[pull] master from mozilla:master

src/core/evaluator.js

@@ -1720,13 +1720,20 @@ class PartialEvaluator {
     const stateManager = new StateManager(initialState);
     const preprocessor = new EvaluatorPreprocessor(stream, xref, stateManager);
     const timeSlotManager = new TimeSlotManager();
+    let markedContentLevel = 0;
 
     function closePendingRestoreOPS(argument) {
       for (let i = 0, ii = preprocessor.savedStatesDepth; i < ii; i++) {
         operatorList.addOp(OPS.restore, []);
       }
     }
 
+    function closePendingMarkedContentOPS() {
+      for (; markedContentLevel > 0; markedContentLevel--) {
+        operatorList.addOp(OPS.endMarkedContent, []);
+      }
+    }
+
     return new Promise(function promiseBody(resolve, reject) {
       const next = function (promise) {
         Promise.all([promise, operatorList.ready]).then(function () {
@@ -1741,7 +1748,7 @@ class PartialEvaluator {
       timeSlotManager.reset();
 
       const operation = {};
-      let stop, i, ii, cs, name, isValidName;
+      let stop, cs, name, isValidName;
       while (!(stop = timeSlotManager.check())) {
         // The arguments parsed by read() are used beyond this loop, so we
         // cannot reuse the same array on each iteration. Therefore we pass
@@ -2299,6 +2306,7 @@ class PartialEvaluator {
             // but doing so is meaningless without knowing the semantics.
             continue;
           case OPS.beginMarkedContentProps:
+            markedContentLevel++;
             if (!(args[0] instanceof Name)) {
               warn(`Expected name for beginMarkedContentProps arg0=${args[0]}`);
               operatorList.addOp(OPS.beginMarkedContentProps, ["OC", null]);
@@ -2333,29 +2341,35 @@ class PartialEvaluator {
               );
               return;
             }
-            // Other marked content types aren't supported yet.
+            // Preserve only the MCID from non-OC property dictionaries.
             args = [
               args[0].name,
               args[1] instanceof Dict ? args[1].get("MCID") : null,
             ];
 
             break;
           case OPS.beginMarkedContent:
+            if (args?.some(arg => arg instanceof Dict)) {
+              warn(`getOperatorList - ignoring operator: ${fn}`);
+              continue;
+            }
+            markedContentLevel++;
+            break;
           case OPS.endMarkedContent:
+            if (args?.some(arg => arg instanceof Dict)) {
+              warn(`getOperatorList - ignoring operator: ${fn}`);
+              continue;
+            }
+            if (markedContentLevel === 0) {
+              continue;
+            }
+            markedContentLevel--;
+            break;
           default:
-            // Note: Ignore the operator if it has `Dict` arguments, since
-            // those are non-serializable, otherwise postMessage will throw
-            // "An object could not be cloned.".
-            if (args !== null) {
-              for (i = 0, ii = args.length; i < ii; i++) {
-                if (args[i] instanceof Dict) {
-                  break;
-                }
-              }
-              if (i < ii) {
-                warn("getOperatorList - ignoring operator: " + fn);
-                continue;
-              }
+            // Avoid postMessage errors from `Dict` arguments.
+            if (args?.some(arg => arg instanceof Dict)) {
+              warn(`getOperatorList - ignoring operator: ${fn}`);
+              continue;
             }
         }
         operatorList.addOp(fn, args);
@@ -2364,8 +2378,8 @@ class PartialEvaluator {
         next(deferred);
         return;
       }
-      // Some PDFs don't close all restores inside object/form.
-      // Closing those for them.
+      // Close marked content and graphics states left open by this stream.
+      closePendingMarkedContentOPS();
       closePendingRestoreOPS();
       resolve();
     }).catch(reason => {
@@ -2378,6 +2392,7 @@ class PartialEvaluator {
             `task: "${reason}".`
         );
 
+        closePendingMarkedContentOPS();
         closePendingRestoreOPS();
         return;
       }
@@ -2395,7 +2410,6 @@ class PartialEvaluator {
     seenStyles = new Set(),
     viewBox,
     lang = null,
-    markedContentData = null,
     disableNormalization = false,
     keepWhiteSpace = false,
     prevRefs = null,
@@ -2425,9 +2439,8 @@ class PartialEvaluator {
     resources ||= Dict.empty;
     stateManager ||= new StateManager(new TextState());
 
-    if (includeMarkedContent) {
-      markedContentData ||= { level: 0 };
-    }
+    let markedContentLevel = 0;
+    let textMarkedContentLevel = null;
 
     const textContent = {
       items: [],
@@ -3173,6 +3186,19 @@ class PartialEvaluator {
       textContentItem.str.length = 0;
     }
 
