Support rich content in markup annotation

- use the xfa parser but in the xhtml namespace.

src/core/annotation.js

@@ -55,6 +55,7 @@ import { ObjectLoader } from "./object_loader.js";
 import { OperatorList } from "./operator_list.js";
 import { StringStream } from "./stream.js";
 import { writeDict } from "./writer.js";
+import { XFAFactory } from "./xfa/factory.js";
 
 class AnnotationFactory {
   /**
@@ -1098,6 +1099,10 @@ class MarkupAnnotation extends Annotation {
         this.data.color = null;
       }
     }
+
+    if (dict.has("RC")) {
+      this.data.richText = XFAFactory.getRichTextAsHtml(dict.get("RC"));
+    }
   }
 
   /**
@@ -2545,6 +2550,10 @@ class PopupAnnotation extends Annotation {
 
     this.setContents(parentItem.get("Contents"));
     this.data.contentsObj = this._contents;
+
+    if (parentItem.has("RC")) {
+      this.data.richText = XFAFactory.getRichTextAsHtml(parentItem.get("RC"));
+    }
   }
 }
 

src/core/xfa/builder.js

@@ -66,7 +66,7 @@ class Empty extends XFAObject {
 }
 
 class Builder {
-  constructor() {
+  constructor(rootNameSpace = null) {
     this._namespaceStack = [];
     this._nsAgnosticLevel = 0;
 
@@ -76,7 +76,8 @@ class Builder {
     this._nextNsId = Math.max(
       ...Object.values(NamespaceIds).map(({ id }) => id)
     );
-    this._currentNamespace = new UnknownNamespace(++this._nextNsId);
+    this._currentNamespace =
+      rootNameSpace || new UnknownNamespace(++this._nextNsId);
   }
 
   buildRoot(ids) {

src/core/xfa/factory.js

@@ -13,13 +13,20 @@
  * limitations under the License.
  */
 
-import { $globalData, $toHTML } from "./xfa_object.js";
+import {
+  $appendChild,
+  $globalData,
+  $nodeName,
+  $text,
+  $toHTML,
+} from "./xfa_object.js";
 import { Binder } from "./bind.js";
 import { DataHandler } from "./data.js";
 import { FontFinder } from "./fonts.js";
 import { stripQuotes } from "./utils.js";
 import { warn } from "../../shared/util.js";
 import { XFAParser } from "./parser.js";
+import { XhtmlNamespace } from "./xhtml.js";
 
 class XFAFactory {
   constructor(data) {
@@ -106,6 +113,43 @@ class XFAFactory {
     }
     return Object.values(data).join("");
   }
+
+  static getRichTextAsHtml(rc) {
+    if (!rc || typeof rc !== "string") {
+      return null;
+    }
+
+    try {
+      let root = new XFAParser(XhtmlNamespace, /* richText */ true).parse(rc);
+      if (!["body", "xhtml"].includes(root[$nodeName])) {
+        // No body, so create one.
+        const newRoot = XhtmlNamespace.body({});
+        newRoot[$appendChild](root);
+        root = newRoot;
+      }
+
+      const result = root[$toHTML]();
+      if (!result.success) {
+        return null;
+      }
+
+      const { html } = result;
+      const { attributes } = html;
+      if (attributes) {
+        if (attributes.class) {
+          attributes.class = attributes.class.filter(
+            attr => !attr.startsWith("xfa")
+          );
+        }
+        attributes.dir = "auto";
+      }
+
+      return { html, str: root[$text]() };
+    } catch (e) {
+      warn(`XFA - an error occurred during parsing of rich text: ${e}`);
+    }
+    return null;
+  }
 }
 
 export { XFAFactory };

src/core/xfa/html_utils.js

@@ -606,10 +606,16 @@ function setPara(node, nodeStyle, value) {
 }
 
 function setFontFamily(xfaFont, node, fontFinder, style) {
-  const name = stripQuotes(xfaFont.typeface);
-  const typeface = fontFinder.find(name);
+  if (!fontFinder) {
+    // The font cannot be found in the pdf so use the default one.
+    delete style.fontFamily;
+    return;
+  }
 
