From 049848ba009014d5056dd79676028cec91bd2a5c Mon Sep 17 00:00:00 2001
From: Jonas Jenwald <jonas.jenwald@gmail.com>
Date: Sun, 21 Apr 2024 11:29:03 +0200
Subject: [PATCH] Unify the `ReadableStream` and `TextContent` code-paths in
 `src/display/text_layer.js`

The only reason that this code still accepts `TextContent` is for backward-compatibility purposes, so we can simplify the implementation by always using a `ReadableStream` internally.
---
 src/display/text_layer.js    | 83 ++++++++++++++++++------------------
 test/driver.js               | 14 +++---
 test/unit/text_layer_spec.js | 42 ++++++++++++++++++
 3 files changed, 88 insertions(+), 51 deletions(-)

diff --git a/src/display/text_layer.js b/src/display/text_layer.js
index 5394f0dc2..03a139cdf 100644
--- a/src/display/text_layer.js
+++ b/src/display/text_layer.js
@@ -250,9 +250,7 @@ function appendText(task, geom, styles) {
     textDivProperties.canvasWidth = style.vertical ? geom.height : geom.width;
   }
   task._textDivProperties.set(textDiv, textDivProperties);
-  if (task._isReadableStream) {
-    task._layoutText(textDiv);
-  }
+  task._layoutText(textDiv);
 }
 
 function layout(params) {
@@ -298,16 +296,14 @@ function render(task) {
     capability.resolve();
     return;
   }
-
-  if (!task._isReadableStream) {
-    for (const textDiv of textDivs) {
-      task._layoutText(textDiv);
-    }
-  }
   capability.resolve();
 }
 
 class TextLayerRenderTask {
+  #reader = null;
+
+  #textContentSource = null;
+
   constructor({
     textContentSource,
     container,
@@ -316,14 +312,26 @@ class TextLayerRenderTask {
     textDivProperties,
     textContentItemsStr,
   }) {
-    this._textContentSource = textContentSource;
-    this._isReadableStream = textContentSource instanceof ReadableStream;
+    if (textContentSource instanceof ReadableStream) {
+      this.#textContentSource = textContentSource;
+    } else if (
+      (typeof PDFJSDev === "undefined" || PDFJSDev.test("GENERIC")) &&
+      typeof textContentSource === "object"
+    ) {
+      this.#textContentSource = new ReadableStream({
+        start(controller) {
+          controller.enqueue(textContentSource);
+          controller.close();
+        },
+      });
+    } else {
+      throw new Error('No "textContentSource" parameter specified.');
+    }
     this._container = this._rootContainer = container;
     this._textDivs = textDivs || [];
     this._textContentItemsStr = textContentItemsStr || [];
     this._fontInspectorEnabled = !!globalThis.FontInspector?.enabled;
 
-    this._reader = null;
     this._textDivProperties = textDivProperties || new WeakMap();
     this._canceled = false;
     this._capability = Promise.withResolvers();
@@ -365,15 +373,14 @@ class TextLayerRenderTask {
    */
   cancel() {
     this._canceled = true;
-    if (this._reader) {
-      this._reader
-        .cancel(new AbortException("TextLayer task cancelled."))
-        .catch(() => {
-          // Avoid "Uncaught promise" messages in the console.
-        });
-      this._reader = null;
-    }
-    this._capability.reject(new AbortException("TextLayer task cancelled."));
+    const abortEx = new AbortException("TextLayer task cancelled.");
+
+    this.#reader?.cancel(abortEx).catch(() => {
+      // Avoid "Uncaught promise" messages in the console.
+    });
+    this.#reader = null;
+
+    this._capability.reject(abortEx);
   }
 
   /**
@@ -429,29 +436,21 @@ class TextLayerRenderTask {
     const { promise, resolve, reject } = Promise.withResolvers();
     let styleCache = Object.create(null);
 
-    if (this._isReadableStream) {
-      const pump = () => {
-        this._reader.read().then(({ value, done }) => {
-          if (done) {
-            resolve();
-            return;
-          }
+    const pump = () => {
+      this.#reader.read().then(({ value, done }) => {
+        if (done) {
+          resolve();
+          return;
+        }
 
-          Object.assign(styleCache, value.styles);
-          this._processItems(value.items, styleCache);
-          pump();
-        }, reject);
-      };
+        Object.assign(styleCache, value.styles);
+        this._processItems(value.items, styleCache);
+        pump();
+      }, reject);
+    };
 
-      this._reader = this._textContentSource.getReader();
-      pump();
-    } else if (this._textContentSource) {
-      const { items, styles } = this._textContentSource;
-      this._processItems(items, styles);
-      resolve();
-    } else {
-      throw new Error('No "textContentSource" parameter specified.');
-    }
+    this.#reader = this.#textContentSource.getReader();
+    pump();
 
     promise.then(() => {
       styleCache = null;
diff --git a/test/driver.js b/test/driver.js
index e47a48a5d..d0a91ce48 100644
--- a/test/driver.js
+++ b/test/driver.js
@@ -335,15 +335,11 @@ class Rasterize {
 
       await task.promise;
 
-      const { _pageWidth, _pageHeight, _textContentSource, _textDivs } = task;
+      const { _pageWidth, _pageHeight, _textDivs } = task;
       const boxes = [];
-      let posRegex;
-      for (
-        let i = 0, j = 0, ii = _textContentSource.items.length;
-        i < ii;
-        i++
-      ) {
-        const { width, height, type } = _textContentSource.items[i];
+      let j = 0,
+        posRegex;
+      for (const { width, height, type } of textContent.items) {
         if (type) {
           continue;
         }
@@ -396,7 +392,7 @@ class Rasterize {
 
       drawLayer.destroy();
     } catch (reason) {
-      throw new Error(`Rasterize.textLayer: "${reason?.message}".`);
+      throw new Error(`Rasterize.highlightLayer: "${reason?.message}".`);
     }
   }
 
diff --git a/test/unit/text_layer_spec.js b/test/unit/text_layer_spec.js
index ca4a400e9..b6d93eab7 100644
--- a/test/unit/text_layer_spec.js
+++ b/test/unit/text_layer_spec.js
@@ -58,5 +58,47 @@ describe("textLayer", function () {
       "",
       "page 1 / 3",
     ]);
+
+    await loadingTask.destroy();
+  });
+
+  it("creates textLayer from TextContent", async function () {
+    if (isNodeJS) {
+      pending("document.createElement is not supported in Node.js.");
+    }
+    const loadingTask = getDocument(buildGetDocumentParams("basicapi.pdf"));
+    const pdfDocument = await loadingTask.promise;
+    const page = await pdfDocument.getPage(1);
+
+    const textContentItemsStr = [];
+
+    const textLayerRenderTask = renderTextLayer({
+      textContentSource: await page.getTextContent(),
+      container: document.createElement("div"),
+      viewport: page.getViewport({ scale: 1 }),
+      textContentItemsStr,
+    });
+    expect(textLayerRenderTask instanceof TextLayerRenderTask).toEqual(true);
+
+    await textLayerRenderTask.promise;
+    expect(textContentItemsStr).toEqual([
+      "Table Of Content",
+      "",
+      "Chapter 1",
+      " ",
+      "..........................................................",
+      " ",
+      "2",
+      "",
+      "Paragraph 1.1",
+      " ",
+      "......................................................",
+      " ",
+      "3",
+      "",
+      "page 1 / 3",
+    ]);
+
+    await loadingTask.destroy();
   });
 });