hgvs/diff.go

   1 // Copyright (C) The Lightning Authors. All rights reserved.
   2 //
   3 // SPDX-License-Identifier: AGPL-3.0
   4
   5 package hgvs
   6
   7 import (
   8         "fmt"
   9         "strings"
  10         "time"
  11
  12         "github.com/sergi/go-diff/diffmatchpatch"
  13 )
  14
  15 type Variant struct {
  16         Position int
  17         Ref      string
  18         New      string
  19         Left     string // base preceding an indel, if Ref or New is empty
  20 }
  21
  22 func (v *Variant) String() string {
  23         switch {
  24         case len(v.New) == 0 && len(v.Ref) == 0:
  25                 return fmt.Sprintf("%d=", v.Position)
  26         case len(v.New) == 1 && v.New == v.Ref:
  27                 return fmt.Sprintf("%d=", v.Position)
  28         case v.New == v.Ref:
  29                 return fmt.Sprintf("%d_%d=", v.Position, v.Position+len(v.Ref)-1)
  30         case len(v.New) == 0 && len(v.Ref) == 1:
  31                 return fmt.Sprintf("%ddel", v.Position)
  32         case len(v.New) == 0:
  33                 return fmt.Sprintf("%d_%ddel", v.Position, v.Position+len(v.Ref)-1)
  34         case len(v.Ref) == 1 && len(v.New) == 1:
  35                 return fmt.Sprintf("%d%s>%s", v.Position, v.Ref, v.New)
  36         case len(v.Ref) == 0:
  37                 return fmt.Sprintf("%d_%dins%s", v.Position-1, v.Position, v.New)
  38         case len(v.Ref) == 1 && len(v.New) > 0:
  39                 return fmt.Sprintf("%ddelins%s", v.Position, v.New)
  40         default:
  41                 return fmt.Sprintf("%d_%ddelins%s", v.Position, v.Position+len(v.Ref)-1, v.New)
  42         }
  43 }
  44
  45 // PadLeft returns a Variant that is equivalent to v but (if possible)
  46 // uses the stashed preceding base (the Left field) to avoid having a
  47 // non-empty Ref or New part, even for an insertion or deletion.
  48 //
  49 // For example, if v is {Position: 45, Ref: "", New: "A"}, PadLeft
  50 // might return {Position: 44, Ref: "T", New: "TA"}.
  51 func (v *Variant) PadLeft() Variant {
  52         if len(v.Ref) == 0 || len(v.New) == 0 {
  53                 return Variant{
  54                         Position: v.Position - len(v.Left),
  55                         Ref:      v.Left + v.Ref,
  56                         New:      v.Left + v.New,
  57                 }
  58         } else {
  59                 return *v
  60         }
  61 }
  62
  63 func Diff(a, b string, timeout time.Duration) ([]Variant, bool) {
  64         dmp := diffmatchpatch.New()
  65         var deadline time.Time
  66         if timeout > 0 {
  67                 deadline = time.Now().Add(timeout)
  68         }
  69         diffs := dmp.DiffBisect(a, b, deadline)
  70         timedOut := false
  71         if timeout > 0 && time.Now().After(deadline) {
  72                 timedOut = true
  73         }
  74         diffs = cleanup(dmp.DiffCleanupEfficiency(diffs))
  75         pos := 1
  76         var variants []Variant
  77         for i := 0; i < len(diffs); {
  78                 left := "" // last char before an insertion or deletion
  79                 for ; i < len(diffs) && diffs[i].Type == diffmatchpatch.DiffEqual; i++ {
  80                         pos += len(diffs[i].Text)
  81                         if tlen := len(diffs[i].Text); tlen > 0 {
  82                                 left = diffs[i].Text[tlen-1:]
  83                         }
  84                 }
  85                 if i >= len(diffs) {
  86                         break
  87                 }
  88                 v := Variant{Position: pos, Left: left}
  89                 for ; i < len(diffs) && diffs[i].Type != diffmatchpatch.DiffEqual; i++ {
  90                         if diffs[i].Type == diffmatchpatch.DiffDelete {
  91                                 v.Ref += diffs[i].Text
  92                         } else {
  93                                 v.New += diffs[i].Text
  94                         }
  95                 }
  96                 if len(v.Ref) == 2 && len(v.New) == 2 {
  97                         v1 := v
  98                         v1.Ref = v1.Ref[:1]
  99                         v1.New = v1.New[:1]
 100                         v.Ref = v.Ref[1:]
 101                         v.New = v.New[1:]
 102                         v.Position++
