query-orchestration/internal/database/repository/text_test.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 TestTextExtraction(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()

	clientId := "EXAMPLE"
	err := queries.CreateClient(ctx, &repository.CreateClientParams{
		Name:     "example_client",
		Clientid: clientId,
	})
	require.NoError(t, err)

	hash := "example_hash"
	id, err := queries.CreateDocument(ctx, &repository.CreateDocumentParams{
		Clientid: clientId,
		Hash:     hash,
	})
	require.NoError(t, err)
	assert.NotEmpty(t, id)

	bucket := "example_bucket"
	key := "example_key"
	cleanid, err := queries.AddDocumentClean(ctx, &repository.AddDocumentCleanParams{
		Documentid: id,
		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)

	isextract, err := queries.IsDocumentTextExtracted(ctx, id)
	require.NoError(t, err)
	assert.False(t, isextract)

	textId, err := queries.AddDocumentText(ctx, &repository.AddDocumentTextParams{
		Cleanid: cleanid,
		Part:    0,
		Bucket:  bucket,
		Key:     key,
		Hash:    "example",
		Createdat: pgtype.Timestamp{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)
	assert.NotEqual(t, uuid.UUID{}, textId)

	isextract, err = queries.IsDocumentTextExtracted(ctx, id)
	require.NoError(t, err)
	assert.False(t, isextract)

	isextract, err = queries.IsDocumentTextExtracted(ctx, id)
	require.NoError(t, err)
	assert.False(t, isextract)

	err = queries.AddDocumentTextEntry(ctx, &repository.AddDocumentTextEntryParams{
		Textid:  textId,
		Version: 543,
	})
	require.NoError(t, err)

	isextract, err = queries.IsDocumentTextExtracted(ctx, id)
	require.NoError(t, err)
	assert.True(t, isextract)

	text, err := queries.GetTextEntryByDocId(ctx, id)
	require.NoError(t, err)
	assert.Equal(t, textId, text.ID)
	assert.Equal(t, id, text.Documentid)
	assert.Equal(t, bucket, text.Bucket)
	assert.Equal(t, key, text.Key)
	assert.Equal(t, "example", text.Hash)
	assert.Equal(t, cleanid, text.Cleanid)
	assert.Equal(t, int64(543), text.Extractionversion)
	assert.Equal(t, textId, text.ID)

	uniqueEntry, err := queries.GetDocumentTextExtractionByHash(ctx, &repository.GetDocumentTextExtractionByHashParams{
		Hash:         "example",
		Cleanentryid: cleanid,
	})
	require.NoError(t, err)
	assert.EqualExportedValues(t, text.ID, uniqueEntry)

	_, err = queries.GetDocumentTextExtractionByHash(ctx, &repository.GetDocumentTextExtractionByHashParams{
		Hash:         "definitely invalid",
		Cleanentryid: cleanid,
	})
	require.EqualError(t, err, "no rows in result set")
}

func TestTextTextractPart(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()

	clientId := "EXAMPLE"
	err := queries.CreateClient(ctx, &repository.CreateClientParams{
		Name:     "example_client",
		Clientid: clientId,
	})
	require.NoError(t, err)

	hash := "example_hash"
	id, err := queries.CreateDocument(ctx, &repository.CreateDocumentParams{
		Clientid: clientId,
		Hash:     hash,
	})
	require.NoError(t, err)
	assert.NotEmpty(t, id)

	bucket := "example_bucket"
	key := "example_key"
	cleanid, err := queries.AddDocumentClean(ctx, &repository.AddDocumentCleanParams{
		Documentid: id,
		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)

	part, err := queries.GetTextractOutputCurrentPart(ctx, &repository.GetTextractOutputCurrentPartParams{
		Clientid: &clientId,
		Querydate: pgtype.Timestamptz{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)
	assert.Equal(t, &repository.GetTextractOutputCurrentPartRow{
		Count: 0,
		Part:  0,
	}, part)

	_, err = queries.AddDocumentText(ctx, &repository.AddDocumentTextParams{
		Cleanid: cleanid,
		Part:    0,
		Createdat: pgtype.Timestamp{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)

	part, err = queries.GetTextractOutputCurrentPart(ctx, &repository.GetTextractOutputCurrentPartParams{
		Clientid: &clientId,
		Querydate: pgtype.Timestamptz{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)
	assert.Equal(t, &repository.GetTextractOutputCurrentPartRow{
		Count: 1,
		Part:  0,
	}, part)

