ranking/post-ranking.go

   1 // vim:ts=4:sw=4:noexpandtab
   2
   3 // Post-ranking happens on the source backend (because it has the source files
   4 // in the kernel’s page cache). In the post-ranking phase we can do (limited)
   5 // source file level analysis, such as in which scope the query string was
   6 // matched (comment, top-level, sub-level).
   7 package ranking
   8
   9 import (
  10         "github.com/Debian/dcs/regexp"
  11         "unicode"
  12 )
  13
  14 //var packageLocation *regexp.Regexp = regexp.MustCompile(`debian-source-mirror/unpacked/([^/]+)_`)
  15
  16 func countSpaces(line string) int32 {
  17         spaces := int32(0)
  18         for _, r := range line {
  19                 if !unicode.IsSpace(r) {
  20                         break
  21                 }
  22                 spaces += 1
  23         }
  24         return spaces
  25 }
  26
  27 func PostRank(opts RankingOpts, match *regexp.Match, querystr *QueryStr) float32 {
  28         totalRanking := float32(1)
  29
  30         line := match.Context
  31
  32         if opts.Scope || opts.Weighted {
  33                 // Ranking: In which scope is the match? The higher the scope, the more
  34                 // important it is.
  35                 scopeRanking := 1.0 - (float32(countSpaces(line)) / 100.0)
  36                 totalRanking *= scopeRanking
  37         }
  38
  39         if opts.Linematch || opts.Weighted {
  40                 // Ranking: Does the search query (with enforced word boundaries) match the
  41                 // line? If yes, earlier matches are better (such as function names versus
  42                 // parameter types).
  43                 index := querystr.boundaryRegexp.FindStringIndex(line)
  44                 if index != nil {
  45                         matchRanking := 0.75 + (0.25 * (1.0 - float32(index[0]) / float32(len(line))))
  46                         totalRanking *= matchRanking
  47                 } else {
  48                         // Punish the lines in which there was no word boundary match.
  49                         totalRanking *= 0.5
  50                 }
  51         }
  52
  53         return totalRanking
  54 }