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, timeout time.Duration) ([]Variant, bool) {
  34         dmp := diffmatchpatch.New()
  35         var deadline time.Time
  36         if timeout > 0 {
  37                 deadline = time.Now().Add(timeout)
  38         }
  39         diffs := dmp.DiffBisect(a, b, deadline)
  40         timedOut := false
  41         if timeout > 0 && time.Now().After(deadline) {
  42                 timedOut = true
  43         }
  44         diffs = cleanup(dmp.DiffCleanupEfficiency(diffs))
  45         pos := 1
  46         var variants []Variant
  47         for i := 0; i < len(diffs); i++ {
  48                 switch diffs[i].Type {
  49                 case diffmatchpatch.DiffEqual:
  50                         pos += len(diffs[i].Text)
  51                 case diffmatchpatch.DiffDelete:
  52                         if i+1 < len(diffs) && diffs[i+1].Type == diffmatchpatch.DiffInsert {
  53                                 // deletion followed by insertion
  54                                 variants = append(variants, Variant{Position: pos, Ref: diffs[i].Text, New: diffs[i+1].Text})
  55                                 pos += len(diffs[i].Text)
  56                                 i++
  57                         } else {
  58                                 variants = append(variants, Variant{Position: pos, Ref: diffs[i].Text})
  59                                 pos += len(diffs[i].Text)
  60                         }
  61                 case diffmatchpatch.DiffInsert:
  62                         if i+1 < len(diffs) && diffs[i+1].Type == diffmatchpatch.DiffDelete {
  63                                 // insertion followed by deletion
  64                                 variants = append(variants, Variant{Position: pos, Ref: diffs[i+1].Text, New: diffs[i].Text})
  65                                 pos += len(diffs[i+1].Text)
  66                                 i++
  67                         } else {
  68                                 variants = append(variants, Variant{Position: pos, New: diffs[i].Text})
  69                         }
  70                 }
  71         }
  72         return variants, timedOut
  73 }
  74
  75 func cleanup(in []diffmatchpatch.Diff) (out []diffmatchpatch.Diff) {
  76         for i := 0; i < len(in); i++ {
  77                 d := in[i]
  78                 for i < len(in)-1 && in[i].Type == in[i+1].Type {
  79                         d.Text += in[i+1].Text
  80                         i++
  81                 }
  82                 out = append(out, d)
  83         }
  84         return
  85 }