+    function closePendingMarkedContentItems(level = 0) {
+      if (!includeMarkedContent || markedContentLevel <= level) {
+        return;
+      }
+      flushTextContentItem();
+
+      for (; markedContentLevel > level; markedContentLevel--) {
+        textContent.items.push({
+          type: "endMarkedContent",
+        });
+      }
+    }
+
     function enqueueChunk(batch = false) {
       const length = textContent.items.length;
       if (length === 0) {
@@ -3291,6 +3317,13 @@ class PartialEvaluator {
           case OPS.beginText:
             textState.textMatrix = IDENTITY_MATRIX.slice();
             textState.textLineMatrix = IDENTITY_MATRIX.slice();
+            textMarkedContentLevel = markedContentLevel;
+            break;
+          case OPS.endText:
+            if (textMarkedContentLevel !== null) {
+              closePendingMarkedContentItems(textMarkedContentLevel);
+              textMarkedContentLevel = null;
+            }
             break;
           case OPS.showSpacedText:
             if (!stateManager.state.font) {
@@ -3451,7 +3484,6 @@ class PartialEvaluator {
                     seenStyles,
                     viewBox,
                     lang,
-                    markedContentData,
                     disableNormalization,
                     keepWhiteSpace,
                     prevRefs: seenRefs,
@@ -3535,7 +3567,7 @@ class PartialEvaluator {
           case OPS.beginMarkedContent:
             flushTextContentItem();
             if (includeMarkedContent) {
-              markedContentData.level++;
+              markedContentLevel++;
 
               textContent.items.push({
                 type: "beginMarkedContent",
@@ -3546,7 +3578,7 @@ class PartialEvaluator {
           case OPS.beginMarkedContentProps:
             flushTextContentItem();
             if (includeMarkedContent) {
-              markedContentData.level++;
+              markedContentLevel++;
 
               const mcid = args[1] instanceof Dict ? args[1].get("MCID") : null;
               textContent.items.push({
@@ -3561,12 +3593,11 @@ class PartialEvaluator {
           case OPS.endMarkedContent:
             flushTextContentItem();
             if (includeMarkedContent) {
-              if (markedContentData.level === 0) {
-                // Handle unbalanced beginMarkedContent/endMarkedContent
-                // operators (fixes issue15629.pdf).
+              if (markedContentLevel === 0) {
+                // Ignore unmatched EMC operators (issue 15629).
                 break;
               }
-              markedContentData.level--;
+              markedContentLevel--;
 
               textContent.items.push({
                 type: "endMarkedContent",
@@ -3585,6 +3616,7 @@ class PartialEvaluator {
         return;
       }
       flushTextContentItem();
+      closePendingMarkedContentItems();
       enqueueChunk();
       resolve();
     }).catch(reason => {
@@ -3599,6 +3631,7 @@ class PartialEvaluator {
         );
 
         flushTextContentItem();
+        closePendingMarkedContentItems();
         enqueueChunk();
         return;
       }

src/core/jpeg_stream.js

@@ -179,7 +179,7 @@ class JpegStream extends DecodeStream {
         // the height is only known from a DNL marker or because the scan simply
         // ends early (issue15492.pdf). `ImageDecoder` reports and scales the
         // frame according to the SOF, so let our own decoder, which honours the
-        // dictionary, handle the image instead.
+        // actual JPEG image data, handle the image instead.
         return null;
       }
       if (useImageDecoder.exifStart) {

src/core/jpg.js

@@ -811,11 +811,14 @@ class JpegImage {
 
   static canUseImageDecoder(data, colorTransform = -1) {
     const view = new DataView(data.buffer, data.byteOffset, data.byteLength);
-    let exifOffsets = null;
+    const info = {
+      width: 0,
+      height: 0,
+      exifStart: 0,
+      exifEnd: 0,
+    };
     let offset = 0;
     let numComponents = null;
-    let scanLines = 0,
-      samplesPerLine = 0;
     let fileMarker = view.getUint16(offset);
     offset += 2;
     if (fileMarker !== /* SOI (Start of Image) = */ 0xffd8) {
@@ -845,12 +848,13 @@ class JpegImage {
             appData[4] === 0 &&
             appData[5] === 0
           ) {
-            if (exifOffsets) {
+            if (info.exifStart) {
               throw new JpegError("Duplicate EXIF-blocks found.");
             }
             // Don't do the EXIF-block replacement here, see `JpegStream`,
             // since that can modify the original PDF document.
-            exifOffsets = { exifStart: oldOffset + 6, exifEnd: newOffset };
+            info.exifStart = oldOffset + 6;
+            info.exifEnd = newOffset;
           }
           fileMarker = view.getUint16(offset);
           offset += 2;
@@ -860,8 +864,8 @@ class JpegImage {
         case 0xffc2: // SOF2 (Start of Frame, Progressive DCT)
           // Skip marker length.
           // Skip precision.
-          scanLines = view.getUint16(offset + (2 + 1));
-          samplesPerLine = view.getUint16(offset + (2 + 1 + 2));
+          info.height = view.getUint16(offset + (2 + 1)); // scanLines
+          info.width = view.getUint16(offset + (2 + 1 + 2)); // samplesPerLine
           numComponents = data[offset + (2 + 1 + 2 + 2)];
           break markerLoop;
         case 0xffff: // Fill bytes
@@ -882,7 +886,7 @@ class JpegImage {
       return null;
     }
     // A zero SOF height means that a later DNL marker defines it.
-    return { width: samplesPerLine, height: scanLines, ...exifOffsets };
+    return info;
   }
 
