func addReverseEdges(ctx context.Context, gs graphstore.Service) error {
log.Println("Adding reverse edges")
var (
totalEntries int
addedEdges int
)
startTime := time.Now()
err := gs.Scan(ctx, new(spb.ScanRequest), func(entry *spb.Entry) error {
kind := entry.EdgeKind
if kind != "" && schema.EdgeDirection(kind) == schema.Forward {
if err := gs.Write(ctx, &spb.WriteRequest{
Source: entry.Target,
Update: []*spb.WriteRequest_Update{{
Target: entry.Source,
EdgeKind: schema.MirrorEdge(kind),
FactName: entry.FactName,
FactValue: entry.FactValue,
}},
}); err != nil {
return fmt.Errorf("Failed to write reverse edge: %v", err)
}
addedEdges++
}
totalEntries++
return nil
})
log.Printf("Wrote %d reverse edges to GraphStore (%d total entries): %v", addedEdges, totalEntries, time.Since(startTime))
return err
}
// EnsureReverseEdges checks if gs contains reverse edges. If it doesn't, it
// will scan gs for all forward edges, adding a reverse for each back into the
// GraphStore. This is necessary for a GraphStoreService to work properly.
func EnsureReverseEdges(ctx context.Context, gs graphstore.Service) error {
var edge *spb.Entry
if err := gs.Scan(ctx, &spb.ScanRequest{}, func(e *spb.Entry) error {
if graphstore.IsEdge(e) {
edge = e
return io.EOF
}
return nil
}); err != nil {
return err
}
if edge == nil {
log.Println("No edges found in GraphStore")
return nil
} else if schema.EdgeDirection(edge.EdgeKind) == schema.Reverse {
return nil
}
var foundReverse bool
if err := gs.Read(ctx, &spb.ReadRequest{
Source: edge.Target,
EdgeKind: schema.MirrorEdge(edge.EdgeKind),
}, func(entry *spb.Entry) error {
foundReverse = true
return nil
}); err != nil {
return fmt.Errorf("error checking for reverse edge: %v", err)
}
if foundReverse {
return nil
}
return addReverseEdges(ctx, gs)
}
func writeEntries(ctx context.Context, s graphstore.Service, reqs <-chan *spb.WriteRequest) (uint64, error) {
var num uint64
for req := range reqs {
num += uint64(len(req.Update))
if err := s.Write(ctx, req); err != nil {
return 0, err
}
}
return num, nil
}
func readEntries(ctx context.Context, gs graphstore.Service, entryFunc graphstore.EntryFunc, edgeKind string, tickets []string) error {
for _, ticket := range tickets {
src, err := kytheuri.ToVName(ticket)
if err != nil {
return fmt.Errorf("error parsing ticket %q: %v", ticket, err)
}
if err := gs.Read(ctx, &spb.ReadRequest{
Source: src,
EdgeKind: edgeKind,
}, entryFunc); err != nil {
return fmt.Errorf("GraphStore Read error for ticket %q: %v", ticket, err)
}
}
return nil
}
// getEdges returns edgeTargets with the given node as their source. Only edge
// entries that return true when applied to pred are returned.
func getEdges(ctx context.Context, gs graphstore.Service, node *spb.VName, pred func(*spb.Entry) bool) ([]*edgeTarget, error) {
var targets []*edgeTarget
if err := gs.Read(ctx, &spb.ReadRequest{
Source: node,
EdgeKind: "*",
}, func(entry *spb.Entry) error {
if graphstore.IsEdge(entry) && pred(entry) {
targets = append(targets, &edgeTarget{entry.EdgeKind, entry.Target})
}
return nil
}); err != nil {
return nil, fmt.Errorf("read error: %v", err)
}
return targets, nil
}
// Populate adds each file node in gs to m.
