE} Side #2 ({conflict_info} ends)")?;

    Ok(())
}

fn materialize_jj_style_conflict(
    hunk: &Merge<BString>,
    conflict_info: &str,
    conflict_marker_style: ConflictMarkerStyle,
    output: &mut dyn Write,
) -> io::Result<()> {
    // Write a positive snapshot (side) of a conflict
    fn write_side(add_index: usize, data: &[u8], output: &mut dyn Write) -> io::Result<()> {
        writeln!(
            output,
            "{CONFLICT_PLUS_LINE} Contents of side #{}",
            add_index + 1
        )?;
        output.write_all(data)
    }

    // Write a negative snapshot (base) of a conflict
    fn write_base(base_str: &str, data: &[u8], output: &mut dyn Write) -> io::Result<()> {
        writeln!(output, "{CONFLICT_MINUS_LINE} Contents of {base_str}")?;
        output.write_all(data)
    }

    // Write a diff from a negative term to a positive term
    fn write_diff(
        base_str: &str,
        add_index: usize,
        diff: &[DiffHunk],
        output: &mut dyn Write,
    ) -> io::Result<()> {
        writeln!(
            output,
            "{CONFLICT_DIFF_LINE} Changes from {base_str} to side #{}",
            add_index + 1
        )?;
        write_diff_hunks(diff, output)
    }

    writeln!(output, "{CONFLICT_START_LINE} {conflict_info}")?;
    let mut add_index = 0;
    for (base_index, left) in hunk.removes().enumerate() {
        // The vast majority of conflicts one actually tries to resolve manually have 1
        // base.
        let base_str = if hunk.removes().len() == 1 {
            "base".to_string()
        } else {
            format!("base #{}", base_index + 1)
        };

        let Some(right1) = hunk.get_add(add_index) else {
            // If we have no more positive terms, emit the remaining negative terms as
            // snapshots.
            write_base(&base_str, left, output)?;
            continue;
        };

        // For any style other than "diff", always emit sides and bases separately
        if conflict_marker_style != ConflictMarkerStyle::Diff {
            write_side(add_index, right1, output)?;
            write_base(&base_str, left, output)?;
            add_index += 1;
            continue;
        }

        let diff1 = Diff::by_line([&left, &right1]).hunks().collect_vec();
        // Check if the diff against the next positive term is better. Since we want to
        // preserve the order of the terms, we don't match against any later positive
        // terms.
        if let Some(right2) = hunk.get_add(add_index + 1) {
            let diff2 = Diff::by_line([&left, &right2]).hunks().collect_vec();
            if diff_size(&diff2) < diff_size(&diff1) {
                // If the next positive term is a better match, emit the current positive term
                // as a snapshot and the next positive term as a diff.
                write_side(add_index, right1, output)?;
                write_diff(&base_str, add_index + 1, &diff2, output)?;
                add_index += 2;
                continue;
            }
        }

        write_diff(&base_str, add_index, &diff1, output)?;
        add_index += 1;
    }

    // Emit the remaining positive terms as snapshots.
    for (add_index, slice) in hunk.adds().enumerate().skip(add_index) {
        write_side(add_index, slice, output)?;
    }
    writeln!(output, "{CONFLICT_END_LINE} {conflict_info} ends")?;
    Ok(())
}

fn diff_size(hunks: &[DiffHunk]) -> usize {
    hunks
        .iter()
        .map(|hunk| match hunk.kind {
            DiffHunkKind::Matching => 0,
            DiffHunkKind::Different => hunk.contents.iter().map(|content| content.len()).sum(),
        })
        .sum()
}

pub struct MaterializedTreeDiffEntry {
    pub path: CopiesTreeDiffEntryPath,
    pub values: BackendResult<(MaterializedTreeValue, MaterializedTreeValue)>,
}

