Phase 4: serialize outer pipeline; remove channels and goroutines from WriteEvents; add TestEnforcePolicyRules

This commit is contained in:
Jay
2026-05-22 16:31:52 -04:00
parent bce69a146b
commit 04e6c4fc57
2 changed files with 129 additions and 289 deletions
+36 -148
View File
@@ -8,7 +8,6 @@ import (
roots "git.wisehodl.dev/jay/go-roots/events" roots "git.wisehodl.dev/jay/go-roots/events"
"github.com/boltdb/bolt" "github.com/boltdb/bolt"
"github.com/neo4j/neo4j-go-driver/v6/neo4j" "github.com/neo4j/neo4j-go-driver/v6/neo4j"
"sync"
"time" "time"
) )
@@ -60,69 +59,11 @@ func WriteEvents(
return WriteReport{Error: fmt.Errorf("error setting up bolt db: %w", err)} return WriteReport{Error: fmt.Errorf("error setting up bolt db: %w", err)}
} }
var wg sync.WaitGroup travellers := createEventTravellers(events)
parsed, parseExcluded := parseEventJSON(travellers)
// Create Event Travellers queued, policyExcluded := enforcePolicyRules(parsed, boltdb, opts.BoltReadBatchSize)
jsonChan := make(chan []byte) converted := convertEventsToSubgraphs(queued, opts.Expanders)
eventChan := make(chan EventTraveller) writeResult := writeEventsToDatabases(driver, boltdb, converted)
wg.Add(1)
go createEventTravellers(&wg, jsonChan, eventChan)
// Parse Event JSON
parsedChan := make(chan EventTraveller)
parseExcludedChan := make(chan EventTraveller)
wg.Add(1)
go parseEventJSON(&wg, eventChan, parsedChan, parseExcludedChan)
// Collect Rejected Events
collectedParseExcludedChan := make(chan []EventTraveller, 1)
wg.Add(1)
go collectTravellers(&wg, parseExcludedChan, collectedParseExcludedChan)
// Enforce Policy Rules
queuedChan := make(chan EventTraveller)
policyExcludedChan := make(chan EventTraveller)
wg.Add(1)
go enforcePolicyRules(&wg, driver, boltdb, opts.BoltReadBatchSize,
parsedChan, queuedChan, policyExcludedChan)
// Collect Skipped Events
collectedPolicyExcludedChan := make(chan []EventTraveller, 1)
wg.Add(1)
go collectTravellers(&wg, policyExcludedChan, collectedPolicyExcludedChan)
// Convert Events To Subgraphs
convertedChan := make(chan EventTraveller)
wg.Add(1)
go convertEventsToSubgraphs(&wg, opts.Expanders, queuedChan, convertedChan)
// Write Events To Databases
writeResultChan := make(chan WriteResult, 1)
wg.Add(1)
go writeEventsToDatabases(&wg, driver, boltdb, convertedChan, writeResultChan)
// Send event jsons into pipeline
go func() {
for _, raw := range events {
jsonChan <- raw
}
close(jsonChan)
}()
// Wait for pipeline to complete
wg.Wait()
// Collect results
parseExcluded := <-collectedParseExcludedChan
policyExcluded := <-collectedPolicyExcludedChan
writeResult := <-writeResultChan
excluded := append(parseExcluded, policyExcluded...) excluded := append(parseExcluded, policyExcluded...)