func (m *Map) Populate(ctx context.Context, gs graphstore.Service) error {
start := time.Now()
log.Println("Populating in-memory file tree")
var total int
if err := gs.Scan(ctx, &spb.ScanRequest{FactPrefix: schema.NodeKindFact},
func(entry *spb.Entry) error {
if entry.FactName == schema.NodeKindFact && string(entry.FactValue) == schema.FileKind {
m.AddFile(entry.Source)
total++
}
return nil
}); err != nil {
return fmt.Errorf("failed to Scan GraphStore for directory structure: %v", err)
}
log.Printf("Indexed %d files in %s", total, time.Since(start))
return nil
}
func scanEntries(ctx context.Context, gs graphstore.Service, entryFunc graphstore.EntryFunc, edgeKind, targetTicket, factPrefix string) error {
var target *spb.VName
var err error
if targetTicket != "" {
target, err = kytheuri.ToVName(targetTicket)
if err != nil {
return fmt.Errorf("error parsing --target %q: %v", targetTicket, err)
}
}
if err := gs.Scan(ctx, &spb.ScanRequest{
EdgeKind: edgeKind,
FactPrefix: factPrefix,
Target: target,
}, entryFunc); err != nil {
return fmt.Errorf("GraphStore Scan error: %v", err)
}
return nil
}
func writeWithReverses(ctx context.Context, gs graphstore.Service, req *spb.WriteRequest) error {
if err := gs.Write(ctx, req); err != nil {
return fmt.Errorf("error writing edges: %v", err)
}
for _, u := range req.Update {
if err := gs.Write(ctx, &spb.WriteRequest{
Source: u.Target,
Update: []*spb.WriteRequest_Update{{
Target: req.Source,
EdgeKind: schema.MirrorEdge(u.EdgeKind),
FactName: u.FactName,
FactValue: u.FactValue,
}},
}); err != nil {
return fmt.Errorf("error writing rev edge: %v", err)
}
}
return nil
}
func getSourceText(ctx context.Context, gs graphstore.Service, fileVName *spb.VName) (text []byte, encoding string, err error) {
if err := gs.Read(ctx, &spb.ReadRequest{Source: fileVName}, func(entry *spb.Entry) error {
switch entry.FactName {
case schema.TextFact:
text = entry.FactValue
case schema.TextEncodingFact:
encoding = string(entry.FactValue)
default:
// skip other file facts
}
return nil
}); err != nil {
return nil, "", fmt.Errorf("read error: %v", err)
}
if text == nil {
err = fmt.Errorf("file not found: %+v", fileVName)
}
return
}
// Run writes the xrefs and filetree serving tables to db based on the given
// graphstore.Service.
func Run(ctx context.Context, gs graphstore.Service, db keyvalue.DB) error {
log.Println("Starting serving pipeline")
tbl := &table.KVProto{db}
// TODO(schroederc): for large corpora, this won't fit in memory
var files []string
entries := make(chan *spb.Entry)
ftIn, nIn, eIn := make(chan *spb.VName), make(chan *spb.Entry), make(chan *spb.Entry)
go func() {
for entry := range entries {
if entry.EdgeKind == "" {
nIn <- entry
if entry.FactName == schema.NodeKindFact && string(entry.FactValue) == "file" {
ftIn <- entry.Source
files = append(files, kytheuri.ToString(entry.Source))
}
} else {
eIn <- entry
}
}
close(ftIn)
close(nIn)
close(eIn)
}()
log.Println("Scanning GraphStore")
var sErr error
go func() {
sErr = gs.Scan(ctx, &spb.ScanRequest{}, func(e *spb.Entry) error {
entries <- e
return nil
})
close(entries)
}()
var (
ftErr, nErr, eErr error
ftWG, edgeNodeWG sync.WaitGroup
)
ftWG.Add(1)
go func() {
defer ftWG.Done()
ftErr = writeFileTree(ctx, tbl, ftIn)
log.Println("Wrote FileTree")
}()
edgeNodeWG.Add(2)
nodes := make(chan *srvpb.Node)
go func() {
defer edgeNodeWG.Done()
nErr = writeNodes(tbl, nIn, nodes)
log.Println("Wrote Nodes")
}()
go func() {
defer edgeNodeWG.Done()
eErr = writeEdges(ctx, tbl, eIn)
log.Println("Wrote Edges")
}()
var (
idxWG sync.WaitGroup
idxErr error
)
idxWG.Add(1)
go func() {
defer idxWG.Done()
