func testBytesFind(t *testing.T, line, token []byte, startat int, exp []int) {
got := []int{}
tokenlen := len(token)
for {
foundat := tekstus.BytesFind(line, token, startat)
if foundat < 0 {
break
}
got = append(got, foundat)
startat = foundat + tokenlen
}
assert(t, exp, got, true)
}
func searchForward(atx, aty int, x, y *int, oldleft, newleft *[]byte) (
adds, dels tekstus.Chunks,
) {
oldleftlen := len(*oldleft)
newleftlen := len(*newleft)
minlen := DefMatchLen
if oldleftlen < minlen {
minlen = oldleftlen
}
// Loop through old line to find matching token
xaty := -1
xx := 1
for ; xx < oldleftlen-minlen; xx++ {
token := (*oldleft)[xx : xx+minlen]
xaty = tekstus.BytesFind(*newleft, token, 0)
if xaty > 0 {
break
}
}
minlen = DefMatchLen
if newleftlen < minlen {
minlen = newleftlen
}
yatx := -1
yy := 1
for ; yy < newleftlen-minlen; yy++ {
token := (*newleft)[yy : yy+minlen]
yatx = tekstus.BytesFind(*oldleft, token, 0)
if yatx > 0 {
break
}
}
if xaty < 0 && yatx < 0 {
// still no token found, means whole chunk has been replaced.
dels = append(dels, tekstus.Chunk{StartAt: atx + *x, V: *oldleft})
adds = append(adds, tekstus.Chunk{StartAt: aty + *y, V: *newleft})
*oldleft = []byte{}
*newleft = []byte{}
return adds, dels
}
// Some chunk has been replaced.
v := (*oldleft)[:xx]
dels = append(dels, tekstus.Chunk{StartAt: atx + *x, V: v})
*oldleft = (*oldleft)[xx:]
*x = *x + xx
v = (*newleft)[:yy]
adds = append(adds, tekstus.Chunk{StartAt: aty + *y, V: v})
*newleft = (*newleft)[yy:]
*y = *y + yy
return adds, dels
}
/*
Lines given two similar lines, find and return the differences (additions and
deletion) between them.
Case 1: addition on new or deletion on old.
old: 00000
new: 00000111
or
old: 00000111
new: 00000
Case 2: addition on new line
old: 000000
new: 0001000
Case 3: deletion on old line (reverse of case 2)
old: 0001000
new: 000000
Case 4: change happened in the beginning
old: 11000
new: 22000
Case 5: both changed
old: 0001000
new: 0002000
*/
func Lines(old, new []byte, atx, aty int) (adds, dels tekstus.Chunks) {
oldlen := len(old)
newlen := len(new)
minlen := 0
if oldlen < newlen {
minlen = oldlen
} else {
minlen = newlen
}
// Find the position of unmatched byte from the beginning.
x, y := 0, 0
for ; x < minlen; x++ {
if old[x] != new[x] {
break
}
}
y = x
// Case 1: Check if addition or deletion is at the end.
if x == minlen {
if oldlen < newlen {
v := new[y:]
adds = append(adds, tekstus.Chunk{StartAt: atx + y, V: v})
} else {
v := old[x:]
dels = append(dels, tekstus.Chunk{StartAt: atx + x, V: v})
}
return
}
// Find the position of unmatched byte from the end
xend := oldlen - 1
yend := newlen - 1
for xend >= x && yend >= y {
if old[xend] != new[yend] {
break
}
xend--
yend--
}
// Case 2: addition in new line.
if x == xend+1 {
v := new[y : yend+1]
adds = append(adds, tekstus.Chunk{StartAt: aty + y, V: v})
return
}
// Case 3: deletion in old line.
if y == yend+1 {
v := old[x : xend+1]
dels = append(dels, tekstus.Chunk{StartAt: atx + x, V: v})
return
}
// Calculate possible match len.
