fix(bbb-web): Fallback to Rasterization if SVG has (lots of) Mask Tags

bbb-common-web/src/main/java/org/bigbluebutton/presentation/handlers/SvgConversionHandler.java

@@ -18,6 +18,9 @@ public class SvgConversionHandler extends AbstractCommandHandler {
     private static String USE_TAG_OUTPUT = "<use";
     private static String USE_TAG_PATTERN = "\\d+\\s" + USE_TAG_OUTPUT;
 
+    private static String MASK_TAG_OUTPUT = "<mask";
+    private static String MASK_TAG_PATTERN = "\\d+\\s" + MASK_TAG_OUTPUT;
+
     private final String id;
 
     public SvgConversionHandler(String id) {
@@ -84,6 +87,26 @@ public int numberOfUseTags() {
         return 0;
     }
 
+    /**
+     *
+     * @return The number of <mask/> tags in the generated SVG.
+     */
+    public int numberOfMaskTags() {
+        if (stdoutContains(MASK_TAG_OUTPUT)) {
+            try {
+                String out = stdoutBuilder.toString();
+                Pattern r = Pattern.compile(MASK_TAG_PATTERN);
+                Matcher m = r.matcher(out);
+                m.find();
+                return Integer.parseInt(m.group(0).replace(MASK_TAG_OUTPUT, "").trim());
+            } catch (Exception e) {
+                log.error("Exception counting the number of mask tags", e);
+                return 0;
+            }
+        }
+        return 0;
+    }
+
     @Override
     protected String getIdTag() {
         return id;

bbb-common-web/src/main/java/org/bigbluebutton/presentation/imp/SvgImageCreatorImp.java

@@ -32,6 +32,7 @@ public class SvgImageCreatorImp implements SvgImageCreator {
     private long imageTagThreshold;
     private long useTagThreshold;
     private long pathsThreshold;
+    private long maskTagThreshold = 0;
     private int convPdfToSvgTimeout = 60;
     private int pdfFontsTimeout = 3;
     private int svgResolutionPpi = 300;
@@ -226,6 +227,7 @@ private boolean generateSvgImage(File imagePresentationDir, UploadedPresentation
                 pHandler.numberOfImageTags() > imageTagThreshold ||
                 pHandler.numberOfPaths() > pathsThreshold ||
                 pHandler.numberOfUseTags() > useTagThreshold ||
+                (maskTagThreshold > 0 && pHandler.numberOfMaskTags() >= maskTagThreshold) ||
                 rasterizeCurrSlide) {
 
             // We need t delete the destination file as we are starting a
@@ -246,6 +248,7 @@ private boolean generateSvgImage(File imagePresentationDir, UploadedPresentation
                 logData.put("fileExists", destsvg.exists());
                 logData.put("numberOfImages", pHandler.numberOfImageTags());
                 logData.put("numberOfPaths", pHandler.numberOfPaths());
+                logData.put("numberOfMasks", pHandler.numberOfMaskTags());
                 logData.put("logCode", "potential_problem_with_svg");
                 logData.put("message", "Potential problem with generated SVG");
                 Gson gson = new Gson();
@@ -412,7 +415,7 @@ private NuProcessBuilder createConversionProcess(String format, int page, String
 
         rawCommand  += " -q -f " + String.valueOf(page) + " -l " + String.valueOf(page) + " " + source + " " + destFile;
         if (analyze) {
-            rawCommand += " && grep -oE '<image|<path|<use' "+destFile+" | sort | uniq -c ";
+            rawCommand += " && grep -oE '<image|<path|<use|<mask' "+destFile+" | sort | uniq -c ";
         }
 
         return new NuProcessBuilder(Arrays.asList("/usr/share/bbb-web/run-in-systemd.sh", timeout + "s", "/bin/sh", "-c", rawCommand));
@@ -494,6 +497,10 @@ public void setUseTagThreshold(long threshold) {
     public void setPathsThreshold(long threshold) {
         pathsThreshold = threshold;
     }
+
+    public void setMaskTagThreshold(long threshold) {
+        maskTagThreshold = threshold;
+    }
 
     public void setSlidesGenerationProgressNotifier(
         SlidesGenerationProgressNotifier notifier) {

bbb-common-web/src/test/scala/org/bigbluebutton/presentation/handlers/SvgConversionHandlerTest.scala