idxErr = writeIndex(&table.KVInverted{db}, nodes)
log.Println("Wrote Search Index")
}()
edgeNodeWG.Wait()
if eErr != nil {
return eErr
} else if nErr != nil {
return nErr
}
es := xrefs.NodesEdgesService(&xsrv.Table{tbl})
if err := writeDecorations(ctx, tbl, es, files); err != nil {
return err
}
ftWG.Wait()
if ftErr != nil {
return ftErr
}
idxWG.Wait()
if idxErr != nil {
return idxErr
}
return sErr
}
func writeEdgePages(ctx context.Context, t table.Proto, gs graphstore.Service) error {
// TODO(schroederc): spill large PagedEdgeSets into EdgePages
log.Println("Writing EdgeSets")
var (
lastSrc *spb.VName
pes *srvpb.PagedEdgeSet
grp *srvpb.EdgeSet_Group
pesTotal int
)
if err := gs.Scan(ctx, new(spb.ScanRequest), func(e *spb.Entry) error {
if e.EdgeKind == "" {
panic("non-edge entry")
}
if pes != nil && !compare.VNamesEqual(lastSrc, e.Source) {
if grp != nil {
pes.EdgeSet.Group = append(pes.EdgeSet.Group, grp)
pesTotal += len(grp.TargetTicket)
}
pes.TotalEdges = int32(pesTotal)
if err := t.Put(xsrv.EdgeSetKey(pes.EdgeSet.SourceTicket), pes); err != nil {
return err
}
pes = nil
grp = nil
pesTotal = 0
}
if pes == nil {
pes = &srvpb.PagedEdgeSet{
EdgeSet: &srvpb.EdgeSet{
SourceTicket: kytheuri.ToString(e.Source),
},
}
}
if grp != nil && grp.Kind != e.EdgeKind {
pes.EdgeSet.Group = append(pes.EdgeSet.Group, grp)
pesTotal += len(grp.TargetTicket)
grp = nil
}
if grp == nil {
grp = &srvpb.EdgeSet_Group{
Kind: e.EdgeKind,
}
}
grp.TargetTicket = append(grp.TargetTicket, kytheuri.ToString(e.Target))
lastSrc = e.Source
return nil
}); err != nil {
return err
}
if pes != nil {
if grp != nil {
pes.EdgeSet.Group = append(pes.EdgeSet.Group, grp)
pesTotal += len(grp.TargetTicket)
}
pes.TotalEdges = int32(pesTotal)
if err := t.Put(xsrv.EdgeSetKey(pes.EdgeSet.SourceTicket), pes); err != nil {
return err
}
}
return nil
}
// Run writes the xrefs and filetree serving tables to db based on the given
// graphstore.Service.
func Run(ctx context.Context, gs graphstore.Service, db keyvalue.DB, opts *Options) error {
if opts == nil {
opts = new(Options)
}
log.Println("Starting serving pipeline")
out := &servingOutput{
xs: table.ProtoBatchParallel{&table.KVProto{DB: db}},
idx: &table.KVInverted{DB: db},
}
entries := make(chan *spb.Entry, chBuf)
var cErr error
var wg sync.WaitGroup
var sortedEdges disksort.Interface
wg.Add(1)
go func() {
sortedEdges, cErr = combineNodesAndEdges(ctx, opts, out, entries)
if cErr != nil {
cErr = fmt.Errorf("error combining nodes and edges: %v", cErr)
}
wg.Done()
}()
err := gs.Scan(ctx, &spb.ScanRequest{}, func(e *spb.Entry) error {
if graphstore.IsNodeFact(e) || schema.EdgeDirection(e.EdgeKind) == schema.Forward {
entries <- e
}
return nil
})
close(entries)
if err != nil {
return fmt.Errorf("error scanning GraphStore: %v", err)
}
wg.Wait()
if cErr != nil {
return cErr
}
pesIn, dIn := make(chan *srvpb.Edge, chBuf), make(chan *srvpb.Edge, chBuf)
var pErr, fErr error
wg.Add(2)
go func() {
defer wg.Done()
if err := writePagedEdges(ctx, pesIn, out.xs, opts); err != nil {
pErr = fmt.Errorf("error writing paged edge sets: %v", err)
}
}()
go func() {
defer wg.Done()
if err := writeDecorAndRefs(ctx, opts, dIn, out); err != nil {
fErr = fmt.Errorf("error writing file decorations: %v", err)
}
}()
err = sortedEdges.Read(func(x interface{}) error {
e := x.(*srvpb.Edge)
pesIn <- e
dIn <- e
return nil
})
close(pesIn)
close(dIn)
if err != nil {
return fmt.Errorf("error reading edges table: %v", err)
}
wg.Wait()
if pErr != nil {
return pErr
}
return fErr
}
// Run writes the xrefs and filetree serving tables to db based on the given
// graphstore.Service.