+  const name = stripQuotes(xfaFont.typeface);
   style.fontFamily = `"${name}"`;
+
+  const typeface = fontFinder.find(name);
   if (typeface) {
     const { fontFamily } = typeface.regular.cssFontInfo;
     if (fontFamily !== name) {

src/core/xfa/parser.js

@@ -30,9 +30,9 @@ import { Builder } from "./builder.js";
 import { warn } from "../../shared/util.js";
 
 class XFAParser extends XMLParserBase {
-  constructor() {
+  constructor(rootNameSpace = null, richText = false) {
     super();
-    this._builder = new Builder();
+    this._builder = new Builder(rootNameSpace);
     this._stack = [];
     this._globalData = {
       usedTypefaces: new Set(),
@@ -42,6 +42,7 @@ class XFAParser extends XMLParserBase {
     this._errorCode = XMLParserErrorCode.NoError;
     this._whiteRegex = /^\s+$/;
     this._nbsps = /\xa0+/g;
+    this._richText = richText;
   }
 
   parse(data) {
@@ -60,8 +61,8 @@ class XFAParser extends XMLParserBase {
     // Normally by definition a &nbsp is unbreakable
     // but in real life Acrobat can break strings on &nbsp.
     text = text.replace(this._nbsps, match => match.slice(1) + " ");
-    if (this._current[$acceptWhitespace]()) {
-      this._current[$onText](text);
+    if (this._richText || this._current[$acceptWhitespace]()) {
+      this._current[$onText](text, this._richText);
       return;
     }
 

src/core/xfa/xhtml.js

@@ -20,6 +20,7 @@ import {
   $content,
   $extra,
   $getChildren,
+  $getParent,
   $globalData,
   $nodeName,
   $onText,
@@ -38,6 +39,7 @@ import {
 import { getMeasurement, HTMLResult, stripQuotes } from "./utils.js";
 
 const XHTML_NS_ID = NamespaceIds.xhtml.id;
+const $richText = Symbol();
 
 const VALID_STYLES = new Set([
   "color",
@@ -109,6 +111,7 @@ const StyleMapping = new Map([
 
 const spacesRegExp = /\s+/g;
 const crlfRegExp = /[\r\n]+/g;
+const crlfForRichTextRegExp = /\r\n?/g;
 
 function mapStyle(styleStr, node) {
   const style = Object.create(null);
@@ -185,6 +188,7 @@ const NoWhites = new Set(["body", "html"]);
 class XhtmlObject extends XmlObject {
   constructor(attributes, name) {
     super(XHTML_NS_ID, name);
+    this[$richText] = false;
     this.style = attributes.style || "";
   }
 
@@ -197,11 +201,16 @@ class XhtmlObject extends XmlObject {
     return !NoWhites.has(this[$nodeName]);
   }
 
-  [$onText](str) {
-    str = str.replace(crlfRegExp, "");
-    if (!this.style.includes("xfa-spacerun:yes")) {
-      str = str.replace(spacesRegExp, " ");
+  [$onText](str, richText = false) {
+    if (!richText) {
+      str = str.replace(crlfRegExp, "");
+      if (!this.style.includes("xfa-spacerun:yes")) {
+        str = str.replace(spacesRegExp, " ");
+      }
+    } else {
+      this[$richText] = true;
     }
+
     if (str) {
       this[$content] += str;
     }
@@ -311,6 +320,15 @@ class XhtmlObject extends XmlObject {
       return HTMLResult.EMPTY;
     }
 
+    let value;
+    if (this[$richText]) {
+      value = this[$content]
+        ? this[$content].replace(crlfForRichTextRegExp, "\n")
+        : undefined;
+    } else {
+      value = this[$content] || undefined;
+    }
+
     return HTMLResult.success({
       name: this[$nodeName],
       attributes: {
@@ -318,7 +336,7 @@ class XhtmlObject extends XmlObject {
         style: mapStyle(this.style, this),
       },
       children,
-      value: this[$content] || "",
+      value,
     });
   }
 }
@@ -457,6 +475,10 @@ class P extends XhtmlObject {
   }
 