@@ -144,75 +85,47 @@ func setDefaultWriteOptions(opts *WriteOptions) {
} }
} }
func createEventTravellers(wg *sync.WaitGroup, jsonChan chan []byte, eventChan chan EventTraveller) { func createEventTravellers(jsons [][]byte) []EventTraveller {
defer wg.Done() travellers := make([]EventTraveller, 0, len(jsons))
for json := range jsonChan { for _, j := range jsons {
eventChan <- EventTraveller{JSON: json} travellers = append(travellers, EventTraveller{JSON: j})
} }
close(eventChan) return travellers
} }
func parseEventJSON(wg *sync.WaitGroup, inChan, parsedChan, excludedChan chan EventTraveller) { func parseEventJSON(in []EventTraveller) (parsed []EventTraveller, excluded []EventTraveller) {
defer wg.Done() for _, traveller := range in {
for traveller := range inChan {
var event roots.Event var event roots.Event
jsonBytes := traveller.JSON err := json.Unmarshal(traveller.JSON, &event)
err := json.Unmarshal(jsonBytes, &event)
if err != nil { if err != nil {
traveller.Error = fmt.Errorf("rejected: %w: %w", ErrMalformedJSON, err) traveller.Error = fmt.Errorf("rejected: %w: %w", ErrMalformedJSON, err)
excludedChan <- traveller excluded = append(excluded, traveller)
continue continue
} }
err = roots.Validate(event) err = roots.Validate(event)
if err != nil { if err != nil {
traveller.Error = fmt.Errorf("rejected: %w: %w", ErrInvalidEvent, err) traveller.Error = fmt.Errorf("rejected: %w: %w", ErrInvalidEvent, err)
excludedChan <- traveller excluded = append(excluded, traveller)
continue continue
} }
traveller.ID = event.ID traveller.ID = event.ID
traveller.Event = event traveller.Event = event
parsedChan <- traveller parsed = append(parsed, traveller)
}
return parsed, excluded
} }
close(parsedChan) func enforcePolicyRules(in []EventTraveller, boltdb *bolt.DB, batchSize int) (queued []EventTraveller, excluded []EventTraveller) {
close(excludedChan) for i := 0; i < len(in); i += batchSize {
end := i + batchSize
if end > len(in) {
end = len(in)
} }
batch := in[i:end]
func enforcePolicyRules(
wg *sync.WaitGroup,
driver neo4j.Driver, boltdb *bolt.DB,
batchSize int,
inChan, queuedChan, excludedChan chan EventTraveller,
) {
defer wg.Done()
var batch []EventTraveller
for traveller := range inChan {
batch = append(batch, traveller)
if len(batch) >= batchSize {
processPolicyRulesBatch(boltdb, batch, queuedChan, excludedChan)
batch = []EventTraveller{}
}
}
if len(batch) > 0 {
processPolicyRulesBatch(boltdb, batch, queuedChan, excludedChan)
}
close(queuedChan)
close(excludedChan)
}
func processPolicyRulesBatch(
boltdb *bolt.DB,
batch []EventTraveller,
queuedChan, skippedChan chan EventTraveller,
) {
eventIDs := make([]string, 0, len(batch)) eventIDs := make([]string, 0, len(batch))
for _, traveller := range batch { for _, traveller := range batch {
eventIDs = append(eventIDs, traveller.ID) eventIDs = append(eventIDs, traveller.ID)
} }
@@ -222,51 +135,34 @@ func processPolicyRulesBatch(
for _, traveller := range batch { for _, traveller := range batch {
if existsMap[traveller.ID] { if existsMap[traveller.ID] {
traveller.Error = fmt.Errorf("skipped: %w", ErrDuplicate) traveller.Error = fmt.Errorf("skipped: %w", ErrDuplicate)
skippedChan <- traveller excluded = append(excluded, traveller)
} else { } else {
queuedChan <- traveller queued = append(queued, traveller)
} }
} }
} }
return queued, excluded
}
func convertEventsToSubgraphs( func convertEventsToSubgraphs(in []EventTraveller, expanders ExpanderPipeline) []EventTraveller {
wg *sync.WaitGroup, expanders ExpanderPipeline, for i, traveller := range in {
inChan, convertedChan chan EventTraveller,
) {
defer wg.Done()
for traveller := range inChan {
// TODO: temporary adapter — removed in Phase 5 // TODO: temporary adapter — removed in Phase 5
validated, _ := roots.NewValidatedEvent(traveller.Event) validated, _ := roots.NewValidatedEvent(traveller.Event)
subgraph := EventToSubgraph(validated, expanders) in[i].Subgraph = EventToSubgraph(validated, expanders)
