chore: project assets for text files are now documents (not artifacts) (

#1992)

packages/client/hmi-client/src/components/documents/tera-document-asset.vue

@@ -20,13 +20,23 @@
 					:active="documentView === DocumentView.EXRACTIONS"
 				/>
 				<Button
+					v-if="documentType === DocumentView.PDF"
 					class="p-button-secondary p-button-sm"
 					label="PDF"
 					icon="pi pi-file"
 					:loading="!pdfLink"
 					@click="documentView = DocumentView.PDF"
 					:active="documentView === DocumentView.PDF"
 				/>
+				<Button
+					v-if="documentType === DocumentView.TXT"
+					class="p-button-secondary p-button-sm"
+					label="txt"
+					icon="pi pi-file"
+					:loading="!pdfLink"
+					@click="documentView = DocumentView.TXT"
+					:active="documentView === DocumentView.TXT"
+				/>
 			</span>
 		</template>
 		<template #info-bar>
@@ -106,6 +116,7 @@
 			:pdf-link="pdfLink"
 			:title="doc.name || ''"
 		/>
+		<code-editor v-else-if="documentView === DocumentView.TXT" :initial-code="code" />
 	</tera-asset>
 </template>
 
@@ -121,13 +132,19 @@ import TeraPdfEmbed from '@/components/widgets/tera-pdf-embed.vue';
 import { DocumentAsset } from '@/types/Types';
 import * as textUtil from '@/utils/text';
 import TeraAsset from '@/components/asset/tera-asset.vue';
-import { downloadDocumentAsset, getDocumentAsset } from '@/services/document-assets';
+import {
+	downloadDocumentAsset,
+	getDocumentAsset,
+	getDocumentFileAsText
+} from '@/services/document-assets';
 import Image from 'primevue/image';
 import TeraShowMoreText from '@/components/widgets/tera-show-more-text.vue';
+import CodeEditor from '@/page/project/components/code-editor.vue';
 
 enum DocumentView {
 	EXRACTIONS = 'extractions',
-	PDF = 'pdf'
+	PDF = 'pdf',
+	TXT = 'txt'
 }
 
 const props = defineProps<{
@@ -140,6 +157,7 @@ const props = defineProps<{
 const doc = ref<DocumentAsset | null>(null);
 const pdfLink = ref<string | null>(null);
 const documentView = ref(DocumentView.EXRACTIONS);
+const code = ref<string>();
 
 const docLink = computed(() =>
 	doc.value?.fileNames && doc.value.fileNames.length > 0 ? doc.value.fileNames[0] : null
@@ -155,6 +173,13 @@ const equations = computed(
 	() => doc.value?.assets?.filter((asset) => asset.assetType === 'equation') || []
 );
 
+const documentType = computed(() => {
+	if (doc.value?.fileNames?.at(0)?.endsWith('.pdf')) {
+		return DocumentView.PDF;
+	}
+	return DocumentView.TXT;
+});
+
 const emit = defineEmits(['close-preview', 'asset-loaded']);
 
 // Highlight strings based on props.highlight
@@ -165,13 +190,21 @@ function highlightSearchTerms(text: string | undefined): string {
 	return text ?? '';
 }
 
+async function openTextDocument() {
+	const filename: string | undefined = doc.value?.fileNames?.at(0);
+	const res: string | null = await getDocumentFileAsText(props.assetId!, filename!);
+	if (!res) return;
+	code.value = res;
+}
+
 watch(
 	() => props.assetId,
 	async () => {
 		if (props.assetId) {
 			const document = await getDocumentAsset(props.assetId);
 			if (document) {
 				doc.value = document;
+				openTextDocument();
 			}
 		} else {
 			doc.value = null;

packages/client/hmi-client/src/page/project/components/tera-upload-resources-modal.vue

