Files
query-orchestration/internal/database/repository/result_test.go
T
Michael McGuinness efe87321b2 Merged in feature/parallellogqueries (pull request #132)
Testing Tweaks

* parallel

* slowestquery

* split

* cleanup

* health

* dynamiccores

* go

* profile

* timeout

* commitmychanges

* taskfile

* sqlcheckstart

* client

* cost

* ctxtimeout
2025-05-05 09:31:21 +00:00

538 lines
16 KiB
Go

package repository_test
import (
"context"
"testing"
"time"
"queryorchestration/internal/database/repository"
"queryorchestration/internal/serviceconfig"
"queryorchestration/internal/test"
"github.com/google/uuid"
"github.com/jackc/pgx/v5/pgtype"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
func TestResults(t *testing.T) {
t.Parallel()
if testing.Short() {
t.SkipNow()
}
ctx := context.Background()
cfg := &serviceconfig.BaseConfig{}
net := test.GetNetwork(t)
test.CreateDB(t, cfg, net, &test.CreateDatabaseConfig{})
queries := cfg.GetDBQueries()
contextQueryID, jsonQueryID := createDependentQueries(t, ctx, queries)
clientId := createClientWithCollector(t, t.Context(), queries)
documentID, _, _ := createDocumentWithCollectorAndResults(t, ctx, queries, clientId, contextQueryID, jsonQueryID)
qv := int32(1)
res, err := queries.GetResultValueWithVersion(ctx, &repository.GetResultValueWithVersionParams{
Queryid: &jsonQueryID,
Queryversion: &qv,
Documentid: &documentID,
})
require.NoError(t, err)
assert.NotNil(t, res.Value)
assert.Equal(t, "json_value", *res.Value)
}
func TestResultValues(t *testing.T) {
t.Parallel()
if testing.Short() {
t.SkipNow()
}
ctx := context.Background()
cfg := &serviceconfig.BaseConfig{}
net := test.GetNetwork(t)
test.CreateDB(t, cfg, net, &test.CreateDatabaseConfig{})
queries := cfg.GetDBQueries()
jsonQueryID, err := queries.CreateQuery(ctx, repository.Querytype(repository.QuerytypeJsonExtractor))
require.NoError(t, err)
clientId := "EXAMPLE"
err = queries.CreateClient(ctx, &repository.CreateClientParams{
Name: "example_client",
Clientid: clientId,
})
require.NoError(t, err)
documentID, err := queries.CreateDocument(ctx, &repository.CreateDocumentParams{
Clientid: clientId,
Hash: "example_hash",
})
require.NoError(t, err)
documentTwoID, err := queries.CreateDocument(ctx, &repository.CreateDocumentParams{
Clientid: clientId,
Hash: "example_hash_two",
})
require.NoError(t, err)
contextQueryID, err := queries.CreateQuery(ctx, repository.Querytype(repository.QuerytypeContextFull))
require.NoError(t, err)
_, err = queries.AddLatestQueryVersion(ctx, jsonQueryID)
require.NoError(t, err)
err = queries.AddActiveQueryVersion(ctx, &repository.AddActiveQueryVersionParams{
Queryid: jsonQueryID,
Versionid: 1,
})
require.NoError(t, err)
_, err = queries.AddLatestQueryVersion(ctx, contextQueryID)
require.NoError(t, err)
_, err = queries.AddLatestQueryVersion(ctx, contextQueryID)
require.NoError(t, err)
err = queries.AddActiveQueryVersion(ctx, &repository.AddActiveQueryVersionParams{
Queryid: contextQueryID,
Versionid: 2,
})
require.NoError(t, err)
jsonVersion := int32(1)
err = queries.AddRequiredQuery(ctx, &repository.AddRequiredQueryParams{
Queryid: jsonQueryID,
Requiredqueryid: contextQueryID,
Addedversion: jsonVersion,
})
require.NoError(t, err)
contextQuery, err := queries.GetQuery(ctx, contextQueryID)
require.NoError(t, err)
qResults, err := queries.ListQueryRequirementValues(ctx, &repository.ListQueryRequirementValuesParams{
Queryid: &jsonQueryID,
Documentid: &documentID,
Version: &jsonVersion,
})
require.NoError(t, err)
assert.Len(t, qResults, 0)
qResults, err = queries.ListQueryRequirementValues(ctx, &repository.ListQueryRequirementValuesParams{
Queryid: &jsonQueryID,
Documentid: &documentTwoID,
Version: &jsonVersion,
})
require.NoError(t, err)
assert.Len(t, qResults, 0)
