Extract operators for files without apply fields

Author: physikerwelt

mathml-core/src/main/java/com/formulasearchengine/mathmltools/exceptions/ParsingErrorHandler.java

@@ -48,7 +48,7 @@ public void error(SAXParseException exception) throws SAXException {
                 LOG.debug(exception.getMessage() + " Error handler is on level 'silent' that ignores this error.");
                 return; // ignore
             case NOTIFY:
-                LOG.error(exception.getMessage() + " Error handler is on level 'notify', no exception was thrown.");
+                LOG.warn(exception.getMessage() + " Error handler is on level 'notify', no exception was thrown.");
                 break;
             case SEVERE:
             case THROW_ALL:

mathml-core/src/main/java/com/formulasearchengine/mathmltools/helper/CMMLHelper.java

@@ -24,7 +24,7 @@ private CMMLHelper() {
     }
 
     /**
-     * Get the first node of the MathML-Content annotations within a MathML document.
+     * Get the first apply node of the MathML-Content annotations within a MathML document.
      *
      * @param mathml full MathML document
      * @return first node of the MathML-Content annotations within a MathML document
@@ -39,6 +39,22 @@ public static Node getFirstApplyNode(String mathml) {
         }
     }
 
+    /**
+     * Get the first node of the MathML-Content annotations within a MathML document.
+     *
+     * @param mathml full MathML document
+     * @return first node of the MathML-Content annotations within a MathML document
+     */
+    public static Node getFirstNode(String mathml) {
+        try {
+            // get the apply node of the ContentMathML root
+            return getFirstNode(new CMMLInfo(mathml));
+        } catch (Exception e) {
+            logger.error("failed to get apply node", e);
+            return null;
+        }
+    }
+
     /**
      * Get the first node of the MathML-Content annotations within a MathML document.
      * Before, the MathML document was converted to strict CMML and subsequently also converted
@@ -91,6 +107,31 @@ public static Node getFirstApplyNode(CMMLInfo cmmlInfo) throws XPathExpressionEx
         return applyRoot;
     }
 
+
+
+    /**
+     * Get the root node from the content mathml.
+     * It will search for the first node of the MathML-Content annotations
+     * or the semantics/apply node within a CMMLInfo document.
+     *
+     * @param cmmlInfo CMMLInfo document
+     * @return first node of the MathML-Content annotations within a MathML document
+     * @throws XPathExpressionException parser exception
+     */
+    public static Node getFirstNode(CMMLInfo cmmlInfo) throws XPathExpressionException {
+        // 1. search for a separate cmml semantic
+        XPath xpath = XMLHelper.namespaceAwareXpath("m", CMMLInfo.NS_MATHML);
+        Node applyRoot = getElement(cmmlInfo, "m:math/m:semantics/m:annotation-xml[@encoding='MathML-Content']/*[1]", xpath);
+        if (applyRoot == null) {
+            // 2. search for a main cmml semantic
+            applyRoot = getElement(cmmlInfo, "*//m:semantics/*[1]", xpath);
+            if (applyRoot == null) {
+                // 3. try to take the apply right beneath the math elements
+                applyRoot = getElement(cmmlInfo, "m:math/*[1]", xpath);
+            }
+        }
+        return applyRoot;
+    }
     /**
      * Extracts a single node for the specified XPath expression.
      *

mathml-core/src/test/java/com/formulasearchengine/mathmltools/mml/CMMLInfoTest.java

@@ -360,4 +360,10 @@ public final void testDtQueryTest() throws Exception {
         assertEquals(s, mml.getXQueryString());
     }
 
-}
+
+    @Test
+    public void testAFormula() throws Exception {
+        Node mainElement = new CMMLInfo(getFileContents(MML_TEST_DIR + "A.mml"));
+        assertThat(mainElement, notNullValue());
+    }
+}

mathml-core/src/test/resources/com/formulasearchengine/mathmltools/mml/tests/A.mml

@@ -0,0 +1,9 @@
+<math xmlns="http://www.w3.org/1998/Math/MathML" alttext="A">
+  <semantics>
+    <ci id="p51.m5.1" xref="p51.m5.1.pmml">A</ci>
+    <annotation-xml encoding="MathML-Presentation">
+      <mi id="p51.m5.1.pmml" xref="p51.m5.1">A</mi>
+    </annotation-xml>
+    <annotation encoding="application/x-tex">A</annotation>
+  </semantics>
+</math>

mathml-similarity/src/test/java/com/formulasearchengine/mathmltools/similarity/node/MathNodeGeneratorTest.java

@@ -38,11 +38,23 @@ public void generateMathNode_mathoid_strict() throws Exception {
         // mathml from mathoid and get the strict cmml semantics
         testConverterOnFile(CMMLHelper::getStrictCmml, "mathnode4", true);
     }
+    @Test
+    public void generateMathNodeMathoidStrictAny() throws Exception {
+        // mathml from mathoid and get the strict cmml semantics
+        testConverterOnFile(CMMLHelper::getFirstNode, "mathnode4", true);
+    }
+
+    @Test
+    public void generateMathNodeHyplag() throws Exception {
+        // mathml from hyplag and get the strict cmml semantics
+        testConverterOnFile(CMMLHelper::getFirstNode, "mathnode5", true);
+    }
 
     private void testConverterOnFile(Function<String, Node> convert, String basicFilename, boolean checkAbstract) throws Exception {
         String expected = IOUtils.toString(this.getClass().getResourceAsStream(basicFilename + "_expected.txt"), "UTF-8");
         String mathml = IOUtils.toString(this.getClass().getResourceAsStream(basicFilename + "_test.txt"), "UTF-8");
-        MathNode mathNode = MathNodeGenerator.generateMathNode(convert.apply(mathml));
+        final Node applyRoot = convert.apply(mathml);
+        MathNode mathNode = MathNodeGenerator.generateMathNode(applyRoot);
 
         if (checkAbstract) {
             mathNode = MathNodeGenerator.toAbstract(mathNode);
@@ -51,4 +63,4 @@ private void testConverterOnFile(Function<String, Node> convert, String basicFil
         assertThat(MathNodeGenerator.printMathNode(mathNode, ""), CoreMatchers.equalTo(expected));
     }
 
-}
+}

mathml-similarity/src/test/resources/com/formulasearchengine/mathmltools/similarity/node/mathnode5_expected.txt

@@ -0,0 +1 @@
+ci:A

mathml-similarity/src/test/resources/com/formulasearchengine/mathmltools/similarity/node/mathnode5_test.txt

@@ -0,0 +1,10 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<math xmlns="http://www.w3.org/1998/Math/MathML" xmlns:m="http://www.w3.org/1998/Math/MathML" alttext="A">
+   <semantics>
+      <ci id="p51.m5.1" xref="p51.m5.1.pmml">A</ci>
+      <annotation-xml encoding="MathML-Presentation">
+         <mi id="p51.m5.1.pmml" xref="p51.m5.1">A</mi>
+      </annotation-xml>
+      <annotation encoding="application/x-tex">A</annotation>
+   </semantics>
+</math>