 103                         v.Left = v1.Ref
 104                         pos++
 105                         variants = append(variants, v1)
 106                 }
 107                 pos += len(v.Ref)
 108                 variants = append(variants, v)
 109                 left = ""
 110         }
 111         return variants, timedOut
 112 }
 113
 114 func cleanup(in []diffmatchpatch.Diff) (out []diffmatchpatch.Diff) {
 115         out = make([]diffmatchpatch.Diff, 0, len(in))
 116         for i := 0; i < len(in); i++ {
 117                 d := in[i]
 118                 // Merge consecutive entries of same type (e.g.,
 119                 // "insert A; insert B")
 120                 for i < len(in)-1 && in[i].Type == in[i+1].Type {
 121                         d.Text += in[i+1].Text
 122                         i++
 123                 }
 124                 out = append(out, d)
 125         }
 126         in, out = out, make([]diffmatchpatch.Diff, 0, len(in))
 127         for i := 0; i < len(in); i++ {
 128                 d := in[i]
 129                 // diffmatchpatch solves diff("AAX","XTX") with
 130                 // [delAA,=X,insTX] but we prefer to spell it
 131                 // [delAA,insXT,=X].
 132                 //
 133                 // So, when we see a [del,=,ins] sequence where the
 134                 // "=" part is a suffix of the "ins" part -- e.g.,
 135                 // [delAAA,=CGG,insTTTCGG] -- we rearrange it to the
 136                 // equivalent spelling [delAAA,insCGGTTT,=CGG].
 137                 if i < len(in)-2 &&
 138                         d.Type == diffmatchpatch.DiffDelete &&
 139                         in[i+1].Type == diffmatchpatch.DiffEqual &&
 140                         in[i+2].Type == diffmatchpatch.DiffInsert &&
 141                         strings.HasSuffix(in[i+2].Text, in[i+1].Text) {
 142                         eq, ins := in[i+1], in[i+2]
 143                         ins.Text = eq.Text + ins.Text[:len(ins.Text)-len(eq.Text)]
 144                         in[i+1] = ins
 145                         in[i+2] = eq
 146                 }
 147                 // when diffmatchpatch says [=yyyyXXXX, delX, =zzz],
 148                 // we really want [=yyyy, delX, =XXXXzzz] (ditto for
 149                 // ins instead of del)
 150                 if i < len(in)-2 &&
 151                         d.Type == diffmatchpatch.DiffEqual &&
 152                         in[i+1].Type != diffmatchpatch.DiffEqual &&
 153                         in[i+2].Type == diffmatchpatch.DiffEqual &&
 154                         len(in[i+1].Text) <= len(d.Text) {
 155                         for cut := 0; cut < len(d.Text); cut++ {
 156                                 skip := strings.Index(d.Text[cut:], in[i+1].Text)
 157                                 if skip < 0 {
 158                                         break
 159                                 }
 160                                 cut += skip
 161                                 if d.Text[cut:]+in[i+1].Text == in[i+1].Text+d.Text[cut:] {
 162                                         in[i+2].Text = d.Text[cut:] + in[i+2].Text
 163                                         d.Text = d.Text[:cut]
 164                                         break
 165                                 }
 166                         }
 167                 }
 168                 // diffmatchpatch solves diff("AXX","XXX") with
 169                 // [delA,=XX,insX] but we prefer to spell it
 170                 // [delA,insX,=XX].
 171                 //
 172                 // So, when we see a [del,=,ins] sequence that has the
 173                 // same effect after swapping the "=" and "ins" parts,
 174                 // we swap them.
 175                 if i < len(in)-2 &&
 176                         d.Type == diffmatchpatch.DiffDelete &&
 177                         in[i+1].Type == diffmatchpatch.DiffEqual &&
 178                         in[i+2].Type == diffmatchpatch.DiffInsert &&
 179                         in[i+1].Text+in[i+2].Text == in[i+2].Text+in[i+1].Text {
 180                         in[i+2], in[i+1] = in[i+1], in[i+2]
 181                 }
 182                 // when diffmatchpatch says [delAAA, insXAY] and
 183                 // len(X)==1, we prefer to treat the A>X as a snp.
