chore: copy x/tools/internal/diff

Author: ldez

internal/x/tools/diff/diff.go

@@ -0,0 +1,176 @@
+// Copyright 2019 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// Package diff computes differences between text files or strings.
+package diff
+
+import (
+	"fmt"
+	"sort"
+	"strings"
+)
+
+// An Edit describes the replacement of a portion of a text file.
+type Edit struct {
+	Start, End int    // byte offsets of the region to replace
+	New        string // the replacement
+}
+
+func (e Edit) String() string {
+	return fmt.Sprintf("{Start:%d,End:%d,New:%q}", e.Start, e.End, e.New)
+}
+
+// Apply applies a sequence of edits to the src buffer and returns the
+// result. Edits are applied in order of start offset; edits with the
+// same start offset are applied in they order they were provided.
+//
+// Apply returns an error if any edit is out of bounds,
+// or if any pair of edits is overlapping.
+func Apply(src string, edits []Edit) (string, error) {
+	edits, size, err := validate(src, edits)
+	if err != nil {
+		return "", err
+	}
+
+	// Apply edits.
+	out := make([]byte, 0, size)
+	lastEnd := 0
+	for _, edit := range edits {
+		if lastEnd < edit.Start {
+			out = append(out, src[lastEnd:edit.Start]...)
+		}
+		out = append(out, edit.New...)
+		lastEnd = edit.End
+	}
+	out = append(out, src[lastEnd:]...)
+
+	if len(out) != size {
+		panic("wrong size")
+	}
+
+	return string(out), nil
+}
+
+// ApplyBytes is like Apply, but it accepts a byte slice.
+// The result is always a new array.
+func ApplyBytes(src []byte, edits []Edit) ([]byte, error) {
+	res, err := Apply(string(src), edits)
+	return []byte(res), err
+}
+
+// validate checks that edits are consistent with src,
+// and returns the size of the patched output.
+// It may return a different slice.
+func validate(src string, edits []Edit) ([]Edit, int, error) {
+	if !sort.IsSorted(editsSort(edits)) {
+		edits = append([]Edit(nil), edits...)
+		SortEdits(edits)
+	}
+
+	// Check validity of edits and compute final size.
+	size := len(src)
+	lastEnd := 0
+	for _, edit := range edits {
+		if !(0 <= edit.Start && edit.Start <= edit.End && edit.End <= len(src)) {
+			return nil, 0, fmt.Errorf("diff has out-of-bounds edits")
+		}
+		if edit.Start < lastEnd {
+			return nil, 0, fmt.Errorf("diff has overlapping edits")
+		}
+		size += len(edit.New) + edit.Start - edit.End
+		lastEnd = edit.End
+	}
+
+	return edits, size, nil
+}
+
+// SortEdits orders a slice of Edits by (start, end) offset.
+// This ordering puts insertions (end = start) before deletions
+// (end > start) at the same point, but uses a stable sort to preserve
+// the order of multiple insertions at the same point.
