Hide empty Text view for documents with no text content

frontend/src/js/components/viewer/PreviewSwitcher.js

@@ -3,6 +3,7 @@ import PropTypes from "prop-types";
 import { resourcePropType } from "../../types/Resource";
 import _ from "lodash";
 
+import { hasTextContent } from "../../util/resourceUtils";
 import { keyboardShortcuts } from "../../util/keyboardShortcuts";
 import { KeyboardShortcut } from "../UtilComponents/KeyboardShortcut";
 
@@ -34,7 +35,7 @@ class PreviewSwitcher extends React.Component {
       return false;
     }
 
-    if (this.props.view === "text" && !resource.text) {
+    if (this.props.view === "text" && !hasTextContent(resource)) {
       return false;
     }
 
@@ -65,8 +66,8 @@ class PreviewSwitcher extends React.Component {
       this.props.view &&
       !this.currentViewModeIsValid(this.props.resource)
     ) {
-      // Automatically switch to a preview if you can preview but there's no extracted text or OCR
-      if (this.props.resource.text) {
+      // Automatically switch to a valid view when the current one is not available
+      if (hasTextContent(this.props.resource)) {
         this.props.setResourceView("text");
       } else if (this.props.resource.ocr) {
         const languages = Object.keys(this.props.resource.ocr);
@@ -136,7 +137,8 @@ class PreviewSwitcher extends React.Component {
           shortcut={keyboardShortcuts.showPreview}
           func={this.showPreview}
         />
-        {this.props.resource.text && !this.props.resource.transcript ? (
+        {hasTextContent(this.props.resource) &&
+        !this.props.resource.transcript ? (
           <PreviewLink
             current={current}
             text="Text"

frontend/src/js/types/Resource.ts

@@ -70,6 +70,7 @@ export type Resource = BasicResource & {
   children: BasicResource[];
   comments: CommentData[];
   selection?: ResourceRange;
+  previewStatus: string;
   text: HighlightableText;
   ocr?: {
     [lang: string]: HighlightableText;

frontend/src/js/util/resourceUtils.spec.ts

@@ -0,0 +1,129 @@
+import { Resource, HighlightableText } from "../types/Resource";
+import { getDefaultView, hasTextContent } from "./resourceUtils";
+
+function makeHighlightableText(contents: string): HighlightableText {
+  return { contents, highlights: [] };
+}
+
+function makeResource(overrides?: Partial<Resource>): Resource {
+  return {
+    uri: "test/doc.pdf",
+    type: "blob",
+    isExpandable: false,
+    processingStage: { type: "processed" },
+    extracted: true,
+    mimeTypes: ["application/pdf"],
+    fileSize: 1024,
+    parents: [],
+    children: [],
+    comments: [],
+    previewStatus: "disabled",
+    text: makeHighlightableText(""),
+    ...overrides,
+  } as Resource;
+}
+
+describe("hasTextContent", () => {
+  test("returns true when text has content", () => {
+    const resource = makeResource({
+      text: makeHighlightableText("Hello world"),
+    });
+    expect(hasTextContent(resource)).toBe(true);
+  });
+
+  test("returns false when text is empty", () => {
+    const resource = makeResource({
+      text: makeHighlightableText(""),
+    });
+    expect(hasTextContent(resource)).toBe(false);
+  });
+
+  test("returns false when text is only whitespace", () => {
+    const resource = makeResource({
+      text: makeHighlightableText("   \n\t  "),
+    });
+    expect(hasTextContent(resource)).toBe(false);
+  });
+});
+
+describe("getDefaultView", () => {
+  test("returns undefined for non-blob resources", () => {
+    const resource = makeResource({
+      text: makeHighlightableText("content"),
+      type: "file",
+    });
+    expect(getDefaultView(resource)).toBeUndefined();
+  });
+
+  test("returns 'text' when document has text content", () => {
+    const resource = makeResource({
+      text: makeHighlightableText("Hello world"),
+    });
+    expect(getDefaultView(resource)).toBe("text");
+  });
+
+  test("returns transcript view when transcript exists", () => {
+    const resource = makeResource({
+      text: makeHighlightableText(""),
+      transcript: {
+        english: makeHighlightableText("transcript content"),
+      },
+    });
+    expect(getDefaultView(resource)).toBe("transcript.english");
+  });
+
+  test("returns first OCR language when text is empty but OCR has content", () => {
+    const resource = makeResource({
+      text: makeHighlightableText(""),
+      ocr: {
+        english: makeHighlightableText("OCR text"),
+      },
+    });
+    expect(getDefaultView(resource)).toBe("ocr.english");
+  });
+
+  test("skips empty OCR entries and returns first non-empty one", () => {
+    const resource = makeResource({
+      text: makeHighlightableText(""),
+      ocr: {
+        french: makeHighlightableText(""),
+        english: makeHighlightableText("OCR text"),
+      },
+    });
+    expect(getDefaultView(resource)).toBe("ocr.english");
+  });
+
+  test("returns 'preview' when text is empty, no usable OCR, but preview is available", () => {
+    const resource = makeResource({
+      text: makeHighlightableText(""),
+      previewStatus: "pass_through",
+    });
+    expect(getDefaultView(resource)).toBe("preview");
+  });
+
+  test("returns 'preview' when text is empty and all OCR entries are also empty", () => {
+    const resource = makeResource({
+      text: makeHighlightableText(""),
+      ocr: {
+        english: makeHighlightableText(""),
+      },
+      previewStatus: "pdf_generated",
+    });
+    expect(getDefaultView(resource)).toBe("preview");
+  });
+
+  test("returns 'text' as last resort when text is empty and preview is disabled", () => {
+    const resource = makeResource({
+      text: makeHighlightableText(""),
+      previewStatus: "disabled",
+    });
+    expect(getDefaultView(resource)).toBe("text");
+  });
+
+  test("returns undefined when resource.text is undefined", () => {
+    const resource = makeResource({
+      text: undefined as unknown as HighlightableText,
+    });
+    expect(getDefaultView(resource)).toBeUndefined();
+  });
+});

frontend/src/js/util/resourceUtils.ts

@@ -77,6 +77,10 @@ export function getCurrentResource(prefix: string): string {
   return window.location.pathname.split("/").slice(2).join("/");
 }
 
+export function hasTextContent(resource: Resource): boolean {
+  return resource.text.contents.trim() !== "";
+}
+
 export function getDefaultView(resource: Resource): string | undefined {
   if (resource.type !== "blob") {
     return undefined;
@@ -93,17 +97,22 @@ export function getDefaultView(resource: Resource): string | undefined {
     return undefined;
   }
 
-  if (resource.text.contents.trim().length === 0 && resource.ocr) {
+  if (!hasTextContent(resource) && resource.ocr) {
     const ocrEntries = Object.entries(resource.ocr);
     const firstNonEmptyEntry = ocrEntries.find(
-      ([_, { contents }]) => contents.trim().length > 0,
+      ([_, { contents }]) => contents.trim() !== "",
     );
 
     if (firstNonEmptyEntry) {
       return `ocr.${firstNonEmptyEntry[0]}`;
     }
   }
 
+  // If text is empty and there's no usable OCR, prefer preview over an empty text view
+  if (!hasTextContent(resource) && resource.previewStatus !== "disabled") {
+    return "preview";
+  }
+
   return "text";
 }