Enable automatic URL linking (bug 1019475)

test/pdfs/.gitignore

@@ -692,3 +692,4 @@
 !issue19182.pdf
 !issue18911.pdf
 !issue19207.pdf
+!link.pdf

web/annotation_layer_builder.js

@@ -25,7 +25,7 @@
 // eslint-disable-next-line max-len
 /** @typedef {import("../src/display/editor/tools.js").AnnotationEditorUIManager} AnnotationEditorUIManager */
 
-import { AnnotationLayer } from "pdfjs-lib";
+import { AnnotationLayer, Util } from "pdfjs-lib";
 import { PresentationModeState } from "./ui_utils.js";
 
 /**
@@ -97,7 +97,7 @@ class AnnotationLayerBuilder {
    * @returns {Promise<void>} A promise that is resolved when rendering of the
    *   annotations is complete.
    */
-  async render(viewport, options, intent = "display") {
+  async render(viewport, options, intent = "display", linkAnnotations) {
     if (this.div) {
       if (this._cancelled || !this.annotationLayer) {
         return;
@@ -119,6 +119,25 @@ class AnnotationLayerBuilder {
       return;
     }
 
+    const uniqueLinks = linkAnnotations.filter(link => {
+      for (const annotation of annotations) {
+        function area(rect) {
+          return Math.abs(rect[2] - rect[0]) * Math.abs(rect[3] - rect[1]);
+        }
+        const intersect = Util.intersect(annotation.rect, link.rect); // Find the intersection between the annotation and the link.
+        if (
+          annotation.subtype === "Link" &&
+          annotation.url === link.url &&
+          intersect !== null &&
+          area(intersect) / area(link.rect) > 0.5 // If the overlap is more than 50%.
+        ) {
+          return false;
+        }
+      }
+      return true;
+    });
+    annotations.push(...uniqueLinks);
+
     // Create an annotation layer div and render the annotations
     // if there is at least one annotation.
     const div = (this.div = document.createElement("div"));

web/app.js

@@ -493,6 +493,7 @@ const PDFViewerApplication = {
       abortSignal: this._globalAbortController.signal,
       enableHWA,
       supportsPinchToZoom: this.supportsPinchToZoom,
+      enableAutolinking: AppOptions.get("enableAutolinking"),
     });
     this.pdfViewer = pdfViewer;
 

web/app_options.js

@@ -236,6 +236,12 @@ const defaultOptions = {
     value: false,
     kind: OptionKind.VIEWER + OptionKind.PREFERENCE,
   },
+  enableAutolinking: {
+    // TODO: remove it when unnecessary.
+    /** @type {boolean} */
+    value: false,
+    kind: OptionKind.VIEWER,
+  },
   externalLinkRel: {
     /** @type {string} */
     value: "noopener noreferrer nofollow",

web/autolinker.js

@@ -0,0 +1,94 @@
+import { createValidAbsoluteUrl, Util } from "../src/shared/util.js";
+import { getOriginalIndex, normalize } from "./pdf_find_controller.js";
+
+class Autolinker {
+  static #urlRegex =
+    /\b(?:https?:\/\/|mailto:|www.)(?:[[\S--\[]--\p{P}]|\/|[\p{P}--\[]+[[\S--\[]--\p{P}])+/gmv; // Regex can be tested and verified at https://regex101.com/r/zuMeQX/3.
+
+  static #addLinkAnnotations(url, index, length, pdfPageView) {
+    // TODO refactor out the logic for a single match from this function
+    const convertedMatch = pdfPageView._textHighlighter._convertMatches(
+      [index],
+      [length]
+    )[0];
+
+    const range = new Range();
+    range.setStart(
+      pdfPageView._textHighlighter.textDivs[convertedMatch.begin.divIdx]
+        .firstChild,
+      convertedMatch.begin.offset
+    );
+    range.setEnd(
+      pdfPageView._textHighlighter.textDivs[convertedMatch.end.divIdx]
+        .firstChild,
+      convertedMatch.end.offset
+    );
+
+    const pageBox = pdfPageView.textLayer.div.getBoundingClientRect();
+    const linkAnnotations = [];
+    for (const linkBox of range.getClientRects()) {
+      if (linkBox.width === 0 || linkBox.height === 0) {
+        continue;
+      }
+
+      const bottomLeft = pdfPageView.getPagePoint(
+        linkBox.left - pageBox.left,
+        linkBox.top - pageBox.top
+      );
+      const topRight = pdfPageView.getPagePoint(
+        linkBox.left - pageBox.left + linkBox.width,
+        linkBox.top - pageBox.top + linkBox.height
+      );
+
+      const rect = Util.normalizeRect([
+        bottomLeft[0],
+        bottomLeft[1],
+        topRight[0],
+        topRight[1],
+      ]);
+
+      linkAnnotations.push({
+        unsafeUrl: url,
+        url,
+        rect,
+        annotationType: 2,
+        rotation: 0,
+        // This is just the default for AnnotationBorderStyle. At some point we
+        // should switch to something better like `new LinkAnnotation` here.
+        borderStyle: {
+          width: 1,
+          rawWidth: 1,
+          style: 1, // SOLID
+          dashArray: [3],
+          horizontalCornerRadius: 0,
+          verticalCornerRadius: 0,
+        },
+      });
+    }
+    return linkAnnotations;
+  }
+
+  static processLinks(pdfPageView) {
+    const [text, diffs] = normalize(
+      pdfPageView._textHighlighter.textContentItemsStr.join("\n")
+    );
+    const matches = text.matchAll(Autolinker.#urlRegex);
+    const links = [];
+    for (const match of matches) {
+      const url = createValidAbsoluteUrl(match[0]);
+      if (url) {
+        const [index, length] = getOriginalIndex(
+          diffs,
+          match.index,
+          match[0].length
+        );
+        links.push(
+          ...this.#addLinkAnnotations(url.href, index, length, pdfPageView)
+        );
+      }
+    }
+    return links;
+  }
+}
+
+export { Autolinker };