+// (Apply detects multiple deletions at the same point as an error.)
+func SortEdits(edits []Edit) {
+	sort.Stable(editsSort(edits))
+}
+
+type editsSort []Edit
+
+func (a editsSort) Len() int { return len(a) }
+func (a editsSort) Less(i, j int) bool {
+	if cmp := a[i].Start - a[j].Start; cmp != 0 {
+		return cmp < 0
+	}
+	return a[i].End < a[j].End
+}
+func (a editsSort) Swap(i, j int) { a[i], a[j] = a[j], a[i] }
+
+// lineEdits expands and merges a sequence of edits so that each
+// resulting edit replaces one or more complete lines.
+// See ApplyEdits for preconditions.
+func lineEdits(src string, edits []Edit) ([]Edit, error) {
+	edits, _, err := validate(src, edits)
+	if err != nil {
+		return nil, err
+	}
+
+	// Do all deletions begin and end at the start of a line,
+	// and all insertions end with a newline?
+	// (This is merely a fast path.)
+	for _, edit := range edits {
+		if edit.Start >= len(src) || // insertion at EOF
+			edit.Start > 0 && src[edit.Start-1] != '\n' || // not at line start
+			edit.End > 0 && src[edit.End-1] != '\n' || // not at line start
+			edit.New != "" && edit.New[len(edit.New)-1] != '\n' { // partial insert
+			goto expand // slow path
+		}
+	}
+	return edits, nil // aligned
+
+expand:
+	if len(edits) == 0 {
+		return edits, nil // no edits (unreachable due to fast path)
+	}
+	expanded := make([]Edit, 0, len(edits)) // a guess
+	prev := edits[0]
+	// TODO(adonovan): opt: start from the first misaligned edit.
+	// TODO(adonovan): opt: avoid quadratic cost of string += string.
+	for _, edit := range edits[1:] {
+		between := src[prev.End:edit.Start]
+		if !strings.Contains(between, "\n") {
+			// overlapping lines: combine with previous edit.
+			prev.New += between + edit.New
+			prev.End = edit.End
+		} else {
+			// non-overlapping lines: flush previous edit.
+			expanded = append(expanded, expandEdit(prev, src))
+			prev = edit
+		}
+	}
+	return append(expanded, expandEdit(prev, src)), nil // flush final edit
+}
+
+// expandEdit returns edit expanded to complete whole lines.
+func expandEdit(edit Edit, src string) Edit {
+	// Expand start left to start of line.
+	// (delta is the zero-based column number of start.)
+	start := edit.Start
+	if delta := start - 1 - strings.LastIndex(src[:start], "\n"); delta > 0 {
+		edit.Start -= delta
+		edit.New = src[start-delta:start] + edit.New
+	}
+
+	// Expand end right to end of line.
+	end := edit.End
+	if end > 0 && src[end-1] != '\n' ||
+		edit.New != "" && edit.New[len(edit.New)-1] != '\n' {
+		if nl := strings.IndexByte(src[end:], '\n'); nl < 0 {
+			edit.End = len(src) // extend to EOF
+		} else {
+			edit.End = end + nl + 1 // extend beyond \n
+		}
+	}
+	edit.New += src[end:edit.End]
+
+	return edit
+}