   [$text]() {
+    const siblings = this[$getParent]()[$getChildren]();
+    if (siblings[siblings.length - 1] === this) {
+      return super[$text]();
+    }
     return super[$text]() + "\n";
   }
 }

src/display/annotation_layer.js

@@ -30,6 +30,7 @@ import {
 } from "./display_utils.js";
 import { AnnotationStorage } from "./annotation_storage.js";
 import { ColorConverters } from "../shared/scripting_utils.js";
+import { XfaLayer } from "./xfa_layer.js";
 
 const DEFAULT_TAB_INDEX = 1000;
 const GetElementsByNameSet = new WeakSet();
@@ -322,6 +323,7 @@ class AnnotationElement {
       titleObj: data.titleObj,
       modificationDate: data.modificationDate,
       contentsObj: data.contentsObj,
+      richText: data.richText,
       hideWrapper: true,
     });
     const popup = popupElement.render();
@@ -676,7 +678,8 @@ class TextAnnotationElement extends AnnotationElement {
     const isRenderable = !!(
       parameters.data.hasPopup ||
       parameters.data.titleObj?.str ||
-      parameters.data.contentsObj?.str
+      parameters.data.contentsObj?.str ||
+      parameters.data.richText?.str
     );
     super(parameters, { isRenderable });
   }
@@ -1546,7 +1549,9 @@ class ChoiceWidgetAnnotationElement extends WidgetAnnotationElement {
 class PopupAnnotationElement extends AnnotationElement {
   constructor(parameters) {
     const isRenderable = !!(
-      parameters.data.titleObj?.str || parameters.data.contentsObj?.str
+      parameters.data.titleObj?.str ||
+      parameters.data.contentsObj?.str ||
+      parameters.data.richText?.str
     );
     super(parameters, { isRenderable });
   }
@@ -1582,6 +1587,7 @@ class PopupAnnotationElement extends AnnotationElement {
       titleObj: this.data.titleObj,
       modificationDate: this.data.modificationDate,
       contentsObj: this.data.contentsObj,
+      richText: this.data.richText,
     });
 
     // Position the popup next to the parent annotation's container.
@@ -1614,6 +1620,7 @@ class PopupElement {
     this.titleObj = parameters.titleObj;
     this.modificationDate = parameters.modificationDate;
     this.contentsObj = parameters.contentsObj;
+    this.richText = parameters.richText;
     this.hideWrapper = parameters.hideWrapper || false;
 
     this.pinned = false;
@@ -1655,6 +1662,7 @@ class PopupElement {
     const dateObject = PDFDateString.toDateObject(this.modificationDate);
     if (dateObject) {
       const modificationDate = document.createElement("span");
+      modificationDate.className = "popupDate";
       modificationDate.textContent = "{{date}}, {{time}}";
       modificationDate.dataset.l10nId = "annotation_date_string";
       modificationDate.dataset.l10nArgs = JSON.stringify({
@@ -1664,8 +1672,20 @@ class PopupElement {
       popup.appendChild(modificationDate);
     }
 
-    const contents = this._formatContents(this.contentsObj);
-    popup.appendChild(contents);
+    if (
+      this.richText?.str &&
+      (!this.contentsObj?.str || this.contentsObj.str === this.richText.str)
+    ) {
+      XfaLayer.render({
+        xfa: this.richText.html,
+        intent: "richText",
+        div: popup,
+      });
+      popup.lastChild.className = "richText popupContent";
+    } else {
+      const contents = this._formatContents(this.contentsObj);
+      popup.appendChild(contents);
+    }
 
