From 93e5417a4e1dfcdbe7a2a9adab9ae7d917445681 Mon Sep 17 00:00:00 2001
From: Ujjwal Sharma <ryzokuken@igalia.com>
Date: Wed, 11 Dec 2024 00:09:32 +0100
Subject: [PATCH] make processLinks sync by using textContent from textLayer

---
 web/autolinker.js          | 41 +++++++++++++++++---------------------
 web/pdf_find_controller.js | 31 ++++++++++------------------
 web/pdf_page_view.js       | 11 ++++++----
 web/text_layer_builder.js  |  4 +++-
 4 files changed, 39 insertions(+), 48 deletions(-)

diff --git a/web/autolinker.js b/web/autolinker.js
index 9932fb2db173e1..80d94fc555cf7a 100644
--- a/web/autolinker.js
+++ b/web/autolinker.js
@@ -1,8 +1,5 @@
 import { createValidAbsoluteUrl, Util } from "../src/shared/util.js";
-import {
-  getOriginalIndex,
-  normalizedTextContent,
-} from "./pdf_find_controller.js";
+import { getOriginalIndex, normalize } from "./pdf_find_controller.js";
 
 class Autolinker {
   static #urlRegex =
@@ -71,25 +68,23 @@ class Autolinker {
     return linkAnnotations;
   }
 
-  static processLinks(pdfPageView) {
-    return pdfPageView.pdfPage.getTextContent().then(content => {
-      const [text, diffs] = normalizedTextContent(content);
-      const matches = text.matchAll(Autolinker.#urlRegex);
-      return Array.from(matches, match => {
-        const url = createValidAbsoluteUrl(match[0]);
-        if (url) {
-          const [index, length] = getOriginalIndex(
-            diffs,
-            match.index,
-            match[0].length
-          );
-          return this.#addLinkAnnotations(url.href, index, length, pdfPageView);
-        }
-        return url;
-      })
-        .filter(annotation => annotation !== null)
-        .flat();
-    });
+  static processLinks(pdfPageView, textContent) {
+    const [text, diffs] = normalize(textContent.join(""));
+    const matches = text.matchAll(Autolinker.#urlRegex);
+    return Array.from(matches, match => {
+      const url = createValidAbsoluteUrl(match[0]);
+      if (url) {
+        const [index, length] = getOriginalIndex(
+          diffs,
+          match.index,
+          match[0].length
+        );
+        return this.#addLinkAnnotations(url.href, index, length, pdfPageView);
+      }
+      return url;
+    })
+      .filter(annotation => annotation !== null)
+      .flat();
   }
 }
 
diff --git a/web/pdf_find_controller.js b/web/pdf_find_controller.js
index 56692bce9245de..3cec7702f2a398 100644
--- a/web/pdf_find_controller.js
+++ b/web/pdf_find_controller.js
@@ -384,19 +384,6 @@ function getOriginalIndex(diffs, pos, len) {
   return [oldStart, oldLen];
 }
 
-function normalizedTextContent(textContent) {
-  const strBuf = [];
-
-  for (const textItem of textContent.items) {
-    strBuf.push(textItem.str);
-    if (textItem.hasEOL) {
-      strBuf.push("\n");
-    }
-  }
-
-  return normalize(strBuf.join(""));
-}
-
 /**
  * @typedef {Object} PDFFindControllerOptions
  * @property {IPDFLinkService} linkService - The navigation/linking service.
@@ -892,12 +879,21 @@ class PDFFindController {
           .then(pdfPage => pdfPage.getTextContent(textOptions))
           .then(
             textContent => {
+              const strBuf = [];
+
+              for (const textItem of textContent.items) {
+                strBuf.push(textItem.str);
+                if (textItem.hasEOL) {
+                  strBuf.push("\n");
+                }
+              }
+
               // Store the normalized page content (text items) as one string.
               [
                 this._pageContents[i],
                 this._pageDiffs[i],
                 this._hasDiacritics[i],
-              ] = normalizedTextContent(textContent);
+              ] = normalize(strBuf.join(""));
               resolve();
             },
             reason => {
@@ -1175,9 +1171,4 @@ class PDFFindController {
   }
 }
 
-export {
-  FindState,
-  getOriginalIndex,
-  normalizedTextContent,
-  PDFFindController,
-};
+export { FindState, getOriginalIndex, normalize, PDFFindController };
diff --git a/web/pdf_page_view.js b/web/pdf_page_view.js
index e92b434a17b77d..8ac0d14aea4c06 100644
--- a/web/pdf_page_view.js
+++ b/web/pdf_page_view.js
@@ -463,15 +463,16 @@ class PDFPageView {
 
   async #renderTextLayer() {
     if (!this.textLayer) {
-      return;
+      return [];
     }
 
     let error = null;
+    let textContent;
     try {
-      await this.textLayer.render(this.viewport);
+      textContent = await this.textLayer.render(this.viewport);
     } catch (ex) {
       if (ex instanceof AbortException) {
-        return;
+        return [];
       }
       console.error("#renderTextLayer:", ex);
       error = ex;
@@ -479,6 +480,7 @@ class PDFPageView {
     this.#dispatchLayerRendered("textlayerrendered", error);
 
     this.#renderStructTreeLayer();
+    return textContent;
   }
 
   /**
@@ -1098,7 +1100,8 @@ class PDFPageView {
         if (this.annotationLayer) {
           await textLayerP;
           if (this.#enableAutolinking) {
-            this.#linkAnnotations = await Autolinker.processLinks(this);
+            const textContent = await textLayerP;
+            this.#linkAnnotations = Autolinker.processLinks(this, textContent);
           }
           await this.#renderAnnotationLayer();
         }
diff --git a/web/text_layer_builder.js b/web/text_layer_builder.js
index f642382b606431..89adcd967c9841 100644
--- a/web/text_layer_builder.js
+++ b/web/text_layer_builder.js
@@ -72,6 +72,7 @@ class TextLayerBuilder {
    * Renders the text layer.
    * @param {PageViewport} viewport
    * @param {Object} [textContentParams]
+   * @returns {Array<string>}
    */
   async render(viewport, textContentParams = null) {
     if (this.#renderingDone && this.#textLayer) {
@@ -80,7 +81,7 @@ class TextLayerBuilder {
         onBefore: this.hide.bind(this),
       });
       this.show();
-      return;
+      return [];
     }
 
     this.cancel();
@@ -112,6 +113,7 @@ class TextLayerBuilder {
     this.#onAppend?.(this.div);
     this.highlighter?.enable();
     this.accessibilityManager?.enable();
+    return textContentItemsStr;
   }
 
   hide() {