	_, err = queries.AddDocumentText(ctx, &repository.AddDocumentTextParams{
		Cleanid: cleanid,
		Part:    0,
		Createdat: pgtype.Timestamp{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)

	part, err = queries.GetTextractOutputCurrentPart(ctx, &repository.GetTextractOutputCurrentPartParams{
		Clientid: &clientId,
		Querydate: pgtype.Timestamptz{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)
	assert.Equal(t, &repository.GetTextractOutputCurrentPartRow{
		Count: 2,
		Part:  0,
	}, part)

	_, err = queries.AddDocumentText(ctx, &repository.AddDocumentTextParams{
		Cleanid: cleanid,
		Part:    1,
		Createdat: pgtype.Timestamp{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)

	part, err = queries.GetTextractOutputCurrentPart(ctx, &repository.GetTextractOutputCurrentPartParams{
		Clientid: &clientId,
		Querydate: pgtype.Timestamptz{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)
	assert.Equal(t, &repository.GetTextractOutputCurrentPartRow{
		Count: 1,
		Part:  1,
	}, part)
}

func TestTextOutPart(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()

	clientId := "EXAMPLE"
	err := queries.CreateClient(ctx, &repository.CreateClientParams{
		Name:     "example_client",
		Clientid: clientId,
	})
	require.NoError(t, err)

	hash := "example_hash"
	id, err := queries.CreateDocument(ctx, &repository.CreateDocumentParams{
		Clientid: clientId,
		Hash:     hash,
	})
	require.NoError(t, err)
	assert.NotEmpty(t, id)

	bucket := "example_bucket"
	key := "example_key"
	cleanid, err := queries.AddDocumentClean(ctx, &repository.AddDocumentCleanParams{
		Documentid: id,
		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:     "hi",
		Hash:    "hi",
		Part:    0,
		Createdat: pgtype.Timestamp{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)
	err = queries.AddDocumentTextEntry(ctx, &repository.AddDocumentTextEntryParams{
		Textid:  textId,
		Version: 123,
	})
	require.NoError(t, err)

	part, err := queries.GetTextOutCurrentPart(ctx, &repository.GetTextOutCurrentPartParams{
		Clientid: &clientId,
		Querydate: pgtype.Timestamptz{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)
	assert.Equal(t, &repository.GetTextOutCurrentPartRow{
		Count: 1,
		Part:  0,
	}, part)

	textId, err = queries.AddDocumentText(ctx, &repository.AddDocumentTextParams{
		Cleanid: cleanid,
		Bucket:  "hi",
		Key:     "hi",
		Hash:    "hi",
		Part:    0,
		Createdat: pgtype.Timestamp{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)
	err = queries.AddDocumentTextEntry(ctx, &repository.AddDocumentTextEntryParams{
		Textid:  textId,
		Version: 123,
	})
	require.NoError(t, err)

	part, err = queries.GetTextOutCurrentPart(ctx, &repository.GetTextOutCurrentPartParams{
		Clientid: &clientId,
		Querydate: pgtype.Timestamptz{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)
	assert.Equal(t, &repository.GetTextOutCurrentPartRow{
		Count: 2,
		Part:  0,
	}, part)

	textId, err = queries.AddDocumentText(ctx, &repository.AddDocumentTextParams{
		Cleanid: cleanid,
		Bucket:  "hi",
		Key:     "hi",
		Hash:    "hi",
		Part:    0,
		Createdat: pgtype.Timestamp{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)
	err = queries.AddDocumentTextEntry(ctx, &repository.AddDocumentTextEntryParams{
		Textid:  textId,
		Version: 123,
	})
	require.NoError(t, err)

	part, err = queries.GetTextOutCurrentPart(ctx, &repository.GetTextOutCurrentPartParams{
		Clientid: &clientId,
		Querydate: pgtype.Timestamptz{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)
	assert.Equal(t, &repository.GetTextOutCurrentPartRow{
		Count: 3,
		Part:  0,
	}, part)

	textId, err = queries.AddDocumentText(ctx, &repository.AddDocumentTextParams{
		Cleanid: cleanid,
		Bucket:  "hi",
		Key:     "hi",
		Hash:    "hi",
		Part:    1,
		Createdat: pgtype.Timestamp{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)
	err = queries.AddDocumentTextEntry(ctx, &repository.AddDocumentTextEntryParams{
		Textid:  textId,
		Version: 123,
	})
	require.NoError(t, err)

	part, err = queries.GetTextOutCurrentPart(ctx, &repository.GetTextOutCurrentPartParams{
		Clientid: &clientId,
		Querydate: pgtype.Timestamptz{
			Time:  time.Now().UTC(),
			Valid: true,
		},
	})
	require.NoError(t, err)
	assert.Equal(t, &repository.GetTextOutCurrentPartRow{
		Count: 1,
		Part:  1,
	}, part)
}