version, err := queries.AddLatestCollectorVersion(ctx, clientId)
require.NoError(t, err)
err = queries.SetActiveCollectorVersion(ctx, &repository.SetActiveCollectorVersionParams{
Versionid: version,
Clientid: clientId,
})
require.NoError(t, err)
bucket := "example_bucket"
key := "example_key"
hash := "hash"
cleanid, err := queries.AddDocumentClean(ctx, &repository.AddDocumentCleanParams{
Documentid: documentID,
Bucket: &bucket,
Key: &key,
Hash: &hash,
Mimetype: repository.NullCleanmimetype{
Valid: true,
Cleanmimetype: repository.CleanmimetypeApplicationPdf,
},
})
require.NoError(t, err)
err = queries.AddDocumentCleanEntry(ctx, &repository.AddDocumentCleanEntryParams{
Cleanid: cleanid,
Version: 1,
})
require.NoError(t, err)
textId, err := queries.AddDocumentText(ctx, &repository.AddDocumentTextParams{
Cleanid: cleanid,
Bucket: "hi",
Key: "hello",
Hash: "example",
Part: 0,
Createdat: pgtype.Timestamp{
Time: time.Now().UTC(),
Valid: true,
},
})
require.NoError(t, err)
err = queries.AddDocumentTextEntry(ctx, &repository.AddDocumentTextEntryParams{
Version: 1,
Textid: textId,
})
require.NoError(t, err)
result := repository.AddResultParams{
Queryid: contextQueryID,
Value: "context_value_1",
Textentryid: textId,
Queryversion: contextQuery.Activeversion,
}
_, err = queries.AddResult(ctx, &result)
require.NoError(t, err)
qResults, err = queries.ListQueryRequirementValues(ctx, &repository.ListQueryRequirementValuesParams{
Queryid: &jsonQueryID,
Documentid: &documentID,
Version: &jsonVersion,
})
require.NoError(t, err)
assert.Len(t, qResults, 1)
assert.Equal(t, contextQueryID, qResults[0].Queryid)
assert.Equal(t, repository.QuerytypeContextFull, qResults[0].Querytype)
assert.Equal(t, "context_value_1", *qResults[0].Value)
assert.NotEqual(t, uuid.UUID{}, qResults[0].ID)
qResults, err = queries.ListQueryRequirementValues(ctx, &repository.ListQueryRequirementValuesParams{
Queryid: &jsonQueryID,
Documentid: &documentTwoID,
Version: &jsonVersion,
})
require.NoError(t, err)
assert.Len(t, qResults, 0)
_, err = queries.AddResult(ctx, &repository.AddResultParams{
Queryid: contextQueryID,
Value: "context_value_2",
Textentryid: textId,
Queryversion: contextQuery.Activeversion - 1,
})
require.NoError(t, err)
qResults, err = queries.ListQueryRequirementValues(ctx, &repository.ListQueryRequirementValuesParams{
Queryid: &jsonQueryID,
Documentid: &documentID,
Version: &jsonVersion,
})
require.NoError(t, err)
assert.Len(t, qResults, 1)
assert.Equal(t, contextQueryID, qResults[0].Queryid)
assert.Equal(t, repository.QuerytypeContextFull, qResults[0].Querytype)
assert.Equal(t, "context_value_1", *qResults[0].Value)
assert.NotEqual(t, uuid.UUID{}, qResults[0].ID)
_, err = queries.AddResult(ctx, &repository.AddResultParams{
Queryid: contextQueryID,
Value: "context_value_3",
Textentryid: textId,
Queryversion: contextQuery.Activeversion,
})
require.NoError(t, err)
qResults, err = queries.ListQueryRequirementValues(ctx, &repository.ListQueryRequirementValuesParams{
Queryid: &jsonQueryID,
Documentid: &documentID,
Version: &jsonVersion,
})
require.NoError(t, err)
assert.Len(t, qResults, 1)
assert.Equal(t, contextQueryID, qResults[0].Queryid)
assert.Equal(t, repository.QuerytypeContextFull, qResults[0].Querytype)
assert.Equal(t, "context_value_3", *qResults[0].Value)
assert.NotEqual(t, uuid.UUID{}, qResults[0].ID)
_, err = queries.AddResult(ctx, &repository.AddResultParams{
Queryid: jsonQueryID,
Value: "json_value_1",
Textentryid: textId,
Queryversion: jsonVersion,
})
require.NoError(t, err)
qResults, err = queries.ListQueryRequirementValues(ctx, &repository.ListQueryRequirementValuesParams{
Queryid: &jsonQueryID,
Documentid: &documentID,
Version: &jsonVersion,
})
require.NoError(t, err)
assert.Len(t, qResults, 1)