     if (!Array.isArray(this.trigger)) {
       this.trigger = [this.trigger];
@@ -1693,6 +1713,7 @@ class PopupElement {
    */
   _formatContents({ str, dir }) {
     const p = document.createElement("p");
+    p.className = "popupContent";
     p.dir = dir;
     const lines = str.split(/(?:\r\n?|\n)/);
     for (let i = 0, ii = lines.length; i < ii; ++i) {
@@ -1759,7 +1780,8 @@ class FreeTextAnnotationElement extends AnnotationElement {
     const isRenderable = !!(
       parameters.data.hasPopup ||
       parameters.data.titleObj?.str ||
-      parameters.data.contentsObj?.str
+      parameters.data.contentsObj?.str ||
+      parameters.data.richText?.str
     );
     super(parameters, { isRenderable, ignoreBorder: true });
   }
@@ -1779,7 +1801,8 @@ class LineAnnotationElement extends AnnotationElement {
     const isRenderable = !!(
       parameters.data.hasPopup ||
       parameters.data.titleObj?.str ||
-      parameters.data.contentsObj?.str
+      parameters.data.contentsObj?.str ||
+      parameters.data.richText?.str
     );
     super(parameters, { isRenderable, ignoreBorder: true });
   }
@@ -1824,7 +1847,8 @@ class SquareAnnotationElement extends AnnotationElement {
     const isRenderable = !!(
       parameters.data.hasPopup ||
       parameters.data.titleObj?.str ||
-      parameters.data.contentsObj?.str
+      parameters.data.contentsObj?.str ||
+      parameters.data.richText?.str
     );
     super(parameters, { isRenderable, ignoreBorder: true });
   }
@@ -1871,7 +1895,8 @@ class CircleAnnotationElement extends AnnotationElement {
     const isRenderable = !!(
       parameters.data.hasPopup ||
       parameters.data.titleObj?.str ||
-      parameters.data.contentsObj?.str
+      parameters.data.contentsObj?.str ||
+      parameters.data.richText?.str
     );
     super(parameters, { isRenderable, ignoreBorder: true });
   }
@@ -1918,7 +1943,8 @@ class PolylineAnnotationElement extends AnnotationElement {
     const isRenderable = !!(
       parameters.data.hasPopup ||
       parameters.data.titleObj?.str ||
-      parameters.data.contentsObj?.str
+      parameters.data.contentsObj?.str ||
+      parameters.data.richText?.str
     );
     super(parameters, { isRenderable, ignoreBorder: true });
 
@@ -1983,7 +2009,8 @@ class CaretAnnotationElement extends AnnotationElement {
     const isRenderable = !!(
       parameters.data.hasPopup ||
       parameters.data.titleObj?.str ||
-      parameters.data.contentsObj?.str
+      parameters.data.contentsObj?.str ||
+      parameters.data.richText?.str
     );
     super(parameters, { isRenderable, ignoreBorder: true });
   }
@@ -2003,7 +2030,8 @@ class InkAnnotationElement extends AnnotationElement {
     const isRenderable = !!(
       parameters.data.hasPopup ||
       parameters.data.titleObj?.str ||
-      parameters.data.contentsObj?.str
+      parameters.data.contentsObj?.str ||
+      parameters.data.richText?.str
     );
     super(parameters, { isRenderable, ignoreBorder: true });
 
@@ -2062,7 +2090,8 @@ class HighlightAnnotationElement extends AnnotationElement {
     const isRenderable = !!(
       parameters.data.hasPopup ||
       parameters.data.titleObj?.str ||
-      parameters.data.contentsObj?.str
+      parameters.data.contentsObj?.str ||
+      parameters.data.richText?.str
     );
     super(parameters, {
       isRenderable,
@@ -2090,7 +2119,8 @@ class UnderlineAnnotationElement extends AnnotationElement {
     const isRenderable = !!(
       parameters.data.hasPopup ||
       parameters.data.titleObj?.str ||
-      parameters.data.contentsObj?.str
+      parameters.data.contentsObj?.str ||
+      parameters.data.richText?.str
     );
     super(parameters, {
       isRenderable,
@@ -2118,7 +2148,8 @@ class SquigglyAnnotationElement extends AnnotationElement {
     const isRenderable = !!(
       parameters.data.hasPopup ||
       parameters.data.titleObj?.str ||
-      parameters.data.contentsObj?.str
+      parameters.data.contentsObj?.str ||
+      parameters.data.richText?.str
     );
     super(parameters, {
       isRenderable,
@@ -2146,7 +2177,8 @@ class StrikeOutAnnotationElement extends AnnotationElement {
     const isRenderable = !!(
       parameters.data.hasPopup ||
       parameters.data.titleObj?.str ||
-      parameters.data.contentsObj?.str
+      parameters.data.contentsObj?.str ||
+      parameters.data.richText?.str
     );
     super(parameters, {
       isRenderable,
@@ -2174,7 +2206,8 @@ class StampAnnotationElement extends AnnotationElement {
     const isRenderable = !!(
       parameters.data.hasPopup ||
       parameters.data.titleObj?.str ||
-      parameters.data.contentsObj?.str
+      parameters.data.contentsObj?.str ||
+      parameters.data.richText?.str
     );
     super(parameters, { isRenderable, ignoreBorder: true });
   }
@@ -2215,7 +2248,9 @@ class FileAttachmentAnnotationElement extends AnnotationElement {
 