 184                 if i < len(in)-1 &&
 185                         d.Type == diffmatchpatch.DiffDelete &&
 186                         in[i+1].Type == diffmatchpatch.DiffInsert &&
 187                         len(d.Text) >= 2 &&
 188                         len(in[i+1].Text) >= 2 &&
 189                         d.Text[1] == in[i+1].Text[1] {
 190                         eqend := 2
 191                         for ; eqend < len(d.Text) && eqend < len(in[i+1].Text) && d.Text[eqend] == in[i+1].Text[eqend]; eqend++ {
 192                         }
 193                         out = append(out,
 194                                 diffmatchpatch.Diff{diffmatchpatch.DiffDelete, d.Text[:1]},
 195                                 diffmatchpatch.Diff{diffmatchpatch.DiffInsert, in[i+1].Text[:1]},
 196                                 diffmatchpatch.Diff{diffmatchpatch.DiffEqual, d.Text[1:eqend]})
 197                         in[i].Text, in[i+1].Text = in[i].Text[eqend:], in[i+1].Text[eqend:]
 198                         i--
 199                         continue
 200                 }
 201                 // when diffmatchpatch says [delAAA, insXaY] and
 202                 // len(Y)==1, we prefer to treat the A>Y as a snp.
 203                 if i < len(in)-1 &&
 204                         d.Type == diffmatchpatch.DiffDelete &&
 205                         in[i+1].Type == diffmatchpatch.DiffInsert &&
 206                         len(d.Text) >= 2 &&
 207                         len(in[i+1].Text) >= 2 &&
 208                         d.Text[len(d.Text)-2] == in[i+1].Text[len(in[i+1].Text)-2] {
 209                         // eqstart will be the number of equal chars
 210                         // before the terminal snp, plus 1 for the snp
 211                         // itself. Example, for [delAAAA, insTTAAG],
 212                         // eqstart will be 3.
 213                         eqstart := 2
 214                         for ; eqstart < len(d.Text) && eqstart < len(in[i+1].Text) && d.Text[len(d.Text)-eqstart] == in[i+1].Text[len(in[i+1].Text)-eqstart]; eqstart++ {
 215                         }
 216                         eqstart--
 217                         out = append(out,
 218                                 diffmatchpatch.Diff{diffmatchpatch.DiffDelete, d.Text[:len(d.Text)-eqstart]},
 219                                 diffmatchpatch.Diff{diffmatchpatch.DiffInsert, in[i+1].Text[:len(in[i+1].Text)-eqstart]},
 220                                 diffmatchpatch.Diff{diffmatchpatch.DiffEqual, d.Text[len(d.Text)-eqstart : len(d.Text)-1]},
 221                                 diffmatchpatch.Diff{diffmatchpatch.DiffDelete, d.Text[len(d.Text)-1:]},
 222                                 diffmatchpatch.Diff{diffmatchpatch.DiffInsert, in[i+1].Text[len(in[i+1].Text)-1:]})
 223                         i++
 224                         continue
 225                 }
 226                 out = append(out, d)
 227         }
 228         // for i := 0; i < len(out)-1; i++ {
 229         //      if out[i].Type == diffmatchpatch.DiffDelete && len(out[i].Text) == 2 &&
 230         //              out[i+1].Type == diffmatchpatch.DiffInsert && len(out[i+1].Text) == 2 {
 231         //              out = append(out, diffmatchpatch.Diff{}, diffmatchpatch.Diff{})
 232         //              copy(out[i+4:], out[i+2:])
 233         //              out[i+2] = diffmatchpatch.Diff{diffmatchpatch.DiffDelete, out[i].Text[1:]}
 234         //              out[i+3] = diffmatchpatch.Diff{diffmatchpatch.DiffInsert, out[i+1].Text[1:]}
 235         //              out[i].Text = out[i].Text[:1]
 236         //              out[i+1].Text = out[i+1].Text[:1]
 237         //      }
 238         // }
 239         return
 240 }
 241
 242 func Less(a, b Variant) bool {
 243         if a.Position != b.Position {
 244                 return a.Position < b.Position
 245         } else if a.New != b.New {
 246                 return a.New < b.New
 247         } else {
 248                 return a.Ref < b.Ref
 249         }
 250 }