assert.Equal(t, contextQueryID, qResults[0].Queryid)
assert.Equal(t, repository.QuerytypeContextFull, qResults[0].Querytype)
assert.Equal(t, "context_value_3", *qResults[0].Value)
assert.NotEqual(t, uuid.UUID{}, qResults[0].ID)
}
func TestUnsyncedNoDepsQueries(t *testing.T) {
if testing.Short() {
t.SkipNow()
}
t.Run("single document", func(t *testing.T) {
t.Parallel()
ctx := t.Context()
cfg := &serviceconfig.BaseConfig{}
net := test.GetNetwork(t)
test.CreateDB(t, cfg, net, &test.CreateDatabaseConfig{})
queries := cfg.GetDBQueries()
clientId := createClientWithCollector(t, ctx, queries)
contextQueryID, jsonQueryID := createDependentQueries(t, ctx, queries)
err := queries.AddCollectorQuery(ctx, &repository.AddCollectorQueryParams{
Clientid: clientId,
Name: "example_name",
Queryid: jsonQueryID,
Addedversion: 1,
})
require.NoError(t, err)
documentID, err := queries.CreateDocument(ctx, &repository.CreateDocumentParams{
Clientid: clientId,
Hash: "example_hash",
})
require.NoError(t, err)
qs, err := queries.ListUnsyncedNoDepsQueriesByDocId(ctx, &documentID)
require.NoError(t, err)
assert.Len(t, qs, 0)
bucket := "example_bucket"
key := "example_key"
hash := "hahs"
cleanid, err := queries.AddDocumentClean(ctx, &repository.AddDocumentCleanParams{
Documentid: documentID,
Bucket: &bucket,
Key: &key,
Hash: &hash,
Mimetype: repository.NullCleanmimetype{
Valid: true,
Cleanmimetype: repository.CleanmimetypeApplicationPdf,
},
})
require.NoError(t, err)
err = queries.AddDocumentCleanEntry(ctx, &repository.AddDocumentCleanEntryParams{
Cleanid: cleanid,
Version: 1,
})
require.NoError(t, err)
qs, err = queries.ListUnsyncedNoDepsQueriesByDocId(ctx, &documentID)
require.NoError(t, err)
assert.Len(t, qs, 0)
textId, err := queries.AddDocumentText(ctx, &repository.AddDocumentTextParams{
Cleanid: cleanid,
Bucket: "hi",
Key: "hello",
Part: 0,
Createdat: pgtype.Timestamp{
Time: time.Now().UTC(),
Valid: true,
},
})
require.NoError(t, err)
err = queries.AddDocumentTextEntry(ctx, &repository.AddDocumentTextEntryParams{
Version: 1,
Textid: textId,
})
require.NoError(t, err)
qs, err = queries.ListUnsyncedNoDepsQueriesByDocId(ctx, &documentID)
require.NoError(t, err)
assert.Len(t, qs, 1)
assert.ElementsMatch(t, []*uuid.UUID{&contextQueryID}, qs)
_, err = queries.AddResult(ctx, &repository.AddResultParams{
Queryid: contextQueryID,
Value: "context_value",
Textentryid: textId,
Queryversion: 1,
})
require.NoError(t, err)
qs, err = queries.ListUnsyncedNoDepsQueriesByDocId(ctx, &documentID)
require.NoError(t, err)
assert.Len(t, qs, 1)
assert.ElementsMatch(t, []*uuid.UUID{&jsonQueryID}, qs)
_, err = queries.AddResult(ctx, &repository.AddResultParams{
Queryid: jsonQueryID,
Value: "context_value",
Textentryid: textId,
Queryversion: 1,
})
require.NoError(t, err)
qs, err = queries.ListUnsyncedNoDepsQueriesByDocId(ctx, &documentID)
require.NoError(t, err)
assert.Len(t, qs, 0)
_, err = queries.AddLatestQueryVersion(ctx, jsonQueryID)
require.NoError(t, err)
err = queries.AddActiveQueryVersion(ctx, &repository.AddActiveQueryVersionParams{
Queryid: jsonQueryID,
Versionid: 2,
})
require.NoError(t, err)
qs, err = queries.ListUnsyncedNoDepsQueriesByDocId(ctx, &documentID)
require.NoError(t, err)
assert.Len(t, qs, 1)
assert.ElementsMatch(t, []*uuid.UUID{&jsonQueryID}, qs)
})
t.Run("two files", func(t *testing.T) {
t.Parallel()
ctx := context.Background()
cfg := &serviceconfig.BaseConfig{}
net := test.GetNetwork(t)
test.CreateDB(t, cfg, net, &test.CreateDatabaseConfig{})
queries := cfg.GetDBQueries()
clientId := createClientWithCollector(t, ctx, queries)
contextQueryID, jsonQueryID := createDependentQueries(t, ctx, queries)
documentID, _, _ := createDocumentWithCollectorAndResults(t, ctx, queries, clientId, contextQueryID, jsonQueryID)