// After we found similar bytes in the beginning and end of line, now
// we have `n` number of bytes left in old and new.
oldleft := old[x : xend+1]
newleft := new[y : yend+1]
oldleftlen := len(oldleft)
newleftlen := len(newleft)
// Get minimal token to search in the new left over.
minlen = DefMatchLen
if oldleftlen < DefMatchLen {
minlen = oldleftlen
}
xtoken := oldleft[:minlen]
xaty := tekstus.BytesFind(newleft, xtoken, 0)
// Get miniminal token to search in the old left over.
minlen = DefMatchLen
if newleftlen < DefMatchLen {
minlen = newleftlen
}
ytoken := newleft[:minlen]
yatx := tekstus.BytesFind(oldleft, ytoken, 0)
// Case 4:
// We did not find matching token of x in y, its mean the some chunk
// in x and y has been replaced.
if xaty < 0 && yatx < 0 {
addsleft, delsleft := searchForward(atx, aty, &x, &y, &oldleft,
&newleft)
if len(addsleft) > 0 {
adds = append(adds, addsleft...)
}
if len(delsleft) > 0 {
dels = append(dels, delsleft...)
}
// Check for possible empty left
if len(oldleft) == 0 {
if len(newleft) > 0 {
adds = append(adds, tekstus.Chunk{
StartAt: atx + x,
V: newleft,
})
}
return
}
if len(newleft) == 0 {
if len(oldleft) > 0 {
dels = append(dels, tekstus.Chunk{
StartAt: aty + y,
V: oldleft,
})
}
return
}
}
// Case 5: is combination of case 2 and 3.
// Case 2: We found x token at y: xaty. Previous byte before that must
// be an addition.
if xaty >= 0 {
v := new[y : y+xaty]
adds = append(adds, tekstus.Chunk{StartAt: aty + y, V: v})
newleft = new[y+xaty : yend+1]
} else {
if yatx >= 0 {
// Case 3: We found y token at x: yatx. Previous byte before that must
// be a deletion.
v := old[x : x+yatx]
dels = append(dels, tekstus.Chunk{StartAt: atx + x, V: v})
oldleft = old[x+yatx : xend+1]
}
}
addsleft, delsleft := Lines(oldleft, newleft, atx+x, aty+y)
if len(addsleft) > 0 {
adds = append(adds, addsleft...)
}
if len(delsleft) > 0 {
dels = append(dels, delsleft...)
}
return
}
/*
BytesRatio compare two slice of bytes and return ratio of matching bytes.
The ratio in in range of 0.0 to 1.0, where 1.0 if both are similar, and 0.0 if
no matchs even found.
`minTokenLen` define the minimum length of token for searching in both of
slice.
*/
func BytesRatio(old, new []byte, minTokenLen int) (
ratio float32, m int, maxlen int,
) {
x, y := 0, 0
oldlen := len(old)
newlen := len(new)
minlen := oldlen
maxlen = newlen
if newlen < oldlen {
minlen = newlen
maxlen = oldlen
}
if minTokenLen < 0 {
minTokenLen = DefMatchLen
}
for {
// Count matching bytes from beginning of slice.
for x < minlen {
if old[x] != new[y] {
break
}
m++
x++
y++
}
if x == minlen {
// All bytes is matched but probably some trailing in
// one of them.
break
}
// Count matching bytes from end of slice
xend := oldlen - 1
yend := newlen - 1
for xend >= x && yend >= y {
if old[xend] != new[yend] {
break
}
m++
xend--
yend--
}
// One of the line have changes in the middle.
if xend == x || yend == y {
break
}
// Cut the matching bytes
old = old[x : xend+1]
new = new[y : yend+1]
oldlen = len(old)
newlen = len(new)
// Get minimal token to search in the new left over.
minlen = minTokenLen
if oldlen < minlen {
minlen = oldlen
}
// Search old token in new, chunk by chunk.
x = 0
y = -1
max := oldlen - minlen
for ; x < max; x++ {
token := old[x : x+minlen]
y = tekstus.BytesFind(new, token, 0)
if y > 0 {
break
}
}
if y < 0 {
// We did not found anything.
break
}
// Cut the changes
old = old[x:]
new = new[y:]
oldlen = len(old)
newlen = len(new)
minlen = oldlen
if newlen < minlen {
minlen = newlen
}
x, y = 0, 0
// start again from begining...
}
ratio = float32(m) / float32(maxlen)
return ratio, m, maxlen
}