+40 -37

api.go

Reviewed

··· 34 34 35 35 // FileMatch contains all the matches within a file. 36 36 type FileMatch struct { 37 37 - // Ranking; the higher, the better. 38 38 - Score float64 // TODO - hide this field? 37 37 + FileName string 38 38 + 39 39 + // Repository is the globally unique name of the repo of the 40 40 + // match 41 41 + Repository string 42 42 + 43 43 + // SubRepositoryName is the globally unique name of the repo, 44 44 + // if it came from a subrepository 45 45 + SubRepositoryName string 46 46 + 47 47 + // SubRepositoryPath holds the prefix where the subrepository 48 48 + // was mounted. 49 49 + SubRepositoryPath string 50 50 + 51 51 + // Commit SHA1 (hex) of the (sub)repo holding the file. 52 52 + Version string 53 53 + 54 54 + // Detected language of the result. 55 55 + Language string 39 56 40 57 // For debugging. Needs DebugScore set, but public so tests in 41 58 // other packages can print some diagnostics. 42 59 Debug string 43 60 44 44 - FileName string 45 45 - 46 46 - // Repository is the globally unique name of the repo of the 47 47 - // match 48 48 - Repository string 49 49 - Branches []string 61 61 + Branches []string 50 62 51 63 // One of LineMatches or ChunkMatches will be returned depending on whether 52 64 // the SearchOptions.ChunkMatches is set. 53 65 LineMatches []LineMatch 54 66 ChunkMatches []ChunkMatch 55 67 56 56 - // RepositoryID is a Sourcegraph extension. This is the ID of Repository in 57 57 - // Sourcegraph. 58 58 - RepositoryID uint32 59 59 - 60 60 - // RepositoryPriority is a Sourcegraph extension. It is used by Sourcegraph to 61 61 - // order results from different repositories relative to each other. 62 62 - RepositoryPriority float64 63 63 - 64 68 // Only set if requested 65 69 Content []byte 66 70 67 71 // Checksum of the content. 68 72 Checksum []byte 69 73 70 70 - // Detected language of the result. 71 71 - Language string 74 74 + // Ranking; the higher, the better. 75 75 + Score float64 // TODO - hide this field? 72 76 73 73 - // SubRepositoryName is the globally unique name of the repo, 74 74 - // if it came from a subrepository 75 75 - SubRepositoryName string 76 76 - 77 77 - // SubRepositoryPath holds the prefix where the subrepository 78 78 - // was mounted. 79 79 - SubRepositoryPath string 77 77 + // RepositoryPriority is a Sourcegraph extension. It is used by Sourcegraph to 78 78 + // order results from different repositories relative to each other. 79 79 + RepositoryPriority float64 80 80 81 81 - // Commit SHA1 (hex) of the (sub)repo holding the file. 82 82 - Version string 81 81 + // RepositoryID is a Sourcegraph extension. This is the ID of Repository in 82 82 + // Sourcegraph. 83 83 + RepositoryID uint32 83 84 } 84 85 85 86 func (m *FileMatch) sizeBytes() (sz uint64) { ··· 134 135 // ChunkMatch is a set of non-overlapping matches within a contiguous range of 135 136 // lines in the file. 136 137 type ChunkMatch struct { 138 138 + DebugScore string 139 139 + 137 140 // Content is a contiguous range of complete lines that fully contains Ranges. 138 141 Content []byte 139 139 - // ContentStart is the location (inclusive) of the beginning of content 140 140 - // relative to the beginning of the file. It will always be at the 141 141 - // beginning of a line (Column will always be 1). 142 142 - ContentStart Location 143 143 - 144 144 - // FileName indicates whether this match is a match on the file name, in 145 145 - // which case Content will contain the file name. 146 146 - FileName bool 147 142 148 143 // Ranges is a set of matching ranges within this chunk. Each range is relative 149 144 // to the beginning of the file (not the beginning of Content). ··· 153 148 // its length will equal that of Ranges. Any of its elements may be nil. 154 149 SymbolInfo []*Symbol 155 150 156 156 - Score float64 157 157 - DebugScore string 151 151 + // FileName indicates whether this match is a match on the file name, in 152 152 + // which case Content will contain the file name. 153 153 + FileName bool 154 154 + 155 155 + // ContentStart is the location (inclusive) of the beginning of content 156 156 + // relative to the beginning of the file. It will always be at the 157 157 + // beginning of a line (Column will always be 1). 158 158 + ContentStart Location 159 159 + 160 160 + Score float64 158 161 } 159 162 160 163 func (cm *ChunkMatch) sizeBytes() (sz uint64) {