pub fn materialized_diff_stream<'a>(
    store: &'a Store,
    tree_diff: BoxStream<'a, CopiesTreeDiffEntry>,
) -> impl Stream<Item = MaterializedTreeDiffEntry> + 'a {
    tree_diff
        .map(|CopiesTreeDiffEntry { path, values }| async {
            match values {
                Err(err) => MaterializedTreeDiffEntry {
                    path,
                    values: Err(err),
                },
                Ok((before, after)) => {
                    let before_future = materialize_tree_value(store, path.source(), before);
                    let after_future = materialize_tree_value(store, path.target(), after);
                    let values = try_join!(before_future, after_future);
                    MaterializedTreeDiffEntry { path, values }
                }
            }
        })
        .buffered((store.concurrency() / 2).max(1))
}

/// Parses conflict markers from a slice.
///
/// Returns `None` if there were no valid conflict markers. The caller
/// has to provide the expected number of merge sides (adds). Conflict
/// markers that are otherwise valid will be considered invalid if
/// they don't have the expected arity.
// TODO: "parse" is not usually the opposite of "materialize", so maybe we
// should rename them to "serialize" and "deserialize"?
pub fn parse_conflict(input: &[u8], num_sides: usize) -> Option<Vec<Merge<BString>>> {
    if input.is_empty() {
        return None;
    }
    let mut hunks = vec![];
    let mut pos = 0;
    let mut resolved_start = 0;
    let mut conflict_start = None;
    let mut conflict_start_len = 0;
    for line in input.split_inclusive(|b| *b == b'\n') {
        if is_conflict_marker_line(line) {
            if line[0] == CONFLICT_START_LINE_CHAR {
                conflict_start = Some(pos);
                conflict_start_len = line.len();
            } else if conflict_start.is_some() && line[0] == CONFLICT_END_LINE_CHAR {
                let conflict_body = &input[conflict_start.unwrap() + conflict_start_len..pos];
                let hunk = parse_conflict_hunk(conflict_body);
                if hunk.num_sides() == num_sides {
                    let resolved_slice = &input[resolved_start..conflict_start.unwrap()];
                    if !resolved_slice.is_empty() {
                        hunks.push(Merge::resolved(BString::from(resolved_slice)));
                    }
                    hunks.push(hunk);
                    resolved_start = pos + line.len();
                }
                conflict_start = None;
            }
        }
        pos += line.len();
    }

    if hunks.is_empty() {
        None
    } else {
        if resolved_start < input.len() {
            hunks.push(Merge::resolved(BString::from(&input[resolved_start..])));
        }
        Some(hunks)
    }
}

/// This method handles parsing both JJ-style and Git-style conflict markers,
/// meaning that switching conflict marker styles won't prevent existing files
/// with other conflict marker styles from being parsed successfully. The
/// conflict marker style to use for parsing is determined based on the first
/// line of the hunk.
fn parse_conflict_hunk(input: &[u8]) -> Merge<BString> {
    // If the hunk starts with a conflict marker, find its first character
    let initial_conflict_marker_char = input
        .lines_with_terminator()
        .next()
        .filter(|line| is_conflict_marker_line(line))
        .map(|line| line[0]);

    match initial_conflict_marker_char {
        // JJ-style conflicts must start with one of these 3 conflict marker lines
        Some(CONFLICT_DIFF_LINE_CHAR | CONFLICT_MINUS_LINE_CHAR | CONFLICT_PLUS_LINE_CHAR) => {
            parse_jj_style_conflict_hunk(input)
        }
        // Git-style conflicts either must not start with a conflict marker line, or must start with
        // the "|||||||" conflict marker line (if the first side was empty)
        None | Some(CONFLICT_GIT_ANCESTOR_LINE_CHAR) => parse_git_style_conflict_hunk(input),
        // No other conflict markers are allowed at the start of a hunk
        Some(_) => Merge::resolved(BString::new(vec![])),
    }
}