internal/x/tools/diff/lcs/common.go

@@ -0,0 +1,179 @@
+// Copyright 2022 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package lcs
+
+import (
+	"log"
+	"sort"
+)
+
+// lcs is a longest common sequence
+type lcs []diag
+
+// A diag is a piece of the edit graph where A[X+i] == B[Y+i], for 0<=i<Len.
+// All computed diagonals are parts of a longest common subsequence.
+type diag struct {
+	X, Y int
+	Len  int
+}
+
+// sort sorts in place, by lowest X, and if tied, inversely by Len
+func (l lcs) sort() lcs {
+	sort.Slice(l, func(i, j int) bool {
+		if l[i].X != l[j].X {
+			return l[i].X < l[j].X
+		}
+		return l[i].Len > l[j].Len
+	})
+	return l
+}
+
+// validate that the elements of the lcs do not overlap
+// (can only happen when the two-sided algorithm ends early)
+// expects the lcs to be sorted
+func (l lcs) valid() bool {
+	for i := 1; i < len(l); i++ {
+		if l[i-1].X+l[i-1].Len > l[i].X {
+			return false
+		}
+		if l[i-1].Y+l[i-1].Len > l[i].Y {
+			return false
+		}
+	}
+	return true
+}
+
+// repair overlapping lcs
+// only called if two-sided stops early
+func (l lcs) fix() lcs {
+	// from the set of diagonals in l, find a maximal non-conflicting set
+	// this problem may be NP-complete, but we use a greedy heuristic,
+	// which is quadratic, but with a better data structure, could be D log D.
+	// indepedent is not enough: {0,3,1} and {3,0,2} can't both occur in an lcs
+	// which has to have monotone x and y
+	if len(l) == 0 {
+		return nil
+	}
+	sort.Slice(l, func(i, j int) bool { return l[i].Len > l[j].Len })
+	tmp := make(lcs, 0, len(l))
+	tmp = append(tmp, l[0])
+	for i := 1; i < len(l); i++ {
+		var dir direction
+		nxt := l[i]
+		for _, in := range tmp {
+			if dir, nxt = overlap(in, nxt); dir == empty || dir == bad {
+				break
+			}
+		}
+		if nxt.Len > 0 && dir != bad {
+			tmp = append(tmp, nxt)
+		}
+	}
+	tmp.sort()
+	if false && !tmp.valid() { // debug checking
+		log.Fatalf("here %d", len(tmp))
+	}
+	return tmp
+}
+
+type direction int
+
+const (
+	empty    direction = iota // diag is empty (so not in lcs)
+	leftdown                  // proposed acceptably to the left and below
+	rightup                   // proposed diag is acceptably to the right and above
+	bad                       // proposed diag is inconsistent with the lcs so far
+)
+
+// overlap trims the proposed diag prop  so it doesn't overlap with
+// the existing diag that has already been added to the lcs.
+func overlap(exist, prop diag) (direction, diag) {
+	if prop.X <= exist.X && exist.X < prop.X+prop.Len {
+		// remove the end of prop where it overlaps with the X end of exist
+		delta := prop.X + prop.Len - exist.X
+		prop.Len -= delta
+		if prop.Len <= 0 {
+			return empty, prop
+		}
+	}
+	if exist.X <= prop.X && prop.X < exist.X+exist.Len {
+		// remove the beginning of prop where overlaps with exist
+		delta := exist.X + exist.Len - prop.X
+		prop.Len -= delta
+		if prop.Len <= 0 {
+			return empty, prop
+		}
+		prop.X += delta
+		prop.Y += delta
+	}
+	if prop.Y <= exist.Y && exist.Y < prop.Y+prop.Len {
+		// remove the end of prop that overlaps (in Y) with exist
+		delta := prop.Y + prop.Len - exist.Y
+		prop.Len -= delta
+		if prop.Len <= 0 {
+			return empty, prop
+		}
+	}
+	if exist.Y <= prop.Y && prop.Y < exist.Y+exist.Len {
+		// remove the beginning of peop that overlaps with exist
+		delta := exist.Y + exist.Len - prop.Y
+		prop.Len -= delta
+		if prop.Len <= 0 {
+			return empty, prop
+		}
+		prop.X += delta // no test reaches this code
+		prop.Y += delta
+	}
+	if prop.X+prop.Len <= exist.X && prop.Y+prop.Len <= exist.Y {
+		return leftdown, prop
+	}
+	if exist.X+exist.Len <= prop.X && exist.Y+exist.Len <= prop.Y {
+		return rightup, prop
+	}
+	// prop can't be in an lcs that contains exist
+	return bad, prop
+}
+
+// manipulating Diag and lcs
+
+// prepend a diagonal (x,y)-(x+1,y+1) segment either to an empty lcs
+// or to its first Diag. prepend is only called to extend diagonals
+// the backward direction.
+func (lcs lcs) prepend(x, y int) lcs {
+	if len(lcs) > 0 {
+		d := &lcs[0]
+		if int(d.X) == x+1 && int(d.Y) == y+1 {
+			// extend the diagonal down and to the left
+			d.X, d.Y = int(x), int(y)
+			d.Len++
+			return lcs
+		}
+	}
+
+	r := diag{X: int(x), Y: int(y), Len: 1}
+	lcs = append([]diag{r}, lcs...)
+	return lcs
+}
+
+// append appends a diagonal, or extends the existing one.
+// by adding the edge (x,y)-(x+1.y+1). append is only called
+// to extend diagonals in the forward direction.
+func (lcs lcs) append(x, y int) lcs {
+	if len(lcs) > 0 {
+		last := &lcs[len(lcs)-1]
+		// Expand last element if adjoining.
+		if last.X+last.Len == x && last.Y+last.Len == y {
+			last.Len++
+			return lcs
+		}
+	}
+
+	return append(lcs, diag{X: x, Y: y, Len: 1})
+}
+
+// enforce constraint on d, k
+func ok(d, k int) bool {
+	return d >= 0 && -d <= k && k <= d
+}