Merged in feature/checkbox (pull request #145)

Checkbox Primary Edge Cases

* hascheckboxlogic

* gen

* preppinggen

* exploringcheckbox

* removeunselected

* tests

* failingtest

* failintests

* nomockqueryapi

* db

* name

* nocontainer

* deleterow

* cnc

* extradocsandupdatetextracts

* moretables

* appndedtables

* baseversion
This commit is contained in:
Michael McGuinness
2025-05-20 12:47:22 +00:00
parent beb221937b
commit 61d12bf8df
69 changed files with 223611 additions and 78482 deletions
+64 -53
View File
@@ -1,7 +1,6 @@
package doctextrunner_test
import (
"context"
"fmt"
"io"
"strings"
@@ -25,7 +24,6 @@ import (
"github.com/aws/aws-sdk-go-v2/service/sqs"
awstextract "github.com/aws/aws-sdk-go-v2/service/textract"
"github.com/google/uuid"
"github.com/pashagolub/pgxmock/v3"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/mock"
"github.com/stretchr/testify/require"
@@ -39,14 +37,12 @@ type DocTextConfig struct {
objectstore.ObjectStoreConfig
}
func TestDocCleanRunner(t *testing.T) {
ctx := context.Background()
pool, err := pgxmock.NewPool()
require.NoError(t, err)
func TestDocTextRunner(t *testing.T) {
t.Parallel()
cfg := &DocTextConfig{}
cfg.DBPool = pool
cfg.DBQueries = repository.New(pool)
net := test.GetNetwork(t)
test.CreateDB(t, cfg, net, &test.CreateDatabaseConfig{})
mockStore := objectstoremock.NewMockS3Client(t)
cfg.StoreClient = mockStore
mockTextract := textractmock.NewMockTextractClient(t)
@@ -58,84 +54,99 @@ func TestDocCleanRunner(t *testing.T) {
runner := doctextrunner.New(&doctextrunner.Services{
Text: documenttext.New(cfg),
})
doc := doctextrunner.Body{
DocumentID: uuid.New(),
}
cleanId := uuid.New()
bucket := "bucket"
hash := `"a06d5e0e795adeb6d0b3732330dbcc15"`
textId := uuid.New()
err := cfg.GetDBQueries().CreateClient(t.Context(), &repository.CreateClientParams{
Clientid: "clientid",
Name: "client_name",
})
require.NoError(t, err)
docId, err := cfg.GetDBQueries().CreateDocument(t.Context(), &repository.CreateDocumentParams{
Clientid: "clientid",
Hash: "hash",
})
require.NoError(t, err)
part := uint16(1)
filetype := "pdf"
key := objectstore.BucketKey{
ClientID: "aa",
Location: objectstore.Import,
CreatedAt: time.Now().UTC(),
EntityID: cleanId,
ClientID: "clientid",
EntityID: uuid.New(),
Location: objectstore.Import,
Part: &part,
FileType: &filetype,
}
err = cfg.GetDBQueries().AddDocumentEntry(t.Context(), &repository.AddDocumentEntryParams{
Documentid: docId,
Bucket: "bucket",
Key: key.String(),
})
require.NoError(t, err)
bucket := "bucket"
hash := "hash"
keyStr := key.String()
filename := "../../assets/textract/helloWorld.gen"
textractBaseOut := test.GetTextractFileResponse(t, filename)
cleanId, err := cfg.GetDBQueries().AddDocumentClean(t.Context(), &repository.AddDocumentCleanParams{
Documentid: docId,
Bucket: &bucket,
Hash: &hash,
Key: &keyStr,
Mimetype: repository.NullCleanmimetype{
Valid: true,
Cleanmimetype: repository.CleanmimetypeApplicationPdf,
},
})
require.NoError(t, err)
err = cfg.GetDBQueries().AddDocumentCleanEntry(t.Context(), &repository.AddDocumentCleanEntryParams{
Cleanid: cleanId,
Version: 1,
})
require.NoError(t, err)
mimetype := repository.NullCleanmimetype{
Valid: true,
Cleanmimetype: repository.CleanmimetypeApplicationPdf,
}
pool.ExpectQuery("name: IsDocumentTextExtracted :one").WithArgs(doc.DocumentID).WillReturnRows(
pgxmock.NewRows([]string{"isextracted"}).
AddRow(false),
)
pool.ExpectQuery("name: GetCleanEntryByDocId :one").WithArgs(doc.DocumentID).
WillReturnRows(
pgxmock.NewRows([]string{"id", "documentId", "bucket", "key", "version", "hash", "mimetype", "fail", "clientId"}).
AddRow(cleanId, doc.DocumentID, &bucket, &keyStr, int64(123), &hash, mimetype, nil, key.ClientID),
)
bodyMsg := io.NopCloser(strings.NewReader(pdfHelloWorld))
mockStore.EXPECT().
GetObject(
mock.Anything,
mock.MatchedBy(func(in *s3.GetObjectInput) bool {
return *in.Bucket == bucket && *in.Key == keyStr
return *in.Bucket == "bucket" && *in.Key == key.String()
}),
mock.Anything,
).
Return(&s3.GetObjectOutput{
Body: bodyMsg,
}, nil)
sent := 0
filename := "../../assets/textract/helloWorld.gen"
textractBaseOut := test.GetTextractFileResponse(t, filename)
mockTextract.EXPECT().
AnalyzeDocument(
mock.Anything,
mock.MatchedBy(func(in *awstextract.AnalyzeDocumentInput) bool {
if sent == 0 {
sent++
return true
}
return false
return true
}),
mock.Anything,
).
Return(textractBaseOut[0]["base"], nil).Once()
pool.ExpectBegin()
pool.ExpectQuery("name: GetDocumentTextExtractionByHash :one").WithArgs(cleanId, hash).WillReturnRows(
pgxmock.NewRows([]string{"id"}).
AddRow(textId),
)
pool.ExpectExec("name: AddDocumentTextEntry :exec").WithArgs(textId, pgxmock.AnyArg()).
WillReturnResult(pgxmock.NewResult("", 1))
pool.ExpectCommit()
mockStore.EXPECT().
PutObject(
mock.Anything,
mock.MatchedBy(func(in *s3.PutObjectInput) bool {
return *in.Bucket == "bucket"
}),
mock.Anything,
).
Return(&s3.PutObjectOutput{}, nil)
mockQueue.EXPECT().
SendMessage(
mock.Anything,
mock.MatchedBy(func(in *sqs.SendMessageInput) bool {
return *in.QueueUrl == cfg.QuerySyncURL && *in.MessageBody == fmt.Sprintf("{\"id\":\"%s\"}", doc.DocumentID.String())
return *in.QueueUrl == cfg.QuerySyncURL && *in.MessageBody == fmt.Sprintf("{\"id\":\"%s\"}", docId.String())
}),
mock.Anything,
).
Return(&sqs.SendMessageOutput{}, nil)
assert.True(t, runner.Process(ctx, doc))
doc := doctextrunner.Body{
DocumentID: docId,
}
assert.True(t, runner.Process(t.Context(), doc))
}
const pdfHelloWorld = `%PDF-1.4