fn parse_jj_style_conflict_hunk(input: &[u8]) -> Merge<BString> {
    enum State {
        Diff,
        Minus,
        Plus,
        Unknown,
    }
    let mut state = State::Unknown;
    let mut removes = vec![];
    let mut adds = vec![];
    for line in input.lines_with_terminator() {
        if is_conflict_marker_line(line) {
            match line[0] {
                CONFLICT_DIFF_LINE_CHAR => {
                    state = State::Diff;
                    removes.push(BString::new(vec![]));
                    adds.push(BString::new(vec![]));
                    continue;
                }
                CONFLICT_MINUS_LINE_CHAR => {
                    state = State::Minus;
                    removes.push(BString::new(vec![]));
                    continue;
                }
                CONFLICT_PLUS_LINE_CHAR => {
                    state = State::Plus;
                    adds.push(BString::new(vec![]));
                    continue;
                }
                _ => {}
            }
        }
        match state {
            State::Diff => {
                if let Some(rest) = line.strip_prefix(b"-") {
                    removes.last_mut().unwrap().extend_from_slice(rest);
                } else if let Some(rest) = line.strip_prefix(b"+") {
                    adds.last_mut().unwrap().extend_from_slice(rest);
                } else if let Some(rest) = line.strip_prefix(b" ") {
                    removes.last_mut().unwrap().extend_from_slice(rest);
                    adds.last_mut().unwrap().extend_from_slice(rest);
                } else if line == b"\n" || line == b"\r\n" {
                    // Some editors strip trailing whitespace, so " \n" might become "\n". It would
                    // be unfortunate if this prevented the conflict from being parsed, so we add
                    // the empty line to the "remove" and "add" as if there was a space in front
                    removes.last_mut().unwrap().extend_from_slice(line);
                    adds.last_mut().unwrap().extend_from_slice(line);
                } else {
                    // Doesn't look like a valid conflict
                    return Merge::resolved(BString::new(vec![]));
                }
            }
            State::Minus => {
                removes.last_mut().unwrap().extend_from_slice(line);
            }
            State::Plus => {
                adds.last_mut().unwrap().extend_from_slice(line);
            }
            State::Unknown => {
                // Doesn't look like a valid conflict
                return Merge::resolved(BString::new(vec![]));
            }
        }
    }

    if adds.len() == removes.len() + 1 {
        Merge::from_removes_adds(removes, adds)
    } else {
        // Doesn't look like a valid conflict
        Merge::resolved(BString::new(vec![]))
    }
}

fn parse_git_style_conflict_hunk(input: &[u8]) -> Merge<BString> {
    #[derive(PartialEq, Eq)]
    enum State {
        Left,
        Base,
        Right,
    }
    let mut state = State::Left;
    let mut left = BString::new(vec![]);
    let mut base = BString::new(vec![]);
    let mut right = BString::new(vec![]);
    for line in input.lines_with_terminator() {
        if is_conflict_marker_line(line) {
            match line[0] {
                CONFLICT_GIT_ANCESTOR_LINE_CHAR => {
                    if state == State::Left {
                        state = State::Base;
                        continue;
                    } else {
                        // Base must come after left
                        return Merge::resolved(BString::new(vec![]));
                    }
                }
                CONFLICT_GIT_SEPARATOR_LINE_CHAR => {
                    if state == State::Base {
                        state = State::Right;
                        continue;
                    } else {
                        // Right must come after base
                        return Merge::resolved(BString::new(vec![]));
                    }
                }
                _ => {}
            }
        }
        match state {
            State::Left => left.extend_from_slice(line),
            State::Base => base.extend_from_slice(line),
            State::Right => right.extend_from_slice(line),
        }
    }

    if state == State::Right {
        Merge::from_vec(vec![left, base, right])
    } else {
        // Doesn't look like a valid conflict
        Merge::resolved(BString::new(vec![]))
    }
}