@@ -0,0 +1,76 @@
+package org.bigbluebutton.presentation.handlers
+
+import java.nio.ByteBuffer
+import java.nio.charset.StandardCharsets
+import java.nio.file.{ Files, Paths }
+
+import org.bigbluebutton.api.util.UnitSpec
+
+/**
+ * Tests the tag-count parsing of SvgConversionHandler.
+ *
+ * During SVG slide analysis the conversion process runs
+ *   grep -oE '<image|<path|<use|<mask' slideN.svg | sort | uniq -c
+ * and the handler parses the resulting stdout. These tests replicate that
+ * pipeline against src/test/resources/sample-with-mask.svg, which was
+ * generated by pdftocairo 24.02.0 (Ubuntu 24.04) from
+ * src/test/resources/sample-with-mask.pdf, a minimal PDF containing an
+ * image XObject with an /SMask (soft mask) drawn twice - soft masks are
+ * what pdftocairo turns into SVG <mask> elements.
+ */
+class SvgConversionHandlerTest extends UnitSpec {
+
+  val sampleSvgFile = "src/test/resources/sample-with-mask.svg"
+  val analyzedTags = List("<image", "<mask", "<path", "<use")
+
+  private def handlerFedWith(stdout: String): SvgConversionHandler = {
+    val handler = new SvgConversionHandler("test")
+    handler.onStdout(ByteBuffer.wrap(stdout.getBytes(StandardCharsets.UTF_8)), true)
+    handler
+  }
+
+  private def countOccurrences(content: String, token: String): Int = {
+    var count = 0
+    var idx = content.indexOf(token)
+    while (idx > -1) {
+      count += 1
+      idx = content.indexOf(token, idx + token.length)
+    }
+    count
+  }
+
+  // Replicates the analysis pipeline: grep -oE '<image|<path|<use|<mask' | sort | uniq -c
+  private def uniqCountOutput(content: String): String = {
+    analyzedTags.map(tag => "%7d %s\n".format(countOccurrences(content, tag), tag)).mkString
+  }
+
+  it should "count mask tags from the analysis output of a generated svg with soft masks" in {
+    val svgContent = new String(Files.readAllBytes(Paths.get(sampleSvgFile)), StandardCharsets.UTF_8)
+    val handler = handlerFedWith(uniqCountOutput(svgContent))
+
+    assert(handler.numberOfMaskTags() == countOccurrences(svgContent, "<mask"))
+    assert(handler.numberOfMaskTags() == 2)
+    assert(handler.numberOfImageTags() == 3)
+    assert(handler.numberOfPaths() == 2)
+    assert(handler.numberOfUseTags() == 4)
+  }
+
+  it should "count zero mask tags when the analysis output has no mask line" in {
+    val handler = handlerFedWith("      2 <image\n    159 <path\n    444 <use\n")
+
+    assert(handler.numberOfMaskTags() == 0)
+    assert(handler.numberOfImageTags() == 2)
+    assert(handler.numberOfPaths() == 159)
+    assert(handler.numberOfUseTags() == 444)
+  }
+
+  it should "count zero tags when the analysis output is not uniq -c formatted" in {
+    val handler = handlerFedWith("pdftocairo: unknown error\n")
+
+    assert(handler.numberOfMaskTags() == 0)
+    assert(handler.numberOfImageTags() == 0)
+    assert(handler.numberOfPaths() == 0)
+    assert(handler.numberOfUseTags() == 0)
+  }
+
+}

bigbluebutton-tests/playwright/core/elements.ts

@@ -410,6 +410,7 @@ export const elements = {
   uploadPresentationFileName: 'uploadTest.png',
   presentationPPTX: 'BBB.pptx',
   presentationTXT: 'helloWorld.txt',
+  maskSamplePdf: 'sample-with-mask.pdf',
   startScreenSharing: 'button[data-test="startScreenShare"]',
   stopScreenSharing: 'button[data-test="stopScreenShare"]',
   managePresentations: 'div[data-test="managePresentations"]',

bigbluebutton-tests/playwright/presentation/presentation.spec.ts

@@ -159,6 +159,20 @@ test.describe.parallel('Presentation', { tag: '@ci' }, () => {
       await presentation.uploadSinglePresentationTest();
     });
 
