Merge pull request #17941 from Snuffleupagus/getTextContent-lang

[api-minor] Include the document /Lang attribute in the textContent-data

src/core/catalog.js

@@ -166,7 +166,7 @@ class Catalog {
     return shadow(
       this,
       "lang",
-      typeof lang === "string" ? stringToPDFString(lang) : null
+      lang && typeof lang === "string" ? stringToPDFString(lang) : null
     );
   }
 

src/core/document.js

@@ -395,10 +395,9 @@ class Page {
   }
 
   loadResources(keys) {
-    if (!this.resourcesPromise) {
-      // TODO: add async `_getInheritableProperty` and remove this.
-      this.resourcesPromise = this.pdfManager.ensure(this, "resources");
-    }
+    // TODO: add async `_getInheritableProperty` and remove this.
+    this.resourcesPromise ||= this.pdfManager.ensure(this, "resources");
+
     return this.resourcesPromise.then(() => {
       const objectLoader = new ObjectLoader(this.resources, keys, this.xref);
       return objectLoader.load();
@@ -625,7 +624,7 @@ class Page {
     });
   }
 
-  extractTextContent({
+  async extractTextContent({
     handler,
     task,
     includeMarkedContent,
@@ -639,31 +638,35 @@ class Page {
       "Properties",
       "XObject",
     ]);
+    const langPromise = this.pdfManager.ensureCatalog("lang");
 
-    const dataPromises = Promise.all([contentStreamPromise, resourcesPromise]);
-    return dataPromises.then(([contentStream]) => {
-      const partialEvaluator = new PartialEvaluator({
-        xref: this.xref,
-        handler,
-        pageIndex: this.pageIndex,
-        idFactory: this._localIdFactory,
-        fontCache: this.fontCache,
-        builtInCMapCache: this.builtInCMapCache,
-        standardFontDataCache: this.standardFontDataCache,
-        globalImageCache: this.globalImageCache,
-        systemFontCache: this.systemFontCache,
-        options: this.evaluatorOptions,
-      });
+    const [contentStream, , lang] = await Promise.all([
+      contentStreamPromise,
+      resourcesPromise,
+      langPromise,
+    ]);
+    const partialEvaluator = new PartialEvaluator({
+      xref: this.xref,
+      handler,
+      pageIndex: this.pageIndex,
+      idFactory: this._localIdFactory,
+      fontCache: this.fontCache,
+      builtInCMapCache: this.builtInCMapCache,
+      standardFontDataCache: this.standardFontDataCache,
+      globalImageCache: this.globalImageCache,
+      systemFontCache: this.systemFontCache,
+      options: this.evaluatorOptions,
+    });
 
-      return partialEvaluator.getTextContent({
-        stream: contentStream,
-        task,
-        resources: this.resources,
-        includeMarkedContent,
-        disableNormalization,
-        sink,
-        viewBox: this.view,
-      });
+    return partialEvaluator.getTextContent({
+      stream: contentStream,
+      task,
+      resources: this.resources,
+      includeMarkedContent,
+      disableNormalization,
+      sink,
+      viewBox: this.view,
+      lang,
     });
   }
 

src/core/evaluator.js

@@ -2307,6 +2307,7 @@ class PartialEvaluator {
     sink,
     seenStyles = new Set(),
     viewBox,
+    lang = null,
     markedContentData = null,
     disableNormalization = false,
     keepWhiteSpace = false,
@@ -2323,6 +2324,7 @@ class PartialEvaluator {
     const textContent = {
       items: [],
       styles: Object.create(null),
+      lang,
     };
     const textContentItem = {
       initialized: false,
@@ -3296,6 +3298,7 @@ class PartialEvaluator {
                     sink: sinkWrapper,
                     seenStyles,
                     viewBox,
+                    lang,
                     markedContentData,
                     disableNormalization,
                     keepWhiteSpace,

src/display/api.js

@@ -1160,6 +1160,7 @@ class PDFDocumentProxy {
  *   items are included when includeMarkedContent is true.
  * @property {Object<string, TextStyle>} styles - {@link TextStyle} objects,
  *   indexed by font name.
+ * @property {string | null} lang - The document /Lang attribute.
  */
 
 /**
@@ -1677,6 +1678,7 @@ class PDFPageProxy {
             resolve(textContent);
             return;
           }
+          textContent.lang ??= value.lang;
           Object.assign(textContent.styles, value.styles);
           textContent.items.push(...value.items);
           pump();
@@ -1687,6 +1689,7 @@ class PDFPageProxy {
       const textContent = {
         items: [],
         styles: Object.create(null),
+        lang: null,
       };
       pump();
     });

src/display/text_layer.js

@@ -64,7 +64,7 @@ const DEFAULT_FONT_ASCENT = 0.8;
 const ascentCache = new Map();
 let _canvasContext = null;
 
