Merge pull request #845 from bjwswang/main

chore: use alpine as base image and install packages to support pdf-t…

Dockerfile

@@ -21,12 +21,20 @@ COPY apiserver/ apiserver/
 # Build
 RUN CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -a -o manager main.go
 RUN CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -o apiserver-bin apiserver/main.go
-# Use distroless as minimal base image to package the manager binary
-# Refer to https://github.com/GoogleContainerTools/distroless for more details
-FROM gcr.io/distroless/static:nonroot
+
+# Use alpine as minimal base image to package the manager binary
+FROM alpine:3.19.1
+
+RUN apk update \
+    # Install packages to support pdf to text conversion
+    && apk add --no-cache  poppler-utils wv unrtf tidyhtml
+
 WORKDIR /
 COPY --from=builder /workspace/manager .
 COPY --from=builder /workspace/apiserver-bin ./apiserver
-USER 65532:65532
+
+RUN adduser -D -u 1000 1000
+
+USER 1000
 
 ENTRYPOINT ["/manager"]

controllers/base/knowledgebase_controller.go

@@ -499,15 +499,6 @@ func (r *KnowledgeBaseReconciler) reconcileFileGroup(ctx context.Context, log lo
 
 func (r *KnowledgeBaseReconciler) handleFile(ctx context.Context, log logr.Logger, file io.ReadCloser, fileName string, tags map[string]string, kb *arcadiav1alpha1.KnowledgeBase, store *arcadiav1alpha1.VectorStore, embedder *arcadiav1alpha1.Embedder) (err error) {
 	log = log.WithValues("fileName", fileName, "tags", tags)
-	if tags == nil {
-		log.Info("file tags is nil, ignore")
-		return fmt.Errorf("file tags is nil, %w", errFileSkipped)
-	}
-	v, ok := tags[arcadiav1alpha1.ObjectTypeTag]
-	if !ok {
-		log.Info("file tags object type not found, ignore")
-		return fmt.Errorf("file tags object type not found, %w", errFileSkipped)
-	}
 	if !embedder.Status.IsReady() {
 		return errEmbedderNotReady
 	}
@@ -530,7 +521,8 @@ func (r *KnowledgeBaseReconciler) handleFile(ctx context.Context, log logr.Logge
 	case ".txt":
 		loader = documentloaders.NewText(dataReader)
 	case ".csv":
-		if v == arcadiav1alpha1.ObjectTypeQA {
+		v, ok := tags[arcadiav1alpha1.ObjectTypeTag]
+		if ok && v == arcadiav1alpha1.ObjectTypeQA {
 			// for qa csv,we skip the text splitter
 			loader = pkgdocumentloaders.NewQACSV(dataReader, fileName)
 		} else {