web/pdf_find_controller.js

@@ -1171,4 +1171,4 @@ class PDFFindController {
   }
 }
 
-export { FindState, PDFFindController };
+export { FindState, getOriginalIndex, normalize, PDFFindController };

web/pdf_page_view.js

@@ -43,6 +43,7 @@ import {
 import { AnnotationEditorLayerBuilder } from "./annotation_editor_layer_builder.js";
 import { AnnotationLayerBuilder } from "./annotation_layer_builder.js";
 import { AppOptions } from "./app_options.js";
+import { Autolinker } from "./autolinker.js";
 import { DrawLayerBuilder } from "./draw_layer_builder.js";
 import { GenericL10n } from "web-null_l10n";
 import { SimpleLinkService } from "./pdf_link_service.js";
@@ -120,6 +121,8 @@ class PDFPageView {
 
   #enableHWA = false;
 
+  #enableAutolinking = false;
+
   #hasRestrictedScaling = false;
 
   #isEditing = false;
@@ -148,6 +151,8 @@ class PDFPageView {
     regularAnnotations: true,
   };
 
+  #linkAnnotations = [];
+
   #layers = [null, null, null, null];
 
   /**
@@ -177,6 +182,7 @@ class PDFPageView {
       options.maxCanvasPixels ?? AppOptions.get("maxCanvasPixels");
     this.pageColors = options.pageColors || null;
     this.#enableHWA = options.enableHWA || false;
+    this.#enableAutolinking = options.enableAutolinking || false;
 
     this.eventBus = options.eventBus;
     this.renderingQueue = options.renderingQueue;
@@ -399,7 +405,8 @@ class PDFPageView {
       await this.annotationLayer.render(
         this.viewport,
         { structTreeLayer: this.structTreeLayer },
-        "display"
+        "display",
+        this.#linkAnnotations
       );
     } catch (ex) {
       console.error("#renderAnnotationLayer:", ex);
@@ -1086,9 +1093,13 @@ class PDFPageView {
           viewport.rawDims
         );
 
-        this.#renderTextLayer();
+        const textLayerP = this.#renderTextLayer();
 
         if (this.annotationLayer) {
+          if (this.#enableAutolinking) {
+            await textLayerP;
+            this.#linkAnnotations = Autolinker.processLinks(this);
+          }
           await this.#renderAnnotationLayer();
         }
 

web/pdf_viewer.js

@@ -228,6 +228,8 @@ class PDFViewer {
 
   #enableNewAltTextWhenAddingImage = false;
 
+  #enableAutolinking = false;
+
   #eventAbortController = null;
 
   #mlManager = null;
@@ -321,6 +323,7 @@ class PDFViewer {
     this.#mlManager = options.mlManager || null;
     this.#enableHWA = options.enableHWA || false;
     this.#supportsPinchToZoom = options.supportsPinchToZoom !== false;
+    this.#enableAutolinking = options.enableAutolinking || false;
 
     this.defaultRenderingQueue = !options.renderingQueue;
     if (
@@ -990,6 +993,7 @@ class PDFViewer {
             l10n: this.l10n,
             layerProperties: this._layerProperties,
             enableHWA: this.#enableHWA,
+            enableAutolinking: this.#enableAutolinking,
           });
           this._pages.push(pageView);
         }