-function getCtx() {
+function getCtx(lang = null) {
   if (!_canvasContext) {
     // We don't use an OffscreenCanvas here because we use serif/sans serif
     // fonts with it and they depends on the locale.
@@ -89,13 +89,13 @@ function cleanupTextLayer() {
   _canvasContext = null;
 }
 
-function getAscent(fontFamily) {
+function getAscent(fontFamily, lang) {
   const cachedAscent = ascentCache.get(fontFamily);
   if (cachedAscent) {
     return cachedAscent;
   }
 
-  const ctx = getCtx();
+  const ctx = getCtx(lang);
 
   const savedFont = ctx.font;
   ctx.canvas.width = ctx.canvas.height = DEFAULT_FONT_SIZE;
@@ -162,7 +162,7 @@ function getAscent(fontFamily) {
   return DEFAULT_FONT_ASCENT;
 }
 
-function appendText(task, geom) {
+function appendText(task, geom, lang) {
   // Initialize all used properties to keep the caches monomorphic.
   const textDiv = document.createElement("span");
   const textDivProperties = {
@@ -184,7 +184,7 @@ function appendText(task, geom) {
   const fontFamily =
     (task._fontInspectorEnabled && style.fontSubstitution) || style.fontFamily;
   const fontHeight = Math.hypot(tx[2], tx[3]);
-  const fontAscent = fontHeight * getAscent(fontFamily);
+  const fontAscent = fontHeight * getAscent(fontFamily, lang);
 
   let left, top;
   if (angle === 0) {
@@ -324,7 +324,7 @@ class TextLayerRenderTask {
       div: null,
       scale: viewport.scale * (globalThis.devicePixelRatio || 1),
       properties: null,
-      ctx: getCtx(),
+      ctx: null,
     };
     this._styleCache = Object.create(null);
     const { pageWidth, pageHeight, pageX, pageY } = viewport.rawDims;
@@ -371,7 +371,11 @@ class TextLayerRenderTask {
   /**
    * @private
    */
-  _processItems(items) {
+  _processItems(items, lang) {
+    if (!this._layoutTextParams.ctx) {
+      this._textDivProperties.set(this._rootContainer, { lang });
+      this._layoutTextParams.ctx = getCtx(lang);
+    }
     const textDivs = this._textDivs,
       textContentItemsStr = this._textContentItemsStr;
 
@@ -403,7 +407,7 @@ class TextLayerRenderTask {
         continue;
       }
       textContentItemsStr.push(item.str);
-      appendText(this, item);
+      appendText(this, item, lang);
     }
   }
 
@@ -440,7 +444,7 @@ class TextLayerRenderTask {
         }
 
         Object.assign(styleCache, value.styles);
-        this._processItems(value.items);
+        this._processItems(value.items, value.lang);
         pump();
       }, this._capability.reject);
     };
@@ -476,7 +480,7 @@ function updateTextLayer({
   }
 
   if (mustRescale) {
-    const ctx = getCtx();
+    const ctx = getCtx(textDivProperties.get(container)?.lang);
     const scale = viewport.scale * (globalThis.devicePixelRatio || 1);
     const params = {
       prevFontSize: null,

test/unit/api_spec.js

@@ -3128,10 +3128,11 @@ describe("api", function () {
     });
 
     it("gets text content", async function () {
-      const { items, styles } = await page.getTextContent();
+      const { items, styles, lang } = await page.getTextContent();
 
       expect(items.length).toEqual(15);
       expect(objectSize(styles)).toEqual(5);
+      expect(lang).toEqual("en");
 
       const text = mergeText(items);
       expect(text).toEqual(`Table Of Content
@@ -3146,13 +3147,14 @@ page 1 / 3`);
       );
       const pdfDoc = await loadingTask.promise;
       const pdfPage = await pdfDoc.getPage(1);
-      const { items, styles } = await pdfPage.getTextContent({
+      const { items, styles, lang } = await pdfPage.getTextContent({
         disableNormalization: true,
       });
       expect(items.length).toEqual(1);
       // Font name will be a random object id.
       const fontName = items[0].fontName;
       expect(Object.keys(styles)).toEqual([fontName]);
+      expect(lang).toEqual(null);
 
       expect(items[0]).toEqual({
         dir: "ltr",