/// Check whether a line is a conflict marker. Removes trailing whitespace
/// before checking against regex to ensure it parses CRLF endings correctly.
fn is_conflict_marker_line(line: &[u8]) -> bool {
    let line = line.trim_end_with(|ch| ch.is_ascii_whitespace());
    CONFLICT_MARKER_REGEX.is_match_at(line, 0)
}

/// Parses conflict markers in `content` and returns an updated version of
/// `file_ids` with the new contents. If no (valid) conflict markers remain, a
/// single resolves `FileId` will be returned.
pub async fn update_from_content(
    file_ids: &Merge<Option<FileId>>,
    store: &Store,
    path: &RepoPath,
    content: &[u8],
    conflict_marker_style: ConflictMarkerStyle,
) -> BackendResult<Merge<Option<FileId>>> {
    let simplified_file_ids = file_ids.clone().simplify();
    let simplified_file_ids = &simplified_file_ids;

    // First check if the new content is unchanged compared to the old content. If
    // it is, we don't need parse the content or write any new objects to the
    // store. This is also a way of making sure that unchanged tree/file
    // conflicts (for example) are not converted to regular files in the working
    // copy.
    let mut old_content = Vec::with_capacity(content.len());
    let merge_hunk = extract_as_single_hunk(simplified_file_ids, store, path).await?;
    materialize_merge_result(&merge_hunk, conflict_marker_style, &mut old_content).unwrap();
    if content == old_content {
        return Ok(file_ids.clone());
    }

    // Parse conflicts from the new content using the arity of the simplified
    // conflicts initially. If unsuccessful, attempt to parse conflicts from with
    // the arity of the unsimplified conflicts since such a conflict may be
    // present in the working copy if written by an earlier version of jj.
    let (used_file_ids, hunks) = 'hunks: {
        if let Some(hunks) = parse_conflict(content, simplified_file_ids.num_sides()) {
            break 'hunks (simplified_file_ids, hunks);
        };
        if simplified_file_ids.num_sides() != file_ids.num_sides() {
            if let Some(hunks) = parse_conflict(content, file_ids.num_sides()) {
                break 'hunks (file_ids, hunks);
            };
        };
        // Either there are no markers or they don't have the expected arity
        let file_id = store.write_file(path, &mut &content[..]).await?;
        return Ok(Merge::normal(file_id));
    };

    let mut contents = used_file_ids.map(|_| vec![]);
    for hunk in hunks {
        if let Some(slice) = hunk.as_resolved() {
            for content in contents.iter_mut() {
                content.extend_from_slice(slice);
            }
        } else {
            for (content, slice) in zip(contents.iter_mut(), hunk.into_iter()) {
                content.extend(Vec::from(slice));
            }
        }
    }

    // If the user edited the empty placeholder for an absent side, we consider the
    // conflict resolved.
    if zip(contents.iter(), used_file_ids.iter())
        .any(|(content, file_id)| file_id.is_none() && !content.is_empty())
    {
        let file_id = store.write_file(path, &mut &content[..]).await?;
        return Ok(Merge::normal(file_id));
    }

    // Now write the new files contents we found by parsing the file with conflict
    // markers.
    // TODO: Write these concurrently
    let new_file_ids: Vec<Option<FileId>> = zip(contents.iter(), used_file_ids.iter())
        .map(|(content, file_id)| -> BackendResult<Option<FileId>> {
            match file_id {
                Some(_) => {
                    let file_id = store.write_file(path, &mut content.as_slice()).block_on()?;
                    Ok(Some(file_id))
                }
                None => {
                    // The missing side of a conflict is still represented by
                    // the empty string we materialized it as
                    Ok(None)
                }
            }
        })
        .try_collect()?;

    // If the conflict was simplified, expand the conflict to the original
    // number of sides.
    let new_file_ids = if new_file_ids.len() != file_ids.iter().len() {
        file_ids
            .clone()
            .update_from_simplified(Merge::from_vec(new_file_ids))
    } else {
        Merge::from_vec(new_file_ids)
    };
    Ok(new_file_ids)
}