traveller.Subgraph = subgraph
convertedChan <- traveller
} }
close(convertedChan) return in
}
func writeEventsToDatabases(
wg *sync.WaitGroup,
driver neo4j.Driver, boltdb *bolt.DB,
inChan chan EventTraveller,
resultChan chan WriteResult,
) {
defer wg.Done()
var travellers []EventTraveller
for traveller := range inChan {
travellers = append(travellers, traveller)
} }
func writeEventsToDatabases(driver neo4j.Driver, boltdb *bolt.DB, travellers []EventTraveller) WriteResult {
boltErr := writeEventsToBoltDB(boltdb, travellers) boltErr := writeEventsToBoltDB(boltdb, travellers)
if boltErr != nil { if boltErr != nil {
resultChan <- WriteResult{ return WriteResult{
Error: fmt.Errorf("boltdb write failed, aborting graph write: %w", boltErr), Error: fmt.Errorf("boltdb write failed, aborting graph write: %w", boltErr),
} }
return
} }
summaries, err := writeEventsToGraphDB(driver, travellers) summaries, err := writeEventsToGraphDB(driver, travellers)
resultChan <- WriteResult{ return WriteResult{
ResultSummaries: summaries, ResultSummaries: summaries,
Error: err, Error: err,
} }
@@ -297,12 +193,4 @@ func writeEventsToGraphDB(driver neo4j.Driver, travellers []EventTraveller) ([]n
return MergeSubgraph(context.Background(), driver, batch) return MergeSubgraph(context.Background(), driver, batch)
} }
func collectTravellers(wg *sync.WaitGroup, inChan chan EventTraveller, resultChan chan []EventTraveller) {
defer wg.Done()
var collected []EventTraveller
for traveller := range inChan {
collected = append(collected, traveller)
}
resultChan <- collected
close(resultChan)
}
+83 -131
View File
@@ -1,9 +1,8 @@
package heartwood package heartwood
import ( import (
roots "git.wisehodl.dev/jay/go-roots/events"
"github.com/stretchr/testify/assert" "github.com/stretchr/testify/assert"
"sync" "github.com/stretchr/testify/require"
"testing" "testing"
) )
@@ -54,26 +53,7 @@ func TestCreateEventTravellers(t *testing.T) {
for _, tc := range cases { for _, tc := range cases {
t.Run(tc.name, func(t *testing.T) { t.Run(tc.name, func(t *testing.T) {
var wg sync.WaitGroup result := createEventTravellers(tc.input)
jsonChan := make(chan []byte)
eventChan := make(chan EventTraveller)
wg.Add(1)
go createEventTravellers(&wg, jsonChan, eventChan)
go func() {
for _, raw := range tc.input {
jsonChan <- raw
}
close(jsonChan)
}()
var result []EventTraveller
for traveller := range eventChan {
result = append(result, traveller)
}
wg.Wait()
assert.Equal(t, len(tc.expected), len(result)) assert.Equal(t, len(tc.expected), len(result))
for i := range tc.expected { for i := range tc.expected {
@@ -81,7 +61,6 @@ func TestCreateEventTravellers(t *testing.T) {
} }
}) })
} }
} }
func TestParseEventJSON(t *testing.T) { func TestParseEventJSON(t *testing.T) {
@@ -136,43 +115,7 @@ func TestParseEventJSON(t *testing.T) {
for _, tc := range cases { for _, tc := range cases {
t.Run(tc.name, func(t *testing.T) { t.Run(tc.name, func(t *testing.T) {
var wg sync.WaitGroup parsed, rejected := parseEventJSON(tc.input)
inChan := make(chan EventTraveller)
parsedChan := make(chan EventTraveller)
rejectedChan := make(chan EventTraveller)
wg.Add(1)
go parseEventJSON(&wg, inChan, parsedChan, rejectedChan)
go func() {
for _, traveller := range tc.input {
inChan <- traveller
}
close(inChan)
}()
var parsed []EventTraveller
var rejected []EventTraveller
var collectWg sync.WaitGroup
collectWg.Add(2)
go func() {
defer collectWg.Done()
for f := range parsedChan {
parsed = append(parsed, f)
}
}()
go func() {
defer collectWg.Done()
for f := range rejectedChan {
rejected = append(rejected, f)
}
}()
collectWg.Wait()
wg.Wait()
assert.Equal(t, tc.wantParsed, len(parsed)) assert.Equal(t, tc.wantParsed, len(parsed))
assert.Equal(t, tc.wantRejected, len(rejected)) assert.Equal(t, tc.wantRejected, len(rejected))
@@ -196,20 +139,92 @@ func TestParseEventJSON(t *testing.T) {
} }
} }
// Skip `enforcePolicyRules` -- requires BoltDB func TestEnforcePolicyRules(t *testing.T) {
db := tempDB(t)
