wip own search

indexdata.go

@@ -22,6 +22,7 @@ import (
 	"math/bits"
 	"unicode/utf8"
 
+	"github.com/sourcegraph/zoekt/own"
 	"github.com/sourcegraph/zoekt/query"
 )
 
@@ -100,6 +101,9 @@ type indexData struct {
 
 	// rawConfigMasks contains the encoded RawConfig for each repository
 	rawConfigMasks []uint8
+
+	// own per repo
+	own []own.Own
 }
 
 type symbolData struct {

matchtree.go

@@ -1022,6 +1022,21 @@ func (d *indexData) newMatchTree(q query.Q, opt matchTreeOpt) (matchTree, error)
 			},
 		}, nil
 
+	case *query.Own:
+		var compiled []func([]byte) bool
+		for _, o := range d.own {
+			compiled = append(compiled, o.CompileForAuthor(s.SearchTerm))
+		}
+
+		return &docMatchTree{
+			reason:  "Own",
+			numDocs: d.numDocs(),
+			predicate: func(docID uint32) bool {
+				fileName := d.fileName(docID)
+				return compiled[d.repos[docID]](fileName)
+			},
+		}, nil
+
 	case *query.BranchesRepos:
 		reposBranchesWant := make([]uint64, len(d.repoMetaData))
 		for repoIdx := range d.repoMetaData {

own/own.go

@@ -0,0 +1,42 @@
+package own
+
+import (
+	"bytes"
+	"fmt"
+	"os"
+)
+
+func Load(ownPath string) (Own, error) {
+	b, err := os.ReadFile(ownPath)
+	if err != nil {
+		return nil, fmt.Errorf("failed to read owner file: %w", err)
+	}
+
+	owners := make(own)
+	fields := bytes.Fields(b)
+	for i := 0; i < len(fields); i += 2 {
+		owners[string(fields[i])] = fields[i+1]
+	}
+	return owners, nil
+}
+
+var Empty = make(own)
+
+type Own interface {
+	// This is the way to go until we are proven it is too slow.
+	CompileForAuthor(author string) func(path []byte) bool
+}
+
+type own map[string][]byte
+
+func (o own) CompileForAuthor(author string) func(path []byte) bool {
+	pattern, ok := o[author]
+	if ok {
+		return func(path []byte) bool {
+			return bytes.Contains(path, pattern)
+		}
+	}
+	return func(path []byte) bool {
+		return false
+	}
+}

query/parse.go

@@ -191,6 +191,11 @@ func parseExpr(in []byte) (Q, int, error) {
 		}
 
 		expr = &Symbol{q}
+
+	case tokOwn:
+		// TODO support any and none
+		expr = &Own{SearchTerm: text}
+
 	case tokParenClose:
 		// Caller must consume paren.
 		expr = nil
@@ -393,6 +398,7 @@ const (
 	tokArchived   = 15
 	tokPublic     = 16
 	tokFork       = 17
+	tokOwn        = 18
 )
 
 var tokNames = map[int]string{
@@ -404,6 +410,7 @@ var tokNames = map[int]string{
 	tokFork:       "Fork",
 	tokNegate:     "Negate",
 	tokOr:         "Or",
+	tokOwn:        "Own",
 	tokParenClose: "ParenClose",
 	tokParenOpen:  "ParenOpen",
 	tokPublic:     "Public",
@@ -425,6 +432,7 @@ var prefixes = map[string]int{
 	"f:":        tokFile,
 	"file:":     tokFile,
 	"fork:":     tokFork,
+	"own:":      tokOwn,
 	"public:":   tokPublic,
 	"r:":        tokRepo,
 	"regex:":    tokRegex,

query/parse_test.go

@@ -114,6 +114,9 @@ func TestParseQuery(t *testing.T) {
 		{"type:file abc def", &Type{Type: TypeFileName, Child: NewAnd(&Substring{Pattern: "abc"}, &Substring{Pattern: "def"})}},
 		{"(type:repo abc) def", NewAnd(&Type{Type: TypeRepo, Child: &Substring{Pattern: "abc"}}, &Substring{Pattern: "def"})},
 
+		// own
+		{"foo own:cezary", NewAnd(&Substring{Pattern: "foo"}, &Own{SearchTerm: "cezary"})},
+
 		// errors.
 		{"--", nil},
 		{"\"abc", nil},

query/query.go

@@ -498,6 +498,26 @@ func (q *GobCache) String() string {
 	return fmt.Sprintf("GobCache(%s)", q.Q)
 }
 
+// Own searches ownership. The fields are mutually exclusive.
+type Own struct {
+	SearchTerm string
+	NoOwner    bool
+	AnyOwner   bool
+}
+
+func (q *Own) String() string {
+	if q.SearchTerm != "" {
+		return fmt.Sprintf("(own %q)", q.SearchTerm)
+	}
+	if q.NoOwner {
+		return "(own none)"
+	}
+	if q.AnyOwner {
+		return "(own any)"
+	}
+	return "(own malformed)"
+}
+
 // Or is matched when any of its children is matched.
 type Or struct {
 	Children []Q

read.go

@@ -20,10 +20,13 @@ import (
 	"fmt"
 	"hash/crc64"
 	"log"
+	"net/url"
 	"os"
+	"path/filepath"
 	"sort"
 
 	"github.com/rs/xid"
+	"github.com/sourcegraph/zoekt/own"
 )
 
 // IndexFile is a file suitable for concurrent read access. For performance
@@ -397,6 +400,18 @@ func (r *reader) readIndexData(toc *indexTOC) (*indexData, error) {
 		}
 	}
 
+	{
+		dir := filepath.Dir(r.r.Name())
+		for _, md := range d.repoMetaData {
+			o, err := own.Load(filepath.Join(dir, url.QueryEscape(md.Name)) + ".own")
+			if err != nil {
+				log.Printf("ignoring error for loading own for %s: %s", md.Name, err)
+				o = own.Empty
+			}
+			d.own = append(d.own, o)
+		}
+	}
+
 	if d.metaData.IndexFormatVersion >= 17 {
 		blob, err := d.readSectionBlob(toc.repos)
 		if err != nil {