func diffLineSet(orgLines, newLines []string, format string) {
sort.Strings(orgLines)
sort.Strings(newLines)
_, matA, matB := ed.EditDistanceFFull(len(orgLines), len(newLines), func(iA, iB int) int {
return tm.DiffOfStrings(orgLines[iA], newLines[iB], 4000)
}, ed.ConstCost(1000), ed.ConstCost(1000))
for i, j := 0, 0; i < len(orgLines) || j < len(newLines); {
switch {
case j >= len(newLines) || i < len(orgLines) && matA[i] < 0:
showDelLine(fmt.Sprintf(format, orgLines[i]))
i++
case i >= len(orgLines) || j < len(newLines) && matB[j] < 0:
showInsLine(fmt.Sprintf(format, newLines[j]))
j++
default:
if strings.TrimSpace(orgLines[i]) != strings.TrimSpace(newLines[j]) {
showDiffLine(fmt.Sprintf(format, orgLines[i]), fmt.Sprintf(format, newLines[j]))
} // if
i++
j++
}
} // for i, j
}
func MatchTokens(delT, insT []string) (matA, matB []int) {
delL, delR := nearChecks(delT)
_, matA, matB = ed.EditDistanceFFull(len(delT), len(insT), func(iA, iB int) int {
if delT[iA] == insT[iB] {
c := 0
if delL[iA] {
c += diffAt(delT, iA-1, insT, iB-1)
}
if delR[iA] {
c += diffAt(delT, iA+1, insT, iB+1)
}
return c
} // if
return len(delT[iA]) + len(insT[iB]) + 5
}, func(iA int) int {
if delT[iA] == " " {
return 0
} // if
return len(delT[iA]) + 2
}, func(iB int) int {
if insT[iB] == " " {
return 0
} // if
return len(insT[iB]) + 2
})
delP, insP := findPairs(delT), findPairs(insT)
alignPairs(matA, matB, delP, insP)
return matA, matB
}
func TestDiffLine(t *testing.T) {
delT := tm.LineToTokens("func g(src string, dst string) error {")
insT := tm.LineToTokens("func (m *monitor) g(gsp string) error {")
_, matA, matB := ed.EditDistanceFFull(len(delT), len(insT), func(iA, iB int) int {
if delT[iA] == insT[iB] {
return 0
} // if
return len(delT[iA]) + len(insT[iB]) + 1
}, func(iA int) int {
if delT[iA] == " " {
return 0
} // if
return len(delT[iA])
}, func(iB int) int {
if insT[iB] == " " {
return 0
} // if
return len(insT[iB])
})
ShowDelTokens(delT, matA, insT)
ShowInsTokens(insT, matB, delT)
}
func AssertStringsEqual(t *testing.T, name string, act, exp []string) {
if villa.StringSlice(exp).Equals(act) {
return
}
t.Errorf("%s unexpected(exp: %d lines, act %d lines)!", name, len(exp), len(act))
t.Logf("exp --- act +++")
t.Logf("Difference:")
_, matA, matB := ed.EditDistanceFFull(len(exp), len(act), func(iA, iB int) int {
sa, sb := exp[iA], act[iB]
if sa == sb {
return 0
}
return ed.String(sa, sb)
}, func(iA int) int {
return len(exp[iA]) + 1
}, func(iB int) int {
return len(act[iB]) + 1
})
for i, j := 0, 0; i < len(exp) || j < len(act); {
switch {
case j >= len(act) || i < len(exp) && matA[i] < 0:
t.Logf("--- %3d: %s", i+1, showText(exp[i]))
i++
case i >= len(exp) || j < len(act) && matB[j] < 0:
t.Logf("+++ %3d: %s", j+1, showText(act[j]))
j++
default:
if exp[i] != act[j] {
t.Logf("--- %3d: %s", i+1, showText(exp[i]))
t.Logf("+++ %3d: %s", j+1, showText(act[j]))
} // else
i++
j++
}
} // for i, j
}
func DiffLines(orgLines, newLines []string, format string) {
if len(orgLines)+len(newLines) == 0 {
return
} // if
_, matA, matB := ed.EditDistanceFFull(len(orgLines), len(newLines), func(iA, iB int) int {