documentTwoID, err := queries.CreateDocument(ctx, &repository.CreateDocumentParams{
Clientid: clientId,
Hash: "example_hash_two",
})
require.NoError(t, err)
bucket := "example_bucket"
key := "example_key"
hash := "hahs"
cleantwoid, err := queries.AddDocumentClean(ctx, &repository.AddDocumentCleanParams{
Documentid: documentTwoID,
Bucket: &bucket,
Key: &key,
Hash: &hash,
Mimetype: repository.NullCleanmimetype{
Valid: true,
Cleanmimetype: repository.CleanmimetypeApplicationPdf,
},
})
require.NoError(t, err)
err = queries.AddDocumentCleanEntry(ctx, &repository.AddDocumentCleanEntryParams{
Cleanid: cleantwoid,
Version: 1,
})
require.NoError(t, err)
qs, err := queries.ListUnsyncedNoDepsQueriesByDocId(ctx, &documentID)
require.NoError(t, err)
assert.Len(t, qs, 0)
qs, err = queries.ListUnsyncedNoDepsQueriesByDocId(ctx, &documentTwoID)
require.NoError(t, err)
assert.Len(t, qs, 0)
textTwoId, err := queries.AddDocumentText(ctx, &repository.AddDocumentTextParams{
Cleanid: cleantwoid,
Bucket: "hi",
Key: "hello",
Part: 0,
Createdat: pgtype.Timestamp{
Time: time.Now().UTC(),
Valid: true,
},
})
require.NoError(t, err)
err = queries.AddDocumentTextEntry(ctx, &repository.AddDocumentTextEntryParams{
Version: 1,
Textid: textTwoId,
})
require.NoError(t, err)
_, err = queries.AddResult(ctx, &repository.AddResultParams{
Queryid: contextQueryID,
Value: "context_value",
Textentryid: textTwoId,
Queryversion: 1,
})
require.NoError(t, err)
qs, err = queries.ListUnsyncedNoDepsQueriesByDocId(ctx, &documentID)
require.NoError(t, err)
assert.Len(t, qs, 0)
qs, err = queries.ListUnsyncedNoDepsQueriesByDocId(ctx, &documentTwoID)
require.NoError(t, err)
assert.Len(t, qs, 1)
assert.ElementsMatch(t, []*uuid.UUID{&jsonQueryID}, qs)
_, err = queries.AddResult(ctx, &repository.AddResultParams{
Queryid: jsonQueryID,
Value: "context_value",
Textentryid: textTwoId,
Queryversion: 1,
})
require.NoError(t, err)
qs, err = queries.ListUnsyncedNoDepsQueriesByDocId(ctx, &documentID)
require.NoError(t, err)
assert.Len(t, qs, 0)
qs, err = queries.ListUnsyncedNoDepsQueriesByDocId(ctx, &documentTwoID)
require.NoError(t, err)
assert.Len(t, qs, 0)
})
t.Run("update query", func(t *testing.T) {
t.Parallel()
ctx := context.Background()
cfg := &serviceconfig.BaseConfig{}
net := test.GetNetwork(t)
test.CreateDB(t, cfg, net, &test.CreateDatabaseConfig{})
queries := cfg.GetDBQueries()
clientId := createClientWithCollector(t, ctx, queries)
contextQueryID, jsonQueryID := createDependentQueries(t, ctx, queries)
documentID, _, _ := createDocumentWithCollectorAndResults(t, ctx, queries, clientId, contextQueryID, jsonQueryID)
_, err := queries.AddLatestQueryVersion(ctx, jsonQueryID)
require.NoError(t, err)
err = queries.AddActiveQueryVersion(ctx, &repository.AddActiveQueryVersionParams{
Queryid: jsonQueryID,
Versionid: 2,
})
require.NoError(t, err)
qs, err := queries.ListUnsyncedNoDepsQueriesByDocId(ctx, &documentID)
require.NoError(t, err)
assert.Len(t, qs, 1)
assert.ElementsMatch(t, []*uuid.UUID{&jsonQueryID}, qs)
})
}
func BenchmarkListUnsynced(b *testing.B) {
ctx := context.Background()
cfg := &serviceconfig.BaseConfig{}
net := test.GetNetwork(b)
test.CreateDB(b, cfg, net, &test.CreateDatabaseConfig{})
queries := cfg.GetDBQueries()
clientId := createClientWithCollector(b, ctx, queries)
_, jsonQueryID := createDependentQueries(b, ctx, queries)
err := queries.AddCollectorQuery(ctx, &repository.AddCollectorQueryParams{
Clientid: clientId,
Name: "example_name",
Queryid: jsonQueryID,
Addedversion: 1,
})
require.NoError(b, err)
documentID, err := queries.CreateDocument(ctx, &repository.CreateDocumentParams{
Clientid: clientId,
Hash: "example_hash",
})
require.NoError(b, err)
b.ResetTimer()
for b.Loop() {
_, _ = queries.ListUnsyncedNoDepsQueriesByDocId(ctx, &documentID)
}
}