+    // maskTagThreshold is a SERVER-SIDE bbb-web setting with no client-settings hook, so nothing
+    // applies it automatically: this test requires bbb-web to be restarted with `maskTagThreshold=1`
+    // in /etc/bigbluebutton/bbb-web.properties (rasterize any slide whose generated SVG contains a
+    // <mask> tag). The @setting-required tag keeps it out of the default CI gate.
+    test(
+      'Masked slide is rasterized when maskTagThreshold is set',
+      { tag: '@setting-required:maskTagThreshold' },
+      async ({ browser, context, page }, testInfo) => {
+        const presentation = new Presentation(browser, context);
+        await presentation.initModPage(page, { testInfo });
+        await presentation.maskRasterizationFallbackTest();
+      },
+    );
+
     test('Upload Other Presentations Format', async ({ browser, context, page }, testInfo) => {
       const presentation = new Presentation(browser, context);
       await presentation.initPages(page, testInfo);

bigbluebutton-tests/playwright/presentation/presentation.ts

@@ -333,6 +333,45 @@ export class Presentation extends MultiUsers {
     }
   }
 
+  async maskRasterizationFallbackTest() {
+    // wait for whiteboard to load and no notifications
+    await this.modPage.waitForSelector(e.whiteboard, ELEMENT_WAIT_LONGER_TIME);
+    await this.modPage.waitForSelector(e.skipSlide);
+    await this.modPage.closeAllToastNotifications();
+
+    await uploadSinglePresentation(this.modPage, e.maskSamplePdf, UPLOAD_PDF_WAIT_TIME);
+
+    // secondary check: the uploaded slide is visibly present
+    await this.modPage.hasElement(e.currentSlideImg, 'should display the uploaded slide as the current slide image');
+
+    // 4.0 renders slides as tldraw image assets, so a visual snapshot cannot tell a rasterized
+    // (embedded-PNG) slide apart from a vector slide of the same content. Inspect the served
+    // slide SVG file instead, deriving its URL from the current tl-image asset.
+    const slideSvgUrl = await this.modPage.page.evaluate(
+      ([selector]) => {
+        const element = document.querySelector(selector) as HTMLElement | null;
+        return element?.style?.backgroundImage?.split('"')[1] ?? null;
+      },
+      [e.currentSlideImg],
+    );
+    // the asset URL carries pageToken/sessionToken query params, e.g. .../svg/1?pageToken=...
+    expect(slideSvgUrl, 'should resolve the served slide SVG url from the current slide asset').toMatch(
+      /\/svg\/\d+(\?|$)/,
+    );
+
+    const slideSvgResponse = await this.modPage.page.request.get(slideSvgUrl as string);
+    expect(slideSvgResponse.ok(), 'should fetch the served slide SVG').toBeTruthy();
+    const slideSvgContent = await slideSvgResponse.text();
+
+    // A rasterized slide is the embedded-PNG SVG produced by SvgImageCreatorImp.createSvgWithEmbeddedPng():
+    // a single <image href="data:image/png;base64,..."> and no vector elements. pdftocairo vector output
+    // instead uses xlink:href for embedded bitmaps and contains <path> elements.
+    expect(slideSvgContent, 'served slide SVG should be the rasterized embedded-PNG form').toMatch(
+      /<image href="data:image\/png;base64,/,
+    );
+    expect(slideSvgContent, 'rasterized slide SVG should contain no vector <path> elements').not.toMatch(/<path[\s>]/);
+  }
+
   async uploadOtherPresentationsFormat() {
     // wait for whiteboard to load and no notifications
     await this.modPage.waitForSelector(e.whiteboard, ELEMENT_WAIT_LONGER_TIME);

bigbluebutton-web/grails-app/conf/bigbluebutton.properties

@@ -72,6 +72,12 @@ imageTagThreshold=800
 # Maximum allowed number of <use> tags in generated svg, if exceeded the conversion will fallback to full BMP (default 10k)
 useTagThreshold=10000
 
+# Minimum number of <mask> tags in generated svg that triggers a fallback to full BMP rasterization (default 0, disabled).
+# Masks are common in ordinary PDFs (soft-masked/alpha images; the bundled default.pdf title page produces 3), and
+# pdftocairo shipped with Ubuntu 24.04 (poppler 24.02.0) generates correct mask values, so this check is disabled by default (0).
+# Set to 1 to rasterize any slide whose svg contains a mask, or to N to rasterize only slides with N or more masks.
+maskTagThreshold=0
+
 #------------------------------------
 # Number of threads in the pool to do the presentation conversion.
 #------------------------------------

bigbluebutton-web/grails-app/conf/spring/doc-conversion.xml

@@ -107,6 +107,7 @@ with BigBlueButton; if not, see <http://www.gnu.org/licenses/>.
         <property name="imageTagThreshold" value="${imageTagThreshold}"/>
         <property name="useTagThreshold" value="${useTagThreshold}"/>
         <property name="pathsThreshold" value="${placementsThreshold}"/>
+        <property name="maskTagThreshold" value="${maskTagThreshold}"/>
         <property name="blankSvg" value="${BLANK_SVG}"/>
         <property name="convPdfToSvgTimeout" value="${svgConversionTimeout}"/>
         <property name="pdfFontsTimeout" value="${pdfFontsTimeout}"/>