     if (
       !this.data.hasPopup &&
-      (this.data.titleObj?.str || this.data.contentsObj?.str)
+      (this.data.titleObj?.str ||
+        this.data.contentsObj?.str ||
+        this.data.richText)
     ) {
       this._createPopup(trigger, this.data);
     }

src/display/xfa_layer.js

@@ -106,7 +106,9 @@ class XfaLayer {
         if (key === "textContent") {
           html.textContent = value;
         } else if (key === "class") {
-          html.setAttribute(key, value.join(" "));
+          if (value.length) {
+            html.setAttribute(key, value.join(" "));
+          }
         } else {
           if (isHTMLAnchorElement && (key === "href" || key === "newWindow")) {
             continue; // Handled below.
@@ -159,11 +161,16 @@ class XfaLayer {
 
     const rootDiv = parameters.div;
     rootDiv.appendChild(rootHtml);
-    const transform = `matrix(${parameters.viewport.transform.join(",")})`;
-    rootDiv.style.transform = transform;
+
+    if (parameters.viewport) {
+      const transform = `matrix(${parameters.viewport.transform.join(",")})`;
+      rootDiv.style.transform = transform;
+    }
 
     // Set defaults.
-    rootDiv.setAttribute("class", "xfaLayer xfaFont");
+    if (intent !== "richText") {
+      rootDiv.setAttribute("class", "xfaLayer xfaFont");
+    }
 
     // Text nodes used for the text highlighter.
     const textDivs = [];

test/pdfs/.gitignore

@@ -479,3 +479,4 @@
 !pr12564.pdf
 !pr12828.pdf
 !secHandler.pdf
+!rc_annotation.pdf

test/pdfs/issue13915.pdf.link

@@ -0,0 +1 @@
+https://github.com/mozilla/pdf.js/files/7024938/2102.12353.pdf

test/test_manifest.json

@@ -137,6 +137,24 @@
        "type": "eq",
        "annotations": true
     },
+    {  "id": "issue2966",
+       "file": "pdfs/rc_annotation.pdf",
+       "md5": "7b978a8c2871b8902656adb67f7bd117",
+       "rounds": 1,
+       "lastPage": 1,
+       "type": "eq",
+       "annotations": true
+    },
+    {  "id": "issue13915",
+      "file": "pdfs/issue13915.pdf",
+      "md5": "fef3108733bbf80ea8551feedb427b1c",
+      "rounds": 1,
+      "firstPage": 51,
+      "lastPage": 51,
+      "link": true,
+      "type": "eq",
+      "annotations": true
+   },
     {  "id": "bug946506",
        "file": "pdfs/bug946506.pdf",
        "md5": "c28911b5c31bdc337c2ce404c5971cfc",

test/unit/xfa_parser_spec.js

@@ -414,7 +414,7 @@ describe("XFAParser", function () {
         [
           " The first line of this paragraph is indented a half-inch.\n",
           " Successive lines are not indented.\n",
-          " This is the last line of the paragraph.\n \n",
+          " This is the last line of the paragraph.\n ",
         ].join("")
       );
     });

web/annotation_layer_builder.css

@@ -192,17 +192,21 @@
   display: inline-block;
 }
 
-.annotationLayer .popup span {
+.annotationLayer .popupDate {
   display: inline-block;
   margin-left: 5px;
 }
 
-.annotationLayer .popup p {
+.annotationLayer .popupContent {
   border-top: 1px solid rgba(51, 51, 51, 1);
   margin-top: 2px;
   padding-top: 2px;
 }
 
+.annotationLayer .richText > * {
+  white-space: pre-wrap;
+}
+
 .annotationLayer .highlightAnnotation,
 .annotationLayer .underlineAnnotation,
 .annotationLayer .squigglyAnnotation,