func Run(ctx context.Context, gs graphstore.Service, db keyvalue.DB, opts *Options) error {
if opts == nil {
opts = new(Options)
}
log.Println("Starting serving pipeline")
tbl := table.ProtoBatchParallel{&table.KVProto{db}}
entries := make(chan *spb.Entry, chBuf)
ftIn := make(chan *spb.VName, chBuf)
nIn, eIn, dIn := make(chan *spb.Entry, chBuf), make(chan *spb.Entry, chBuf), make(chan *spb.Entry, chBuf)
go func() {
for entry := range entries {
if graphstore.IsNodeFact(entry) {
nIn <- entry
if entry.FactName == schema.NodeKindFact && string(entry.FactValue) == "file" {
ftIn <- entry.Source
}
} else {
eIn <- entry
}
dIn <- entry
}
close(ftIn)
close(nIn)
close(eIn)
close(dIn)
}()
log.Println("Scanning GraphStore")
var sErr error
go func() {
sErr = gs.Scan(ctx, &spb.ScanRequest{}, func(e *spb.Entry) error {
entries <- e
return nil
})
if sErr != nil {
sErr = fmt.Errorf("error scanning GraphStore: %v", sErr)
}
close(entries)
}()
var (
ftErr, nErr, eErr, dErr error
ftWG, xrefsWG sync.WaitGroup
)
ftWG.Add(1)
go func() {
defer ftWG.Done()
ftErr = writeFileTree(ctx, tbl, ftIn)
if ftErr != nil {
ftErr = fmt.Errorf("error writing FileTree: %v", ftErr)
} else {
log.Println("Wrote FileTree")
}
}()
xrefsWG.Add(3)
nodes := make(chan *srvpb.Node)
go func() {
defer xrefsWG.Done()
nErr = writeNodes(ctx, tbl, nIn, nodes)
if nErr != nil {
nErr = fmt.Errorf("error writing Nodes: %v", nErr)
} else {
log.Println("Wrote Nodes")
}
}()
go func() {
defer xrefsWG.Done()
eErr = writeEdges(ctx, tbl, eIn, opts.MaxEdgePageSize)
if eErr != nil {
eErr = fmt.Errorf("error writing Edges: %v", eErr)
} else {
log.Println("Wrote Edges")
}
}()
go func() {
defer xrefsWG.Done()
dErr = writeDecorations(ctx, tbl, dIn)
if dErr != nil {
dErr = fmt.Errorf("error writing FileDecorations: %v", dErr)
} else {
log.Println("Wrote Decorations")
}
}()
var (
idxWG sync.WaitGroup
idxErr error
)
idxWG.Add(1)
go func() {
defer idxWG.Done()
idxErr = writeIndex(ctx, &table.KVInverted{db}, nodes)
if idxErr != nil {
idxErr = fmt.Errorf("error writing Search Index: %v", idxErr)
} else {
log.Println("Wrote Search Index")
}
}()
xrefsWG.Wait()
if eErr != nil {
return eErr
} else if nErr != nil {
return nErr
} else if dErr != nil {
return dErr
}
ftWG.Wait()
if ftErr != nil {
return ftErr
}
idxWG.Wait()
if idxErr != nil {
return idxErr
}
return sErr
}
// Run writes the xrefs and filetree serving tables to db based on the given
// graphstore.Service.
func Run(ctx context.Context, gs graphstore.Service, db keyvalue.DB, opts *Options) error {
if opts == nil {
opts = new(Options)
}
log.Println("Starting serving pipeline")
edges, err := tempTable("complete.edges")
if err != nil {
return err
}
defer func() {
if err := edges.Close(); err != nil {
log.Printf("Error closing edges table: %v", err)
}
}()
out := &servingOutput{
xs: table.ProtoBatchParallel{&table.KVProto{db}},
idx: &table.KVInverted{db},
completeEdges: &table.KVProto{edges},
}
entries := make(chan *spb.Entry, chBuf)
var cErr error
var wg sync.WaitGroup
wg.Add(1)
go func() {
cErr = combineNodesAndEdges(ctx, out, entries)
if cErr != nil {
cErr = fmt.Errorf("error combining nodes and edges: %v", cErr)
}
wg.Done()
}()
err = gs.Scan(ctx, &spb.ScanRequest{}, func(e *spb.Entry) error {
entries <- e
return nil
})
close(entries)
if err != nil {
return fmt.Errorf("error scanning GraphStore: %v", err)
}
wg.Wait()
if cErr != nil {
return cErr
}
pesIn, dIn := make(chan *srvpb.Edge, chBuf), make(chan *srvpb.Edge, chBuf)
var pErr, fErr error
wg.Add(2)
go func() {
defer wg.Done()
if err := writePagedEdges(ctx, pesIn, out.xs, opts.MaxEdgePageSize); err != nil {
pErr = fmt.Errorf("error writing paged edge sets: %v", err)
}
}()
go func() {
defer wg.Done()
if err := writeFileDecorations(ctx, dIn, out); err != nil {
fErr = fmt.Errorf("error writing file decorations: %v", err)
}
}()
if err := readCompletedEdges(ctx, out.completeEdges, pesIn, dIn); err != nil {
return fmt.Errorf("error reading edges table: %v", err)
}
wg.Wait()
if pErr != nil {
return pErr
}
return fErr
}
// LogClose closes gs and logs any resulting error.
func LogClose(ctx context.Context, gs graphstore.Service) {
if err := gs.Close(ctx); err != nil {
log.Printf("GraphStore failed to close: %v", err)
}
}