lazygit/pkg/commands/patch/patch_modifier.go
Stefan Haller 9cc33c479b Use forward patches and --reverse flag for partial patches too
There's no reason to have two different ways of applying patches for whole-file
patches and partial patches; use --reverse for both. Not only does this simplify
the code a bit, but it fixes an actual problem: when reverseOnGenerate and
keepOriginalHeader are both true, the generated patch header is broken (the two
blobs in the line `index 6d1959b..6dc5f84 100644` are swapped). Git fails to do
a proper three-way merge in that case, as it expects the first of the two blobs
to be the common ancestor.

It would be possible to fix this by extending ModifiedPatchForLines to swap the
two blobs in this case; but this would prevent us from concatenating all patches
and apply them in one go, which we are going to do later in the branch.
2023-03-07 09:49:34 +01:00

197 lines
5.2 KiB
Go

package patch
import (
"fmt"
"regexp"
"strings"
"github.com/sirupsen/logrus"
)
var (
hunkHeaderRegexp = regexp.MustCompile(`(?m)^@@ -(\d+)[^\+]+\+(\d+)[^@]+@@(.*)$`)
patchHeaderRegexp = regexp.MustCompile(`(?ms)(^diff.*?)^@@`)
)
type PatchOptions struct {
// Create a reverse patch; in other words, flip all the '+' and '-' while
// generating the patch.
Reverse bool
// If true, we're building a patch that we are going to apply using
// "git apply --reverse". In other words, we are not flipping the '+' and
// '-' ourselves while creating the patch, but git is going to do that when
// applying. This has consequences for which lines we need to keep or
// discard when filtering lines from partial hunks.
//
// Currently incompatible with Reverse.
WillBeAppliedReverse bool
// Whether to keep or discard the original diff header including the
// "index deadbeef..fa1afe1 100644" line.
KeepOriginalHeader bool
}
func GetHeaderFromDiff(diff string) string {
match := patchHeaderRegexp.FindStringSubmatch(diff)
if len(match) <= 1 {
return ""
}
return match[1]
}
func GetHunksFromDiff(diff string) []*PatchHunk {
hunks := []*PatchHunk{}
firstLineIdx := -1
var hunkLines []string //nolint:prealloc
pastDiffHeader := false
lines := strings.SplitAfter(diff, "\n")
for lineIdx, line := range lines {
isHunkHeader := strings.HasPrefix(line, "@@ -")
if isHunkHeader {
if pastDiffHeader { // we need to persist the current hunk
hunks = append(hunks, newHunk(hunkLines, firstLineIdx))
}
pastDiffHeader = true
firstLineIdx = lineIdx
hunkLines = []string{line}
continue
}
if !pastDiffHeader { // skip through the stuff that precedes the first hunk
continue
}
if lineIdx == len(lines)-1 && line == "" { // skip the trailing newline
continue
}
hunkLines = append(hunkLines, line)
}
if pastDiffHeader {
hunks = append(hunks, newHunk(hunkLines, firstLineIdx))
}
return hunks
}
type PatchModifier struct {
Log *logrus.Entry
filename string
hunks []*PatchHunk
header string
}
func NewPatchModifier(log *logrus.Entry, filename string, diffText string) *PatchModifier {
return &PatchModifier{
Log: log,
filename: filename,
hunks: GetHunksFromDiff(diffText),
header: GetHeaderFromDiff(diffText),
}
}
func (d *PatchModifier) ModifiedPatchForLines(lineIndices []int, opts PatchOptions) string {
if opts.Reverse && opts.KeepOriginalHeader {
panic("reverse and keepOriginalHeader are not compatible")
}
// step one is getting only those hunks which we care about
hunksInRange := []*PatchHunk{}
outer:
for _, hunk := range d.hunks {
// if there is any line in our lineIndices array that the hunk contains, we append it
for _, lineIdx := range lineIndices {
if lineIdx >= hunk.FirstLineIdx && lineIdx <= hunk.LastLineIdx() {
hunksInRange = append(hunksInRange, hunk)
continue outer
}
}
}
// step 2 is collecting all the hunks with new headers
startOffset := 0
formattedHunks := ""
var formattedHunk string
for _, hunk := range hunksInRange {
startOffset, formattedHunk = hunk.formatWithChanges(
lineIndices, opts.Reverse, opts.WillBeAppliedReverse, startOffset)
formattedHunks += formattedHunk
}
if formattedHunks == "" {
return ""
}
var fileHeader string
// for staging/unstaging lines we don't want the original header because
// it makes git confused e.g. when dealing with deleted/added files
// but with building and applying patches the original header gives git
// information it needs to cleanly apply patches
if opts.KeepOriginalHeader {
fileHeader = d.header
} else {
fileHeader = fmt.Sprintf("--- a/%s\n+++ b/%s\n", d.filename, d.filename)
}
return fileHeader + formattedHunks
}
func (d *PatchModifier) ModifiedPatchForRange(firstLineIdx int, lastLineIdx int, opts PatchOptions) string {
// generate array of consecutive line indices from our range
selectedLines := []int{}
for i := firstLineIdx; i <= lastLineIdx; i++ {
selectedLines = append(selectedLines, i)
}
return d.ModifiedPatchForLines(selectedLines, opts)
}
func (d *PatchModifier) OriginalPatchLength() int {
if len(d.hunks) == 0 {
return 0
}
return d.hunks[len(d.hunks)-1].LastLineIdx()
}
func ModifiedPatchForRange(log *logrus.Entry, filename string, diffText string, firstLineIdx int, lastLineIdx int, opts PatchOptions) string {
p := NewPatchModifier(log, filename, diffText)
return p.ModifiedPatchForRange(firstLineIdx, lastLineIdx, opts)
}
func ModifiedPatchForLines(log *logrus.Entry, filename string, diffText string, includedLineIndices []int, opts PatchOptions) string {
p := NewPatchModifier(log, filename, diffText)
return p.ModifiedPatchForLines(includedLineIndices, opts)
}
// I want to know, given a hunk, what line a given index is on
func (hunk *PatchHunk) LineNumberOfLine(idx int) int {
n := idx - hunk.FirstLineIdx - 1
if n < 0 {
n = 0
} else if n >= len(hunk.bodyLines) {
n = len(hunk.bodyLines) - 1
}
lines := hunk.bodyLines[0:n]
offset := nLinesWithPrefix(lines, []string{"+", " "})
return hunk.newStart + offset
}
func nLinesWithPrefix(lines []string, chars []string) int {
result := 0
for _, line := range lines {
for _, char := range chars {
if line[:1] == char {
result++
}
}
}
return result
}