@@ -61,15 +61,16 @@ import Button from 'primevue/button';
 import { AcceptedTypes, AcceptedExtensions } from '@/types/common';
 import { uploadCodeToProject } from '@/services/code';
 import { useProjects } from '@/composables/project';
-import { Artifact, AssetType, Dataset } from '@/types/Types';
-import { uploadArtifactToProject } from '@/services/artifact';
+import { DocumentAsset, AssetType, Dataset } from '@/types/Types';
+import { uploadDocumentAssetToProject } from '@/services/document-assets';
 import { createNewDatasetFromCSV } from '@/services/dataset';
 import useAuthStore from '@/stores/auth';
 import { ref } from 'vue';
 import TeraDragAndDropImporter from '@/components/extracting/tera-drag-n-drop-importer.vue';
 import InputText from 'primevue/inputtext';
 import { useToastService } from '@/services/toast';
 import TeraImportGithubFile from '@/components/widgets/tera-import-github-file.vue';
+import { extractPDF } from '@/services/knowledge';
 
 defineProps<{
 	visible: boolean;
@@ -90,7 +91,7 @@ async function processFiles(files: File[], csvDescription: string) {
 			case AcceptedTypes.PDF:
 			case AcceptedTypes.TXT:
 			case AcceptedTypes.MD:
-				return processArtifact(file);
+				return processDocument(file);
 			case AcceptedTypes.PY:
 			case AcceptedTypes.R:
 			case AcceptedTypes.JL:
@@ -115,15 +116,23 @@ async function processCode(file: File) {
  * Process a pdf, txt, md file into an artifact
  * @param file
  */
-async function processArtifact(file: File) {
+async function processDocument(file: File) {
 	// This is pdf, txt, md files
-	const artifact: Artifact | null = await uploadArtifactToProject(
+	const document: DocumentAsset | null = await uploadDocumentAssetToProject(
 		file,
 		useProjects().activeProject.value?.username ?? '',
 		'',
 		progress
 	);
-	return { id: artifact?.id ?? '', assetType: AssetType.Artifacts };
+	let newAsset;
+	if (document && document.id) {
+		newAsset = await useProjects().addAsset(AssetType.Documents, document.id);
+	}
+	if (document && newAsset && file.name.toLowerCase().endsWith('.pdf')) {
+		await extractPDF(document);
+		return { file, error: false, response: { text: '', images: [] } };
+	}
+	return { file, error: true, response: { text: '', images: [] } };
 }
 
 /**

packages/client/hmi-client/src/services/document-assets.ts

@@ -157,10 +157,25 @@ async function downloadDocumentAsset(documentId: string, fileName: string): Prom
 	}
 }
 
+async function getDocumentFileAsText(documentId: string, fileName: string): Promise<string | null> {
+	const response = await API.get(
+		`/document-asset/${documentId}/download-document-as-text?filename=${fileName}`,
+		{}
+	);
+
+	if (!response || response.status >= 400) {
+		logger.error('Error getting document file as text');
+		return null;
+	}
+
+	return response.data;
+}
+
 export {
 	getAll,
 	getDocumentAsset,
 	uploadDocumentAssetToProject,
 	downloadDocumentAsset,
-	createNewDocumentFromGithubFile
+	createNewDocumentFromGithubFile,
+	getDocumentFileAsText
 };

packages/server/src/main/java/software/uncharted/terarium/hmiserver/controller/dataservice/DocumentController.java

@@ -22,6 +22,8 @@
 import software.uncharted.terarium.hmiserver.proxies.dataservice.DocumentProxy;
 import software.uncharted.terarium.hmiserver.proxies.jsdelivr.JsDelivrProxy;
 import org.apache.http.entity.StringEntity;
+import org.apache.commons.io.IOUtils;
+import java.nio.charset.StandardCharsets;
 
 import java.io.IOException;
 import java.util.List;
@@ -149,4 +151,30 @@ public ResponseEntity<byte[]> downloadDocument(
 			return ResponseEntity.internalServerError().build();
 		}
 	}
+
+	@GetMapping("/{id}/download-document-as-text")
+	public ResponseEntity<String> getDocumentFileAsText(@PathVariable("id") String documentId, @RequestParam("filename") String filename) {
+
+		log.debug("Downloading document file {} for document {}", filename, documentId);
+
+		try (CloseableHttpClient httpclient = HttpClients.custom()
+			.disableRedirectHandling()
+			.build()) {
+
+			PresignedURL presignedURL = proxy.getDownloadUrl(documentId, filename).getBody();
+			final HttpGet httpGet = new HttpGet(presignedURL.getUrl());
+			final HttpResponse response = httpclient.execute(httpGet);
+			final String textFileAsString = IOUtils.toString(response.getEntity().getContent(), StandardCharsets.UTF_8);
+
+			return ResponseEntity.ok(textFileAsString);
+
+
+		} catch (Exception e) {
+			log.error("Unable to GET document data", e);
+			return ResponseEntity.internalServerError().build();
+		}
+
+	}
+
+
 }