+29

api_test.go

Reviewed

··· 17 17 import ( 18 18 "bytes" 19 19 "encoding/gob" 20 20 + "reflect" 20 21 "strings" 21 22 "testing" 22 23 "time" ··· 136 137 t.Fatalf("want %d, got %d", wantBytes, cm.sizeBytes()) 137 138 } 138 139 } 140 140 + 141 141 + func TestMatchSize(t *testing.T) { 142 142 + cases := []struct { 143 143 + v any 144 144 + size int 145 145 + }{{ 146 146 + v: FileMatch{}, 147 147 + size: 256, 148 148 + }, { 149 149 + v: ChunkMatch{}, 150 150 + size: 112, 151 151 + }, { 152 152 + v: candidateMatch{}, 153 153 + size: 72, 154 154 + }, { 155 155 + v: candidateChunk{}, 156 156 + size: 40, 157 157 + }} 158 158 + for _, c := range cases { 159 159 + got := reflect.TypeOf(c.v).Size() 160 160 + if int(got) != c.size { 161 161 + t.Errorf(`sizeof struct %T has changed from %d to %d. 162 162 + These are match structs that occur a lot in memory, so we optimize size. 163 163 + When changing, please ensure there isn't unnecessary padding via the 164 164 + tool fieldalignment then update this test.`, c.v, c.size, got) 165 165 + } 166 166 + } 167 167 + }

+3 -3

contentprovider.go

Reviewed

··· 369 369 } 370 370 371 371 type candidateChunk struct { 372 372 + candidates []*candidateMatch 372 373 firstLine uint32 // 1-based, inclusive 373 374 lastLine uint32 // 1-based, inclusive 374 375 minOffset uint32 // 0-based, inclusive 375 376 maxOffset uint32 // 0-based, exclusive 376 376 - candidates []*candidateMatch 377 377 } 378 378 379 379 // chunkCandidates groups a set of sorted, non-overlapping candidate matches by line number. Adjacent ··· 566 566 567 567 func (p *contentProvider) chunkMatchScore(secs []DocumentSection, m *ChunkMatch, language string, debug bool) (float64, string) { 568 568 type debugScore struct { 569 569 - score float64 570 569 what string 570 570 + score float64 571 571 } 572 572 573 573 score := &debugScore{} ··· 654 654 655 655 func (p *contentProvider) matchScore(secs []DocumentSection, m *LineMatch, language string, debug bool) (float64, string) { 656 656 type debugScore struct { 657 657 - score float64 658 657 what string 658 658 + score float64 659 659 } 660 660 661 661 score := &debugScore{}

+2 -2

hititer.go

Reviewed

··· 35 35 36 36 // distanceHitIterator looks for hits at a fixed distance apart. 37 37 type distanceHitIterator struct { 38 38 - started bool 39 39 - distance uint32 40 38 i1 hitIterator 41 39 i2 hitIterator 40 40 + distance uint32 41 41 + started bool 42 42 } 43 43 44 44 func (i *distanceHitIterator) String() string {

+10 -6

matchiter.go

Reviewed

··· 20 20 ) 21 21 22 22 // candidateMatch is a candidate match for a substring. 23 23 + // 24 24 + // Note: a lot of these can be in memory, so think about fieldalignment when 25 25 + // modify the fields of this structure. 23 26 type candidateMatch struct { 24 24 - caseSensitive bool 25 25 - fileName bool 26 26 - symbol bool 27 27 - symbolIdx uint32 28 28 - 29 27 substrBytes []byte 30 28 substrLowered []byte 31 29 32 32 - file uint32 30 30 + file uint32 31 31 + symbolIdx uint32 33 32 34 33 // Offsets are relative to the start of the filename or file contents. 35 34 runeOffset uint32 36 35 byteOffset uint32 37 36 byteMatchSz uint32 37 37 + 38 38 + // bools at end for struct field alignment 39 39 + caseSensitive bool 40 40 + fileName bool 41 41 + symbol bool 38 42 } 39 43 40 44 // Matches content against the substring, and populates byteMatchSz on success