sa, sb := strings.TrimSpace(orgLines[iA]), strings.TrimSpace(newLines[iB])
// When sa and sb has 1/3 in common, convertion const is equal to del+ins const
return tm.CalcDiffOfSourceLine(sa, sb, (len(sa)+len(sb))*120)
}, func(iA int) int {
return max(1, len(strings.TrimSpace(orgLines[iA]))*100)
}, func(iB int) int {
return max(1, len(strings.TrimSpace(newLines[iB]))*100)
})
var lo lineOutput
for i, j := 0, 0; i < len(orgLines) || j < len(newLines); {
switch {
case j >= len(newLines) || i < len(orgLines) && matA[i] < 0:
lo.outputDel(fmt.Sprintf(format, orgLines[i]))
i++
case i >= len(orgLines) || j < len(newLines) && matB[j] < 0:
lo.outputIns(fmt.Sprintf(format, newLines[j]))
j++
default:
if strings.TrimSpace(orgLines[i]) != strings.TrimSpace(newLines[j]) {
lo.outputChange(fmt.Sprintf(format, orgLines[i]), fmt.Sprintf(format, newLines[j]))
} else {
lo.outputSame(fmt.Sprintf(format, newLines[j]))
} // else
i++
j++
}
} // for i, j
lo.end()
}
func diffLinesTo(orgLines, newLines []string, format string, lo lineOutputer) int {
if len(orgLines)+len(newLines) == 0 {
return 0
}
start, orgEnd, newEnd := 0, len(orgLines), len(newLines)
if len(orgLines)*len(newLines) > 1024*1024 {
// Use trivial comparison to offset same head and tail lines.
start, orgEnd, newEnd = offsetHeadTails(orgLines, newLines)
}
fastMode := false
if len(orgLines)*len(newLines) > 1024*1024 {
fastMode = true
}
_, matA, matB := ed.EditDistanceFFull(orgEnd-start, newEnd-start, func(iA, iB int) int {
sa, sb := orgLines[iA+start], newLines[iB+start]
rawEqual := sa == sb
sa, sb = strings.TrimSpace(sa), strings.TrimSpace(sb)
posCost := 0
if isBlockStart(sa) {
posCost += (newEnd - start - 1 - iB) * 10 / (newEnd - start)
}
if rawEqual {
return posCost
}
if sa == sb {
return posCost + 1
}
mx := (len(sa) + len(sb)) * 150
var dist int
if fastMode && len(sa) > 10*len(sb) {
dist = 100 * (len(sa) - len(sb))
} else if fastMode && len(sb) > 10*len(sa) {
dist = 100 * (len(sb) - len(sa))
} else {
// When sa and sb has 1/3 in common, convertion const is equal to del+ins const
dist = tm.CalcDiffOfSourceLine(sa, sb, mx)
}
// Even a small change, both lines will be shown, so add a 20% penalty on that.
return (dist*4+mx)/5 + 1 + posCost
}, func(iA int) int {
return mathp.MaxI(1, len(strings.TrimSpace(orgLines[iA+start]))*100)
}, func(iB int) int {
return mathp.MaxI(1, len(strings.TrimSpace(newLines[iB+start]))*100)
})
cnt := 0
for i, j := 0, 0; i < len(orgLines) || j < len(newLines); {
switch {
case i < start || i >= orgEnd && j >= newEnd:
// cut by offsetHeadTails
lo.outputSame(fmt.Sprintf(format, newLines[j]))
i++
j++
case j >= newEnd || i < orgEnd && matA[i-start] < 0:
lo.outputDel(fmt.Sprintf(format, orgLines[i]))
cnt++
i++
case i >= orgEnd || j < newEnd && matB[j-start] < 0:
lo.outputIns(fmt.Sprintf(format, newLines[j]))
cnt++
j++
default:
if strings.TrimSpace(orgLines[i]) != strings.TrimSpace(newLines[j]) {
lo.outputChange(fmt.Sprintf(format, orgLines[i]), fmt.Sprintf(format, newLines[j]))
cnt += 2
} else {
lo.outputSame(fmt.Sprintf(format, newLines[j]))
} // else
i++
j++
}
}
lo.end()
return cnt
}