hgvs/diff.go

   1 package hgvs
   2
   3 import (
   4         "fmt"
   5         "time"
   6
   7         "github.com/sergi/go-diff/diffmatchpatch"
   8 )
   9
  10 type Variant struct {
  11         Position int
  12         Ref      string
  13         New      string
  14 }
  15
  16 func (v *Variant) String() string {
  17         switch {
  18         case len(v.New) == 0 && len(v.Ref) == 1:
  19                 return fmt.Sprintf("%ddel", v.Position)
  20         case len(v.New) == 0:
  21                 return fmt.Sprintf("%d_%ddel", v.Position, v.Position+len(v.Ref)-1)
  22         case len(v.Ref) == 1 && len(v.New) == 1:
  23                 return fmt.Sprintf("%d%s>%s", v.Position, v.Ref, v.New)
  24         case len(v.Ref) == 0:
  25                 return fmt.Sprintf("%d_%dins%s", v.Position-1, v.Position, v.New)
  26         case len(v.Ref) == 1 && len(v.New) > 0:
  27                 return fmt.Sprintf("%ddelins%s", v.Position, v.New)
  28         default:
  29                 return fmt.Sprintf("%d_%ddelins%s", v.Position, v.Position+len(v.Ref)-1, v.New)
  30         }
  31 }
  32
  33 func Diff(a, b string) []Variant {
  34         dmp := diffmatchpatch.New()
  35         diffs := cleanup(dmp.DiffCleanupEfficiency(dmp.DiffBisect(a, b, time.Time{})))
  36         pos := 1
  37         var variants []Variant
  38         for i := 0; i < len(diffs); i++ {
  39                 switch diffs[i].Type {
  40                 case diffmatchpatch.DiffEqual:
  41                         pos += len(diffs[i].Text)
  42                 case diffmatchpatch.DiffDelete:
  43                         if i+1 < len(diffs) && diffs[i+1].Type == diffmatchpatch.DiffInsert {
  44                                 // deletion followed by insertion
  45                                 variants = append(variants, Variant{Position: pos, Ref: diffs[i].Text, New: diffs[i+1].Text})
  46                                 pos += len(diffs[i].Text)
  47                                 i++
  48                         } else {
  49                                 variants = append(variants, Variant{Position: pos, Ref: diffs[i].Text})
  50                                 pos += len(diffs[i].Text)
  51                         }
  52                 case diffmatchpatch.DiffInsert:
  53                         if i+1 < len(diffs) && diffs[i+1].Type == diffmatchpatch.DiffDelete {
  54                                 // insertion followed by deletion
  55                                 variants = append(variants, Variant{Position: pos, Ref: diffs[i+1].Text, New: diffs[i].Text})
  56                                 pos += len(diffs[i+1].Text)
  57                                 i++
  58                         } else {
  59                                 variants = append(variants, Variant{Position: pos, New: diffs[i].Text})
  60                         }
  61                 }
  62         }
  63         return variants
  64 }
  65
  66 func cleanup(in []diffmatchpatch.Diff) (out []diffmatchpatch.Diff) {
  67         for i := 0; i < len(in); i++ {
  68                 d := in[i]
  69                 for i < len(in)-1 && in[i].Type == in[i+1].Type {
  70                         d.Text += in[i+1].Text
  71                         i++
  72                 }
  73                 out = append(out, d)
  74         }
  75         return
  76 }