fix: MathML combines multidigit numbers with sup/subscript, comma separators, and multicharacter text when outputting to DOM (#3999)

edemaine · web-flow · commit 7d79e220f465 · 2024-12-17T10:07:35.000-05:00
* fix: MathML combines multidigit numbers and text when outputting to DOM Fixes #3995 * Push numbers into sup/sub base, handle {,} separators * Improve comments and merging efficiency
diff --git a/src/buildMathML.js b/src/buildMathML.js
@@ -128,6 +128,30 @@ export const getVariant = function(
     return null;
 };
 
+/**
+ * Check for <mi>.</mi> which is how a dot renders in MathML,
+ * or <mo separator="true" lspace="0em" rspace="0em">,</mo>
+ * which is how a braced comma {,} renders in MathML
+ */
+function isNumberPunctuation(group: ?MathNode): boolean {
+    if (!group) {
+        return false;
+    }
+    if (group.type === 'mi' && group.children.length === 1) {
+        const child = group.children[0];
+        return child instanceof TextNode && child.text === '.';
+    } else if (group.type === 'mo' && group.children.length === 1 &&
+        group.getAttribute('separator') === 'true' &&
+        group.getAttribute('lspace') === '0em' &&
+        group.getAttribute('rspace') === '0em'
+    ) {
+        const child = group.children[0];
+        return child instanceof TextNode && child.text === ',';
+    } else {
+        return false;
+    }
+}
+
 /**
  * Takes a list of nodes, builds them, and returns a list of the generated
  * MathML nodes.  Also combine consecutive <mtext> outputs into a single
@@ -165,13 +189,25 @@ export const buildExpression = function(
                 lastGroup.children.push(...group.children);
                 continue;
             // Concatenate <mn>...</mn> followed by <mi>.</mi>
-            } else if (group.type === 'mi' && group.children.length === 1 &&
-                       lastGroup.type === 'mn') {
-                const child = group.children[0];
-                if (child instanceof TextNode && child.text === '.') {
-                    lastGroup.children.push(...group.children);
-                    continue;
+            } else if (isNumberPunctuation(group) && lastGroup.type === 'mn') {
+                lastGroup.children.push(...group.children);
+                continue;
+            // Concatenate <mi>.</mi> followed by <mn>...</mn>
+            } else if (group.type === 'mn' && isNumberPunctuation(lastGroup)) {
+                group.children = [...lastGroup.children, ...group.children];
+                groups.pop();
+            // Put preceding <mn>...</mn> or <mi>.</mi> inside base of
+            // <msup><mn>...base...</mn>...exponent...</msup> (or <msub>)
+            } else if ((group.type === 'msup' || group.type === 'msub') &&
+                group.children.length >= 1 &&
+                (lastGroup.type === 'mn' || isNumberPunctuation(lastGroup))
+            ) {
+                const base = group.children[0];
+                if (base instanceof MathNode && base.type === 'mn') {
+                    base.children = [...lastGroup.children, ...base.children];
+                    groups.pop();
                 }
+            // \not
             } else if (lastGroup.type === 'mi' && lastGroup.children.length === 1) {
                 const lastChild = lastGroup.children[0];
                 if (lastChild instanceof TextNode && lastChild.text === '\u0338' &&
diff --git a/src/mathMLTree.js b/src/mathMLTree.js
@@ -95,7 +95,18 @@ export class MathNode implements MathDomNode {
         }
 
         for (let i = 0; i < this.children.length; i++) {
-            node.appendChild(this.children[i].toNode());
+            // Combine multiple TextNodes into one TextNode, to prevent
+            // screen readers from reading each as a separate word [#3995]
+            if (this.children[i] instanceof TextNode &&
+                this.children[i + 1] instanceof TextNode) {
+                let text = this.children[i].toText() + this.children[++i].toText();
+                while (this.children[i + 1] instanceof TextNode) {
+                    text += this.children[++i].toText();
+                }
+                node.appendChild(new TextNode(text).toNode());
+            } else {
+                node.appendChild(this.children[i].toNode());
+            }
         }
 
         return node;
diff --git a/test/__snapshots__/mathml-spec.js.snap b/test/__snapshots__/mathml-spec.js.snap
@@ -428,9 +428,35 @@ exports[`A MathML builder should concatenate digits into single <mn> 1`] = `
       <mn>
         0.34
       </mn>
+      <mo>
+        =
+      </mo>
+      <msup>
+        <mn>
+          .34
+        </mn>
+        <mn>
+          1
+        </mn>
+      </msup>
+    </mrow>
+    <annotation encoding="application/x-tex">
+      \\sin{\\alpha}=0.34=.34^1
+    </annotation>
+  </semantics>
+</math>
+`;
+
+exports[`A MathML builder should concatenate digits into single <mn> 2`] = `
+<math xmlns="https://www.w3.org/1998/Math/MathML">
+  <semantics>
+    <mrow>
+      <mn>
+        1,000,000
+      </mn>
     </mrow>
     <annotation encoding="application/x-tex">
-      \\sin{\\alpha}=0.34
+      1{,}000{,}000
     </annotation>
   </semantics>
 </math>
diff --git a/test/mathml-spec.js b/test/mathml-spec.js
@@ -29,7 +29,8 @@ describe("A MathML builder", function() {
     });
 
     it('should concatenate digits into single <mn>', () => {
-        expect(getMathML("\\sin{\\alpha}=0.34")).toMatchSnapshot();
+        expect(getMathML("\\sin{\\alpha}=0.34=.34^1")).toMatchSnapshot();
+        expect(getMathML("1{,}000{,}000")).toMatchSnapshot();
     });
 
     it('should make prime operators into <mo> nodes', () => {