   parse(data, { dnlScanLines = null } = {}) {

src/core/string_utils.js

@@ -62,6 +62,7 @@ const PDFStringTranslateTable = [
   0x2019, 0x201a, 0x2122, 0xfb01, 0xfb02, 0x141, 0x152, 0x160, 0x178, 0x17d,
   0x131, 0x142, 0x153, 0x161, 0x17e, 0, 0x20ac,
 ];
+const PDFStringTextDecoders = Object.create(null);
 
 function stringToPDFString(str, keepEscapeSequence = false) {
   // See section 7.9.2.2 Text String Type.
@@ -85,7 +86,10 @@ function stringToPDFString(str, keepEscapeSequence = false) {
 
     if (encoding) {
       try {
-        const decoder = new TextDecoder(encoding, { fatal: true });
+        const decoder = (PDFStringTextDecoders[encoding] ??= new TextDecoder(
+          encoding,
+          { fatal: true }
+        ));
         const buffer = stringToBytes(str);
         const decoded = decoder.decode(buffer);
         if (keepEscapeSequence || !decoded.includes("\x1b")) {

test/integration/text_layer_spec.mjs

@@ -1534,4 +1534,30 @@ describe("Text layer", () => {
       expect(getPercentDiff(rect.height, 12)).toBeLessThan(0.03);
     });
   });
+
+  describe("marked-content nesting (bug 1898053)", () => {
+    let pages;
+
+    beforeAll(async () => {
+      pages = await loadAndWait(
+        "bug1898053_minimal.pdf",
+        ".textLayer .endOfContent"
+      );
+    });
+    afterAll(async () => {
+      await closePages(pages);
+    });
+
+    it("must keep auto-closed sections at the text-layer root", async () => {
+      await Promise.all(
+        pages.map(async ([browserName, page]) => {
+          const count = await page.evaluate(
+            () =>
+              document.querySelectorAll(".textLayer > .markedContent").length
+          );
+          expect(count).toBe(6);
+        })
+      );
+    });
+  });
 });

test/pdfs/.gitignore

@@ -951,3 +951,4 @@
 !signed_verified.pdf
 !function_based_shading_cmyk.pdf
 !large_jpeg_downscale.pdf
+!bug1898053_minimal.pdf

test/unit/api_spec.js

@@ -4480,6 +4480,59 @@ Caron Broadcasting, Inc., an Ohio corporation (“Lessee”).`)
       await loadingTask.destroy();
     });
 
+    it("auto-closes marked content opened in a text object (bug 1898053)", async function () {
+      const loadingTask = getDocument(
+        buildGetDocumentParams("bug1898053_minimal.pdf")
+      );
+      const pdfDoc = await loadingTask.promise;
+      const pdfPage = await pdfDoc.getPage(1);
+      const { items } = await pdfPage.getTextContent({
+        includeMarkedContent: true,
+      });
+
+      expect(items.map(item => item.id ?? item.type ?? item.str)).toEqual([
+        "Hello, world!",
+        "p3R_mc1",
+        "endMarkedContent",
+        "p3R_mc2",
+        "endMarkedContent",
+        "p3R_mc3",
+        "endMarkedContent",
+        "p3R_mc4",
+        "endMarkedContent",
+        "p3R_mc5",
+        "endMarkedContent",
+        "p3R_mc6",
+        "endMarkedContent",
+      ]);
+
+      await loadingTask.destroy();
+    });
+
+    it("preserves marked content spanning text objects (bug 1823296)", async function () {
+      const loadingTask = getDocument(buildGetDocumentParams("bug1823296.pdf"));
+      const pdfDoc = await loadingTask.promise;
+      const pdfPage = await pdfDoc.getPage(1);
+      const { items } = await pdfPage.getTextContent({
+        includeMarkedContent: true,
+        disableNormalization: true,
+      });
+
+      const start = items.findIndex(item => item.id === "p3R_mc8");
+      const end = items.findIndex(
+        (item, index) => index > start && item.type === "endMarkedContent"
+      );
+
+      expect(items.slice(start + 1, end).map(item => item.str)).toEqual([
+        "",
+        "PDF/UA is not a separate file-format but simply a way to use the " +
+          "familiar PDF format invented by Adobe",
+        "Systems and now standardized as ISO 32000.[5]",
+      ]);
+
+      await loadingTask.destroy();
+    });
+
     it("gets text content with multi-byte entries, using predefined CMaps (issue 16176)", async function () {
       const loadingTask = getDocument(
         buildGetDocumentParams("issue16176.pdf", {