require.NoError(t, SetupBoltDB(db))
fx := LoadFixtures(t)
// Pre-write bare and generic_tag as existing events
bareJSON, _ := fx.ValidatedEvent(t, "bare").MarshalJSON()
genericJSON, _ := fx.ValidatedEvent(t, "generic_tag").MarshalJSON()
bareID := fx.ValidatedEvent(t, "bare").ID()
genericID := fx.ValidatedEvent(t, "generic_tag").ID()
err := BatchWriteEvents(db, []EventBlob{
{ID: []byte(bareID), JSON: bareJSON},
{ID: []byte(genericID), JSON: genericJSON},
})
assert.NoError(t, err)
e_tag_id := fx.ValidatedEvent(t, "e_tag_valid").ID()
p_tag_id := fx.ValidatedEvent(t, "p_tag_valid").ID()
cases := []struct {
name string
input []EventTraveller
wantQueued int
wantExcluded int
}{
{
name: "empty input",
input: []EventTraveller{},
wantQueued: 0,
wantExcluded: 0,
},
{
name: "no duplicates",
input: []EventTraveller{
{ID: e_tag_id},
{ID: p_tag_id},
},
wantQueued: 2,
wantExcluded: 0,
},
{
name: "some duplicates",
input: []EventTraveller{
{ID: bareID},
{ID: e_tag_id},
},
wantQueued: 1,
wantExcluded: 1,
},
{
name: "all duplicates",
input: []EventTraveller{
{ID: bareID},
{ID: genericID},
},
wantQueued: 0,
wantExcluded: 2,
},
}
for _, tc := range cases {
t.Run(tc.name, func(t *testing.T) {
queued, excluded := enforcePolicyRules(tc.input, db, 100)
assert.Equal(t, tc.wantQueued, len(queued))
assert.Equal(t, tc.wantExcluded, len(excluded))
for _, ex := range excluded {
assert.ErrorIs(t, ex.Error, ErrDuplicate)
}
})
}
}
func TestConvertEventsToSubgraphs(t *testing.T) { func TestConvertEventsToSubgraphs(t *testing.T) {
fx := LoadFixtures(t) fx := LoadFixtures(t)
cases := []struct { cases := []struct {
name string name string
event roots.ValidatedEvent traveller EventTraveller
wantNodeCount int wantNodeCount int
wantRelCount int wantRelCount int
}{ }{
{ {
name: "event with no tags", name: "event with no tags",
event: fx.ValidatedEvent(t, "bare"), traveller: EventTraveller{Event: fx.ValidatedEvent(t, "bare").Event()},
wantNodeCount: 2, // event + user wantNodeCount: 2, // event + user
wantRelCount: 1, // signed wantRelCount: 1, // signed
}, },
@@ -217,78 +232,15 @@ func TestConvertEventsToSubgraphs(t *testing.T) {
for _, tc := range cases { for _, tc := range cases {
t.Run(tc.name, func(t *testing.T) { t.Run(tc.name, func(t *testing.T) {
var wg sync.WaitGroup
inChan := make(chan EventTraveller)
convertedChan := make(chan EventTraveller)
expanders := NewExpanderPipeline(DefaultExpanders()...) expanders := NewExpanderPipeline(DefaultExpanders()...)
results := convertEventsToSubgraphs([]EventTraveller{tc.traveller}, expanders)
wg.Add(1) assert.Len(t, results, 1)
go convertEventsToSubgraphs(&wg, expanders, inChan, convertedChan) assert.NotNil(t, results[0].Subgraph)
assert.Equal(t, tc.wantNodeCount, len(results[0].Subgraph.Nodes()))
go func() { assert.Equal(t, tc.wantRelCount, len(results[0].Subgraph.Rels()))
inChan <- EventTraveller{Event: tc.event.Event()}
close(inChan)
}()
var result EventTraveller
for f := range convertedChan {
result = f
}
wg.Wait()
assert.NotNil(t, result.Subgraph)
assert.Equal(t, tc.wantNodeCount, len(result.Subgraph.Nodes()))
assert.Equal(t, tc.wantRelCount, len(result.Subgraph.Rels()))
}) })
} }
} }
// Skip `writeEventsToDatabases` tests -- requires BoltDB + Neo4j // Skip `writeEventsToDatabases` tests -- requires BoltDB + Neo4j
func TestCollectEvents(t *testing.T) {
cases := []struct {
name string
input []EventTraveller
expected int
}{
{
name: "empty channel",
input: []EventTraveller{},
expected: 0,
},
{
name: "multiple travellers",
input: []EventTraveller{
{ID: "id1"},
{ID: "id2"},
{ID: "id3"},
},
expected: 3,
},
}
for _, tc := range cases {
t.Run(tc.name, func(t *testing.T) {
var wg sync.WaitGroup
inChan := make(chan EventTraveller)
resultChan := make(chan []EventTraveller)
wg.Add(1)
go collectTravellers(&wg, inChan, resultChan)
go func() {
for _, f := range tc.input {
inChan <- f
}
close(inChan)
}()
result := <-resultChan
wg.Wait()
assert.Equal(t, tc.expected, len(result))
})
}
}